Talk:Tourette syndrome/Archive 5

Contemporary prevalence estimates
The article currently says "Contemporary prevalence estimates range from 1 to 3 per 1,000[52] to 10 per 1,000;[53]" and uses the latter (higher) estimate to derive an estimate of 530,000 school-age children with Tourette's in the US. But the abstract of reference [53 ] says "Depending on the sample characteristics, 0.15% to 1.1% of all children had Tourette's disorder." 1.1% would be 11 per 1000, not 10 per 1000, no? Isn't there an off-by-one error here? Or perhaps the main text should use percentages rather than per-thousand figures, since the source uses percentages? Eubulides 20:05, 17 May 2007 (UTC)


 * The range of 1 to 10 comes from a Scahill and other summary reviews, which cites the Kadesjo number at the higher end; that's a problem from relying on abstracts rather than wider consensus. The ten is a good number. The precision of the eleven isn't needed, and the 530,000 number is based on the more commonly accepted ten. Sandy Georgia (Talk) 20:22, 17 May 2007 (UTC)


 * Well, you got me curious so I went and read the Kadesjö et al. paper. It reports two studies: the first found 5 of 435 11-year-olds, for a prevalence of 1.1% (95% confidence level = 0.4%–2.7%). The second found 58 of 38,586 5- to 15-year olds, for a prevalence of 0.15% (95% CL = 0.11%–0.19%). The paper's discussion section then says "it seems prudent to conclude that the prevalence of Tourette's disorder in Swedish school-age children is in the order of 0.15% to about 1.0%" and perhaps this is where Scahill's "1.0%" came from. But I dunno, what we really have here is two studies whose results disagree, and the first study (which the paper says was "admittedly a small sample", with only 5 cases found) is the one with the much-bigger prevalence figure.


 * I did a bit more reading on this, and found a recent meta-analysis paper on the more-general subject of neurologic disorders: It gives prevalence estimates for autism spectrum disorder, cerebral palsy, etc. For Tourette syndrome it says "the data were insufficient". Its meta-analysis included the two studies mentioned in the Kadesjö et al. paper, along with six other studies (one of which reported an even higher prevalence rate of 1.85%, albeit also with a small sample size). Catherine Paddock wrote an article about the Hirtz et al. paper which has a few more details. It seems to me that the Wikipedia article text could be updated a bit to reflect this recent meta-analysis. Eubulides 07:34, 18 May 2007 (UTC)


 * As Paddock writes:
 * They also acknowledged that these figures do not capture the full extent of the burden of these disorders. For a more realistic assessment of the burden of these illnesses, the figures would have to include estimates of the duration, intensity, degree of disability, frequency, impact on life expectancy, and some measure of the suffering they cause, as well as the wider impact on family and society at large.


 * One of the reasons it can be helpful to rely on meta-analyses by recognized experts within the field, as they are likely to account for and be well aware of these factors when analyzing the methodological flaws of various studies. Within the NIH, Castellanos is about the only name known in TS research, and I'll need to doublecheck this, but I believe he is part of the TS Study Group, which includes Scahill et al, and he doesn't appear to have been part of the Kirtz team. Sandy Georgia (Talk) 13:15, 23 May 2007 (UTC)


 * It would be helpful to cite a good meta-analysis for TS, if one is available. Hirtz et al. is not really a meta-analysis; it's more of a review paper. My impression is that Hirtz et al. had fairly high standards of review (they emphasized studies with large numbers of people, for example) but no study is perfect of course. Eubulides 18:49, 23 May 2007 (UTC)


 * I'm going to have do some extensively digging into four file drawers to see what may satisfy you; unfortunately, that's the only referencing system I've got :-) Sandy Georgia  (Talk) 19:25, 23 May 2007 (UTC)


 * OK, thanks, there's no huge rush here; the important thing is to get it right. Eubulides 22:58, 23 May 2007 (UTC)


 * After some more research, I found the following:
 * The lead says "the incidence may be as high as one in a hundred people", which is wrong on two counts. First, the study in question measures prevalence, not incidence. Second, the study is about 11-year-olds, not the general population.
 * Question; do you understand that 11-yo is the age at which tics can be detected, as they statistically tend to remit after that? Sandy Georgia  (Talk) 14:48, 23 May 2007 (UTC)


 * I do understand that symptoms peak around that age, yes. But the original quote was quite muddled. It's been removed, so this point is now moot. Eubulides 18:49, 23 May 2007 (UTC)


 * OK: are you still concerned about "muddle" in the Epidemiology section? Sandy Georgia  (Talk) 19:25, 23 May 2007 (UTC)


 * Yes, very much so. Please see 'Some muddles in "Epidemiology"' below. Eubulides 22:58, 23 May 2007 (UTC)


 * The 530,000 number is just an estimate taken from a paper that focuses on disruptive behavior problems rather than epidemiology, and it is not well-supported enough to appear in this article anywhere, much less the lead.


 * I concur it shouldn't be in the lead; it was put there during FAC. I disagree it is not worthy of mention anywhere in the article, as apparently the authors of the paper (no slouches in TS research) thought it worthy of mention.  Sandy Georgia  (Talk) 14:48, 23 May 2007 (UTC)


 * I am particularly concerned that the "530,000" number is misleading, since it implies a lot more certainty about prevalence than we actually have. For example, if that number is derived from the Kadesjö & Killberg study, it's extrapolating from a relatively small study of 11-year-olds to a large population aged 5 to 18, which is not a reasonable inference for TS where the prevalence is higher among 11-year-olds. In an uncertain area like this, it's best to cite papers that focus on prevalence, instead of papers that focus on some other issues and where prevalance is a bit of an offhand remark. Reviewers of the former kind of papers are more eagle-eyed about prevalence issues; reviewers of the latter are more likely to let the matter slip, as it's not central to the paper. If we have references that are directly on point (which we do here), we should cite them and rely on them instead. Eubulides 18:49, 23 May 2007 (UTC)


 * Except that, in this case, the author is Scahill, and that's one of his areas. Also, one doesn't "stop" having the TS gene(s) just because tics remit.  Sandy Georgia  (Talk) 19:25, 23 May 2007 (UTC)


 * I think if you went back to Scahill and asked him about that 530,000 number he'd be the first to admit that it's just a handwavy upper-boundish estimate. It's not the first or only number that should be given in a general-interest encyclopedia. It'd be better to avoid this sort of misleading number entirely, and to say something like "tens to hundreds of thousands of school-age children in the US" as an estimate: this would more accurately convey to the non-expert reader the scope of the problem, and our uncertainty about its size. Eubulides 22:58, 23 May 2007 (UTC)


 * I've agreed it need not be in the lead; I disagree that Scahill commits "handwavy" numbers to print in peer-reviewed journals. I also disagree that there is anything misleading about the number. On Wikipedia, we can't just say "tens to hundreds of thousands of school-age children in the US" unless someone else has published that estimate.  We can't do original research.  The 530,000 is the only number I've seen in print. Sandy Georgia  (Talk) 04:04, 24 May 2007 (UTC)


 * Handwavy is handwavy; it doesn't matter how eminent the hand is. The best way to present this information is to avoid using that misleading "530,000", and to word the information in some other way. That's what other responsible web sites in this area do, and we should do the same. It is terribly misleading to report the calculated upper bound (530,000) without also reporting the calculated lower bound (53,000). If we can't report both numbers for some reason, then we should report neither. Eubulides 07:34, 24 May 2007 (UTC)


 * Well, "handwavy" is personal opinion compared to a reliable source; you can call Scahill and Walkup and Mink and Leckman (four of the finest in TS research) and Advances in Neurology "handwavy" if you'd like: I call the publication a highly reliable source.  The 530,000 number is reported in a reliable source, and it makes sense in relation to Kurlan's numbers.  At any rate, now that I have dug through and reorganized my files and gotten my hands on the actual articles again, the problem is resolved.  I see that the article does, in fact, report both bracketing numbers, so we don't have to engage in original research after all &mdash; it reports the 53,000 to 530,000 range, and it gives a bit more data that may clear some other things up (like the ten or eleven &mdash; Scahill mentions 11 here, and 10 in another paper &mdash; that level of precision doesn't concern me, but if we want to keep it all in one place, we can go to 11 as you originally suggested, and then explain the context for the ten.  And we can do some rewording to address your concerns about better explaining epidemiology to laypersons, but I don't want to make the article too stat topheavy.  But still, you opine that Scahill's 530,000 children is "handwavy", when reliable sources report that TS is 5 to 12 times more common in children than adults (the Apter number is based on 17 to 18-year olds).  The article includes a number that not only makes sense vis-a-vis Kurlan's work, Apter's work, etc., but is reported in the most recent peer-reviewed medical tome on Tourette's, edited by three recognized TS experts, published in the Epidemiology section as the only addition on epidemiology, written by several of the most renowned researchers in TS; this is about as reliable as a source can get.  Once I get through your entire list below, I'll reword to include the bracket, but I won't invent numbers that aren't reported by reliable sources.  Sandy Georgia  (Talk) 09:39, 24 May 2007 (UTC)


 * What caused me to call the number "handwavy" was primarily the Wikipedia article's mentioning the 530,000 upper bound without also mentioning the 53,000 lower bound. That's extremely handwavy, even misleading. If both numbers are mentioned then the most important part of my objection goes away, though the "3" part of both numbers is still handwavy: there's no way that Scahill or anyone else knows that the upper bound is 530,000 rather than 500,000 (and likewise for 53,000 versus 50,000). I also dislike using absolute numbers here rather sticking to frequencies, as is done elsewhere in the article. Mixing up numbers like this is a recipe for confusion. I would still prefer leaving both numbers out; but if the 530,000 goes in, the 53,000 must also go in. Eubulides 16:39, 24 May 2007 (UTC)


 * On the 53,000 and 530,000, since I've reread the source and see that it does mention both numbers, I've already agreed we will mention the bracketed range, so that should no longer be a concern. (If I could spend less time typing replies to you and more time typing up the research, we could get this done faster, and I do have to travel next week, so please have patience :-)  In terms of dropping the 3's, again, I worry about trading one problem for another; in the future, another editor will come along and ask why we dropped the 3's, and accuse us of original research.  I will try to find a way to word it that addresses all of this and more, if I ever get a chance to focus on the work to be done.  There are several problems in your proposed text, so I need to answer your queries, and propose my own fixes.  I understand the research quite well; I need time to work on the responses you need, incorporate some other sources I've located, and re-write to address your concerns.  Absolute numbers are necessary to help laypersons interpret the magnitude of the numbers; we've given you (below) reasons why percentages are less helpful and more confusing to readers.  But again, I will try to incorporate both in a way that doesn't overburden the text with stats, percentages, and numbers.  Sandy Georgia  (Talk) 19:59, 24 May 2007 (UTC)


 * Thanks, it sounds like things will be under control then. If the 53,000 and 530,000 are both given, and explained in context, that would be OK with me. The important thing is to communicate both the magnitude and the uncertainty about the magnitude concisely and clearly to inexpert readers. Good luck! Eubulides 20:18, 24 May 2007 (UTC)


 * SInce I've got the text in front of me, I'll put something "stopgap" in for now, while I work on ironing out the larger "muddle" you've described and providing thorough context, which I want to do after I've re-reviewed absolutely everything. Even I am now confused about the per 1,000 and per 10,000, and I want to be carefully review everything.  Sandy Georgia  (Talk) 20:34, 24 May 2007 (UTC)


 * In the "Epidemiology" section, the Bagheri citation does not support the claim that Tourette historically was thought to have 5 to 10 people in 10,000 having the condition; that number comes from the Zinner paper. I don't see the relevance of the Bagheri citation here.


 * I'm not sure that's correct (although it's theoretically possible that I typed the wrong named ref). I will doublecheck tomorrow or the next day, as I don't have time to dig through my files today; whether Zinner or Bagheri, it can be found in numerous other places, at any rate.  Sandy Georgia  (Talk) 14:48, 23 May 2007 (UTC)


 * I read both papers. The Bagheri et al. paper is online (it's cited in the Wikipedia article) so you needn't paw through your paper copies to check it. Bagheri et al. cite other papers giving prevalence estimates ranging from 0.001% to 0.1%. But the Wikipedia article says 0.05% to 0.1%. If we cite Bagheri et al. we should give its lower range, instead of ignoring it and giving the lower range of Zinner only. Eubulides 18:49, 23 May 2007 (UTC)


 * I ignore Bagheri on many issues for the simple reason that, although it has a lot of still current and useful information, it was written before DSM-IV-TR (the text revision). Zinner is more recent.  It's a matter of knowing what is still valid in Bagheri and what's not.  And even some of Zinner is now outdated. Sandy Georgia  (Talk) 19:25, 23 May 2007 (UTC)


 * OK, but if Bagheri et al. is being ignored here, then why is it being cited here, in support for a number that it disagrees with? Removing the citation is a simple fix for the problem. Eubulides 22:58, 23 May 2007 (UTC)


 * I'll recheck that this weekend (and in the context of your proposed text below). Sandy Georgia  (Talk) 04:04, 24 May 2007 (UTC)


 * The Zinner paper talks about lifetime prevalence, which is not the same thing as the number of people having the condition at any given time.


 * When making wording changes, please understand most TS is detected 8 - 12 (today - most of it was missed in the past), while most tics remit as one passes through adolescence. The best estimates are on children, as adults may not be aware that they even had tics as a child.  Sandy Georgia  (Talk) 14:48, 23 May 2007 (UTC)


 * I understand that. But the Zinner paper talks about a lifetime prevalence, while the Wikipedia article incorrectly characterizes this as "5 to 10 people in 10,000 having the condition". That's not the same as lifetime prevalence. Lifetime prevalence would correspond to "5 to 10 people in 10,000 have the condition now or have had it in the past". The wording must get changed to accurately summarize Zinner's paper. The simplest way to change it is to use the phrase "lifetime prevalence". Eubulides 18:49, 23 May 2007 (UTC)


 * The claim that "multiple studies published since 2000 demonstrate that the prevalence is much higher than previously thought" is not supported by any citations, and it doesn't match any of the citations in the Hirtz et al. paper mentioned above, so it should go.


 * If you want me to put a long string of citations on that phrase (which amounts to citing common knowledge), I can do that in a few days, when I have more time. How many citations do you want?  :-)  Sandy Georgia  (Talk) 14:48, 23 May 2007 (UTC)


 * "Multiple" would require at least two citations, preferably three; five would be overkill. Best would be a meta-analysis or review paper, like Hirtz et al. (but for more-recent results). Eubulides 18:49, 23 May 2007 (UTC)


 * I'll provide cites over the next few days; I won't be able to get to it today, and maybe not tomorrow, but certainly by the weekend. I'll put it on the talk page first, so you can decide if it's necessary to add the cites and how many (I consider this would be citing common knowledge, and may result in overciting the article).  Sandy Georgia  (Talk) 19:25, 23 May 2007 (UTC)


 * I don't think it's common knowledge among non-experts, so two or three good cites would be appropriate and welcome; thanks. Or you can use just one cite, if it reviews several other studies that support the claim in question. Thanks. Eubulides 22:58, 23 May 2007 (UTC)


 * I'm not sure what is meant by "contemporary prevalence estimates", perhaps estimates since 1999?, but that seems like somewhat an arbitrary cutoff given that by far the largest published studies were done in the 1990s. I think it's better to say that studies report widely varying numbers and leave it at that.


 * A reference to the Hirtz et al. paper should be added.


 * This last sentence seems to indicate where you may be misunderstanding current research. There *are* natural cutoff dates in terms of the history of the research on TS.  Saying that "most of the studies were done in the 1990s" is misleading, since the seminal work showing that most tics remit during adolescence was published in 1998.  That is why "contemporary" papers (which I can define better if you wish, in a few days, when I can dig through four file drawers of research papers) written by knowledgeable TS experts and recognizing past methodological problems, are more relevant than antiquated numbers subject to ascertainment bias and other flaws.  Sandy Georgia  (Talk) 14:48, 23 May 2007 (UTC)


 * I did not say "most of the studies were done in the 1990s". I said "by far the largest studies". All other things being equal, larger studies give us results that we can be more confident in, for the usual statistical reasons. Eubulides 18:49, 23 May 2007 (UTC)


 * Not necessarily. DMS-IV-TR is from 2000, recognition that most tics remit during adolescence is from 1998, and understanding of the importance of sorting out comorbid conditions started in the late 90s.  Older studies on TS have to be viewed in context of current knowledge.  Sandy Georgia  (Talk) 19:25, 23 May 2007 (UTC)


 * Of course; that's what I meant by "all other things being equal". However, it's still the case that larger studies lead to more statistical confidence, and that larger prevalence figures we've seen more recently are coming from smaller studies, in which we have less statistical confidence. I don't think this point can be entirely overlooked, but the current text doesn't mention it. Eubulides 22:58, 23 May 2007 (UTC)


 * You are correct that the current text doesn't delve into explaining all of the issues behind reporting epidemiological numbers for TS; I can work on (or perhaps you already have in the proposed text below) briefly summarizing the issues, but don't want to get too far into writing a statistical tome for laypersons in an encyclopedia. For example, Kurlan was a large study that reported 4%; explaining all the methodological issues there (and why an analysis of those issues would yield a higher number) is really beyond the scope of this article, and current consensus is in the realm of Kurlan's number anyway, so those numbers are reported, and are based on a large study.  But, yes, more can be done to fine tune the text in this area. Sandy Georgia  (Talk) 04:04, 24 May 2007 (UTC)


 * First, there must be some sort of factor-of-ten confusion here. The current consensus is 0.1% to 1% prevalence. But according to the Wikipedia article, Kurlan is reporting 4%, which is way above the upper bound of the consensus. Second, I suspect that the Kurlan figures are not being summarized correctly in the Wikipedia article. My suspicion is prompted by Table 3E of the data supplement from Hirtz et al., which says that Kurlan is reporting a 0.94% (not 4%) prevalence among 1596 8- to 17-year-old New Yorker subjects. As for "large study", Hirtz et al. reference five studies that are much larger than Kurlan, all of which report lower prevalence rates than Kurlan's 0.94%. Eubulides 07:34, 24 May 2007 (UTC)


