Talk:Surrogate endpoint

Identical topics
The Surrogate marker and Surrogate endpoint articles discuss identical topics. I suggest this article be merged to Surrogate endpoint. Gak 01:35, 22 May 2007 (UTC)

Absolutely (same applies to Surrogate end point, which is merely a discouraged spelling of "endpoint")! I also agree that Surrogate endpoint should be the name of the merged article. --88.77.253.198 23:17, 14 October 2007 (UTC)


 * Went ahead and performed the merge. EAE (Holla!) 03:42, 16 January 2008 (UTC)

A surrogate endpoint and a surrogate marker are absolutely not the same thing; a surrogate marker is an indirect measurement of disease progression (e.g. CD4 count in HIV) used to inform medical treatement. It is a continuous variable. A surrogate endpoint is a pre-determined value of a surrogate marker used to estimate the effectiveness of an intervention in a clinical trial, as described in the article. This is a binary variable. I therefore propose to de-merge these topics unless anyone objects strongly. Tricky (talk) 16:12, 18 June 2009 (UTC)

Low importance of surrogate outcomes
I think this statement must be revised. Could somebody give some feedback about this article? --Truebreath (talk) 12:15, 11 May 2014 (UTC)

Unbalanced flag
This article makes some very valid points about surrogate endpoints. But it completely lacks balance, coming across as a lengthy rant against the use of an important tool for clinical investigations.


 * While I agree (as do most) with the argument that true outcome-based endpoints should always be performed on new treatment modalities, the article completely neglects the role that less expensive, more readily measured endpoints have in the process of moving drugs forward through clinical development, from less expensive to more expensive trials.
 * Antibiotic trials are performed on the basis of the well established surrogate endpoints of physical signs and symptoms of infection. If we required a mortality endpoint in every clinical trial, the shortage of new antibiotics would become more acute, as the cost of developing drugs for infections that have mortality <10% with the current standard of care would rise into the billions of dollars.
 * The new hepatitis C drugs that were developed recently provide 90% viral clearance vs 50% for the old. How many patients should die of cirrhosis while we run 10 to 15 year trials to demonstrate that cirrhosis is prevented in those rendered virus free with the new regimens, just as was true of the old?
 * The article as currently constituted notes that proof of reduction of cardiovascular death was not obtained for many years after the approval of simvastain based on surrogate markers. The implication seems to be that approval should have been withheld until the 4S trial was completed 5 years later.  How many additional deaths would have occurred if the approval of statins had been delayed for an addtional 5 years?  This is a terrible example and I don't understand exactly how the editor who wrote this thought it supported their POV.

Like most things in life, what is needed in clinical development is a thoughtful, balanced approach that balances potential risk with potential benefit. No one waited until the protease inhibitors had demonstrated effects on mortality to approve them or begin using them. They were given to every HIV patient in the country based on serum viral load measurements, and within 2 years the rate of HIV deaths in the US was cut in half. Similarly, no one is going to wait 15 years to make sure that clearing the HCV virus from patients using the new drugs prevents cirrhosis just like curing their infection with the old one does. Diabetes drugs are a very different story, in which the risk of using the old drugs is small, and the risk of approving new ones based on serum HbA1c alone is high.

The article should reflect this. Formerly 98 (talk) 06:16, 21 June 2014 (UTC)

Happy with your imput
There is no doubt that treatment of bloodpressure lowers stroke, and that the use of antihypertensives saves lives. I thought this was obvious. But we need to be critical. Dual RAS didn't. Alpha blockers didn't. Intensive blood pressure lowering under 140/90 didn't. In diabetes lowering blood pressure under 130/80 harmed. The adagio "the lower (surrogate) the better" is dangerous in hypertension.

Otherwise the farmacological treatment of BMI and HbA1c surrogates has costed billions of dollars and didn't apport proven benefit, (excepted for metformin). The farmacological treatment of BMI almost always didn't work or harmed.

The predecessor of the statins the fibrate atromidin costed lives.

The use of surrogate outcomes and compound outcomes in medical litterature may mislead readers (doctors and patients) if the limitations of these endpoints are not adequately explained. I try to explain the limitations of different surrogate and compound endpoints.

I will try too make the text more balanced in developping the first part.(benefits of surrogate endpoints)

I will remove the primary sources. but this will take some time.

I hope then you can remove your flag.....

You said "Antibiotic trials are performed on the basis of the well established surrogate endpoints of physical signs and symptoms of infection. If we required a mortality endpoint in every clinical trial, the shortage of new antibiotics would become more acute, as the cost of developing drugs for infections that have mortality <10% with the current standard of care would rise into the billions of dollars."

I think The absence of physical signs and symptoms of infection is a hard endpoint for resolution of an infection, not a surrogate. The surrogate is the culture and antibiogram. I never said only mortality was a hard endpoint; you can see this in the examples cited in the text.

You said "The new hepatitis C drugs that were developed recently provide 90% viral clearance vs 50% for the old. How many patients should die of cirrhosis while we run 10 to 15 year trials to demonstrate that cirrhosis is prevented in those rendered virus free with the new regimens, just as was true of the old?" I agree; this must be put in the item benefits.

You said "The article as currently constituted notes that proof of reduction of cardiovascular death was not obtained for many years after the approval of simvastain based on surrogate markers. The implication seems to be that approval should have been withheld until the 4S trial was completed 5 years later. How many additional deaths would have occurred if the approval of statins had been delayed for an addtional 5 years? This is a terrible example and I don't understand exactly how the editor who wrote this thought it supported their POV."

