Bias In Clinical Research Research Articles

Dissemination and publication of research findings: an updated review of related biases

To identify and appraise empirical studies on publication and related biases published since 1998; to assess methods to deal with publication and related biases; and to examine, in a random sample of published systematic reviews, measures taken to prevent, reduce and detect dissemination bias. The main literature search, in August 2008, covered the Cochrane Methodology Register Database, MEDLINE, EMBASE, AMED and CINAHL. In May 2009, PubMed, PsycINFO and OpenSIGLE were also searched. Reference lists of retrieved studies were also examined. In Part I, studies were classified as evidence or method studies and data were extracted according to types of dissemination bias or methods for dealing with it. Evidence from empirical studies was summarised narratively. In Part II, 300 systematic reviews were randomly selected from MEDLINE and the methods used to deal with publication and related biases were assessed. Studies with significant or positive results were more likely to be published than those with non-significant or negative results, thereby confirming findings from a previous HTA report. There was convincing evidence that outcome reporting bias exists and has an impact on the pooled summary in systematic reviews. Studies with significant results tended to be published earlier than studies with non-significant results, and empirical evidence suggests that published studies tended to report a greater treatment effect than those from the grey literature. Exclusion of non-English-language studies appeared to result in a high risk of bias in some areas of research such as complementary and alternative medicine. In a few cases, publication and related biases had a potentially detrimental impact on patients or resource use. Publication bias can be prevented before a literature review (e.g. by prospective registration of trials), or detected during a literature review (e.g. by locating unpublished studies, funnel plot and related tests, sensitivity analysis modelling), or its impact can be minimised after a literature review (e.g. by confirmatory large-scale trials, updating the systematic review). The interpretation of funnel plot and related statistical tests, often used to assess publication bias, was often too simplistic and likely misleading. More sophisticated modelling methods have not been widely used. Compared with systematic reviews published in 1996, recent reviews of health-care interventions were more likely to locate and include non-English-language studies and grey literature or unpublished studies, and to test for publication bias. Dissemination of research findings is likely to be a biased process, although the actual impact of such bias depends on specific circumstances. The prospective registration of clinical trials and the endorsement of reporting guidelines may reduce research dissemination bias in clinical research. In systematic reviews, measures can be taken to minimise the impact of dissemination bias by systematically searching for and including relevant studies that are difficult to access. Statistical methods can be useful for sensitivity analyses. Further research is needed to develop methods for qualitatively assessing the risk of publication bias in systematic reviews, and to evaluate the effect of prospective registration of studies, open access policy and improved publication guidelines.

Enduring Contributions of Henry K. Beecher to Medicine, Science, and Society (Vol. 45, No. 4 and Vol. 46, No. 1 of International Anesthesiology Clinics).