 * Do you understand that prevalence rates are, according to one source, 5 to 12 times higher in chidren than adults? I have Kurlan's paper in front of me; do you? I have several papers that discuss Kurlan's paper in front of me.  You have Hirtz, who I've never heard of and don't have a copy of.  So, although multiple reliable sources report Kurlan's data, now I've got to put my hands on Hirtz so I can tell you what the discrepancy is.  Scanning Kurlan's article, I don't see a .94 anywhere.  (I do see for about the fifteenth time since I started digging through my files yet another mention that "Traditionally, TS has been viewed as a rare disorder" &mdash; I can reference that as many times as you'd like as well.)  Kurlan's paper reports that almost 4% of children aged 8 to 17 met criteria for a TS diagnosis; how Hirtz (or anyone else) may convert that number to a number applicable to an entire population, I don't know, but it will by necessity involve some assumptions.  I have the Apter study, the Swedish study and a Scahill summary review of all of the studies, discussing methodological limitations and flaws of each. Scahill's summary table of all of the recent epidemiological studies somehow converts (without explaining how) Kurlan's 4% of 8 to 17 year olds to an 8 per 1,000 prevalence in regular ed, and a 15 per 1,000 prevalence in special ed; I suspect Hirtz used some version of that.   Whatever the conclusion is, I don't want to do their numbers for them; I report what they say (53,000 to 530,000 works for me, now that I see the source does discuss the entire bracket). Sandy Georgia  (Talk) 09:39, 24 May 2007 (UTC)


 * I do understand prevalence rates are much higher in children. It does appear that the 4% number in Kurlan is not comparable to the other prevalence rates mentioned; we need to get to the bottom of this. Until we do, the 4% number should go; we can't justify putting "4%" in the Wikipedia article until we know what it's really measuring. Eubulides 16:39, 24 May 2007 (UTC)


 * I have read and understand the Kurlan report and I know what it's measuring (which amounts to what all TS studies measure &mdash; ascertainment bias); I can get to the bottom of it as soon as I can type it all up. We can fix your concerns in the text, but I want to (again) be careful not to overburden the text with descriptions of statistics, epidemiology, and a lot of numbers.  I agree that the 4% needs context, and needs to be balanced by other numbers.  Sandy Georgia  (Talk) 19:59, 24 May 2007 (UTC)


 * I'll make changes along these lines shortly. Eubulides 07:03, 23 May 2007 (UTC)


 * It would be helpful if you would make changes after discussion; I won't have time for the next few days to cite anything you still may want cited, but please explain further the nature of your concerns and what you intend to accomplish with these changes (I'm really not clear what they aim to do), as it appears you may be giving undue weight to Kirtz, and misunderstanding the history and important events in TS research. Sandy Georgia  (Talk) 14:48, 23 May 2007 (UTC)


 * I made some discussion: I cited Hirtz and Paddock on May 18 here; there was no response for five days; so I went ahead and made changes accordingly. I realize that the TS article must be treated more gingerly than Joe Random Wikipedia article, but a five-day delay does seem more than enough for this sort of thing. Eubulides 18:49, 23 May 2007 (UTC)


 * A lot of review by many editors went into this article. And, I'm at a particularly busy time &mdash; sorry :-)  It always seems to happen with this article (someone put it up at GAC when I was traveling, and someone tried to put it up at FAC as I was about to travel; but if I'm home doing nothing, nothing happens on the article for months (except vandalism).  Happens every time &mdash; I'm sorry for the five-day delay, but I also have a hard time taking the Hirtz work in the same context as work from other TS researchers.   Sandy Georgia  (Talk) 19:25, 23 May 2007 (UTC)


 * I know what you mean—scheduling problems happen to me too. However, my impression is that there is still a reasonable amount of (shall we say) skepticism (perhaps some would call it inertia…) among the larger neurological community about some of these larger prevalence estimates, so we should not let the largest and scariest estimates take over the article. Eubulides 22:58, 23 May 2007 (UTC)


 * I don't share your impression, but I do agree that if you find the citations muddled, there's a problem that needs to be fixed. Earlier, you proposed changes at 07:03, 23 May 2007 and made them at  07:27, May 23, 2007; I'm usually quick, but not that quick :-)  Now, though, I'm seeing something that may help me understand your concern about the only published actual number I'm aware of; why on earth are you calling the 530,000  number "scary"?  If anything, the fact that most TS escapes clinical attention is reassuring, because most TS doesn't require clinical attention.  The number is anything but scary; it shows how common the condition is, and that makes it less "scary".  Sandy Georgia  (Talk) 04:04, 24 May 2007 (UTC)


 * "Scary" in the sense of "oh my goodness, more than half a million US schoolchildren have Tourette's, we must increase funding in that area!" Eubulides 07:34, 24 May 2007 (UTC)


 * Never mind that not at all children with TS require intervention, why, pray tell, should you or Wiki be worried about funding for any disorder ?  We don't worry about other people's conflict of interests when we report what reliable sources say.  I'll start on your list below after I've had a chance to get through and re-read all the papers I just dug out. This has gotten so long that I've created several sections for easy of editing and to try to figure out which new text I've missed.  Sandy Georgia  (Talk) 09:39, 24 May 2007 (UTC)


 * I agree with you about the funding. I'm not worried about the funding. I am worried, however, that the Wikipedia article is inadvertantly making reliable TS estimates look larger than they really are. Eubulides 16:39, 24 May 2007 (UTC)


 * Or, larger than what you and one source believes they are; at any rate, your concerns will be addressed. Since I now have found most of the papers, and have a LOT of typing to do, and pending travel, can you please let me know if for now you are satisfied that the lead is in good shape (considering that I can easily cite the "once considered rare" clause)?  It will really trouble me to travel, leaving concerns in the lead; is there anything in the lead now that concerns you, or are we now only talking about the epidemiology section?   Sandy Georgia  (Talk) 19:59, 24 May 2007 (UTC)


 * When the reliable source estimates 53,000 to 530,000, and Wikipedia gives just the 530,000 estimate, then Wikipedia is making the reliable source's estimate look larger than it really is; that was my concern, but you'll be addressing it so that's cool. As for the lead, my main concern with it now is the sentence "As many as one in a hundred people may have Tourette's or tic disorders, with the more common tics of eye blinking, coughing, throat clearing, sniffing, and facial movements." The "with" clause is hard to parse grammatically; it's not clear what the "1/100" figure actually links up with in the rest of the sentence. More important, the 1/100 figure should square with whatever estimates are in "Epidemiology"; currently it seems somewhat inconsistent to me. I was going to wait until we got Epidemiology straightened out before going back to this sentence, though. It can wait. Eubulides 20:18, 24 May 2007 (UTC)

Arbitrary section break
There are quite a few problem with your analysis and changes, beginning with the sources you have used (unrecognized in the field of TS research; not the best sources). First, you've included the Apter study, which is a very strange choice of a very old study with numerous methodological flaws, and not worthy of mention. Second, you've removed two studies that do summarize current knowledge, by very well-recognized experts in the field (Scahill, et al), in favor of work of lesser importance. I actually can't find a good reason for including the Hirtz paper, since it is about many conditions, and isn't written by experts recognized within the field, focusing on the sum total of knowledge about TS (e.g.; Scahill et al). Hirtz's "guess" should have no place in the article, particularly when the Scahill, Tourette Study Group numbers have been eliminated, and they are based on serious meta-analysis of the studies, including their various methodological flaws. Out of curiosity, why do you accord Kirtz's "guesses" more weight than analysis by Scahill and the TS Study Group? Third, some of your summary of Kurlan's work is lacking &mdash; it's important to distinguish that he looked at a school-age sample, because TS tends to remit as children pass through adolescence, and prevalence rates among children are higher than adults &mdash; I hate to see an extreme level of precision introduced into the article (3.8 vs almost 4), but can live with that, although I think it's misleading to imply these numbers are known to such a high level of precision. Fourth, I feel your summary makes it harder for laypersons to understand how relatively common TS is, by providing no context. I'm not sure if I should revert your changes until I can incorporate your concerns correctly since I don't have time to correct them this week or next, but at minimum I'll save the diffs here for when I can go back and fix. . The changes are going to be hard enough to fix that a revert might be more expeditious. I'm not clear what you intend to accomplish; the level of precision between 10% and 11% &mdash; when knowlegeable sources discount the 11% &mdash; isn't something I'm following. It's unclear to me what you intend to accomplish by blurring the issues with mention of Apter, widely-varying older studies, and percentages which provide no context to the layreader. If you can explain better what your edits are intending to accomplish, I may understand how to re-write in order to incorporate your concerns, but at minimum, the more important Scahill et al papers need to come back, and Apter isn't worthy of mention, while I'm not sure Kirtz is either. Sandy Georgia (Talk) 13:03, 23 May 2007 (UTC)
 * PS, in case you need a list of problems with the Apter study &mdash; and why it's not worthy of mention &mdash; I can supply the following from memory, and more if needed by digging out the article and articles about the article ... which I won't have time to do in the next few weeks, but basically:
 * It is based on 16 to 17 year-olds, screened for induction into the Israel Defense Force. The recruits are of an age when tics have remitted for most persons, who may not even remember or have known they had childhood tics or TS.
 * It is old enough that it comes from an era when people weren't likely to know they had had TS as a child, regardless of whether tics had or had not remitted.
 * It was based on screening; problems with reliability.
 * Because it was for military recruits, recruitees were likely to lie so future promotions wouldn't be affected.
 * There are more methodological issues, but that's the basic gist of why it's results must be viewed in contemporary context (the knowledge that most tics remitted during adolescence dates to about 1998, while the Apter study was published in 1993), and not really worthy of mention in the article at all. Contemporary estimates &mdash; interpreted by knowledgeable researchers &mdash; should be given more weight than older, discounted published results. If Kirtz considers the Apter number part of a reasonable bracket of this "wide variety in estimates", he's using old data which needs to be viewed in the context of how it was gathered.  Apter numbers just aren't that worthy of mention.  Sandy Georgia  (Talk) 14:18, 23 May 2007 (UTC)

I went ahead and reverted, because it was easier to start fresh than to fix all of this. I did remove the 530,000 number from the lead; it was put there during FAC by another editor, probably resulting in undue weight, since the TS study group puts the number at 4 - 5, not 10. Eubulides, if you can better explain what you want to accomplish by adding the Kirtz paper, we can work that out, but please review proposed wording changes here on talk. Sandy Georgia (Talk) 14:31, 23 May 2007 (UTC)


 * I'm well aware of the methdological problems in the Apter et al. study, and agree that its estimates are too low by today's standards (partly because the diagnostic criteria have changed since then). The text I suggested gave Apter et al. as a lower bound and suggested that the true number was much higher. However, Apter et al. does provide a useful lower bound and it has some important virtues that other studies lack (notably, a large sample size and avoidance of self-selection). An advantage of the text that you reverted is that it clearly identifies which studies were for which ages (for example, it said that the Apter study was for 16- and 17-year-olds), whereas the current text is hopelessly muddled in this area. Eubulides 18:49, 23 May 2007 (UTC)
 * I put some "placeholder" text in the lead while we resolve this, so I wouldn't lose the refs. Understanding your concerns better will help work out the text.  While Apter provides a useful lower bound, since many current sources reference the 1 to 10 currently accepted prevalence, I'm wondering why it's specifically necessary to drag Apter up from the dark ages?  Sandy Georgia  (Talk) 19:25, 23 May 2007 (UTC)
 * Please see below for some strong points of of Apter et al. Eubulides 07:34, 24 May 2007 (UTC)


 * I did not remove all references to the Scahill Williams Schwab-Stone paper, only the references that talked about prevalence. This is because it's not a prevalence paper, it's a disruptive-behavior paper. You're right that the reference to the Scahill Tanner Dure paper should remain; I removed that inadvertently (sorry). I don't agree that "the knowledge that most tics remitted during adolescence dates to about 1998"; it was known well before this. See, for example, Erenberg G, Cruse RP, Rothner AD, The natural history of Tourette Syndrome: a follow-up study, Ann Neurl. 1987 Sep;22(3):383-5,, and this is merely the first cite that popped up on my screen. It's fine with me if we remove the guess of Hirtz et al. but I'm not sure it's wise to substitute someone else's guess, particularly if it's not clearly labeled as a guess. I agree that the 3.8% implies undue precision, so 4% is OK, though it would be better yet if we mention a confidence level or a study size that emphasizes that even the "4" is questionable. Eubulides 18:49, 23 May 2007 (UTC)


 * The Erenberg paper is based on a questionnaire; not a good sampling method, didn't firmly establish the concept. It was the Leckman paper in 1998 that firmly established the concept (followed by a Burd Kerbeshian paper a year or two later).  I don't see a good reason to remove the Scahill mention of prevalence from the disruptive paper, since it is the only section in that volume of Advances in Neurology (the most recent) that discusses current prevalence, and since Scahill is at the forefront of reporting on those numbers.  On the level of precision and confidence intervals, I'm not a fan of introducing that level of terminology into an encyclopedic entry, since so few readers are versed in statistics, and those that are or want to know that info are likely to read the actual study. Numerous papers cite the range as 1 to 10; somewhere (which I'd have to dig into my files to find) an average of 4 to 5 is recommended, but the range is often cited as 1 to 10.  Sandy Georgia  (Talk) 19:25, 23 May 2007 (UTC)


 * I don't know what you mean by "firmly established the concept" exactly, but I still maintain that it was common knowledge before 1998 that symptoms remitted during adolescence. Here's another example: Leckman JF, Developmental psychopathology and neurobiology of Tourette's Syndrome. J Am Acad Child Adolesc Psychiatry', 1994 Jan;33(1):2-15, . I can easily come up with more. Maybe the idea wasn't "firmly established" enough to satisfy your reviewers (i.e., not enough high-quality studies exactly on point), but it was conventional wisdom. On the issue of what number to report, I'd still rather see a strongly-supported number (or, more likely, range of numbers), since the 530,000 number is admittedly handwavy and it is not the central point of the citation. We should not be giving naive readers the mistaken impression of far more precision and accuracy than we have. Eubulides 22:58, 23 May 2007 (UTC)


 * OK, and looking at the earlier Leckman/Cohen paper you cite, we find more wording which confirms what I'm telling you about the 1998 publication. This paper says: "The authors present a model of the developmental psychopathology and neurobiology of Tourette's syndrome that provides a framework for ongoing research and treatment. ... The model is based on clinical experience and a selective review of relevant scientific literature. ... In the future, we expect that our understanding will extend from the molecular level of how specific genes and epigenetic factors confer and mediate vulnerability to an understanding of why certain symptoms emerge when they do in the course of central nervous system development."  The future came in their 1998 paper, when they confirmed this model beyond clinical experience, with data, which was later validated by Burd and Kerbeshian. The 1998 paper was seminal.  Sandy Georgia  (Talk) 04:04, 24 May 2007 (UTC)


 * I'm not disputing the seminal 1998 modeling results. All I'm disputing is whether it was common knowledge before 1998 that TS symptoms remit during adolescence. The Apter et al. study was done by people who knew that the symptoms do tend to remit, even if they didn't yet have a model for why. And the number they reported isn't outlandishly outside the current consensus: it's for a higher age range than usual, where you might expect prevalence to be somewhat below the 0.1% at the lower end of the current consensus for childhood prevalence. The study has some strong points: it suffered from less self-selection than other TS studies, and it was a large study, more than 15× larger than Kurlan. Eubulides 07:34, 24 May 2007 (UTC)


 * The current text muddles numbers from different studies without explaining why they vary so much. Part of the explanation is that the various studies measure different-aged groups. Part of it is that different diagnostic criteria were used. And part of it is who knows what? (Nobody really knows.) The text gives way too much confidence in numbers that are very much up in the air, and it gives too much emphasis to numbers from one study that are way higher than consensus estimates. Eubulides 18:49, 23 May 2007 (UTC)


 * OK, now I see what your concerns are, but writing a concise, encyclopedic description of the problems with reporting on epidemiological numbers will be no easy task. I'd not like to see us verge into too much statistical detail, as it will be lost on some of the readership, but I do see what your concerns are now.  In terms of working this out, I have a question that will need consensus.  When I originally reported the 530,000 number, I wanted to "do the math" to explain what the equivalent numbers would be for different prevalence rates.  But, I'm not sure if "doing the math" ourselves amounts to original research or synthesis.  By the way, I disagree that the 530,000 is much higher than consensus estimates, since just about every source mentions the likely very high rate of undetected TS.  As far as I know, only Hirtz is disagreeing, and I haven't read that paper.  (I'm also wondering if you have access to the commentary about his paper, as there may be some discussion there.)  I'm going to be busier than usual over the next few weeks, with some travel, although I always check in regularly.  I'm hoping you'll propose text changes before making them, so we can work together more efficiently.  I'm going to be out the rest of the day, but I'll look at whether I can clear up the muddle tomorrow, unless you propose some new text first.  Best regards, Sandy Georgia  (Talk) 19:25, 23 May 2007 (UTC)


 * If the consensus prevalence estimates are 0.1% to 1%, that would yield a range of 53,000 to 530,000, no? So if we report the "530,000" we should also report the 53,000, and make it clear that the two numbers are part of a range. Personally, though, I think that the "3" in "530,000" gives a misleading impression of accuracy, so we should say "around 50,000 to around 500,000", or better yet, "tens to hundreds of thousands". That's all we really know. If we use a phrase like that, it doesn't count as original research in my view: we're just trying to accurately summarize the wide variety of numbers in the literature. Also, Hirtz et al. is not disagreeing with the consensus: they said that if forced to guess they would guess 0.35%, which is right in the middle of the logarithmic range of the consensus. The only commentary published about the Hirtz paper is that Adam Kaplin complained about their underestimating MS cases (see their correspondence published April 1). Hirtz et al. replied " We strongly support relying on the best possible studies—and not anecdotal evidence or non-population-based data—conducted in the same way over time, to inform us about the true frequency and time trends of diseases with major burden to the US population.", a statement that they'd no doubt make about the TS prevalence estimates as well. Eubulides 22:58, 23 May 2007 (UTC)


 * We have this thing on Wikipedia called original research. All we have is one published "hard" number; I've never encountered others.  I'm not sure we can "do the math" ourselves if no reliable source has published it thusly.  We know it would be accurate to describe the range as in the hundreds of thousands, but 1) I don't believe any credible researcher would ever commit to print that there are only tens of thousands of children with TS in the US (the NIH has already said there are 200,000 adults with "severe" TS in the US, and severe TS in adulthood is a rarity), and 2) doing that will trade one problem for another.  You're objecting to a published number from a reliable source; that is what Wiki is based on.  If I "fuzz it up" with original research (reporting tens to hundreds of thousands), someone else will come along in the future and want to know why we're fuzzing up a reported number and reporting vague made-up numbers, when reliable sources report hard numbers.  Wikipedia reports what has been published by reliable sources; we can't just make it up because you're concerned the 530,000 is "scary".