I agree; but it was not me who made this text. The example is terrible; but it was preceded by the atromidin disaster. Please don't use abreviations; what is POV?

You said "Like most things in life, what is needed in clinical development is a thoughtful, balanced approach that balances potential risk with potential benefit." I agree

You say "No one waited until the protease inhibitors had demonstrated effects on mortality to approve them or begin using them. They were given to every HIV patient in the country based on serum viral load measurements, and within 2 years the rate of HIV deaths in the US was cut in half." I agree, but there were problems with the choice of an adequate surrogate endpoint; initially CD4 counts were used. You said: "Similarly, no one is going to wait 15 years to make sure that clearing the HCV virus from patients using the new drugs prevents cirrhosis just like curing their infection with the old one does." I agree, HIV and HCV are two successes of surrogate treatment. This must come in benefits

You said: "Diabetes drugs are a very different story, in which the risk of using the old drugs is small, and the risk of approving new ones based on serum HbA1c alone is high." i agree; but intensive treatment (glycemic controle below HbA1c 6,5) is harmfull.

Finally could you give some input on the part: problems with surrogate endpoints?

--Truebreath (talk) 10:57, 22 June 2014 (UTC)
 * To clarify, POV stands for point of view, and the Wiki guideline WP:NPOV holds that articles must hold a neutral, balance point of view, giving voice to both sides. -- CFCF  (talk · contribs · email) 17:31, 27 June 2014 (UTC)


 * I still think it reads like an anti-surrogate endpoint editorial, or possibly just a general criticism of the pharmaceutical industry. There are 25 lines explaining the benefits of surrogate endpoints and approximately 250 lines describing why they are a problem.  Wikipedia's neutral point of view policy requires that the articles should devote space to each point of view proportionate to its prominence among experts, and that we describe controversies without taking sides in them.  If the current view among such experts was 10 to 1 against their use, they would not be permitted to be used in approval decisions, and they would not be discussed in meta analyses and systematic reviews.  But they are discussed in 90% of systematic reviews, and they are used for approval under some conditions in every country in the world. The article is unbalanced.


 * A few other comments:
 * The statement "Systematic reviews should not present surrogate outcome data as their primary analysis" is clearly not mainstream as 90% of published meta analyses do exactly that. The article should be mostly representative of mainstream opinion.
 * The problems with rimonabant are unrelated to the subject of surrogate endpoints. It was pulled for CNS side effects, which are unrelated to the question of whether weight loss was an acceptable surrogate for cardiovascular disese.  There is also an extended criticism of the two recently approved weight loss drugs (which I don't really approve of either), but neither of these has been shown to have adverse cardiovascular effects.  So the listing out of the side effects of these drugs is just WP:COAT.  Similarly, neither the cost of orlistat nor any vitamin loss associated with its use has anything to do with the validity of surrogate endpoints.  This is also WP:COAT
 * Teriparatide is criticized even though it used a "real" endpoint for in its trials. It reduced vertebrae fractures by nearly 2/3's (From 14.3% to 5.0%), and reduced combined non-vertebral fractures by half (5.5% vs. 2.6%). Hip fractures were 0.2% vs 0.7% for control, almost reaching statistical signficance in spite of very low incidence.  To read the article, one would get the impression that this drug has not been proven to do anything at all.  Furthermore, as Teriparatide used "real" endpoints for approval, it has nothing to do with the subject of surrogate endpoints, and this paragraph is WP:COAT.
 * This is what the Cochrane group said about alendronate: "At 10 mg per day, both clinically important and statistically significant reductions in vertebral, non-vertebral, hip and wrist fractures were observed for secondary prevention ('gold' level evidence, www.cochranemsk.org). We found no statistically significant results for primary prevention, with the exception of vertebral fractures, for which the reduction was clinically important ('gold' level evidence)." But there is no mention of any on these benefits in the paragraph. Only an extended discussion of what the drug doesn't do and of its side effects.  And given that this drug was approved based on "real" endpoints, none of this has anything to do with "disadvantages of surrogate endpoints".


 * I'll stop here because my comments on the other paragraphs will differ only in the specifics. The problems throughout are WP:OR, WP:OS and WP:COAT.


 * There's lots of reasons to be unhappy with the pharmaceutical industry. But that editorial does not belong here, first because this article is about the much narrower issue of surrogate endpoints, and second because Wikipedia has a neutral point of view policy. Formerly 98 (talk) 06:31, 27 June 2014 (UTC)
 * I have to agree with these points, and clearly the first statement that systematic reviews should not include surrogate endpoint data needs be more balanced. Preferably we should look for a statement from a major international organization.-- CFCF  (talk · contribs · email) 17:30, 27 June 2014 (UTC)