University of Alabama School of Medicine, Birmingham, Alabama. jmckeown@uab.eduEnduring Contributions of Henry K. Beecher to Medicine, Science, and Society (Vol. 45, No. 4 and Vol. 46, No. 1 of International Anesthesiology Clinics). Edited by Edward Lowenstein, M.D., F.R.C.A., and Bucknam McPeek, M.D. Baltimore, Lippincott Williams & Wilkins, 2007. Pages: 249. Price: $131.00 per part.In my short career as an academic anesthesiologist, no unpleasantness has quite matched the deep revulsion elicited when I learned it was time for my yearly recertification with the institutional review board. The first time I muddled through the web-based program took hours and I thought, “surely I did that not even 6 months ago!” I painfully logged on once again for the moderately shortened recertification version and “point-clicked” my way as quickly as possible through the material. However, this time, just like the first, I paused momentarily when I came across the name “Henry Knowles Beecher.” What caught my attention and surprised me for the second time was that this whistle-blower, this champion for informed consent, was an anesthesiologist. This bit of trivia intrigued me, and I skimmed the summarized account of his 1966 landmark paper Ethics and Clinical Research on the computer. In my eagerness to put this computer tutorial out of my misery, it is very likely that I did not comprehend the importance of his work, nor the impact it had on the practice of medicine and medical research. Thus, the tenacious Dr. Beecher and his paper did not cross my mind again until very recently, when I was invited to review Enduring Contributions of Henry K. Beecher to Medicine, Science, and Society , edited by Lowenstein and McPeek. Suffice it to say that my appreciation for Beecher's life and work has changed drastically.This book is a well-organized compilation of Dr. Beecher's papers and archival and intimate biographical materials. His original papers, including 1966's Ethics and Clinical Research are accompanied by excellent commentaries that elaborate on the controversial work of Dr. Beecher and its relevance to the practice of modern medicine. This organization allows the reader to experience Dr. Beecher in his own words, to assess his ideas, and then engage in a discussion of these ideas through the commentaries.Besides his crusade for informed consent and reform in human medical research, Dr. Beecher's diverse volume of work also includes studies of pain and its perception, anesthesia-related mortality, the placebo effect, bias in clinical research, and even hallucinogens for use by the military. The most enduring of his “Enduring Contributions” is, in fact, the paper Ethics and Clinical Research mentioned previously. Having been cited over 500 times in medical literature alone, this exposé of over 20 examples of unethical research in poorly informed or uninformed human subjects sent shockwaves through the medical research community. Considering the specific studies Dr. Beecher cited, it is especially chilling that he asserted that “such breaches of ethical conduct in experimentation are by no means rare but are almost… universal.”1This paper is credited with spurring the National Institutes of Health to empower local institutional review boards to critique all human research proposals. The most fascinating paradox is that Dr. Beecher himself was fully established in the medical research community whose corruption he condemned. His firm belief was that the end never justified the means if a single patient could possibly be harmed. To his fellow researchers, however, he was ultimately faithful; he never exposed by name any of the individuals whose actions he used to kindle a firestorm.This selection will be enjoyed by enthusiasts of anesthesia and medical history alike, because Dr. Henry K. Beecher transcended anesthesiology. He was one of a rare number of clinicians who first excelled in their specialty and then rose through research to affect the practice of medicine globally.University of Alabama School of Medicine, Birmingham, Alabama. jmckeown@uab.edu

Bias in clinical research

The quality of a clinical study depends on internal and external factors. Studies have internal validity when, random error apart, reported differences between exposed and unexposed individuals can be attributed only to the exposure under investigation. Internal validity may be affected by bias, that is, by any systematic error that occurs in the design or in the conduction of a clinical research. Here we focus on two major categories of bias: selection bias and information bias. We describe three types of selection biases (incidence-prevalence bias, loss-to-follow-up bias, and publication bias) and a series of information biases (i.e. misclassification bias--recall bias, interviewer bias, observer bias, and regression dilution bias--and lead-time bias).

Evaluating conflicts of interest in research presented in CME venues

There is much in the literature regarding the potential for commercial bias in clinical research and in continuing medical education (CME), but no studies were found regarding the potential for bias in reporting original research in CME venues. This pilot study investigated the presence of perceived bias in oral and print content of research findings presented in certified CME activities. Research presentations at two national primary care CME activities, where authors had self-reported potential conflicts of interest, were peer reviewed and monitored for perceived commercial bias. Blinded and unblinded peer reviewers' and monitors' analyses of bias were compared to assess whether knowledge of potential conflicts of interest affected perceptions of bias. Knowledge of potential conflicts of interest appeared to increase awareness of potential commercial bias with regard to use of a single product in care and assurance that there was reasonable evidence to support the practice recommendation. A perception of the presenter's strong opinion regarding care did not appear to be influenced by knowledge of a potential conflict of interest. While limited, by study design, this research detected subjectivity and variability in perceiving commercial bias within research findings presented in CME venues. Further study of these questions is required to guide the resolution of conflicts of interest in research and CME.