 * The NIH reference does not estimate 200,000 adults; it estimates 200,000 total. If their estimate is right, then there could indeed be fewer than 100,000 school-age children affected. 53,000, though quite low, does not seem entirely implausible given these estimates. About 15% of the US population is school age children (the exact number depends on how you count "school age"), and about 15% of 200,000 would be 30,000 children; this would be a lower bound of course, given that children exhibit symptoms more often than adults do. Certainly the phrase "hundreds of thousands" would imply a value much higher than the estimated lower bound here, as nobody is seriously saying that there are at minimum 200,000 US children with TS. All this being said, if the number is causing so much trouble, let's just yank it, and talk about percentages only. That's simpler anyway. Eubulides 07:34, 24 May 2007 (UTC)


 * Recognizing that we can now incorporate the range (53,000 to 530,000) since it is specifically mentioned in the source, I'm still not understanding your urgency to "yank" a number reported by a very reliable source (other than a vague concern that it's "scary" because someone may have to "fund" something somewhere), and I don't know who you are referring to when you say "nobody is seriously saying that there are at minimum 200,000 US children with TS". If we use the 53 million number of school-age students per the most recent US census as cited in Scahill, Kurlan's number would have over 2 million students with TS and your Hirtz number would have 185,500 (and they're in the lower range of the bound I see most often used, which is 4-5 and would result in 212,000 to 265,000). I wouldn't be in a hurry to substitute our "original research" math based on NIH estimates over other medical input from reliable sources, as the NIH has gotten it dramatically wrong before.  On Wikipedia, we report reliably sourced numbers rather than doing the math ourselves, but several hundred thousand is not at all a stretch; 53,000 on the other hand is quite a stretch, particularly since it relies on a prevalence rate derived from an adult rather than school-age sample, which I don't think any credible researcher would accept as relevant to school-age. Talking about prevalence percentages not only has confused the issues here, but it will do nothing to convey to Wiki's readership the order of magnitude of the number of children anywhere with TS; a real number will.  Please help me understand why you are so anxious to "yank" the number, as I'm just not following your concern, and I don't think someone's funding concerns should prevent us from using reliable sources.  You're putting a lot of stock into one source &mdash; the NIH &mdash; perhaps they are also concerned about this "funding" issue.  Sandy Georgia  (Talk) 14:50, 24 May 2007 (UTC)


 * Sorry, I guess I'm not following. Are you now disputing Scahill's lower bound estimate of 53,000? If so, in what sense is the 53,000 derived from a "prevalence rate derived from an adult rather than a school-age sample"? Nobody has mentioned adult samples so far in this thread; this is the first I've heard of any numbers based on adult samples. Eubulides 16:46, 24 May 2007 (UTC)


 * No, it's not an issue of "disputing" the numbers (you made a similar mistaken assumption elsewhere on this page when I mentioned that I didn't use something from source X because it was subsequently updated from source Y). Generalizing, it's a matter of knowing the context of the numbers, and subsequent refuting or validating research since something was published, how much consensus surrounds a given topic or number, and other research issues impacting a given topic (for example, the Bagheri paper still has some current and good content, even though it was written before DSM-IV-TR.  We don't throw out the baby with the bathwater; we just have to know context and subsequent research so we don't use outdated info.)  Your preferred method of quoting the numbers has me dizzy and I probably confused the Apter study in my statement above, particularly after pulling an all-nighter to sort through my files before I travel; the TS literature typically reports x per 1,000 or x per 10,000 (I also don't know why they use two different reporting methods). My statement above was assuming that the lower bound was determined by Apter (a study NOT on children), but reviewing some of the other Scahill papers, I think it *may* actually be based on Costello.  Off the top of my head, I don't recall the methodological issues in Costello, and I'll need to review the literature.   I have a LOT of data at hand here, and plenty that suggest that the best range is three to eight per 1000, while others suggest six to eight (putting Hirtz at the lower end in any case), but besides reviewing methodological issues in each study, I also need to make sure I'm clear what age range is being described with each number/study.  This is going to take some time and concentration.  I'm not "disputing" anything as much as I'm providing context that shows that the lower bound is as &mdash; if not more &mdash; unreasonable than the upper bound (which I don't agree is unreasonable at all, but that's neither here nor there at this point), but again, if you'll give me time to process through dozens of papers I now have in front of me, we'll get there.  Hirtz simply isn't a source that I'm going to easily roll over and accept as canonical, since I'm beginning to understand that the NIH does have a potential funding interest/COI in getting lower numbers reported.  I have plenty of analysis here of the various methodological issues that I'll sort through and present, now that I've got it all in front of me.  What I'm not prepared to do is to give undue weight to Hirtz over the entire body of TS research.  It appears to me that you really want the 530,000 out of the article, which isn't reasonable IMO, since those are the only hard numbers I've ever encountered, which will help our readers understand the numbers.  Since I don't know how far I'm going to get on this work by this weekend, I'll temporarily adjust the text to include the lower bound based on the text I've now located; hopefully that will put you at ease until we can better work out everything. Sandy Georgia  (Talk) 20:26, 24 May 2007 (UTC)


 * Hirtz "guesses" .35; multiple reliable sources say the range is 1–10, and Scahill publishes 530,000 in a very reliable peer-reviewed source. Hirtz doesn't even figure on the radar screen of the most highly-regarded TS researchers, so I hope we can avoid relying on his "guess" at all, since his 3.5 is not dramatically different from the consensus of those knowledgeable in the field (assuming I can find the 4–5 reference in my file drawer &mdash; if not, we're back to 530,000).  Sandy Georgia  (Talk) 04:04, 24 May 2007 (UTC)


 * Again, it sounds like you're making an off-by-factor-of-10 error here. Hirtz et al. guesses 0.35%; your "multiple reliable sources" say the range is 1/1000 to 10/1000, which is 0.1% to 1%. So Hirtz et al. is guessing right in the mainstream. By the way, another peeve I have about "Epidemiology" is that it gets all confused about percentages versus per 1000 versus other ways of calculating ratios: it should just use percentages and stick with that, as that is what most non-experts are comfortable with. I'll add that to my list of items to fix below. Eubulides 07:34, 24 May 2007 (UTC)


 * PS, If I'm reading correctly, your main concern is that we not imply that we know more about the 530,000 number than we do; I agree on that score, so we should be able to get this done. My personal bias is that the number is probably accurate, but I acknowledge that a 4–5 number is more oft-cited and more defensible based on the published literature.  I wish we could do the math and cut the 530,000 in half without it being called original research.  Remember, also, that the NIH estimates that 200,000 adults have "severe" TS, which makes the 530,000 number in children look quite reasonable, since severe TS in adulthood is a rarity, while undetected TS in childhood is common. For that reason, I don't believe we need to lose the number, but do agree that better context might help in interpreting the usefulness of the number.  Sandy Georgia  (Talk) 19:36, 23 May 2007 (UTC)


 * Thanks, well, I'm not sure it can be done without losing or paraphrasing the number, but I'll try to come up with alternate wording. This will require some thought, though. For convenience I'll list below some of the muddles I see with the current text in "Epidemiology". Eubulides 22:58, 23 May 2007 (UTC)


 * I'm going to get on this over the next few days; weekend at the latest. I'm responding in part tonight because I notice you called the 530,000 number "scary" twice in your latest response, and I just find that a curious take on that number.  I see that number as reassuring and anything but scary; TS is common, lots of people have it, and it usually requires no clinical attention and often escapes diagnostic attention.  Sandy Georgia  (Talk) 04:04, 24 May 2007 (UTC)


 * That's not the impression I got from reading the Epidemiology section. Perhaps some rewording is in order there, too. I have added that to the list below. Eubulides 07:34, 24 May 2007 (UTC)

Epidemiology

 * Some muddles in "Epidemiology" (the current version of the following list applies to an older version of the main article):
 * 1) The current wording says Tourette was "rare" but doesn't cite information that supports this assertion, at least to the satisfaction of the EU's definition of "rare disorder", which is 0.5 per 1000 prevalence or less; see Rare Disorders. One way to fix this is to remove the word "rare" (it used to say "rare disorder", but even the word "rare" alone, is questionable here). A better way to fix it is to cite an older reference with a lower prevalence estimate.
 * I'm not planning to engage in a point-by-point rebuttal of each point on your "list", but as an example of the problems with many of the issues you've raised, I'll address your first point here, and the rest later in a summary. The sources for the statement that Tourette was once considered rare (before and after recent changes) were the Bagheri and the Zinner articles.  First, here are the exact quotes from those articles:
 * Bagheri: "Once believed to be rare, Tourette's syndrome is now known to be a more common disorder ..."
 * Zinner: "Until recently, TD was considered a very rare psychiatric diagnosis that was witnessed perhaps once in a medical professional's life.  Although no longer considered rare, this disorder often is undetected, partly because of its diverse expression."
 * Second, although you're applying a specific definition of "rare", and I can cite whatever is needed to meet any definition of rare, both sources do support "0.5 per 1000 prevalence or less", as both of them cite a range which includes .5 per 1000. The article was meticulously and accurately cited, and I will similarly cite everything else you've requested, even if it is already cited, and is common knowledge.  I'll address the rest of your points in a summary, so that this section doesn't become unnecessarily long and unreadable.  Sandy Georgia  (Talk) 22:40, 25 May 2007 (UTC)
 * I agree that both sources say TS was formerly thought to be rare, but nevertheless the current sentence in the Wikipedia article gives a misleading impression of these earlier citations and what they thought "rare" meant. Zinner explicitly said TS is "no longer considered rare" and in support of this gave a lower bound of 0.5/1000 and an upper bound of 1/1000, numbers that (a) would make it very unlikely for it to be "rare" according to the EU definition of 0.5/1000 or less and (b) are clearly numbers that Zinner himself thought meant TS was not rare. Bagheri et al. are similar: they said TS "is now known to be a more common disorder" and gave a lower bound of 0.1/1000 and an upper bound of 1/1000, which would mean TS might be rare, and might not be, according to the EU definition, and also suggests that Bagheri et al. themselves thought those numbers were not rare. However you slice it, the current wording in Wikipedia is dicey. One simple way to fix it would be to remove the phrase ", with less than 1 person in 1,000 having the condition"; that would remove the confusion. Eubulides 23:23, 25 May 2007 (UTC)
 * I disagree with the implication that the Wikipedia article can use any definition of rare that it pleases. It should use rare to mean what people in the field ordinarily mean by rare. In marginal cases, where a syndrome is borderline rare, it should say so. Of course in the old days TS was definitely thought to be a non-borderline case, but the 1/1000 number in the current Wikipedia article is a different matter: it's not an "old days" number, and this is what is leading to the confusion. Eubulides 23:23, 25 May 2007 (UTC)
 * 1) The "5 to 10 people in 10,000 having the condition" isn't an accurate rephrasing of the "lifetime prevalence" in Zinner.
 * 2) Bagheri et al. is cited as evidence for a number it disagrees with.
 * 3) The "multiple studies published since 2000" claim isn't supported by citations.
 * 4) Larger studies, which have better statistical confidence intervals, have reported lower prevalence estimates; this should be mentioned.
 * 5) We shouldn't mention the larger conventional-wisdom estimates without also mentioning the smaller ones.
 * 6) Part of the reason that the prevalence is higher now is because we've changed the diagnostic criteria, and that factor isn't clearly stated.
 * 7) The "10 per 1000" prevalence estimate is for what age group again?
 * 8) The "530,000" number is worrisome for reasons I've already mentioned.
 * 9) In the second paragraph, the "4%" is much larger than even the "10 per 1,000" upper bound of the first paragraph; what's up with that?
 * 10) Hirtz et al. say Kurlan estimates 9.4 per 1000 prevalence; why the discrepancy here? Is the "4%" something other than prevalence?
 * 11) And why is the 2nd paragraph citing estimates for "children in regular education" rather than for all children (or for the general population) as the other statistics do?
 * 12) And the "over 19% of school-age children" doesn't match the source paper, which gives over 19% for regular-ed children, not for all school-age children.
 * 13) The "males are affected 3 to 4 times more often than females" cites Zinner, which (a) is a source that paragraph 1 disparages as being old-hat, and (b) doesn't agree with what Zinner actually says: he says "2:1 to 4:1".
 * 14) The section mixes "N per 100" figures with "N per 1000" figures and also gives percentages for the latter, which is confusing. The section should just stick per-1000 figures uniformly (or stick to percentages; I don't care all that much which). It's overkill and confusing to use both percentage and per-1000.
 * 15) The section doesn't say or imply that it's not such a big deal if lots of US children have TS, since it usually requires no clinical attention.
 * 16) The comparison between the old "1 person in 1,000" estimate and the newer estimates is misleading. The old number was for the general population. The newer numbers are all for children, where everyone agrees the prevalence is higher. The text should compare old apples to new apples, not old oranges to new apples. One possibility is the 1986 Burd et al. study of North Dakota children.
 * 17) The "As many as 1 in 100 people may experience some form of tic disorder" seems to contradict the earlier "emerging consensus is between 1 and 11 per 1,000". Presumably the former number is for the general population, the latter just for children; this point should be made more clearly.
 * 18) The text currently reads "the emerging consensus is between 1 and 11 per 1,000" but then cites Scahill, which uses the figures 1 and 10 per 1,000, not 1 and 11 per 1,000, to derive its bounds of 53,000 and 530,000. The text should match the citation. (Ironically, this whole thread started with the opposite problem, where the text used to say 10 per 1,000 but cited a paper reporting 11 per 1,000. The difference between 10 and 11 doesn't matter, but the point is that numbers in the text should match numbers in the corresponding citations.)
 * 19) The first sentence "Tourette syndrome is frequently misdiagnosed or underdiagnosed…" belongs under "Diagnosis", not "Epidemiology".
 * 20) "prevalence rates in pediatric populations are higher than those in adult populations". This is supported by a citation that talks about prevalence rates in children, not in pediatric populations. In some countries, pediatrics goes as high as age 21, which few people would call children. Let's use the language of the reference rather than trying to paraphrase it. Besides, "children" is a lot easier to read than "pediatric populations".
 * 21) Kurlan et al. is described as a "large, community-based study", which is peacock talk. The study size is already given, so don't say "large". I'm not sure what "community-based" is doing here. Just call it a "study".
 * 22) The current text places a lot of emphasis on Kurlan et al., which is a smaller study that seems to be reporting much higher numbers than the consensus. Even if its prevalence rate is brought down to the 9.4/1000 level that Hirtz et al. are saying for it, why focus only on this seeming outlier? Let's present the consensus rather than pushing something at (or well above) the high end of the consensus. Or, if we do talk about Kurlan et al. at length, for balance we should also talk about (say) the much-larger Kadesjö & Gillberg study that reported 1.5/1000 prevalence.
 * 23) Wording nit: "between 1 and 11 per 1,000" is confusingly phrased; it should be "between 1 per 1,000 and 11 per 1,000", for the same reason that "between 1 and 10%" should be "between 1% and 10%".
 * 24) Wording nit: "1 person in 1,000" versus "10 per 1000". It's a bit confusing to use in some places and per others. Let's standardize on one word. I vote for per since per 1000 is analogous to percent.
 * 25) Wording nit: No need to cite Scahill et al. twice in adjacent sentences; rephrase and just cite once.
 * 26) Wording nit: The wording in the 53,000/530,000 sentence is repetitive and can easily be tightened up.
 * Eubulides 07:14, 25 May 2007 (UTC)


 * I switched the above section to numbered rather than bulleted, so I can refer to each point below without having to chop the text above. Sandy Georgia  (Talk) 08:08, 24 May 2007 (UTC)

I have not yet finished my review of all of the sources I have that discuss the epidemiological studies, but I have now read the full Hirtz study, and find some mistakes, misunderstandings, and/or misrepresentations in your analyses of that paper and in your proposed text. I know with our heads buried in these numbers, things can get confusing (like dropped zeros and so on). Before I prepare a complete summary of where we stand vis-vis your list, can you please tell me:
 * 1) exactly where in Hirtz you find the 9.4 that you mention for Kurlan? I can't find it anywhere.
 * 2) where you are getting your sample size for Kadesjo? Do you have the full report?  If so, please quote to me the exact text which specifies that sample size for a Tourette's study, since I have sources that say N=435 on Kadesjo.  If you can't verify an N=435, I will have to purchase the paper.  Thanks, Sandy Georgia  (Talk) 03:31, 26 May 2007 (UTC)