Sourcing for all the myriad examples?
I see someone deleting a source because it's primary. No matter what MEDRS pushes, that makes me cringe -- and I'd argue that an article about surrogate endpoints is by definition about research, not treatment, so MEDRS should not apply to it. Meanwhile, I see a bazillion examples like "Any hospitalization (composite outcome) / cost, disease burden". What's the source for, say, this one? Better a primary source than no source, that's for sure. I can't even parse whether this is "hospitalization = surrogate, cost = real" or "cost = surrogate, disease burden = real". Nor could I believe it either way. Wnt (talk) 16:46, 27 June 2014 (UTC)
 * I have to disagree, this article discusses treatments and treatment options. Therefore we need to abide by WP:MEDRS when it comes to such statements. You are correct in that many of the statements here are purely research related, but those should also abide by MEDRS, which if paraphrased specifically states that clinically relevant statements need to follow certain rules. Therefor an article can comply with MEDRS, without all of the statements needing to be in accordance to the extra rules put on medical content. Each statement needs to be taken in context and even in a research related article do we need to stick to MEDRS when it applies. -- CFCF  (talk · contribs · email) 17:30, 27 June 2014 (UTC)
 * Well, to begin with, a surrogate endpoint is a concept, not a clinical observation in itself, and has broad applicability. In ecological studies you can have things like "An evaluation of the relationship between salmon farm biomass, organic inputs to sediments, physicochemical changes associated with those inputs and the infaunal response – with emphasis on total sediment sulfides, total volatile solids, and oxidation reduction potential as surrogate endpoints for biological monitoring".  After all, the numbers of various organisms you can catch are absurdly variable due to experimental error and natural cycles, but simple chemical tests can sometimes predict the effect on them. Wnt (talk) 18:56, 27 June 2014 (UTC)


 * http://www.ncbi.nlm.nih.gov/pubmed/24779115 may be a good start for any unsourced claims concerning cardiovascular disease/cardiac surgery. (2014) -- CFCF  (talk · contribs · email) 17:45, 27 June 2014 (UTC)

So somewhone found it necessary to order sourcing for the examples of surrogates; the examples are used for the thousands of ordinary people who read encyclopedias. Whithout those examples the text is unreadable for most people. I thought it was enough to put hundreds of crossreferences. So i have put some hunderd of references. Maybe someone find better ones. --Truebreath (talk) 13:09, 3 July 2014 (UTC)

Primary research sources on the use of surrogate, composite, and simple outcome endpoints
I've removed the following from the article:


 * "Three meta epidemiologic studies found trials reporting surrogate primary outcomes were more likely to report larger treatment effects or more positive outcomes than trials reporting final patient relevant primary outcomes.[non-primary source needed][14][15][16] Trials using surrogate endpoints were more likely to report positive results, while those using mortality endpoints were less likely to be positive. Trials using surrogate endpoints (and disease-specific mortality) as an endpoint were more likely to be exclusively commercially funded.[non-primary source needed][17]


 * ...Only about one third of authors of randomised clinical trials that used a surrogate as a primary outcome reported clearly that the primary outcome was a surrogate. Better reporting was needed.[non-primary source needed][19]"

These are primary research articles that pass over important statistical issues and potential confounders. To be specific:
 * The hurdle to showing statistical signficance is often higher in an outcomes trial. While the power of a trial using blood pressure as an endpoint is determined by the number of patients, the power of a trial examining cardiovascular outcomes is determined by the number of events, which is typically a low single digit percentage of the number of subjects in the trial. Thus part of the problem is the greater difficulty of demonstrating an effect based on low frequency binary events, and not simply the use of surrogate endpoints to cover up lack of efficacy as implied by the text.
 * Similar confounding factors apply to the observation that trials sponsored by for profits are more likely to show superiority of new treatments than those by non-profits. While the non-profit organizations are indifferent to showing superiority, no sensible for profit company will run a head to head trial against an older treatment unless required to do so by regulatory authorities or confident of success. Similarly, publication bias rears its ugly head here. This is an important problem, but one that is independent of the issue of surrogate endpoints, and thus off-topic.
 * Reference 18 points out that commercial sponsors are more likely to use disease-specific mortality than all-cause mortality. This is not, as implied by the article, a conspiracy to cover up bad results, but a requirement of the regulatory authorities. See for example: http://www.fda.gov/downloads/Drugs/GuidanceComplianceRegulatoryInformation/Guidances/UCM071627.pdf


 * "Sponsors should establish an independent cardiovascular endpoints committee to prospectively adjudicate, in a blinded fashion, cardiovascular events during all phase 2 and phase 3 trials. These events should include cardiovascular mortality, myocardial infarction, and stroke, and can include hospitalization for acute coronary syndrome, urgent revascularization procedures, and possibly other endpoints..."

Formerly 98 (talk) 22:12, 27 June 2014 (UTC)

Figure problems
"Coronary or cardiac revascularization (above before, below after) is a surrogate endpoint in lie of mortality in heart disease." This is terribly wrong. Cardiac revascularisation is an intervention; not a surrogate. The coronarography is a surrogate for cv events and mortality. Please make the correction. --Truebreath (talk) 14:46, 11 July 2014 (UTC)

Because formerly 98 didn't change the figure i decided to change the text.

--Truebreath (talk) 20:55, 18 July 2014 (UTC)

Good and bad surrogates? or surrogates in the context of treatment? a first response to the point of views of “Formerly 98”
There are more than 250 biomarkers that are causally related to cardiovascular disease and that can be used as surrogates for prevention of cardiovascular disease. Surrogates used for cardiovascular disease evolved from cholesterol, to LDL cholesterol, to HDL cholesterol, to high dense particles in LDL cholesterol, to crp and inflammatory parameters and to visualisation of plaque burden etc. In the context of bypass graft procedures maybe the arteriografy is a good surrogate for the outcome. (the procedure itself is not a surrogate as wrongly added in the figure). The NUMBER of bypass procedures can be a part of a compound outcome as major cardiovascular events. So please correct the figure. But few surrogates have been sufficiently evaluated to justify their use in developing drugs or making treatment decisions. (text IOM 2010)

But wat is the BEST surrogate for an individual treatment, or food, or medical intervention , or medical devise? What surrogate predicts best the effect on hard outcomes? The text of the 2010 US institute of medecine gives us examples:

The use of LDL as a surrogate endpoint for some cardiovascular outcomes, but not all cardiovascular outcomes for statin drug interventions is supported by data. The use of LDL cholesterol as surrogate for cardiovascular outcomes for other cardiovascular interventions and medications, foods, or supplements is not supported by data. HDL-C has not yet been qualified as a surrogate endpoint for cardiovascular risk because there is no evidence that HDL-raising interventions can improve outcomes. The use of LDL cholesterol as cv surrogate for oestrogens of oestro-progestogen associations and for ezetimibe is not supported by data. P164 P165

HDL cholesterol and beta-carotene in serum as surrogates for cv disease  are not supported  by data. Niacin and derivatives didn’t influence cv events in trials; neither did beta-carotene capsules; torcetrapib increased cv risk.