Reporting of Outcomes in Orthopaedic Randomized Trials

Randomization, concealment of treatment allocation, and blinding are all known to limit bias in clinical research. Nonsurgical studies that fail to meet these standards have been reported to inflate the differences between treatment and control groups. While surgical trials can rarely blind surgeons or patients, they can often blind outcome assessors. The aim of this systematic review was threefold: (1) to examine the reporting of outcome measures in orthopaedic trials, (2) to determine the feasibility of blinding in published orthopaedic trials, and (3) to examine the association between the magnitude of treatment differences and the blinding of outcome assessors. We identified and reviewed thirty-two randomized, controlled trials published in The Journal of Bone and Joint Surgery (American Volume) in 2003 and 2004 for the appropriate use of outcome measures. These trials represented 3.4% of all 938 studies published during that time-period. All thirty-two trials were reviewed by two authors for (1) the outcome measures used and (2) the blinding of outcomes assessors. We calculated the magnitude of the treatment effect of the use of blinded compared with unblinded outcome assessors. Ten (31%) of the thirty-two randomized controlled trials used a modified outcome instrument. Of the ten trials, four failed to describe how the outcome instrument was modified. Nine of the ten articles did not describe how the modified instrument was validated and retested. Sixteen of the thirty-two randomized controlled trials did not report blinding of outcome assessors when blinding would have been possible. Among the studies with continuous outcome measure, unblinded outcomes assessment was associated with significantly larger treatment effects than blinded outcomes assessment (standardized mean difference, 0.76 compared with 0.25; p = 0.01). Similarly, in the studies with dichotomous outcomes, unblinded outcomes assessments were associated with significantly greater treatment effects than blinded outcomes assessments (odds ratio, 0.13 compared with 0.42; p < 0.001). The ratio of odds ratios (unblinded to blinded outcomes assessment) was 0.31, suggesting that unblinded outcomes assessment was associated with a potential for exaggeration of the benefit of the effectiveness of a treatment in our cohort of studies. In future orthopaedic randomized controlled trials, emphasis should be placed on detailed reporting of outcome measures to facilitate generalization and the outcome assessors should be blinded, when possible, to limit bias.

Reporting of Outcomes in Orthopaedic Randomized Trials

Background: Randomization, concealment of treatment allocation, and blinding are all known to limit bias in clinical research. Nonsurgical studies that fail to meet these standards have been reported to inflate the differences between treatment and control groups. While surgical trials can rarely blind surgeons or patients, they can often blind outcome assessors. The aim of this systematic review was threefold: (1) to examine the reporting of outcome measures in orthopaedic trials, (2) to determine the feasibility of blinding in published orthopaedic trials, and (3) to examine the association between the magnitude of treatment differences and the blinding of outcome assessors. Methods: We identified and reviewed thirty-two randomized, controlled trials published in The Journal of Bone and Joint Surgery (American Volume) in 2003 and 2004 for the appropriate use of outcome measures. These trials represented 3.4% of all 938 studies published during that time-period. All thirty-two trials were reviewed by two authors for (1) the outcome measures used and (2) the blinding of outcomes assessors. We calculated the magnitude of the treatment effect of the use of blinded compared with unblinded outcome assessors. Results: Ten (31%) of the thirty-two randomized controlled trials used a modified outcome instrument. Of the ten trials, four failed to describe how the outcome instrument was modified. Nine of the ten articles did not describe how the modified instrument was validated and retested. Sixteen of the thirty-two randomized controlled trials did not report blinding of outcome assessors when blinding would have been possible. Among the studies with continuous outcome measure, unblinded outcomes assessment was associated with significantly larger treatment effects than blinded outcomes assessment (standardized mean difference, 0.76 compared with 0.25; p = 0.01). Similarly, in the studies with dichotomous outcomes, unblinded outcomes assessments were associated with significantly greater treatment effects than blinded outcomes assessments (odds ratio, 0.13 compared with 0.42; p < 0.001). The ratio of odds ratios (unblinded to blinded outcomes assessment) was 0.31, suggesting that unblinded outcomes assessment was associated with a potential for exaggeration of the benefit of the effectiveness of a treatment in our cohort of studies. Conclusions: In future orthopaedic randomized controlled trials, emphasis should be placed on detailed reporting of outcome measures to facilitate generalization and the outcome assessors should be blinded, when possible, to limit bias.