 * The 9.4 is column 8 of the last row in Table 3E of the data supplement from Hirtz et al., which is available (no subscription needed) from the table of contents for Neurology 68(5). The sample size for Kadesjö & Gillberg is summarized in the same table. The Kadesjö & Gillberg paper, which I have, reports two different studies of Swedish children. Study 1 had 435 children aged 10.6 through 11.6 years, and found 5 with TS, for a prevalence of 1.1% (95% CL = 0.4%–2.7%). Here's a quote: "A total of 435 children (233 boys, 202 girls) were included." Study 2 had 38,586 children aged 5 to 15 years old, and found 58 children with TS, for a prevalence of 0.15% (95% CL = 0.11%–0.19%). Here's a quote: "These children still lived in Värmland in May 1998, at a time when the population of 5- to 15-year-olds in the county comprised 38,586 individuals (19,970 boys and 18,816 girls) born in the years 1980–1990." Both studies used DSM-IV. The first study was based on a population sample, the second on a county-wide sample referred to the only specialized tic disorder clinic in that county. I can give more details if you like (and if you don't want the hassle of acquiring the paper). Eubulides 04:44, 26 May 2007 (UTC)
 * Thanks for the link; I can now see that 9.4 number on Kurlan, but I still don't know what it is. I can think of two things: in the Hirtz paper, they describe some adjustments they made to data (at the top of page 327) &mdash; maybe they somehow adjusted Kurlan's data?  And, footnote e of the Data Supplement says the 9.4 is a combined rate for SpEd and RegEd; I've not yet encountered any other literature in any of my sources that reports a combined rate, or reports a 9.4, so I can't verify that's what it is, or that they combined correctly.  I don't know what can be done with that number, but since there are much bigger problems with the Hirtz report and methods wrt Tourette's, I'm not too worried about chasing down that one number right now.  Our mention of Kurlan should probably focus on the selection bias in the sample, anyway.
 * On Kadesjo, I need to make sure I'm understanding you, since I do not have that report, and every analysis of it that I do have uses N=435. You said (above) that Study 1 had 435 and Study 2 had 38,586.  No other TS source that I have or have seen ever reports Study 2 &mdash; there has to be a good reason for that.  I'd like to understand why the NIH used it, and why others don't.  Is there any indication in the text? Did you get the two studies confused when you said above, "the second on a county-wide sample referred to the only specialized tic disorder clinic in that county"?  I'm surprised the larger study would be the referred sample, while the smaller study would be a population sample.  And why are we using any sample referred to specialty clinics to determine epidemiology (maybe I'll figure that out when I finish reading a few more sources on the studies &mdash; I haven't gotten through everything yet.)  Besides that it's DSM-IV, so not very useful.  The Hirtz report, for a number of reasons, doesn't even approach a reasonable analysis of Tourette's epidemiology, and yet their numbers (3.8 to 7.2) are still within the range other sources report, so I think we're spending way too much time on this.
 * I've added some things on the list to the article; I hope we can nail down the issues first, and separately (later) decide how we want to handle the whole business about units and percentages. I thought the article was clear in the version that passed FAC and was copyedited by Tony, who edits medical articles all the time, and was a fresh set of eyes, unfamiliar with the text, and TimVickers.  I'm still not comfortable introducing decimals, as they don't make sense to layreaders.  BTW, the increased citations didn't really seem necessary to me, but they've only added 2KB to the article size so far, so not a problem.  That's it for me for the day on the computer; I want to get through more of the reading now.  Sandy Georgia  (Talk) 05:15, 26 May 2007 (UTC)
 * Certainly others have mentioned the results of both Kadesjö & Gillberg studies. For example, Hornsey et al. (2001) refer to both studies' results on the first page of their paper (which is one of the references in the existing Wikipedia article). Eubulides 06:35, 26 May 2007 (UTC)
 * Kadesjö & Gillberg seem to report a combined special+regular-ed rate for both studies. For study 1, the quote is "Children attending special classrooms (e.g., those with mental retardation) were examined separately." but the resulting numbers are not reported separately. For study 2, there's no mention of the distinction between special and regular ed and my assumption is the numbers are lumped together there as well. I just now checked, and I didn't get the two studies confused. The large value of N (38,586) for study 2 is the total number of 5- to 15-year-olds in the county, not the total number referred to the specialized clinic; does that help? By the way, 364 of the 435 students in study 1 also underwent a DSM-III-R based evaluation at age 7, for which study 1 was a followup, so one can't help but wonder whether this colored study 1's results somewhat (i.e., whether it is to some extent a DSM-III-R + DSM-IV study). Eubulides 06:35, 26 May 2007 (UTC)
 * Since it appears you're not done with your current round of edits, I'll hold off on further comments for now. Eubulides 06:35, 26 May 2007 (UTC)

Summary of list
I'm still processing through all of my sources, but want to post some summary info today. The proposal (below) has too many mistakes to work with; I suggest we focus on improving the text currently in the article, which is accurate.

Also, since the current text has no mistakes, I also hope Eubulides will be comfortable taking the time to sort things out correctly, particularly since I want to do more reading, track down some pending issues (like what's going on with Kadesjo), and I have pending travel. There's no need to rush through this, since the current text has no glaring problems.

At any rate, here's a summary vis-a-vis the list of "muddles" in the Epidemiology section above:

I disagree with some of your statements and requests for citations on the following, but I made these changes anyway:
 * Done, although don't agree were necessary.
 * 1 (TS was once thought to be rare)
 * 2, 3 and 4 on prevalence numbers and studies.
 * 8, (in children, not sure it matters, since TS is a condition of childhood onset, but clarified)

I agree with you that changes were warranted on the following items, and I made the changes.
 * Done, agreed.
 * 6 and 9, bracket the 53,000 to 530,000
 * 14, three to four males to females, clarified in the citation
 * 20, moved the first sentence in Epidemiology to Diagnosis, although that info will need to be worked back in to an analysis of all factors affecting prevalence in TS, when that is later added.


 * To do, factors affecting prevalence estimates
 * 7, Your statement that prevalence rates have changed because of DSM-IV-TR is incomplete, and not entirely accurate; I plan to write a paragraph summarizing the issues affecting prevalence estimates. You are correct that it should be clarified, since there is such a range.  I want to keep this paragraph short and not too statistical.

How to deal with Kurlan needs to be worked out. There are numerous pros and cons to his work (selection bias in sample, but perhaps the only study to account for current diagnostic criteria &mdash; I need to check that). Paragraph needs to be sorted out and completed to briefly discuss those factors, since his numbers get good play in the literature and need to be dealt with.
 * To do, Kurlan
 * 10, 11, 12, 13, 22 and 23 &mdash; all Kurlan points to be dealt with in a re-write (not all of which I agree with, but will be sorted out)
 * I added some explanation now of Kurlan's higher results. They did several things right (that most other studies didn't do), that result in higher prevalence:  1) large sample size, 2) highly trained examiners, 3) multiple informants (teachers, parents, examiners), 4) direct classroom observation for a long enough time period to avoid tic suppression, and 5) excluded DSM-IV impairment criterion to reflect current DSM-IV-TR diagnostic criteria.  I'm not sure any other study did all of that, but I'm still reviewing.  On the other hand, their study may have been affected by selection bias, in that respondents with tics were more likely to participate in the study.  However, Kurlan's results agree with Mason's results.  Several pros, several cons, a study that did many things right, gets a lot of publicity, and can't be entirely disounted, although we can later balance this text by analysis of other *good* studies (not some of the older junk with serious methodological issues that Hirtz included).  I hope to have a better sense of another study to include for balance after doing still more reading.  I'm still not sure what to do about Hirtz's 9.4, though, since I don't find it in the Kurlan text, so am reporting the numbers as the study reports them; we can't do original research.  Sandy Georgia  (Talk) 19:15, 26 May 2007 (UTC)
 * Scahill's analysis of the newer studies indicates a tighter range of 3 to 8 per 1,000, meaninig that good studies to focus on are Peterson, Khalifa, and Hornsey. (That statement seems to imply that he discounts Kadesjo and others below three, like Costello and Apter, which makes sense considering the known methodological issues in those.  Some of what I've read from Kadesjo is just way out of line, and it looks like a study highly biased towards autism samples, raising many questions about misdiagnoses and other issues &mdash; not entirely surprising, as Gillberg is involved, but I need to read it and analyses of it.)  Sandy Georgia  (Talk) 19:21, 26 May 2007 (UTC)
 * I found another piece of the 9.4 puzzle, but I still can't get to the bottom of it. This is speculation, so can't be used in the article unless we can verify it, and I don't understand how to get from Kurlan percentages to Scahill/Hirtz prevalences.  A Scahill analysis of the studies states Kurlan's rates as 8/1,000 for RegEd and 15/1,000 for SpecEd.  I don't know where he gets those numbers, since they're not in the study text, but I trust they're right, based on something I just haven't found.  Then, Hirtz's 9.4 could be the combined average between the 8 and 15, since the percentage of children in SpecEd is smaller than those in RegEd.  So, it sounds like the Hirtz 9.4 is probably an accurate, combined number, but I don't know how to relate the percentages to the prevalence.  Sandy Georgia  (Talk) 20:06, 26 May 2007 (UTC)


 * Inaccurate, wrong, misunderstood, irrelevant, or need further clarification and consensus
 * 5, correlation between sample size and prevalence &mdash; completely wrong as far as I can tell. Unsourced, original research.
 * 15, disagree with your preferred method for stating prevalence (my read of the text below is that Colin also disagrees). I'd also prefer to stick with the method used throughout the TS literature.  Historic prevalence rates for TS are usually expressed as x per 10,000, current rates are expressed as x per 1,000, and tic disorders are expressed as x per 100.  The order of magnitude difference there is very clear to laypersons, and introducing decimal points to standardize everything to one unit will not be more clear.  At any rate, hopefully we can sort this out after we've nailed down content issues, with a final copyedit.
 * 17, 18 and 19 also. 17, Burd study is old and not very useful, numbers as now stated are accurate, no need to change.  18, you are mistaken (mixed up tic disorders and TS perhaps?)  19, also mistaken, Scahill's text mentions both, we need to stick to what the sources say.
 * 16, may consider adding something about most not needing clinical attention to this section if I can find some well-sourced text that is relevant to epidemiology, but generally, this is not an epidemiological issue (I believe this point fell out of your concern about funding?)
 * Added the best quote I could find which related to epidemiology, from Leckman Lancet paper. Sandy Georgia  (Talk) 18:14, 26 May 2007 (UTC)


 * 21, disagree with your point about adults vs. children, can't completely plagiarize the text, re-wording is needed, propose a change if you want, but I'd like to avoid using the exact words in the text, rather re-phrase.
 * 24, 25, 26, 27 nitpicks in interim text, that can be worked out on final copy edit pass, after we decide which units to report. Frankly, some of your listed items run the risk of making it appear that you overstate the problems in this article to diminish its credibility and quality, but I will assume good faith and that your intent is to improve clarity, even though three very good and independent copyeditors went through the original text.

If you see anything that you believe urgently needs attention before I travel, please point it out; otherwise, a break from this work for a week might benefit all of us. Sandy Georgia (Talk) 15:37, 26 May 2007 (UTC)


 * Thanks for all that work! This is still work-in-progress, right? so I'll wait before commenting. Eubulides 00:09, 27 May 2007 (UTC)

Proposal

 * Here's a proposed redraft of the first three paragraphs of "Epidemiology". I've tried to address all the problems mentioned above, except for the areas marked "??" where I'll need your help to fill in.

Because the tics of Tourette syndrome tend to remit or subside with maturity, such that a diagnosis may no longer be warranted for many adults, prevalence rates in children are higher than those in adults.

Tourette syndrome was formerly thought to have a prevalence well under 1 per 1,000 school-age children. However, multiple studies published since 1995 demonstrate that the prevalence is higher than previously thought, though part of this is merely due to changes in diagnostic criteria. A study of 38,586 5- to 15-year-old children in Sweden found a prevalence rate of 1.5 per 1,000. A study of 1,596 8- to 17-year-old children in New York found that over 190 per 1,000 children in regular education had tics, and almost 40 per 1,000 met the diagnostic criteria for Tourette syndrome. ((??The "40" is much larger than the upper bound of "10" below, and is also much larger than the 9.4 that Hirtz et al. summarize Kurlan as reporting. Also, shouldn't we be reporting total statistics, not regular-ed? What sort of self-selection happened? What is going on here??)) The children with tic disorders in the latter study were usually undiagnosed. As these two examples suggest, larger studies, which have better confidence intervals, have found lower prevalence rates. A 2007 review of the strongest evidence available concluded that the data were insufficient to accurately estimate the prevalence of Tourette's in the US. A commonly suggested range for school-age children is between 1 per 1,000 and 10 per 1,000; the lower number yields an estimate of 53,000 school-age children with Tourette's in the United States, based on 2000 census data, while the higher yields an estimate of 530,000.

In the general population, as many as 10 per 1000 may experience some form of tic disorder, which includes transient tics, chronic tics, or Tourette syndrome. Tourette syndrome is found among all social, racial and ethnic groups, and males are affected 2 to 4 times more often than females.


 * Eubulides 08:53, 25 May 2007 (UTC)

Comment on above proposed text:
 * I appreciate most of the studies are based in the US but please try to research and maintain a global perspective. A recent change to the above proposed text now says "In the US general population...". The source (NIH, US) says "It is estimated that 200,000 Americans have the most severe form of TS, and as many as one in 100 ...". The way I read this, they only specify "Americans" to give the 200,000 figure, not the ratio. I'm not aware of any studies that show that Americans are more or less likely to have TS than any other group (though I'm sure they are much more likely to be diagnosed with it than, say, the French).


 * The text contradicts itself. Firstly it says "A 2007 review of the strongest evidence available concluded that the data were insufficient to estimate the prevalence of Tourette's in the US." and then it goes on to give estimated prevalence figures. (Note that "However" is one of the Words to avoid). If it is absolutely "insufficient" then you're really saying that any figures are rubbish. Surely this is relative: "insufficient to accurately estimate" or "precisely estimate" would be better. All population prevalence figures for diseases are estimates. Some are only accurate to an order of magnitude. TS isn't particularly unusual in this regard. Colin°Talk 08:11, 25 May 2007 (UTC)


 * Thanks for your comment; I've removed the "US" in the first quote, and have put in an "accurately" and removed the "However" in the second quote. I hope this addresses your objections. Eubulides 08:53, 25 May 2007 (UTC)

Problems with proposal
I'm still processing through all of my sources, but will post some summary info today. The proposal has numerous mistakes, and is wrong or misleading in many areas. The current text in the article may need editing for clearing up some issues (Kurlan, factors affecting prevalence numbers in TS) and decision about what units to report, but at least it's not wrong, so I suggest we stay focused on the current text, which is accurate.

Tourette syndrome was formerly thought to have a prevalence well under 1 per 1,000 school-age children. However, multiple studies published since 1995 demonstrate that the prevalence is higher than previously thought, though part of this is merely due to changes in diagnostic criteria. A study of 38,586 5- to 15-year-old children in Sweden found a prevalence rate of 1.5 per 1,000. A study of 1,596 8- to 17-year-old children in New York found that over 190 per 1,000 children in regular education had tics, and almost 40 per 1,000 met the diagnostic criteria for Tourette syndrome. ((??The "40" is much larger than the upper bound of "10" below, and is also much larger than the 9.4 that Hirtz et al. summarize Kurlan as reporting. Also, shouldn't we be reporting total statistics, not regular-ed? What sort of self-selection happened? What is going on here??)) The children with tic disorders in the latter study were usually undiagnosed. As these two examples suggest, larger studies, which have better confidence intervals, have found lower prevalence rates. A 2007 review of the strongest evidence available concluded that the data were insufficient to accurately estimate the prevalence of Tourette's in the US. A commonly suggested range for school-age children is between 1 per 1,000 and 10 per 1,000; the lower number yields an estimate of 53,000 school-age children with Tourette's in the United States, based on 2000 census data, while the higher yields an estimate of 530,000. ...males are affected 2 to 4 times more often than females.
 * Not necessarily wrong, but not useful. Can do much better. Decide if we want to go into numbers here (at all), or leave them for History.  (Gave enough to reference old examples of "rare", without giving so much that we get into territory of History section.)
 * Incomplete, misleading. In fact, I think it's completely wrong, since most of the studies cited are DSM-IV.  I'm still doublechecking, but I suspect that Kurlan is the only study that accounted for DSM-IV-TR (no impairment), which makes his numbers more accurate in that sense than any others, and explains part of the difference between his numbers and others.  The proposed text gives no mention of increasing awareness, that older estimates were affected by ascertainment bias, the direct interview and observation (better methods) are used in current studies.  Unsourced.  Why 1995? Will address this better later with a summary paragraph of issues affecting prevalence numbers in TS studies.
 * Wrong (??). Where did those numbers come from ? I have two sources that report Kadesjo N=435, although I don't have the study.  These numbers aren't reported in TS reviews &mdash; what's the deal?  I will need to find out.
 * Best to take care with discussion of specific studies, lest we end up writing a tome on the pros, cons, methodological issues in each.
 * Not wrong, but extremely confusing, incomplete and not useful. We can do better, if we decide to use Kurlan at all.  Kurlan needs to be dealt with, separately, since he's widely quoted.  The pros and cons would need to be discussed, including sampling bias, but he did account for no impairment, in line with DSM-IV-TR.
 * Completely wrong, don't even know where this notion comes from, uncited, original research, synthesis, and oversimplification considering all factors. Sample sizes and prevalence, for example :
 * Apter, 28,037 .4/1000 not on children, very old study, doesn't use DSM-IV-TR, numerous acknowledged issues, not even a contender.
 * Khalifa, 4,479 6/1000
 * Kurlan, 1596 8-15, or 9/1000 (has to be worked out, not clear)
 * Hornsey, 918, 8/1000
 * Peterson, 776, 3/1000
 * Kadesjo, 435, 11/1000
 * Numbers above show no correlation between sample size and prevalence, and there are far too many confounding factors in each study to make such an oversimplication of the numbers.
 * Wrong on too many counts to detail here; further, Hirtz is not a particularly useful analysis because of their unselective method of choosing studies, and their criteria which penalized TS for the change to DSM-IV-TR. Nonetheless, the average they reported was 3.5 for one group of studies, 7.2 for another (they gave undue weight to first group because of their method).  Results and methodology misrepresented.  Not al all "stongest evidence available", and at any rate, their numbers are exactly within the range of all Scahill numbers, so there is no need to muddy up the text with old numbers and dubious interpretations of a confusing methodology.
 * To clear up confusion, Scahill's text (and other sources) specifically mention the 1 to 11 range, while his example uses 1 to 10; this is very clear, and we shouldn't be doing the math ourselves or changing what the sources say. Best to stick to reporting what reliable sources wrote, which is 1 to 11 consensus range, 1 to 10 used for sample numbers.
 * Not useful, not consensus, one source only, we can do better with info I dug up above. (Done).