High sensitivity CRP as surrogate for cardiovascular outcomes is not supported by data.

Tumor response rate, tumor shrinkage, decreasing tumor size surrogates are not supported by data. There were inconsistent findings on tumor shrinkage and clinical benefit. The validity of the tests was a problem also.

Bone density measurements are a  surrogate outcome.

Radiographic vertebral fractures is a composite outcome of a surrogate outcome (asymptomatic vertebral fractures) and clinical vertebral fractures. The latter being the meaningful clinical outcome that matters. Nice 2008 “The definition of vertebral fractures is different in different trials. (outcome definition) (15 or 20% reduction in the height of one or more vertebrae). 20% was found to give more reliable results. Some studies used clinical (that is symptomatic) fractures as their endpoint whereas others used fractures that were identified radiographically. The majority of vertebral fractures (50–70%) do not come to clinical attention. Nice 2008 http://www.nice.org.uk/nicemedia/live/11748/42447/42447.pdf In the 3 cochrane analyses about the different biphosphonates they don't make the difference between both. The Canadian therapeutic letter systematic review makes the difference and don't count vertebral fractures as clinical meaningful endpoint. So bdm is a good surrogate for secondary prevention of hip and wrist fractures in postmenopausal women for biphosphonates. During three years. But til know not for clinical vertebral fractures. And not for primary prevention. The alendronate cochrane analysis found only "gold" evidence for primary prevention of vertebral fractures (composite outcome), not for hip and wrist fractures, and not for clinical vertebral fractures.

For the same reason deep venous thromboflebitis (found by duplex scan eg) is a composite outcome, of clinically dvt and not clinicaly detected dvt; clinically dvt is the hard endpoint that matters for patients. See text on grade. And the same can be said about pulmonary embolism.

So there are no good or bad surrogates in se. You must see a surrogate in the context wherein it is used. This is the third step in the surrogate evaluation process.

So I’ am not against use of surrogates for approval, I am not against innovation, I am not against the pharmaceutic industry, like you try to explain. I suggest you make the text on benefits of surrogates. But for me it is not necessary to make 500 lines about this. I am convinced they are useful. You find them overall in medical literature. And during 38 years i have treated surrogate outcomes. So i suggest you make the part on hepatitis C and B. (I see you removed a cochrane review on treatment of hepatitis c, for reasons of copy pasting. I thought wikipedia tried to collaborate more with the cochrane collaboration. I almost always use their summaries, and copy past them, because they are so easy to understand. Is this forbidden????Can we not copy paste from open sources as plos one????)

But recent studies have challenged the assumption that reliance on surrogates can accurately predict the effect of treatment on hard outcomes. There are the oral hypoglycaemic drugs that reduce HbA1c but increase the risk of cardiovascular events, antihypertensive drugs that do not reduce the risk of stroke,  and drugs that improve cholesterol profiles but do not reduce cardiovascular events. Explanations for such phenomena include unwanted effects of the drug or an incomplete understanding of the pathophysiology of the disease.

So when the FDA considers each blood pressure lowering medication as healthy, this is not evidence based. So when the FDA says it is unethical to do trials against placebo for a ca antagonists they are right; but you can do trials against hygroton or against an ace inhibitor. So please remove that. It is in the interest of the Pharmaceutical industry and the overall health of people that the right surrogate is used in the right context. The science about this has recently been developed after 2000, and is still going on.

This discussion is of high interest for patients and doctors. For patients it contributes to de-medicalisation. For doctors it stimulates evidence based thinking to take the right decisions for individual patients. Am I treating a patient or a surrogate?

“The FDA’s current regulatory authority is not sufficient. Inadequate fulfillment of postmarketing studies and incomplete understanding of how consumers interpret food and dietary supplement claims prevent robust protection of public health. “ IOM P 197

Can we agree on what precedes? Then we can go further--Truebreath (talk) 15:30, 11 July 2014 (UTC)

Point by point response to the point of views of “Formerly 98”. Second response.
Formerly 98 said: "Systematic reviews should not present surrogate outcome data as their primary analysis" is clearly not mainstream as 90% of published meta analyses do exactly that.” But a meta analysis is not the same as a “systematic review and meta analysis”. In the analyses of failed surrogate outcomes I tried to find systematic reviews. Those don't look only at the surrogate but also at the clinical endpoints that matter (efficacy AND efficiency outcomes) AND to the safety outcomes. See cochrane “Systematic review (synonym: systematic overview): A review of a clearly formulated question that uses systematic and explicit methods to identify, select and critically appraise relevant research, and to collect and analyse data from the studies that are included in the review. Statistical methods (meta-analysis) may or may not be used to analyse and summarise the results of the included studies. “ And “A crucial step in a cochrane systematic review is the specification of all patient-important outcomes relevant to the intervention strategies under comparison .” Cochrane handbook.