Sandy Georgia (Talk) 14:57, 26 May 2007 (UTC)


 * Again, thanks for all that work! As mentioned above I'll wait before commenting. Eubulides 00:09, 27 May 2007 (UTC)
 * Thanks, Eubulides; then if you don't mind, I'll do the remainder of my reading on the plane/in the airports, and catch up with the rest in a little more than a week. Thanks for all your help in pointing out puddled muddles.  Sandy Georgia  (Talk) 00:23, 27 May 2007 (UTC)

Temporary
Here's what I did for now, while we work on finetuning the rest:
 * Tourette syndrome has historically been described as rare, with less than 1 person in 1,000 having the condition.  However, multiple studies published since 2000 demonstrate that the prevalence is much higher than previously thought, and that Tourette syndrome can no longer be considered rare.  A large, community-based study of 1,596 school-age children suggested that over 19% of school-age children have tics, with almost 4% (40 per 1,000) of children in regular education fulfilling the diagnostic criteria for Tourette syndrome.  The children with tic disorders in that study were usually undiagnosed; other studies place the prevalence of TS in school-age children much lower—the emerging consensus is between 1 and 11 per 1,000 (.1% to 1.1%).  A prevalence rate of 1 per 1,000 yields an estimate of 53,000 school-age children with Tourette's in the United States, based on 2000 US census data, while an estimate of 10 per 1,000 would yield an estimate of 530,000 school-age children with TS.

This:
 * 1) changes the per 10,000 to a per 1,000 number, for consistency.
 * 2) reports percentage as well as per 1,000, hopefully to help keep us all on the same page
 * 3) combines the Bagheri, Zinner mention of older prevalence into one "merged" ("less than 1 per 1,000) number to avoid being burdered by excess detail in an outdated number and to avoid sorting out the differences in what they report
 * 4) removes the older Scahill study, since the newer ones I have update them anyway, and using only one report gives better consistency with less explaining
 * 5) adds in the 1 to 11 Eubulides was concerned about
 * 6) brackets the 53 to 530
 * 7) introduces new numbers which provide context for the high Kurlan estimate
 * 8) changes the wording "rare disorder" to just rare, to hopefully help avoid confusion with formal definitions of rare disorders.

There's still more to be done, but is that a good start for now, without overburdening the text with a lot of numbers and stats? Sandy Georgia (Talk) 21:03, 24 May 2007 (UTC)


 * The temporary version addresses 5 of the muddles listed in so I struck them out; thanks. However, the temporary version introduces some muddles too. I went over it again and wrote down all the issues I found (some of which were already there but I missed them; these were relatively minor). The muddle list now has 27 entries of which 4 are minor wording nits and 5 have already been fixed. I also revised the draft replacement to reflect the changes made in the temporary version along with the extra muddles I found. Enjoy. Eubulides 07:31, 25 May 2007 (UTC)
 * Well, thanks for changing the list, since I did a stack of bedtime reading last night to prepare my summary of where we stood on your list, and your new list invalidates that work. I planned to post a summary today of what was completed, what I had found, and what I planned to do still, referencing the numbers on the list :-)  Moving targets certainly chew up a lot of time.  And I must carve out some time today to keep the Schizophrenia FAR moving along.  I haven't yet reviewed your new new list, but I don't think it's going to be particularly helpful in allowing me to do actually get something done, given that the work to be done is already evident.
 * So since I can't post the summary I carefully prepared last night, I'll just add a few tidbits. First, I found a reference to 750,000 children in the USA with TS, just FYI ("no one is claiming two hundred thousand" ???).  Second &mdash; and again, I want to get through all the studies before making any changes &mdash; I don't see any usefulness in muddying up the text with Hirtz's opinion that a reliable estimate can't be made, followed by him making an estimate that is within the same range as everyone else's estimates.  That adds nothing new to the whole business that all epidemiological estimates ... are estimates.  I'm not in favor of adding any reference to Hirtz, partly because I try to never add an article I haven't read, but mostly because he's coming to no new conclusions, the paper is about a whole lot of other conditions, and I have no papers critiquing him as I do of everything else I use. Third, although I'm reluctant to cite common knowledge, and find some of your citation requests out of line, I'm going to go ahead and cite the items you've raised, avoiding the strings of citations I could add, and choosing only one or two citations for each.  Sandy Georgia  (Talk) 11:43, 25 May 2007 (UTC)
 * Now I've looked at your list: please don't keep changing it, as you've got me chasing my tail.  There are a couple of items on the list that are absurdly outrageous, and I had already prepared responses referencing the earlier numbers, which are now gone.  Second, you're asking for changes that alter the way numbers are commonly expressed in all the literature about TS, just because you have a different preference.  Third, I now see that a lot of your confusion comes from a failure to sort out broader tic disorders from Tourette's.  Fourth, on your issue about rare, you may have dropped a zero (I need to catch up on that still).  Fifth, when numbers are commonly cited a certain way throughout the literature about TS, and the sources I use here also cite them that way, I'm not going to change to decimal point percentages to satisfy your particular style requirements while confusing our readers with decimals;  one in a 100 people are thought to have tic disorders.  It's not necessary to say that 10 in a 1000 have tic disorders, when the sources across the board reference one in a hundred.  One to ten in a thousand may have TS, one in a hundred have tic disorders &mdash; a magnitude of ten difference.  That's as clear as can be.
 * I don't want to intersperse answers into your list above, as this page is already becoming unreadable. As soon as you let me know your list is not going to get re-numbered again, I'll prepare a new summary and begin to address your issues.  BTW, starting the list with an unreasonable request that we leave out mention of something that can be cited to just about everything ever written about TS (that it was once thought to be rare) does not give your list a good start. With some of the things you've added to the list (like citing .1 per 1000 people have tic disorders and citing TS thought to be rare to a specific definition), your demands here are raising my eyebrow, as they did when you suggested Hirtz as a superior source for citing TS epidemiology.  Sandy Georgia  (Talk) 12:03, 25 May 2007 (UTC)
 * There are moving targets on both sides here. You changed the article; I changed the list of problems accordingly. I appended items to the end of the list, and struck out items fixed; the numbers on existing items didn't change. So please post what you've got; you don't have to renumber your reply. You can work on the later numbers later. Eubulides 19:10, 25 May 2007 (UTC)
 * More generally, your reluctance to post interim work is slowing down the discussion, since I am forced to figure out which way you're headed by the hints you're dropping rather than just seeing what you've got. It would be more efficient for both of us if you simply posted what you've done so far, incomplete as it is. I have done that myself in "Epidemiology" above; it's not done, and it's not perfect, but it's a start. Let's move this discussion along as efficiently as possible. Eubulides 19:10, 25 May 2007 (UTC)
 * I suspect the 750,000 estimate comes from a source that uses a prevalence estimate of 10 per 1000, and multiplies that by the total US population aged zero through 18, or something like that. Basing the number on a larger population (i.e., on all children rather than just school-age children) would explain why the source arrived at a 750,000 figure as opposed to a 530,000 figure. But you've already noted that 10 per 1000 is the upper bound of the current consensus. My comment about "nobody is claiming 200,000" was about the lower bound, not the upper bound. Perhaps you've found a reliable source claiming that 10 per 1000 is the lower bound? But then how do you reconcile this with the fact that another reliable source, one you've mentioned multiple times, claims that 1 per 1000 is the lower bound, and 10 per 1000 is the upper bound? Eubulides 19:10, 25 May 2007 (UTC)
 * The current draft in "Epidemiology" does not mention the 3.5 per 1000 "guess" of Hirtz et al. This should address much of your objection about mentioning Hirtz et al. I see no reason to not mention Hirtz et al.: it's an accurate review of high-quality epidemiological studies in this area, it's a reliable source, and it's the most recent review in the published literature. Eubulides 19:10, 25 May 2007 (UTC)
 * If you haven't personally read Hirtz et al., that does not immediately disqualify it from being cited here. It would not be a good precedent if a single editor prevented the addition of reliable citations merely because the editor didn't have the time or access to read them. Please trust the other editors more, in areas where you haven't yet been able to personally review the citations. Eubulides 19:10, 25 May 2007 (UTC)
 * I am not asking for lots of citations. I am asking for citations that actually back up the claims being made in the article. This is a common problem in the current text—the claims do not match the citations. There's only one place where I am asking for at least one new citation: the "multiple studies published since 2000" claim. I don't see why this is an unreasonable request. If it's important that multiple studies published since 2000 are claiming something, then let's see a citation for that. Eubulides 19:10, 25 May 2007 (UTC)
 * The current proposal in "Epidemiology" above suggests using "X per 1,000" figures uniformly, so most of your comment about % versus per-1000 seems out of date. The Wikipedia article should not use multiple notations for prevalence in the same paragraph. If it uses per-1000, it should stick to per-1000 figures uniformly. Mixing notations leads to real confusion, as we've seen multiple times here. It's our obligation to be clear. Eubulides 19:10, 25 May 2007 (UTC)
 * I'm glad to know you didn't change any numbers; that will help me pick up where I left off last night.
 * I don't believe it optimum to do this work piecemeal; I prefer to add to the text, and resolve the Kurlan issue and everything else once I've finished my review of the literature about the epidemiological studies and can resolve the entire issue to the standards I expect of myself and this article. Kurlan is cited often in the press and the lay literature; it would be easiest to lose Kurlan, but I don't think that will work, considering how often those numbers are cited.  I'd like to find something that will help me deal with that &mdash; it's cited too often to sweep it under the rug, but it needs to be balanced with other info.  I hope to be able to finish this before I travel next week, hopefully by the weekend (which is what I told you several days ago and you didn't seem to have a problem with), and I hope to have access to the Hirtz article soon.
 * Based on how far I got in my review last night, I have my pick of dozens of articles I can use for the common knowledge information you want cited; I am going to pick carefully so that I can combine as many sources as possible, shed new or better light on issues, and not clunk up the article with multiple random or unnecessary citations and text. While I've been digging through scores of articles and books, I've also found other items that may be worthy of addition, but I'm always mindful of the article size, and writing in a way that takes daughter articles into account.
 * Careful thought and many months of work and review by other editors went into this article (your disparagement notwithstanding), and I hope the time we spend getting it right was and will continue to be well spent. I hope you can understand that questioning statements like "TS was once thought to be rare" and others have caught me completely by surprise, but it's always good to have fresh eyes to help me realize what others might not take as common knowledge &mdash; it never would have occurred to me that such common knowledge could create a problem, and I want to get it right.  As a numbers person, I worked hard to keep the article as layperson friendly and stat-jargon free as I could; I want to keep the article correctly pitched, without writing a tome on epidemiology, while still figuring out how to solve the Kurlan issue.  I put up an interim version so that you can hopefully relax while I finish my review of the literature; if allowing me the time to do that is going to be a problem, and we need to go through multiple edits and re-edits instead, please let me know, and I'll post something half-assed for now, although working piecemeal is not the way I prefer to work.  Sandy Georgia  (Talk) 19:54, 25 May 2007 (UTC)
 * Eubulides, please be patient. Why does it have to fit your timetable? This isn't a day job and you aren't anyone's manager. I don't like posting half finished work myself, so I understand Sandy's need to take time before posting. Especially when you're waiting to pounce on it like a starved lion. Colin°Talk 22:02, 25 May 2007 (UTC)
 * There must be some misunderstanding here. I have suggested neither a timetable nor a schedule. Eubulides 22:29, 25 May 2007 (UTC)
 * SandyGeorgia, if you feel more comfortable working cathedral-fashion than bazaar-fashion for your next cut then I won't insist on the bazaar; I'm sure you'll do an excellent job. Please bear in mind, though, that the draft above was written in bazaar-style; it is explicitly incomplete and it has obvious flaws and was primarily meant to be raw material for any future version. I would fix some of those flaws now myself except you have asked me to stop making changes to that section while you work on your draft. Eubulides 22:29, 25 May 2007 (UTC)
 * I have never questioned the fact that TS was once thought to be rare; obviously it was. I'm not even insisting on a citation here; if space is tight let's omit the citation. All I'm saying is that if we do provide a citation for this fact, the citation should support the fact rather than undermine it. Eubulides 22:29, 25 May 2007 (UTC)
 * The draft above does not sweep Kurlan et al. under the rug: it cites the paper and discusses its results at some length. It has some "??"s that need fixing, that's all. Eubulides 22:29, 25 May 2007 (UTC)
 * I'm not surprised that Wikipedia currently muddles the epidemiology of Tourette's despite careful review. Such muddles are all too common, even in refereed journal articles. The fundamental message of Tourette syndrome is right; what's muddled are the confusing presentation and the mismatches between citations and text. Let's not worry about how those problems got into the section; let's focus on fixing them. Eubulides 22:29, 25 May 2007 (UTC)
 * We don't have to write a tome to get the numbers and fundamental message right. The draft isn't a tome at all, and it gets the numbers right (outside the "??" sections of course). Eubulides 22:29, 25 May 2007 (UTC)
 * As for "disparagement", I just now reread the list of muddles in in an attempt to find and remove any disparaging phrases, but came up empty. It is my best attempt at honest criticism of a difficult area. If some of the words rub you the wrong way please let me know and I'll do my best to fix them. (I won't renumber anything, I promise!) Eubulides 22:38, 25 May 2007 (UTC)
 * I put up what I acknowledged was an interim version to address your main concerns, until I could get through my files and write a complete version incorporating more epidemiology and citing common knowledge, and because I have travel pending; the first thing you did with it was nitpick it to pieces with trivial and some inaccurate items, extending your "list" to 27, even though you knew it was an interim version. Obviously, I won't be putting up anymore interim work to try to satisfy your own definitions, styles, guidelines, and preferences.  For now, I may have to stop responding to your posts if I am to get this work done before my travel. Sandy Georgia  (Talk) 00:23, 26 May 2007 (UTC)

Percentages vs per XXXX in epidemiology
I agree that mixing styles is confusing but I'm not at all convinced that switching to percentages improves things. I posted just this question to Wikiproject Clinical Medicine asking for opinons, but didn't get much response. Have a read of the Bad Science column and the quoted text from Statistics: What Seems Natural? Science, Vol 292, Issue 5518, 853-855, 4 May 2001.

I can't relate to 0.1%. If you say 1 in 1000 school children then I can visualise that means 1 child in my old secondary school. The EU did not define "rare disorder" as 0.05%, but as 5 per 10,000. I did some maths at university and even I find myself going number blind when I see "0.05%". Try reading it out loud. Does it sink in? Would you ever find yourself saying "zero point zero five percent" when talking to someone? Colin°Talk 08:48, 24 May 2007 (UTC)


 * Absolutely concur; this is an encyclopedia, accessed by laypersons, and we have to carefully work to reflect accurate medical knowledge in a way that is accessible to laypersons (most of whom won't understand statistics or care about confidence intervals and the like). We need to report numbers laypersons can relate to.  When a reliable source says 4% of school-age children have TS, that is something people can understand.  When we say, 53,000 to 530,000 school-age children, people can understand those numbers.  Asking a layperson to relate to .8% isn't going to work.  BTW, the rare disorder thing is a complete red herring, but apparently I'm going to have to provide the fifteen cites I just located which substantiate that text, which is absolutely necessary and pivotal to this article and an understanding of Tourette's (I'll just have to find the time to answer that list above, but I'm not happy about the wild goose chase to cite common facts when, for gosh sakes, it was once believed there were no more than 50 cases in the entire US).  I really abhor that this carefully cited and written article is now going to be another example of overcitation of common knowledge facts on Wikipedia.


 * wrt "rare" I agree a formal definition is a red herring. One purpose of defining a "rare disorder" is for "orphan disease" status, which is important for certain drug development/licensing. The US use 1:1500 (or 200,000 people in the US) and the EU use 1:2000. Most people would regard "rare" as meaning that they are unlikely to know anyone with the condition. I think the statement that TS was once considered rare is common knowledge and doesn't need cited. What isn't (yet) common knowledge and does need (is) cited is the idea that it isn't anything like as rare as once thought. The "common knowledge" that it is rare goes hand-in-hand with the common view that it is "the swearing disease". Colin°Talk 12:22, 24 May 2007 (UTC)


 * Regardless of how someone across the pond defines "rare", TS was once considered rare even by that definition. I believe the 50 number I quoted above was from an old Time magazine, and ... The U.S. National Institutes of Health (NIH) turned down a 1972 grant proposal from the TSA because "the reviewers believed there were probably no more than 100 cases of TS in the entire nation". (Cohen DJ, Jankovic J, Goetz CG, (eds). Advances in neurology, Vol. 85, Tourette syndrome. Lippincott, Williams & Wilkins, Philadelphia, PA, 2001, p. xviii. ISBN 0-7817-2405-8)  (Ah, the NIH gets it dramatically right again.)  At any rate, the multitude of sources that refer to the fact that TS was once considered rare are likely to be the same ones mentioning that it's no longer considered rare, so since I now have to trudge through all of these papers, it's the same difference.  I wish I had easier access to a medical library to make this manual task less onerous and time-consumsing, but removing mention that TS was once considered rare (and still is by most laypersons not in the know, who still think it's not only rare, but "bizarre") from the article just leaves out a very important factor in understanding TS.  (And here I thought I was going to spend the next few weeks helping salvage the featured status of Lesch-Nyhan syndrome and Schizophrenia; I guess I won't be working on those articles at WP:FAR any longer, and those are articles that need dramatic amounts of help.  Sandy Georgia  (Talk) 14:04, 24 May 2007 (UTC)


 * Foul Talker TIME 1949-08-29: "Though extremely rare* ... *About 50 cases (a dozen of them in the U.S.) have been reported since the affliction was described by French Neurologist Georges Gilles de la Tourette in 1885.". Colin°Talk 18:44, 24 May 2007 (UTC)


 * In case it gets lost in the shuffle above (I'm certainly losing track), I'm copying one question down here:


 * "Scary" in the sense of "oh my goodness, more than half a million US schoolchildren have Tourette's, we must increase funding in that area!" Eubulides 07:34, 24 May 2007 (UTC)
 * Never mind that not at all children with TS require intervention, why, pray tell, should you or Wiki be worried about funding for any disorder ?  We don't worry about other people's conflict of interests when we report what reliable sources say.  Sandy Georgia  (Talk) 09:48, 24 May 2007 (UTC)
 * I followed up in the "shuffle above"; the thread is better-placed there than in an unrelated section about percentages versus per-thousands. Eubulides 19:57, 24 May 2007 (UTC)


 * I mildly prefer %, but I agree that per-1000 statistics are also fine. The point is that we should be consistent, and should consistently use per-1000 statistics (or per-cent; I don't care that much which). It is not acceptable to jumble together a lot of statistics with different denominators, as this is so confusing that it even confuses the editors (as witness several such confusions on this web page). I changed the draft revision in above to use per-1000 statistics, to address this objection. Eubulides 19:57, 24 May 2007 (UTC)


 * NO kidding; I'm confusing myself between a night of no sleep, your method of reporting percentages, and the per 1,000 and per 10,000 encountered in the research. We'll pull it all together in a consistent way.  Sandy Georgia  (Talk) 20:30, 24 May 2007 (UTC)


 * As for "rare", my primary objection is not that the current text claims that TS was thought to be rare, but that the numbers cited to back this claim (5/1000 to 10/1000) actually contradict the claim, at least in the EU. It would be fine to fix this problem by citing a (presumably older) reference that gives smaller numbers, i.e., numbers less than 5/1000. Eubulides 19:57, 24 May 2007 (UTC)


 * For the record, it seems you have dropped a zero. Both refs (Zinner and Bagheri) state 5 to 10 per 10,000 &mdash; not 1000 &mdash; which is .5 to 1 per 1000.  It's not really necessary to cite any number, as almost every important text ever written about TS can be used to cite the fact that it was once considered rare, and I don't want to overburden this article with numbers, percentages, and stats which are better dealt with in daughter articles.  If you insist, another sentence about the very old numbers can be added, but we're bordering on the absurd here since, even today, most people still think TS is rare.  Sandy Georgia  (Talk) 12:45, 25 May 2007 (UTC)


 * You're right, I dropped a zero. Thanks for catching that. But my comment is valid even with the zero put back in, so let me rephrase it with the numbers corrected:


 * As for "rare", my primary objection is not that the current text claims that TS was thought to be rare, but that the numbers cited to back this claim (0.5/1000 to 1/1000) actually contradict the claim, at least in the EU. It would be fine to fix this problem by citing a (presumably older) reference that gives smaller numbers, i.e., numbers less than 0.5/1000. Eubulides 19:10, 25 May 2007 (UTC)


 * As for citing old numbers, if the goal is to educate the public that TS is no longer rare, and if the section starts by saying people used to think it was rare, and then gives a citation for that, then the citation should support the argument rather than disagreeing with it. That's all I'm trying to say. It would be fine with me if we simply dropped the citation entirely and just said "People used to think TS was rare" full stop. That would be fine. But we shouldn't follow the claim by a citation that, when one goes to actually read it, does not support the claim. Eubulides 19:10, 25 May 2007 (UTC)


 * It's your opinion that the citation doesn't support it, because you use a specific definition of "rare"; at any rate, it will be addressed. Sandy Georgia  (Talk) 22:03, 25 May 2007 (UTC)


 * All right; I'll see what I can do there, now that I understand your concern (although I think you're applying an overly narrow defintion of rare, but I do understand we don't need that additional "muddle"); that will save me from typing up the gazillion sourcees I found last night with the exact wording that "TS was once considered rare". I hope you understand the volume of paperwork I now have before me to sort through.  Give me time to process it all.  Sandy Georgia  (Talk) 20:30, 24 May 2007 (UTC)

Perspective
Nearly 20,000 words have been written on this talk page concerning epidemiology. That section has grown from 335 words to 514 words. That's between 40 and 60 words of analysis per word in the article. This is probably a record in article-space.

This is an encyclopaedia article for lay readers, not a scientific paper or an application for a research grant. The basic message has been tested and refined a little. The extra text clarifies and expands a few areas but in places it is coming close to tedious when explaining the minutiae of flaws, bias and changes to diagnostic criteria. These things are important to those who care but that is a minority of the readership.

Can we focus on tidying and tightening the text while aiming to draw things to a close? There are plenty other articles on Wikipedia in much greater need of expert analysis. Colin°Talk 11:58, 27 May 2007 (UTC)
 * Now you're making me feel so awful about the other articles and areas of Wiki I've neglected this week :/ I do agree that I should add another paragraph explaining the issues affecting the studies in TS, but I might not get to it until after my travel. Yes, we should focus on finishing up the current text, as a lot of this discussion has resulted in a loss of clarity in the article, with the introduction of excess detail.  More below ...  Sandy Georgia  (Talk) 15:05, 27 May 2007 (UTC)

I'd also be happy with something that is not too abstruse and to the point. My main goal all along has been clarity (for non-experts) and appropriately-cited claims. (I am currently not reading the article proper, since I want SandyGeorgia to have the time to rewrite "Epidemiology" to her satisfaction.) Eubulides 01:33, 28 May 2007 (UTC)


 * I don't think I'll add the remaining paragraph or two until I return from travel; I've got to get some other things done for now. Sandy Georgia  (Talk) 01:49, 28 May 2007 (UTC)

Textual comments
The text says:
 * The results are similar to those of the study by Mason et al. of Grade 9 pupils, which excluded the impairment criterion of DSM-IV to reflect current diagnostic criteria.

Can I suggest some changes:
 * 1) The "by Mason et al is dropped. Say "another study" or "a later/earlier study" or something less academic. There's a citation for those who want to know the particular study.
 * Yes. Sandy Georgia  (Talk) 15:05, 27 May 2007 (UTC)
 * 1) The "Grade 9" pupils is changed to give an age (range). I've no idea what Grade 9 is.
 * Let me see if I can find that. Sandy Georgia  (Talk) 15:05, 27 May 2007 (UTC)
 * 1) The last clause is probably meaningless to the average reader. I know the change to the DSM is covered earlier in the article and is a very important within TS, but I bet most readers won't have absorbed its significance. The finer points of DSM-IV-TR vs DSM-IV may cause eyes to glaze. I wonder if all you are really saying is that this other study is "similar" since it used effectively the same diagnostic criteria as Kurlan.
 * Yes, key here is that it used the current diagnostic criteria, which most other studies don't. (That's one of several reasons why Hirtz's analysis is fairly useless for our purposes, and a partial explanation for Kurlan's higher numbers.) Sandy Georgia (Talk) 15:05, 27 May 2007 (UTC)

The second paragraph repeats the phrase "referred to specialty clinics". The words "specialty clinics" offend my British ears, which is how I noticed it :-) Can the repetition be avoided? Perhaps
 * Researchers at such clinics tend to see only the more severe cases since children with milder symptoms do not get referred.

Hmm. Doesn't cover everything as it doesn't mention the word "bias". Perhaps you can do better?
 * Yes, heading the right direction there. Sandy Georgia  (Talk) 15:05, 27 May 2007 (UTC)

The text varies in saying "Tourette syndrome", "TS" or "Tourette's". Is this variation desirable or not?
 * Tony and I discussed this pre-FAC. Yes, the terms are fairly interchangeable, and there are times where one or the other "rolls off the tongue" more easily.  Mentioning all varieties also helps the Google rank.  Sandy Georgia  (Talk) 15:05, 27 May 2007 (UTC)

I asked Tony a question about the digits vs spelling numbers. What do you think?
 * I don't like the mixing of spelling out versus using digits, even though I know Tony is following WP:MOS. I'd prefer one usage throughout: the one used in most of the TS literature.  Sandy Georgia  (Talk) 15:05, 27 May 2007 (UTC)

Can "A study of 1,596" be changed to "A study of nearly 1,600". I was happy with "large" but Eubulides said it was a peacock term. Changing it to just numbers now means the reader has to judge the size, and without experience of such studies, doesn't know if this is typically "large" or "small". I think rounding the number might help the reader absorb it without going blind. However, if you can find a source calling it a "large" study, then I'd recommend going back to the old text. Colin°Talk 11:58, 27 May 2007 (UTC)
 * I hate burdening the text with specific details of studies, but I hate even more fudging the numbers for readability. If we change it to nearly 1,600, as in my argument to Eubulides above about the 3s in 53 and 530, someone will come along in the future and complain that we fudged, so we'll only exchange one problem for another.  Let me see if any of my reliable sources characterize the study in a way that is useful without being peacock-ish.  Back later today.  Sandy Georgia  (Talk) 15:05, 27 May 2007 (UTC)
 * PS, Colin, if you don't mind, I'll post the details here on the talk page, but leave it to you to make the final text adjustments, because of my pending travel, and also because I'm probably too close to the text and topic, and your extra set of eyes can improve clarity. Sandy Georgia  (Talk) 15:12, 27 May 2007 (UTC)


 * Gathering notes as I work at User:SandyGeorgia/Sandbox2; Mason is 13 to 14 year olds, VERY small sample size. Sandy Georgia  (Talk) 17:11, 27 May 2007 (UTC)
 * How about: "The results confirm those of an earlier smaller study on pupils aged 13–14 years." ?? Colin°Talk 19:33, 27 May 2007 (UTC)
 * I get tangled up in "earlier smaller", and earlier may be redundant to confirm, as they have to be earlier to be confirmed. How about dropping earlier or changing it to a "smaller study on UK pupils aged ... "? I do like the "confirm" word, since Mason's numbers were previously unmatched.  On the "large" Kurlan study, I struck out in my review of the literature. Scahill referred to another larger study than Kurlan's as "relatively large" (Khalifa), so I don't think we can use large. I'm done reviewing literature for now, and hope to be able to turn my attention to Kushner. Moving the Kurlan detail to a note should do the trick. Sandy Georgia  (Talk) 19:43, 27 May 2007 (UTC)

How to handle the little details? Lots of textbooks use footnotes (separate from the refs) to go into the specifics. I don't know if you wanted to go down that road, but it is an option. Alternatively you could combine the footnote comments with the ref. So the actual numbers and age range for Kurlan could be dropped and moved to the footnotes as an extra comment for the curious. Kurlan uses the term "schoolchildren" in the title and abstract so I don't see any problem with using that instead. We're splitting hairs over a year or two, and that's quite irrelevant to the reader. Colin°Talk 15:39, 27 May 2007 (UTC)
 * Good idea. Actually, I'd need to learn how to do it, since I never have, but we can have a Notes section, separate from Footnotes.  For example, the Note Fvasconcellos added about trade names in the US would also be better placed as a Note rather than Footnote.  I need to look at another article to learn how to do this; if you know how, can you set up a sample by moving the Note about drug names into a new section?  (Should we give it an "a" rather than a number?) Otherwise, I'll go dig up another article where I can learn the technique.  Sandy Georgia  (Talk) 16:06, 27 May 2007 (UTC)
 * PS, I just had a new look at WP:GTL, to set this up correctly. There are some new additions at GTL &mdash; we have to make sure WP:MEDMOS is in line with those.    Sandy Georgia  (Talk) 16:11, 27 May 2007 (UTC)
 * LOL, I see Fvasconcellos is riding to the rescue, and even invoking God :-) Sandy Georgia  (Talk) 16:26, 27 May 2007 (UTC)
 * I'm sorry, but mixing footnotes and a bulleted list—too much. Fvasconcellos (t·c) 16:28, 27 May 2007 (UTC)
 * You're right, but GTL actually says to do that &mdash; we need to deal with that. Another thing for the list !  Sandy Georgia  (Talk) 16:30, 27 May 2007 (UTC)

"Epidemiology" paragraphs 1 and 2
It's been a couple of weeks, so I went back and looked at Epidemiology. I started by reviewing its first two paragraphs, making several incremental changes. These changes are relatively minor: they leave the main outline alone, but simplify the text in an attempt to make it clearer to non-experts (suggested in above). They also adjust the text slightly to make it match citations, and remove a couple of stray citations to Schapiro 2002.

I'll look at paragraphs 3 through 5 later.

In rereading the first two paragraphs, it seems to me that they could be simplified further, but that would require more rewriting that would leak into the rest of the section, so I will put that off until I've had time to review the remaining paragraphs more carefully.

Eubulides 07:28, 11 June 2007 (UTC)


 * Thanks, Eubulides. That looks to me like a careful copyedit by someone who has access to the sources and the education to understand them. I think it has made it more readable by the "general reader". I agree with you on trying to use the original phrasings from the sources. Although it seems like stealing a little bit, it is hard not to lose or add something when trying to put things in ones own words. I also agree on replacing the quotation with our own words but it might have lost the point about "educational and other support" (though "other support" is rather vague). Do children with mild TS need educational support? If so what kind? (Such questions are better answered in another section).
 * I wonder about the phrase "a diagnosis is no longer warranted", which remains from the earlier text. To me, a diagnosis is the identification an illness. Perhaps this is a lay POV and maybe the medical folk talk about "having a diagnosis of" rather than "having had a diagnosis of" or "having been diagnosed with"? What do you think? Is there an alternative word or phrasing? Colin°Talk 09:13, 11 June 2007 (UTC)


 * "And other support" was vague, which is why I left it out. I took it to mean professional support other than by the usual medical and educational personnel (a diffuse category) but in rereading it I now think it might also mean support from the patient's family. I agree it'd be better to cover this elsewhere. As for the other phrase, the cited source says "By early adulthood, tic severity may have declined sufficiently that a TS diagnosis may no longer be warranted." So the phrase does appear to be medical-folk talk (at least one "folk" anyway…). Technically, what's going on here is that many patients no longer meet the DSM-IV-TR criteria in adulthood, so they no longer can be accurately diagnosed as having TS. I can't offhand think of a better way to say it than what's already there. Eubulides 17:18, 11 June 2007 (UTC)


 * No, "diagnosis may no longer be warranted" is not "lay POV"; it is the terminology used throughout the medical literature, and the most accurate description for discussing this. Eubulides inquired whether patients identified under DSM-IV-TR required intervention, so I searched out and supplied a direct quote to answer his query, which has now been deleted and replaced with someone vague and not what the text said.  Sandy Georgia  (Talk) 07:01, 14 June 2007 (UTC)


 * Sandy, you may be misreading what I (not Eubulides) said about "lay POV". I questioned whether my understanding of what diagnosis meant was my lay POV. Both you and Eubulides have confirmed this is standard "medical speak" and if you think it is the "most accurate description" then that's fine. Colin°Talk 10:56, 14 June 2007 (UTC)

I've finished my first pass through paragraphs 3 through 5 of Epidemiology and have installed several edits. Again, the goal was to preserve all the points and structure of the current text, while simplifying the discussion and using better or trimmed-down references. I'll now work on redoing the section to make it even shorter (and I hope better) and will put a draft of it in the talk page. Eubulides 22:36, 11 June 2007 (UTC)

Epidemiology, take 3
Here are some thoughts I had for improving Epidemiology. A suggested rewrite is listed below, under.


 * The section does not flow well. It starts with a few general concepts like prevalence is higher in children, then talks about problems in doing studies, then talks about history of epidemiological studies, then goes back to giving general prevalence concepts like prevalence is higher in males. Most likely, people who read the section want to know the general concepts first, then the current best estimates, then history, last problems with epidemiology. Eubulides 23:29, 11 June 2007 (UTC)
 * The original version, which was solidly reviewed and passed FAC not long ago, didn't have flow problems; getting back to the solid basis the article started with before all this chasing of tails to end up with basically the same content would be a good place to begin reconstruction. Sandy Georgia  (Talk) 08:43, 14 June 2007 (UTC)
 * The discussion of Kurlan doesn't fit in well with the rest of the section. The rest of the section talks in general terms and cites review articles, which is the recommended practice for Wikipedia medical articles. But the Kurlan section talks about one study (OK, two if you count Mason but that's just an aside). And this study reports results that are well outside the consensus (a prevalence rate of 38 per 1000, which would actually be higher if special-ed children were included). Kurlan focuses on the distinction between special- and regular-ed schoolchildren, which is a point worth making, so let's make that point and use Kurlan to support it instead. Eubulides 17:32, 12 June 2007 (UTC)
 * Why do you keep miscomparing Kurlan numbers to other studies? Percentages as he expresses are not the prevalence rates as expressed in other studies.  We've been over this several times.  Kurlan reports 8 in regular ed populations, 15 in special ed, and your Hirtz study reports the combined is 9-something.  And, how is Mason an aside?  Sandy Georgia  (Talk) 08:43, 14 June 2007 (UTC)
 * I wonder how much consensus is actually present in the "emerging consensus" that is discussed. If you take a look at you will see an author who estimates a prevalence between 4 per 1,000 and 17.6 per 1,000, which differs quite a bit from the "consensus" 1 per 1,000 to 11 per 10,000 given by Scahill. Perhaps it's better not to say "consensus" just yet. Eubulides 23:29, 11 June 2007 (UTC)
 * You'll also see the other quote from the same source, clearly stating the same thing said on this side of the pond, "that seven recent studies have remarkably consistent findings", and then referencing the same studies we've covered. Why do you want to extend the prevalence range upper bound even higher, from 11 to 17, because one author includes a higher study?  If you want, we can add a phrase after "tighter range of 3 to 8" indicating one saying it's even as high as 17.6, but I don't think we have good reason to do that, when there are consistent good studies in the 3 to 8 range.  Sandy Georgia  (Talk) 08:43, 14 June 2007 (UTC)
 * I can't read the paper you quote but have found a recent review that includes the same author and is UK-centric. (Free text here, requires free registration). There is a whole section on estimating the UK TS population. This paper regards 6–10 per 10,000 as a likely figure based on the current diagnostic criteria. Based on 2000 UK census data, "with a prevalence of 0.6% to 1%, this means that between 64 000 and 106 000 people between the ages of 5 and 18 years in the UK may have Tourette’s syndrome, of all severities." The paper also confirms "the Tourette’s syndrome symptoms are probably mild, fulfilling DSM-IV-TR criteria but unlikely to cause distress or impairment." Finally, it states "The prevalence of Tourette’s syndrome in special educational populations such as those with learning difficulties or autistic spectrum disorder is even higher." If you have problems getting this paper, let me know. Whether you use it or not, a review such as that helps confirm the information from other papers. It might be nice to include the UK population figures in the same manner as the US ones. Colin°Talk 10:54, 15 June 2007 (UTC)

Epidemiology redraft
Tourette syndrome is found among all social, racial and ethnic groups, Males are affected 3 to 4 times more often than females. Prevalence is much higher among children than adults: the tics of Tourette's begin in childhood, but tend to remit or subside with maturity so that a diagnosis is often no longer warranted, and children are 5 to 12 times more likely than adults to be identified as having tic disorders. Special education schoolchildren are more likely to meet diagnostic criteria for Tourette's than regular education schoolchildren are. As many as 1 in 100 people experience chronic motor or vocal tics or transient tics of childhood.