Formerly 98 said: “problems with rimonabant are unrelated to the subject of surrogate endpoints. It was pulled for CNS side effects, which are unrelated to the question of whether weight loss was an acceptable surrogate for cardiovascular disaese” The hardest endpoint is mortality. Rimonobant was not found to decrease cv mortality or cv events; there was found a higher risk of depression and suicide., two clinically meaningfull endpoints, related to morbidity and mortality. So ema found the benefits didn't outweigh the harms. So (small) weight loss is a bad surrogate endpoint for rimonobant. Surrogate endpoints are not absolute, you have to see them in the particular context. Considerable weight loss was a good surrogate for bariatric procedures for cv events as described in the text.

Formerly said: “Teriparatide is criticized even though it used a "real" endpoint for in its trials. It reduced vertebral fractures by nearly 2/3's (From 14.3% to 5.0%), and reduced combined non-vertebral fractures by half (5.5% vs. 2.6%).” As said before vertebral fractures are not a “real” endpoint but a surrogate composite endpoint. It is a composite of clinically radiological vertebral fractures and asymptomatic radiological fractures. Hip fracture was not reduced. You should always describe absolute risk decrease and NNT in your viewpoints, not relative. Cochrane doesn't have a combined systematic review on all biphosphonates ; wrist and hip fractures and clinically vertebral fractures are unequivocally hard endpoints. Clinically vertebral fractures are only occasionally reported in the 3 cochrane analysis. So the conclusion of the therapeutics letter systematic review “ in secondary prevention, the small benefits of bisphosphonates likely outweigh the harms during the first 3 years of therapy, but harms likely outweigh benefits for durations greater than 3 years “ seems right.

Formerly 98 said “The problems throughout are WP:OR, WP:OS and WP:COAT. “

Wikipedia: original research: the item of problems with surrogate outcome is not original research by me; it was found throughout searches about the item in systematic reviews found  by a search strategy essentialy based on  Dare, trip database and cochrane. I tried to confront them with Nice and other guidelines in an international context. I never give my personnel opinion about the evidence. I sometimes confront evidence. I found it interesting to confront two systematic revies of biphosphonates. The reader (patients and doctors) should make individual decisions about this evidence. I try always to refer to cochrane summaries if they exist, because they are easy readable by wikipedia users.

Wikipedia "oversight", (the power to suppress edits, is entrusted to a restricted number of users,). I don't care some edits were deleted because they were not from secundairy sources. But the item of surrogates is not so easy as let say ciprofloxacin. It involves the whole area of medecine. Finding relyable secundary sources about the item is not easy. One example of a deletion was the removal in the lead of an editorial about evidence based making of mini reviews in the field of haematology. I removed it and added excerpts from the cochrane handbook of systematic reviews, which explained exactly the same. Maybe this handbook too is not a secundary source?

Wikipedia Coatrack (articles run against the fundamental neutral point of view policy: in particular the requirement that articles be balanced). I agree and suggest you write the part benefits of surrogates. Maybe beginning with looking at the benefits of hepatitis C treatments, because i had problems to find them in hepatitis B and in hepatitis C. So i have put this part in the problems section. If you find better proof you change it to the benefits.

To make the text “more balanced” formerly 98 made a table, in which he describes 14 medications approved in 2008 on basis of surrogate enpoints, and then he looks for the actual situation on the meaningfull clinical outcome. Why formerly takes 2008? Why does he make a table full of medications that most readers and doctors don't know? The only possible breakthrough medication is the arthemeter combination, included in the WHO essential medecines and maybe etravirine which is an additional option for HIV-infected patients with multiple treatment failure. The rest are me too medications (ca antagonists, statins) and orphin drugs for which we can maybe wait until the end of my life to have clinical meaningfull endpoints, because they are so seldom. So this table ads nothing to the point of benefits of surrogate endpoints. So i ask you to delete it. There are already enough examples of surrogate endpoints from medications that people know.

Another example of deletion by formerly 98 was: "Three meta epidemiologic studies found trials reporting surrogate primary outcomes were more likely to report larger treatment effects or more positive outcomes than trials reporting final patient relevant primary outcomes.[non-primary source needed][14][15][16] Trials using surrogate endpoints were more likely to report positive results, while those using mortality endpoints were less likely to be positive. Trials using surrogate endpoints (and disease-specific mortality) as an endpoint were more likely to be exclusively commercially funded." He rejected those studies with his own theories as “statistical significance” and “confounders” without proof. Only to refute them. This is pure speculation. This is not evidence based. The facts are right or not; it is not important what formerly 98 thinks about those facts or why tose facts are not relevant. Those articles were found in BMJ and   by using searches with trip database. They confer with data found in composite outcomes “that the effect is smallest for the most important component and biggest for the less important components”., which was well mentioned in a systematic review.

Formerly 98 found “Reference 18 points out that commercial sponsors are more likely to use disease-specific mortality than all-cause mortality. This is not, as implied by the article, a conspiracy to cover up bad results, but a requirement of the regulatory authorities .” The text in wikipedia was very neutral:  “Trials using surrogate endpoints (and disease-specific mortality) as an endpoint were more likely to be exclusively commercially funded. “ Those were facts stated in the cohort of trials. No conspiracy theories at all.

A wikipedia editor (formerly 98?) found there were not enough references in the text of examples of surrogate outcomes; they didn't notice the 150 cross references within wikipedia. So i added some 150 references most from systematic cochrane reviews. But still the heading find not enough references. So please can you remove that heading. And please correct the figure.