Tourette's itself was once thought to be rare; a 1973 registry reported only 485 cases worldwide. However, this began to change. By 2000 a review estimated a lifetime prevalence of 0.5 per 1,000 to 1 per 1,000, and multiple studies published since 1996 demonstrate even higher prevalences. A 2006 review estimated that between 1 and 11 children per 1,000 have Tourette's. Using 2000 census data and a prevalence range of 1 per 1,000 to 10 per 1,000, the review estimated that 53,000 to 530,000 US school-age children have Tourette's.

Prevalence estimates vary widely because epidemiologic methods are vulnerable to error: symptoms vary, are often intermittent, and are often not recognized by the patient or family. There are few broad-based population studies. Mild cases of Tourette's are less likely to come to tertiary or clinical attention, raising the problem of ascertainment bias: before 1980 epidemiological studies of Tourette's were based on individuals referred to specialty clinics, but children with milder symptoms are unlikely to be referred, so these studies were biased towards more severe cases. Studies using direct classroom observation or multiple informants yield higher prevalence estimates than other assessment methods do; many patients thus identified have mild characteristics that need no long-term medical intervention.

Eubulides 17:32, 12 June 2007 (UTC)

Review of draft
This review was done on the text here, which has changed slightly as a result.

I've compared the above text with the text from before any of the recent edits. Generally I think it is an improvement. The structure is better and slightly tighter (the main reason it is significantly shorter is the removal of the paragraph on Kurlan's study). I've tried to look for places where something may have been lost or emphasis changed. Of course, tightening the text also gives room to perhaps mention something that wasn't previously mentioned. I haven't attempted that...

The fact that "Prevalence is much higher among children than adults." is kind of lost in the above draft. There are hints that the tics "tend to remit" and that a "diagnosis [in adulthood] is no longer warranted" but this is too subtle for the reader to guess the scale of the difference between adult and childhood. I'd like the above sentence restored, possibly after the "warranted" sentence, beginning with "Therefore, ..."?

The "warranted for many adults" was shortened to just "warranted". Doing this loses the rough qualification that "many" gives. Combined with the vagueness of "tend to", the reader loses the idea of how many. The tics need to remit to such an extent that the diagnosis is dropped: does that happen to a few; a lot; most; nearly all? Can we add some qualification back, based on what the sources say.

I understand the order/structure you've gone for: facts, history, issues. However, it leads to the slightly confusing situation were current prevalence figures are given, followed by several older sets of figures. The reader may end up having the older figures as the ones they remember. A slight change to the ordering would result from moving the "A 2006 review .. with Tourette's" sentence to the end of the history paragraph. The structure then would be: Presenting the prevalence in that form also maintains the idea that this is an evolving figure, not one that anyone has settled on with any permanence. Another advantage is that the review's figures satisfy the reader after seeing the "even higher prevalences" statement.
 * 1) Analysis of affected groups, particularly childhood vs adulthood
 * 2) Population prevalence: history presented, leading up to current figures.
 * 3) Discussion of epidemiological difficulties

The previous text gave several "current" figures (1, 6–8, 10, 11), to which you mention above a range of 4–17.6 and Kurlan's 38 for "regular education children". It also used the word "consensus" for the 1–11 range. I'm not really in a position to argue which ranges we should give (Sandy can deal with that). By giving just one example range we are placing our emphasis solely on one review. Calling it just "A 2006 review" weakens it by making it look like just one author's POV. Using the word "suggested" weakens it further. The result is that the "current" data we present is merely based on a "suggestion" in "a review". Can/should the reader's opinion of the review be strengthened? For example, by stating how many studies were considered before coming up with the range. I take your point about wondering about consensus. Do any of the sources use that word?

The "1972 grant proposal" example has been replaced with a "1973 registry" example. I agree that's better as it is shorter, to the point and worldwide rather than US-centric.

Kurlan's study has been dropped. Sandy points out above that it is widely discussed so perhaps the argument that we are citing one study vs reviews isn't fair: we could cite a review citing Kurlan. But, from your edit history, I see that you've previously elided such intermediate citations. Unless Kurlan's study has been discredited, then it should not be dismissed merely as an outlier. Sandy points out above that it does several unique things which may account for the higher figure. Since it is just one study, we need to be sure it is important enough to mention. Can its importance be mentioned in the text (& cited to a review) along with a short discussion of it?

Colin°Talk 09:38, 12 June 2007 (UTC)


 * Thanks for the review. I have done the simple things you asked for. I put back "Prevalence is much higher in children than adults", put in an "often" near "warranted", put the current figures at the end of the history paragraph, and changed "suggested" to "estimated". The other points are more difficult:


 * I do not have access to the 2006 Scahill paper so I don't know whether it says "consensus" and do not know how many studies it considered nor how strong it thought these studies were.
 * As for what the consensus is, I'm sure we can easily find other papers that give still other ranges. For example, says "Although there has been debate as to how common Tourette’s syndrome is, a prevalence figure of between 0.6% and 1% in mainstream schoolchildren (using DSM-IV-TR criteria) is more likely to be correct than the previous lower figures that have been reported." (But note that this number is for mainstream schoolchildren; presumably the overall number for children would be higher but is not given.) I suspect Scahill's wider range of 1/1000 to 11/1000 is an attempt to cover wider ground and thus gain more consensus, but without access to the paper it's hard to say.
 * I'm not at all saying Kurlan et al. is discredited, only that its numbers are way out of range. Partly this may be due to the selection bias mentioned in the current version of the Wikipedia article. But there is something else going on here: the Kurlan numbers aren't measured the same way that other articles measure them: they aren't given relative to the overall school population, for example. Kurlan et al. focus on whether TS and related tic disorders are more common in special-ed versus regular-ed students; the question of prevalence in the general population is secondary. We could try to shoehorn their numbers into the same sort of statistics others use (that's what Hirtz et al. tried to do, apparently) but it's not clear to me how to do this. There are enough problems with discussing this paper that it's probably best to leave it out, at least in an article at this level.


 * Eubulides 14:28, 12 June 2007 (UTC)


 * There's not much more I can suggest, so I await Sandy's response which I'm sure is brewing somewhere :-). I forgot to mention the first time: the "A 2006 review" sentence is a bit on the long side. Can we split the sentence so the calculation is in a second sentence, while maintaining the fact that the calculation came from the review (which I think is important enough to make explicit)? Another thing that struck me before, and which you reminded me now: is this difference between children with "special educational needs" worth noting in the first paragraph. I can't find anywhere in the article where it is mentioned. Is there any info on the kinds of extra problems the kids had (e.g speech, reading, attention)? Colin°Talk 16:15, 12 June 2007 (UTC)


 * Good point about special-ed! I add a comment about special-ed children in the first paragraph, and used Kurlan to support it. That's what Kurlan is mostly about anyway; I feel much better about using Kurlan in this way. The kinds of extra problems children face is an important issue as well, but it belongs in some other section (not Epidemiology) so let's defer that for now. I also wordsmithed the "2000 review" sentence along the lines of your suggestion. Thanks again. Eubulides 17:32, 12 June 2007 (UTC)


 * We do know what Kurlan's comparable numbers are, because both Scahill and Hirtz report them; Scahill reports 8 for regular ed, 15 for special ed, and Hirtz reports some combined 9ish number as I recall. The 8 regular ed number is in line with consensus and other studies; the 15 is an issue because of the selection bias in the special ed response clearly discussed in both of his papers; it would be wrong to overextend that result, which is why I never included it in the original version. No, Kurlan is not mostly about the difference between regular and special ed unless you misinterpret the study limitations; if you turn mention of Kurlan's study into a special ed/regular ed difference, that will be misleading because of the sampling issues.  The significance of Kurlan's study is that he used trained observers and multiple informants in a large population over a long period of time; you've now removed what his study is known for from the text and are attempting to replace it with the weakest parts of his study, overextending results without accounting for the special ed selection bias. I included that specific info somewhere at one time; not it's gone and I'm chasing my tail again to correct information that was once correct. Sandy Georgia  (Talk) 06:22, 14 June 2007 (UTC)

Just catching up on work done here in the last two days. There are actually more than a few substantial changes of meaning in the edits I just reviewed and possibly some errors, so I'll have to catch up to you all quickly—tonight or tomorrow. Eubulides, which Scahill publication is it that you don't have? Also, note the error above, in (quote):
 * If you take a look at Robertson MM (August 1, 2005). "Tourette syndrome". Psychiatry 4 (8): 92–97. DOI:10.1383/psyt.2005.4.8.92. you will see an author who estimates a prevalence between 4 per 1,000 and 17.6 per 1,000, which differs quite a bit from the "consensus" 1 per 1,000 to 11 per 10,000 given by Scahill.

I think you meant 1,000 on Scahill. I'm not sure where the 17 is coming from, but I don't think it's worthy (yet) of mention, whatever it is, so the difference is only great if we're willing to say the prevalance is as high as 17; I don't think there's enough credible evidence to go there, just yet. Sandy Georgia (Talk) 22:08, 13 June 2007 (UTC)

Another review of proposed draft
A couple of excellent edits were introduced over the last few days (more below) that combined sentences in ways that only one reference was needed or clarified wording. The order/flow can be improved (it got quite twisted around in the last round of debate here), but not at the cost of introducing errors. In addition to the errors introduced over the last few days (see below), here is a list of concerns about the proposed text:
 * I don't know where this statement is coming from; even Kurlan recognized his sample was biased towards special ed, and I don't recall any such statement as this one in his two papers. I don't recall him drawing this conclusion, considering the sampling issues (please provide the quote if I'm wrong), but this looks like synthesis and a misinterpretation, considering the limitations of his sampling methods: "Special education schoolchildren are more likely to meet diagnostic criteria for Tourette's than regular education schoolchildren are."
 * I don't have the full paper. But from the abstract you get
 * "RESULTS: Twenty-seven percent of 341 students classified as receiving special education (SpEd) had tics compared with 19.7% (p = 0.008) of 1,255 students in regular classroom programs (RegEd). The weighted prevalence estimates for tics were 23.4% in SpEd and 18.5% in RegEd. A higher percentage of students in SpEd (7.0%) met diagnostic criteria for TS than students in RegEd (3.8%; p = 0.01)." (my emphasis)
 * I assume, from your comments, that the article goes on to discuss (or other reviewers have) whether those results can be regarded as a fair interpretation of the population at large. Colin°Talk 08:58, 14 June 2007 (UTC)


 * ... "but tend to remit or subside with maturity so that a diagnosis is often no longer warranted" ... diagnosis is no longer warranted by adulthood (the studies specifically deal with the first two decades of life, and leaving this off makes the time frame unclear and the statement vague ...
 * "Tourette's itself was once thought to be rare; ..." "Tourette's itself" isn't optimum prose, the article is about Tourette's itself.
 * "However, this began to change. By 2000 a review estimated a lifetime prevalence of 0.5 per 1,000 to 1 per 1,000, and multiple studies published since 1996 demonstrate even higher prevalences." "However, this began to change" is a throwaway sentence that adds nothing; it didn't just change, studies got better and showed why earlier studies were wrong (previous text stated this but was deleted).  And, in terms of the history of Tourette syndrome, the 1996 date has no significance, so gives no context for things suddenly "changing".  Also, bouncing from the past to 2000 then back to 1996 is confusing.  The good and newer studies, upon which current consensus relies, are all newer than 2000; that's the only reason to mention a date.  Older studies all had issues.  1996 is only coming in to include Hirtz, which isn't a good analysis for TS numbers.  So, this is muddying the issues, which is that studies are getting better, estimates are getting better.  1996 is a red herring; good studies are more recent; the text explaining why more recent studies has been deleted, so there's no context left, just a vague, unexplained "things began to change".
 * Another note confirming text above, new results are zeroing in, and insignificance of 1996 date, Robertson writes similar to all of Scahill text ... Robertson 2005 "Tourette syndrome was once considered to be uncommon, but no less than seven recent studies have remarkably consistent findings ... " Sandy Georgia (Talk) 05:15, 14 June 2007 (UTC)
 * "A 2006 review estimated that between 1 and 11 children per 1,000 have Tourette's." I'm not sure this journal paper is classified as a "review", and it didn't "estimate" the number.  It states the "emerging consensus" is 1 to 11, with a tighter range of agreement between 3 and 8 (which, BTW, is basically exactly the range Hirtz came up with, but by averaging a lot of bad studies, rather than an analysis of newer, better studies), and within the range of Robertson's strange 4 to 17, with an unexplained 17.  In other words, this seems like a lot of verbiage that obscures a fairly simple matter of an emerging consensus, as stated in the original text.
 * I think these issues might be due to Eubulides not having the 2006 paper. If it contains what you say about "emerging consensus" and the 1, 11, 3, 8 figures then lets write a sentence that says that. Note that the version from May 22nd doesn't say "consensus", merely that "contemporary prevalence estimates range from 1 to 3 per 1,000 to 10 per 1,000". Colin°Talk 11:14, 14 June 2007 (UTC)
 * "Prevalence estimates vary widely because epidemiologic methods are vulnerable to error:" ... This doesn't seem quite right or complete; not all of the errors in estimates have to do with epidemiologic issues, there are other differences. And they aren't still "varying widely"; as methods get better, they are focusing in, so the emphasis here doesn't seem right.  We need to explain why estimates change; there are other reasons not covered here, and studies are getting better because those issues are being addressed in newer studies.
 * "Studies using direct classroom observation or multiple informants yield higher prevalence estimates than other assessment methods do; many patients thus identified have mild characteristics that need no long-term medical intervention." This obscures mention that newer studies are better because of direct observation by multiple informants including trained observers - which is what Kurlan did and older studies didn't.  The fact that this makes the higher estimates more likely to be more accurate isn't explained, and the point of raising this info is no longer clear.  Scahill has wording we can adapt, but basically a lot of this proposed draft just isn't correct.  Need no long-term medical intervention is not what the source said; it's a paraphrase which obscures the fact that support is needed.  Sandy Georgia  (Talk) 03:22, 14 June 2007 (UTC)

Older edits to review and fix
Problems in the edits of the last few days; it might be premature to move forward to yet another draft, which introduces error on top of error. The edits of the last few days introduced some vague and choppy prose, some potentially misleading text, unnecessarily deleted citations.

Colin, on your comments above, I've got to first focus on working back through some of the edits done in the last two days. The Hirtz review is just not useful; as a meta-analysis, one of its fundamental flaws is that it will be weighted towards Apter (by the large sample size), which has dramatic, fundamental known errors (we haven't even started on their claim that 8% of people with TS have ADHD and that the male/female ratio is essentially equal; such are the known errors found in Apter, because people had good reason to lie. Since it's the largest study, it just doesn't allow Hirtz to provide for a good meta-analysis). Preferable to reflect more studied analyses of newer data, which are available. Yes, several publications use the word "consensus" on epidemiology. Eubulides, which of the three cited Scahill publications do you have and not have, so I can be more careful to provide exact quotes, and I'll see if I can locate online copies anywhere?

Removing Kurlan will just create a future headache, because it will be re-added as it is so widely quoted and reported. Better to deal with it than just leave it out. It has a problem (selection bias) that should be addressed, but there is no strong evidence that its prevalance is wrong, and it validates Mason and is within the range of several other studies. This 4 to 17 Roberson number is now coming up; I can't see any reason to broaden the range even further than current consensus (to 17 instead of 11) because Robertson suggests it, when several other publications confine the range more reasonably. Robertson just isn't subject to the same broad peer review and consensus that TSA Medical Advisory Members are. There seems to be still be ongoing misunderstandings here on Kurlan; his numbers aren't at all out of range, we just haven't gotten to the bottom of why he reports percentages one way, while Scahill and Hirtz each convert his percentages to a prevalance, which is completely within range of other studies and consensus. Kurlan's number is 8, as reported by Hirtz and Scahill, which is within the 1 to 11 and the 4 to 17. His work has not been discredited (rather, viewed in context), and the reason it warrants mention is specifically that it has been one of the few broad studies to use trained technicians for observation (and since such wording was used in the literature, it is not a "peacock" term ... it's a studied analysis of what he did right that former studies got wrong).

In general, I see several potential misstatements and misquotes, and choppy prose which introduces vagueness, so I'll get through the list below to first correct the current text as soon as I can. Sandy Georgia (Talk) 00:55, 14 June 2007 (UTC)


 * [[Image:Symbol keep vote.svg|15px]] Dude reference was deleted, verified the text, tied older studies together, no reason to delete. (FIXED) http://en.wikipedia.org/w/index.php?title=Tourette_syndrome&diff=next&oldid=137395467


 * Quote not reworded accurately, context and meaning lost, need to dig out sources again and correct this: http://en.wikipedia.org/w/index.php?title=Tourette_syndrome&diff=next&oldid=137397735


 * [[Image:Symbol keep vote.svg|15px]] Connect choppiness: (FIXED) http://en.wikipedia.org/w/index.php?title=Tourette_syndrome&diff=next&oldid=137399490


 * Check this wording, not sure it's accurate, dig out source and fix: http://en.wikipedia.org/w/index.php?title=Tourette_syndrome&diff=next&oldid=137402703


 * Misleading, need to re-read and re-do to recover precision: http://en.wikipedia.org/w/index.php?title=Tourette_syndrome&diff=next&oldid=137403099


 * A Robertson publication aimed at laypersons is not a superior source, nice that it includes mention of international data which helps globalize the article (although I carefully tried to use only the best sources, not older lay publications, and would be much happier if we could cite this to a PMID or a textbook, not a publication aimed at laypersons), but no need to remove US/NIH data, which provide context for other numbers mentioned and are from a superior source. Should restore NIH data, [[Image:Symbol keep vote.svg|15px]] but would like to find that Robertson info in a better source if possible.  http://en.wikipedia.org/w/index.php?title=Tourette_syndrome&diff=next&oldid=137407707
 * "this began to change"? Vague, choppy, uninformative.  Previous text explained why it changed … Because newer studies showed it wrong.  Bagheri doesn't differ; need to recheck that, again, no need to delete sources.  http://en.wikipedia.org/w/index.php?title=Tourette_syndrome&diff=next&oldid=137499827


 * Figure out where I have this source, if I have it online, and if I can send it to Eubulides. Since I don’t receive TSA junk, I must have found it online somewhere - figure out where I got it or whether I have it saved somewhere online.  http://en.wikipedia.org/w/index.php?title=Tourette_syndrome&diff=next&oldid=137510941
 * I see you found the first page, which is most of it (I do have it in electronic form if you need it). The second page is mostly a table summarizing the data in the studies performed since 2000 (the date for the better studies analyzed).  I've also now caught up and realize that it's the 2006 Scahill paper that you don't have.  I do not have it in electronic form; I can just provide the exact quotes wherever needed.  Sandy Georgia  (Talk) 03:29, 14 June 2007 (UTC)


 * Hirtz review is not useful; outdated, doesn't include newer studies, gives undue weight to studies with dramatic/significant known errors, no reason to cite it when superior analysis/sources are available, and certainly no reason to delete all of those sources for future readers/editors who want the detail. Sources should be restored; [[Image:Symbol keep vote.svg|15px]] they aren't hindering the article, and they aid readers who want to do the research. http://en.wikipedia.org/w/index.php?title=Tourette_syndrome&diff=next&oldid=137523553


 * [[Image:Symbol keep vote.svg|15px]] This is a very good edit: http://en.wikipedia.org/w/index.php?title=Tourette_syndrome&diff=next&oldid=137527016


 * Trained technicians is not a peacock term: it is a specific criticism of other studies that explains the difference with Kurlan's study - the first to use trained observers.  Terminology that is specifically used in sources shouldn’t really be branded as “peacock”, especially when it explains why the study was superior (in that sense, anyway).  This is specifically raised in one of the other papers; I’ll find the exact wording and relate this passage to the discussion in the literature that this was a specific shortcoming in earlier studies (although I’m now paraphrasing less and less and having to use direct quotes from sources). Also, I thought we had replaced grade 9 with an age since grade is meaningless, need to backtrack old notes and correct that to include age rather than grade. http://en.wikipedia.org/w/index.php?title=Tourette_syndrome&diff=next&oldid=137528980


 * [[Image:Symbol keep vote.svg|15px]] Other citations aren't needed? The three to four was originally questioned by Eubulides, although it’s basically a “common knowledge” type number, which is why I did the work to put the detail in.  No reason to delete citations which another editor may ask for in the future, for the same reasons that Eubulides did – if other sources quote other numbers, we should explain.   (UNDID) http://en.wikipedia.org/w/index.php?title=Tourette_syndrome&diff=next&oldid=137547530

Sandy Georgia (Talk) 03:37, 14 June 2007 (UTC)


 * Another note to myself; this edit has to be repaired as well. http://en.wikipedia.org/w/index.php?title=Tourette_syndrome&diff=next&oldid=137547530  Sandy Georgia  (Talk) 05:04, 14 June 2007 (UTC)


 * Sandy has alerted me to the fact that this article has become unstable. I worked on the prose some time back, and thus have an investment in it. I see changes that have reduced its quality significantly. I also see a reduction in the quality of referencing. I'd be pleased if people raised their proposed edits on the talk page before plunging in. The entry above is unsigned ... Tony 03:31, 14 June 2007 (UTC) PS By "unstable", I was referring only to Sandy's advice that there has been heavy editing recently and that as a result, the prose will need to be worked on. Tony 07:42, 14 June 2007 (UTC)
 * Yes. The flow is now off, and important concepts are muddled, obscured or lost, replaced by vagueness; before repairing the copyedit issues, the content has to be repaired. Copyedit may be premature, since repairs have to be done.   Sandy Georgia  (Talk) 08:06, 14 June 2007 (UTC)


 * Oops, the entry above is mine, but is a list under my sig. I'll sign the list. I don't think the article has become unstable, though; I'm just spending a lot of time explaining things, fixing errors, adding back deleted citations, and ending up back where we started but with vague prose.  For example, three to four times more males than females was questioned, I extensively cited and explained the number, then the info I spent so much time generating was deleted.  I was specifically asked if cases identified under better sampling methods and with current DSM criterion required services or intervention, I searched for and added an answer to that query, the info was then deleted.  Recent studies came under fire; I extensively cited all of the recent studies; the citations were deleted.  Hard to understand deleting citations, as they don't burden the reader.  Some information is being obscured with vague text and choppy phrases like "However, this began to change", and "Prevalence of Tourette syndrome is much higher among children."  And there have been several proposals for information that is outright wrong.  Not unstable; just a lot of work to keep it on track.  Sandy Georgia  (Talk) 03:37, 14 June 2007 (UTC)

Placing a copy of the last version I knew to be accurate here, so I can get back to it for repair. It flowed fine before all of these changes. It may, again, be easier to revert and rebuild from an accurate version, incorporating the good changes. Sandy Georgia (Talk) 07:38, 14 June 2007 (UTC)

Summarzing all of the accumulated problems (see above), I suggest we go back to the May 22 version, before any of this tail chasing that has resulted in almost no change, and re-incorporate the valuable changes from there. We've typed boatloads of words and I've produced truckloads of research for you, to come back to the same conclusions, just as we did on the three to four males to females, whether TS was historically viewed as a rare disorder, and all the other hoops we've been jumping through to end up where we started (albeit with Eubulides now knowing more about TS than when he started), which was with an accurate, well written, well sourced epidemiology section that has now been reduced to vague muddled concepts and poor flow. If I recall correctly, all of this has come about because Eubulides wants to introduce the Hirtz study, which isn't even primarily about TS, and draws on older, outdated data, doesn't have the advantage of the current data and emerging consensus, and because you wanted the lower bound on the prevalence numbers? You've now found UK numbers as well (in the Robertson 2005 study), so they can be added, the lower bound on US numbers is added, but nothing else of substance has been changed here, and this has been one time-consuming wild goose chase. Sandy Georgia (Talk) 08:43, 14 June 2007 (UTC)

I've printed out the body-text from the 22nd May, 6th June (unchanged since 28th May), the 12th June and the draft. I can find fault in them all so I'm not sure it is useful to revert back so far.

The draft has some weak sentences I didn't spot but those are minor issues. I'm not going to argue with the flaws you found since they require a deeper understanding than I can supply. However, I like the structure of the draft: The first two paragraphs could be swapped if required. I like how the reader can get hold of the current facts without having to read through discussion material on the difficulties of diagnosis or performing epidemiological studies. The text doesn't contain any flab.
 * 1) Analysis of affected groups: child vs adult, male vs female, SpEd vs RegEd, social/racial/ethnic.
 * 2) Population prevalence: history presented, leading up to current figures.
 * 3) Discussion of epidemiological/diagnostic difficulties.

The version from the 22nd May has some repetition and sentences that are longer than they need to be. The first paragraph gives prevalence figures without indicating that these are for children, which is a fairly serious flaw if the figures for adults are much lower. It calls Kurlan's study "large" which may not be a suitable adjective. It doesn't discuss any of the aspects of that study that make it better or indicate its possible flaws. It doesn't mention the 5–12x children vs adult ratio. It doesn't include an example of "rare". Although the sentences flow well enough, the overall structure is IMO a bit weaker than the draft.

The version from the 6th June has expanded from the previous round of discussions. Despite getting a copy-edit, its overall structure is poor. For example, the final paragraph should be right up near the top since those aspects are key epidemiological statements. A good point is that the Kurlan study is dealt with in a paragraph on its own. This separates that study's results and issues from other data that doesn't come from the study (compare 22nd May). It also contains a much better version of the Scahill information than the draft or the 22nd.

I don't want to get bogged down with reviewing the edits that gave the version on 12th June since clearly Eubulides felt the section needed a more radical restructuring – hence the draft.

I also must repeat that I have not reviewed the refs so make no comment on additions/deletions/changes in that regard. If this is where the worst of the crimes have been committed :-) then I'm ignorant. A "general reader" of any of those four versions would IMO come away with pretty much the same knowledge about TS. The exception is the Kurlan study, which is is only briefly mentioned in the 22nd May. It is mentioned in the draft but only to support the SpEd data. Sandy - are you aware of better studies that look at the SpEd population group?

In summary, I think there is something good to take from each of the versions and that we can improve on what was originally there. There's probably also some points still on this talk-page that haven't made there way to the article. Colin°Talk 13:04, 14 June 2007 (UTC)


 * I'm working on it in my sandbox, and will propose a new draft here once I get out all the kinks that are still troubling me. In addition to working out some subtle errors that are introduced/proposed, I also believe the issues in all of the studies should be explained before the numbers are presented, so the discrepancies among estimates and rising prevalence rates can be understood in context.  IMO, explaining the issues affecting TS studies after presenting the numbers will leave the uninitiated reader even more confused about the rising prevalence rates. Sandy Georgia  (Talk) 18:21, 16 June 2007 (UTC)

New draft
From my sandbox, reviewed by Tony and Colin; unless someone objects, I'll add this to the article tomorrow.

Remove and correct subtle misstatements while incorporating good additions and adding some missing info; restore an order that explains the issues in TS studies before introducing the numbers, so the numbers are viewed in correct context. Paragraphs:

To do later: Find exact reference to 1973 registry, to avoid citing Robertson book. Sandy Georgia (Talk) 23:25, 16 June 2007 (UTC)
 * 1) Introduce solid, established basic data, setting the context for the adult/pediatric population issue, relationship to tic disorders, general numbers, etc.
 * 2) Introduce and explain all factors—not just symptom variance—that affect studies.  Expand section on problems with prevalence estimates to fully explain methodological limitations and why prevalence numbers have risen with better methodology.
 * 3) Once rare, no longer, better numbers from improved studies.  Then give the prevalence numbers, within the context now established.
 * Leave out Kurlan (no one gets it) and Hirtz (doesn't include recent better studies, undue weight to outdated Apter study which has issues, not even a TS article). Remove references to 1996, not a significant date in TS history, good studies came out after 2000; keep this article in sync with TS History.
 * I so carefully avoided promoting or using any of the books aimed at laypersons in this article (it troubles me when an article like Asperger syndrome is sourced mostly to books you can find in the pop psychology department of your local bookstore, and I prefer PubMed sources and medical textbooks). I urgently want to replace the Facts booksource.  I think I've located the registry source to replace the Robertson Facts book, but need to get my hands on it—hopefully soon.  I'll see if a friend can view this article in a library.  Abuzzahab FE, Anderson FO. Gilles de la Tourette's syndrome; international registry. Minnesota Medicine. 1973 Jun;56(6):492-6.
 * Sandy Georgia (Talk) 14:45, 17 June 2007 (UTC)

Tourette syndrome is found among all social, racial and ethnic groups, has been reported in all parts of the world, and is three to four times more frequent among males than among females. The tics of Tourette syndrome begin in childhood and tend to remit or subside with maturity; thus, a diagnosis may no longer be warranted for many adults, and prevalence is much higher among children than adults. Children are five to twelve times more likely than adults to be identified as having tic disorders; as many as 1 in 100 people experience tic disorders, including chronic tics and transient tics in childhood.

There were few broad-based community studies published before 2000 and until the 1980s, most epidemiological studies of Tourette syndrome were based on individuals referred to tertiary care or specialty clinics. Children with milder symptoms are unlikely to be referred to specialty clinics, so these studies have an inherent bias towards more severe cases. Epidemiological studies of Tourette syndrome are vulnerable to error, because tics vary in intensity and expression, are often intermittent, and are not always recognized by patients or their families. There are discrepancies across current and prior prevalence estimates due to several factors: ascertainment bias in earlier samples drawn from clinically referred cases, assessment methods that may fail to detect milder cases, and differences in diagnostic criteria and thresholds. As the diagnostic threshold has moved towards milder cases with the removal of the impairment criterion in DSM-IV-TR, the result is an increase in estimated prevalence. Recent studies—recognizing that tics may often be undiagnosed and hard to detect—use direct classroom observation and multiple informants (parent, teacher, and trained observers), and therefore record more cases than older studies relying on referrals.

Tourette syndrome was once thought to be rare: in 1972, the US National Institutes of Health (NIH) believed there were fewer than 100 cases in the United States, and a 1973 registry reported only 485 cases worldwide. By 2000, a review estimated a lifetime prevalence of 0.5–1.0 per 1,000; however, multiple studies published since 2000 have consistently demonstrated that the prevalence is much higher than previously thought. The emerging consensus is that 1–11 children per 1,000 have Tourette's, with several studies supporting a tighter range of 6–8 children per 1,000. Using year 2000 US census data, a prevalence range of 1–10 per 1,000 yields an estimate of 53,000–530,000 school-age children with Tourette's. Using UK population data from the year 2000, a 10 per 1,000 estimate means that 106,000 children aged 5–18 years may have Tourette syndrome. Most of these children are undiagnosed and have mild symptoms without distress or impairment.

Sandy Georgia (Talk) 23:25, 16 June 2007 (UTC)

As Sandy said, I've seen a sneak preview of this text and have made my comments which have been addressed to my satisfaction. I've followed the edits that led to this text, which incorporates the best of all the various texts we've looked at, as well as ensuring very careful wording. I've now got a greater appreciation of how a subtle change of words can give the wrong impression by accident. The casual "general reader" might not spot any significant differences between the texts. However, a reader who wishes to read more studiously will, I think, find these words are faithful not only to the sources cited but also to the larger body of TS research, of which Sandy has intimate knowledge. The three paragraphs share the same structure that I said I liked about the draft proposed by Eubulides. Each paragraph is cohesive and reads well. The order of paragraphs 2 and 3 are swapped, which wasn't my original preferred order as I thought there were advantages to presenting the current facts up front and leaving the discussion to last. However, there are good arguments for this order too. The explanation, in the middle, helps the reader understand why the prevalence figures change throughout the last paragraph. I'm happy to recommend this text be adopted into the article. Colin°Talk 10:20, 17 June 2007 (UTC)
 * As am I. In my humble opinion, this new wording is clearer than the original text at the time of FAC and will add to the article's accuracy. Fvasconcellos (t·c) 12:54, 17 June 2007 (UTC)
 * Yes, it is a real improvement over the original text, and I thank Eubulides for pinpointing the weaknesses in the earlier version; he gets credit for the improvement (even if I am still bristling that he labeled sourced text in an edit summary as "peacock terms" :-) Fv, your input to and vigilance of this article contributes to its integrity, and Colin, the textual clarity of this section is due to your attention to detail, the time you invest in understanding each word, and your superb skills at tightening the wording on copyedit.  I have the knowledge of the research, but I don't have your ability to put it together in tight, brilliant prose; thank you so much for all the help.   I hope to verify the registry data today.  Sandy Georgia  (Talk) 15:01, 17 June 2007 (UTC)

Khalifa concern
I'm honing in on the registry source, but there is still one little subtlety in the proposed draft that is troubling me. What we have is technically accurate, but still may be subtly misleading to the reader. If you look at my sandbox comments on the Khalifa study, you see that we have a large study, that missed cases, that insisted upon impairment (strictly interpreted DSM-IV criterion), yet finds a prevalence in that "remarkably consistent" range of 6–8. The draft says:
 * There are discrepancies across current and prior prevalence estimates due to several factors: ascertainment bias in earlier samples drawn from clinically referred cases, assessment methods that may fail to detect milder cases, and differences in diagnostic criteria and thresholds.  As the diagnostic threshold has moved towards milder cases with the removal of the impairment criterion in DSM-IV-TR, the result is an increase in estimated prevalence.

This is true to Scahill, but he also goes on to explain that Khalifa tried to address the mild/impairment issue, by insisting on impairment. At least one other newer study (Wang) found 5.6/1,000 using DSM-IV, and I believe there are others. I don't want to get into discussing individual studies like Khalifa or Wang again, and don't want to grow the text too much, but my concern is that the text subtly implies that much of the increase in prevalence is due to the DSM change, when in fact, even studies with impairment are honing in on that range. I wish we could fix this with just one or two words or phrases from the Scahill letter (PDF). Sandy Georgia (Talk) 18:23, 17 June 2007 (UTC)
 * How about something like ... the result is an increase in estimated prevalence, although newer studies using the old impairment criterion of DSM-IV but better sampling and assessment methodology also report consistently higher prevalence.  ??  Sandy Georgia  (Talk) 18:38, 17 June 2007 (UTC)

I can't read the full papers. I have found that Khalifa milked that study for two more papers: and. You might find some more information there. Based on your notes and those abstracts, the study used a screening questionnaire which we might assume would remove those children with tics that only a trained eye would spot. I don't know what the questionnaire asked and how much effect it has but it is an example of an "old style" sampling technique so doesn't fit with your new "better sampling and assessment methodology" clause. It is an example of a community study rather than referrals. But it also isn't completely the new style either.

Of the children they found had TS, "All except one with TS had contact with medical services". This is a similar phrase to your notes. They don't explicitly state they already had a diagnosis. I assume their contact with medical services was wrt the tics, which must have been troublesome enough to warrant seeking help. But the text is vague enough that perhaps their medical contact was wrt a comorbid issue.

Your notes say "High response rate to questionnaire (59%), but (IMO) that also means milder tics probably missed". Yet also say "the 41% who didn't respond would be more likely to have non-impairing or unnoticed tics if they had any". I don't under how a good response rate means milder tics were probably missed.

Currently I don't know what to do with it, other than, like Scahill, take it as evidence that "TS is more common in school-age children than previously proposed". Perhaps there is more info in those two other papers to enlighten you? Colin°Talk 19:20, 17 June 2007 (UTC)

I've copied the new draft by Sandy onto the main article. I think we've discussed this section enough. Sandy has gone on wikibreak. I'm going to archive this talk page to draw a line under things. Lets all find other parts of wiki to enjoy for a while. Cheers, Colin°Talk 10:55, 18 June 2007 (UTC)