About all those things i want to have a neutral view from another wikipedia editor. I have made thousands of contributions to wikipedia ; i never had that kind of personal attacs. And if there were critics always there was found a solution. Formerly 98 attacks my personal integrity, by making allusions of conspiracy, and of points of view against the pharmaceutical industry. This must stop. Living in hope.

Yours sincerely --Truebreath (talk) 19:19, 15 July 2014 (UTC)

Deleting content about hepatitis B by "formerly 98"
The whole content was deleted because "there was proof that hepatitis B prevented hepatocellular carcinoma in a 2009 meta analysis." You can find this meta analysis and a discussion about it here: http://ebm.bmj.com/content/14/3/86.extract

It was a metaanalysis of 3 rcts, and 16 cohort studies of which 10 were of good quality. So the outcome of such analysis can only be hypothesis forming. The commentary from ebm.bmj says: "pivotal trials have found no evidence on death, livercancer or liver transplantation. The metaanalysis suggests a potential benefit of treatment of hepatitis B on liver cancer."

Another meta analysis was given http://www.ncbi.nlm.nih.gov/pubmed/23713520 the meta analysis included 6 "studies" (not trials) about lamivudine, which found a lower risk of liver cancer. So again this is suggestive for an effect. But no proof. There was only one trial included on lamivudine: "In the only 2004 randomized, controlled clinical trial including 651 chronic hepatitis B patients (58% HBeAg positive) with biopsy-proven cirrhosis or advanced fibrosis, lamivudine was found to significantly reduce the risk of HCC compared to placebo (3.9% vs 7.4%, P = 0.047)[37]. When HCC cases diagnosed during the first year of treatment were excluded, the risk reduction was marginally non-significant (P = 0.052). So this is a close call....and the study was terminated early after a mean duration of 32.4 mo because of significant beneficial effects in the treatment group (7.8% developed cirrhosis complications vs 17.7% in the placebo group, P = 0.001). http://www.ncbi.nlm.nih.gov/pubmed/15470215 Pooled data from 49 studies found: However, HCC still develops at a rate of 1.3 per 100 patient years in CHB patients receiving an oral anti-viral agent. This finding highlights the need for continued HCC surveillance, particularly in CHB patients with inadequate viral suppression, older age and cirrhosis.

On the contrary a 2013systematic review in BMJ found : http://www.ncbi.nlm.nih.gov/pubmed/23945731

Randomised trials found no effect of antiviral therapy on HCC or mortality. Cohort studies found that antiviral therapy increased the risk of HCC (risk ratio 1.43; 95% CI 1.06 to 1.95), whereas case-control studies found a decreased risk of HCC in the intervention group (risk ratio 0.69; 95% CI 0.54 to 0.88). There was a clear difference between the results of RCTs and observational studies (test for subgroup differences, p<0.001). Antiviral therapy did not affect mortality in cohort studies, but reduced mortality in case-control studies (relative risk 0.71; 95% CI 0.54 to 0.93; test for subgroup differences, p=0.406). CONCLUSIONS:

The effect of antiviral therapy on clinical outcomes in HBV remains to be established. Although there was a positive effect in the sensitivity analyses, the strength of the evidence does not allow for extrapolation to clinical practice as research design plays an essential role in the overall assessment.

So i put back my text and ad the results of the systematic review.

--Truebreath (talk) 12:05, 12 July 2014 (UTC)

So, I think that in order to meet the requirements of NPOV, you need to present this as something other than "The evil pharmaceutical companies pushing drug through the approval process without adequate evidence". Its not just a shortcut, in many cases such as HBV there are ethical issues. Given that
 * HBV infection is clearly associated with cirrhosis and HCC
 * HBeAg seroconversion clearance is associated with reduced risk of cirrhosis and HCC
 * HBeAg levels are directly correlated with viral replication and viral DNA levels in the liver
 * Nucleosides are associated with increased rates of HBeAg clearance. There is no readily identifiable mechanism by which they would do this other than by reducing hepatic viral load.
 * Nucleosides have been shown to reduce liver inflammation

No regulatory agency in the world will approve a multi-year RCT at this point on the grounds that it would be unethical to deny treatment that is highly likely to be efficacious to patients simply to prove that they will develop liver cancer. And very few hepatologists would participate in such a trial. And even assuming that you could somehow get the regulators to approve and find clincial investigators willing to participate in such a trial, one could never successfully enroll the trial with patients.

In essence, you are demanding an RCT of parachutes for gravitational challenge, and furthermore an RCT that regulatory agencies will not permit due to the ethics of denying a treatment that is strongly believed to be beneficial, just for the purpose of demonstrating that those who receive placebo will develop cancer.

I don't expect you to respond, as you never respond to my comments unless you are trying to get me to remove an unbalanced flag or other item that you cannot act on unilaterally. But you should consider that there is more going on there than a run around the burden of proving efficacy. Formerly 98 (talk) 18:10, 12 July 2014 (UTC)

I've adjusted the language in the discussion of the systematic review. The authors are clearly sitting on the fence, and cannot make their own minds up about what the evidence means. But the Wikipedia passage focused exclusively on the negative statements. There was no mention of passages such as:

Formerly 98 (talk) 01:09, 13 July 2014 (UTC)
 * "Sensitivity analyses show a positive effect of treatment on mortality."
 * "The effect of modern nucleos(t)ides could not be assessed as newer trials do not include placebo-treated or untreated patients in the control groups"
 * "our findings are not sufficiently convincing and do not allow for changes in clinical practice"
 * "We also found a decrease in HCC incidence and overall mortality in sensitivity analyses of patients with cirrhosis. "
 * "We found a beneficial effect of interferon and/or NA on mortality in HBV when including RCTs and observational studies. The assessment of mortality is robust to bias."

So you think i never respond??? All this surrogate evidence you mention makes it likely, but don't proof there is an effect on clinical meaningfull endpoints. In 2000 there should have be done a trial adequately powered to find clinical endpoints on e.g. liver cancer for hepatitis B and interferon and first generation nucleosides. It should have included some 600 people and 4 years of follow up, then we should not have the problems of unethical trials against placebo now. So contrary to what you say i don't find it necessary to do a trial now against placebo.

Sincerely yours --Truebreath (talk) 20:01, 15 July 2014 (UTC)

Deletion of section about benzedrine, rainbow pills by "formerly 98 "." Safety issues have litle to do with surrogate outcomes...."
Formerly 98 found: "All these safety issues (of appetite suppressants nd others) are the worthy topic of a separate article, but they have little to do with the use of surrogate endpoints, esp. those drugs introduced prior to widespread use of RCTs."

Formerly 98 removed this section:

Amphetamines (marketed as Benzedrine) became popular for weight loss during the late 1930s. They worked primarily by suppressing appetite, and had other effects such as increased alertness. Use of amphetamines increased over the subsequent decades.Obetrol was a formulation of amphetamine mixed salts that included methamphetamine and was approved by the U.S. FDA in 1960. This culminated in the "rainbow pill" regime. This was a combination of multiple pills, all thought to help with weight loss, taken throughout the day. Typical regimens included stimulants, such as amphetamines, as well as thyroid hormone, diuretics, digitalis, laxatives, and often a barbiturate to suppress the side effects of the stimulants. In 1967/1968 a number of deaths attributed to diet pills triggered a Senate investigation and the gradual implementation of greater restrictions on the market. This culminating in 1979 with the FDA banning the use of amphetamines, then the most effective of the diet drugs, in diet pills. All amphetamine-based appetite suppressants have following adverse events: neuro-psychological disorders (headaches, insomnia, nervousness, depression, etc.) and cardiovascular disorders (arterial hypertension, palpitations, cardiac arrhythmia)

Aminorex, an amphetamine analogue, was an appetite suppressant that caused an epidemic of idiopathic pulmonary hypertension in Switzerland, Austria and Germany between 1967 an 1972. It was approved for obesity in those countries. It was very often fatal. 2% of users got the disease. The incidence of pulmonary hypertension increased 10 times. After withdrawal of the product in 1972 the incidence decreased again.

Safety issues have litle to do with surrogate outcomes....

Formerly 98 always repeat that adverse events of treatments have nothing to do with surrogate outcomes.

Taking amphetamin appetite suppressants for weightlos (surrogate), is entended to avoid comorbities as high HBA1c (diabetes) and high bloodpressure (hypertension) in the long term. Both are surrogates for cardiovascular disease and cv mortality. But by another pathway a few patients get pulmonary hypertension or a heartvalve disease; and half of them dy (in the short term) of cardiovascular disease. Other patients get hypertension and tachyarytmias, (surrogate outcomes we went to avoid), other patients dy suddenly by dependence, overdose and abuse...

So safety issues have everything to do with surrogate outcomes for aminorex, benzedrine and obetroll.

Safety issues are meaningfull clinical outcomes.(hard endpoints) that have everything to do with surrogate endpoints.

Those amfetamine derivatives were historical cases of invalid surrogate outcomes. The trials demonstrated an effect on the surrogate (weight), but harms outweighed the benefits.

So i ask you to undelete all the parts removed for this reason.

"Those appetite suppressants were introduced prior to widespread use of RCTs."

Formerly 98 says that widespread use of rct's should be a criterion for inclusion of the cases of problems with surrogate outcomes. Based on what??? The trials on amphetamines (aminorex, benzedrin, obetrol) found people did lose weight (surrogate). see here: http://www.ncbi.nlm.nih.gov/pubmed/4910309 and here http://www.ncbi.nlm.nih.gov/pubmed/4920753 About benzedrin and obetrol http://onlinelibrary.wiley.com/doi/10.1002/j.1550-8528.1994.tb00061.x/pdf

Surrogate outcomes can be used in clinical trials but also in prospective and retrospective studies, and in meta analyses about them. also in geografic studies, registry studies, in laboratory animals and in vitro models. They can be used in claims for food, novel foods, herbs, vitamins and dietary supplements. Even you brought meta analyses based on mixed results of trials and prospective studies.

You removed all the problems with surrogates from before 1997 because they were historical. "So widespread use of trials begun around 1997...".Could be. But who says that? Where is the evidence?

Wikipedia learns us: "One of the most famous clinical trials was James Lind's demonstration in 1747 that citrus fruits cure scurvy. He compared the effects of various acidic substances, ranging from vinegar to cider, on groups of afflicted sailors, and found the group who were given oranges and lemons had largely recovered from scurvy after six days. After 1750 the discipline began to take its modern shape.

Frederick Akbar Mahomed had his professional life from 1872 to 1884. He initiated the Collective Investigation Record for the British Medical Association; this organization collected data from physicians practicing outside the hospital setting and was the precursor of modern collaborative clinical trials."

Til around 2000 the rainbowpills were prescribed in Europe, by some doctors. They mixed amphetamines, thyroid hormone, diuretics, laxatives, and often a barbiturate. This is not so historical at all.

And if you have doubts about the prevalence of amphetamine use and the amphetamine epidemic from 1940 to 1960 which is almost the same as now in the US i offer you to read: https://www.evidence.nhs.uk/document?ci=http%3A%2F%2Fwww.ajph.org%2Fcgi%2Fcontent%2Fabstract%2F98%2F6%2F974%3Ffromsource%3Dnelm&q=amphetamine&ReturnUrl=%2Fsearch%3Fq%3Damphetamine%26pa%3D2

Sincerely yours

--Truebreath (talk) 13:57, 15 July 2014 (UTC)


 * , I would say that if you want to be intellectually honest, list out the trials that were performed and the role that surrogate biomarkers had on 1) getting these drugs approved, and 2) in their marketing.


 * Because I would say that if you can't point to a published or regulatory agency submitted clinical trial that used a surrogate endpoint that impacted one of these two things, you're blaming surrogate endpoints for a situation in which the real problem was failure to test the compound for efficacy before it was marketed. You're not writing about surrogate endpoints at all in this case. There were no endpoints, surrogate or otherwise, because there were no trials.


 * On the other hand, if your goal is to use this article as a WP:COATRACK for an extended screed about everything you think is wrong with modern medicine, the pharmaceutical industry, and the regulatory agencies that approve drugs, please feel free to do so. You've written 165 KB of rambling, opinionated editorial that no one is going to read anyway, and its not really worth my while to spend any more time trying to introduce some balance.


 * I'll focus my efforts on articles that are likely to be read. Formerly 98 (talk) 22:55, 15 July 2014 (UTC)

So i thought you didn't expect that i should respond you "as i never respond to your comments unless i am trying to get you to remove an unbalanced flag or other item that i cannot act on unilaterally." Now i did respond to much i think.

It is not my purpose to explain the role that surrogate biomarkers have on 1) getting drugs approved, and 2) in their marketing. But it is interesting to know how much of drugs approved on surrogate markers, are really breakthrough drugs. Moreover the title of the article is NOT surrogate outcome in clinical trials; but surrogate outcome.

In the section problems with surrogate enpoints it is my purpose to look for imbalances between surrogate biomarkers and meaningfull clinical outcomes.

Aciclovir was used during twenty years for prevention of postviral neuralgies by shingles. Now after all those years there is no proof of effect and there was no effect at preventing neuralgia at 4 or 6 month. (recent cochrane review 2014). All surrogate evidence maked it likely it should help, but it didn't. This is important for people, doctors and society.

Antibiotics for otitis media: during 20 years there was an expectant observationa approach in the Netherlands. Cochrane says: "It is difficult to balance the small benefits against the small harms of antibiotics in children with AOM. However, for most children with mild disease, an expectant observational approach seems justified."Most guidelines plaid for antibiotics.

I can give you a myriade of that kind of problems.

In the section benefits i will add no more content. I am completely convinced surrogate outcomes are important for medecine development, but at this moment few are validated for the context they are used for. I suggest you will further elaborate this, because there are so many examples: antibiotics, anti malarials, urinary desinfectants, heparin, anticoagulants, asthma and copd medications, hormonal products as insulin and thyroid hormone etcetera etcetera maybe you  can fill 350 lines with that, to make the article balanced.

Sincerely yours --Truebreath (talk) 13:03, 16 July 2014 (UTC)

Source on use of surrogate endpoints in soft-tissue sarcoma clinical trials
[http://www.cancernetwork.com/sarcoma/analysis-validates-use-pfs-surrogate-endpoint-sarcoma-trials Analysis Validates Use of PFS As Surrogate Endpoint in Sarcoma Trials. March 2016] Seems useful and relevant but given the edit warring above I hesitate to edit the article. Source says "In an analysis of surrogacy, the researchers found a highly significant correlation between OS and PFS (R = 0.61), as well as between RR and OS (R = 0.51)." Not clear what R would be an acceptable minimum. Are there guidelines on what correlation is required (eg by FDA) in different circumstances ? - Rod57 (talk) 06:17, 10 March 2016 (UTC)

Does correlation make a surrogate
Article says "A correlate does not make a surrogate. It is a common misconception that if an outcome is a correlate (that is, correlated with the true clinical outcome) it can be used as a valid surrogate end point (that is, a replacement for the true clinical outcome). However, proper justification for such replacement requires that the effect of the intervention on the surrogate end point predicts the effect on the clinical outcome—a much stronger condition than correlation." but this seems nonsense and needs to be better explained.[http://www.cancernetwork.com/sarcoma/analysis-validates-use-pfs-surrogate-endpoint-sarcoma-trials Analysis Validates Use of PFS As Surrogate Endpoint in Sarcoma Trials. March 2016] (quoted in the section above) seems to explicitly evaluate surrogates based on their correlation with outcomes. Help! - Rod57 (talk) 06:40, 10 March 2016 (UTC)

"Relationship is not linear" - is confusing
Article says " While elevated cholesterol levels increase the likelihood for heart disease, the relationship is not linear - many people with normal cholesterol develop heart disease, and many with high cholesterol do not" seems internally inconsistent. By "not linear" perhaps "poorly correlated" is meant. - Can someone clarify ? - Rod57 (talk) 06:45, 10 March 2016 (UTC)

New discussion of FDA use of surrogate endpoints
[http://www.medpagetoday.com/PublicHealthPolicy/FDAGeneral/58156 FDA Slammed for Rush to Approval of Cancer Drugs - Analysis questions use of surrogate endpoints. May 2016]. Does it say anything new ? - Rod57 (talk) 16:58, 4 June 2016 (UTC)