Response to Letter Regarding Article by Khan et al, “Prognostic Value of Troponin T and I Among Asymptomatic Patients With End-Stage Renal Disease: A Meta-Analysis”

HomeCirculationVol. 113, No. 20Response to Letter Regarding Article by Khan et al, “Prognostic Value of Troponin T and I Among Asymptomatic Patients With End-Stage Renal Disease: A Meta-Analysis” Free AccessLetterPDF/EPUBAboutView PDFView EPUBSections ToolsAdd to favoritesDownload citationsTrack citationsPermissions ShareShare onFacebookTwitterLinked InMendeleyReddit Jump toFree AccessLetterPDF/EPUBResponse to Letter Regarding Article by Khan et al, “Prognostic Value of Troponin T and I Among Asymptomatic Patients With End-Stage Renal Disease: A Meta-Analysis” Nadia A. Khan, MD, MSc, FRCPC Brenda R. Hemmelgarn, MD, PhD, FRCPC Marcello Tonelli, MD, MSc, FRCPC Christopher R. Thompson, MD, CM, FRCPC Adeera Levin, MD, FRCPC Nadia A. KhanNadia A. Khan Division of Internal Medicine, University of British Columbia, British Columbia, Canada Search for more papers by this author Brenda R. HemmelgarnBrenda R. Hemmelgarn Division of Nephrology, University of Calgary, Calgary, Alberta, Canada Search for more papers by this author Marcello TonelliMarcello Tonelli Division of Nephrology, University of Alberta, Edmonton, Alberta, Canada Search for more papers by this author Christopher R. ThompsonChristopher R. Thompson Division of Cardiology, University of British Columbia, British Columbia, Canada Search for more papers by this author Adeera LevinAdeera Levin Division of Nephrology, University of British Columbia, British Columbia, Canada Search for more papers by this author Originally published23 May 2006https://doi.org/10.1161/CIRCULATIONAHA.106.619643Circulation. 2006;113:e776We thank Drs Apple, Jaffe, and Herzog for their interest in our article and for drawing attention to the National Kidney Foundation (NKF) Guidelines on Evaluation of Cardiovascular disease.1 As stated in our paper, because the troponin studies used differing assays and variable cut points and reported conflicting results, direct comparison of the results was problematic, and a rigorous meta-analysis was required to clarify this association.2 The review from the NKF did not include all relevant studies (eg, results from 8 trials were not included in the NKF review), reported on results using differing cut points or outcomes, and did not evaluate for study heterogeneity and publication bias, given the significant possibility that negative observational studies may not be published3 (as was indeed found in our analysis). In addition, we respectfully submit that recommendations from professional organizations and regulatory bodies, although valuable, can be usefully informed by additional research—the existence of the NKF and FDA documents does not contradict our statement that the available literature has produced conflicting results.Dr Apple and colleagues correctly point out that the cut point of 0.1 ng/mL is not a part of the ESC/ACC guidelines. We chose this cut point to improve between study comparisons and allow for mathematical synthesis, as the vast majority of troponin T studies evaluated prognosis at this level. Thus, from the evidence base, a cut point of 0.1 ng/mL was associated with reduced long-term survival among asymptomatic hemodialysis patients. The degree of imprecision at this cut point is also close to that recommended by the ESC/ACC. Most troponin assays do not currently meet this standard at the 99th percentile.4–5 However, as more studies accrue and precision of troponin assays at lower thresholds improves, a lower cut point may prove to have greater predictive power in this population.We agree that troponin T elevations are more often seen in patients with chronic kidney disease compared with troponin I. It remains difficult to assess the ability of troponin I to predict mortality, given the variability of assays and assay cut points studied.Thank you for identifying the typographical error in Table 2: The table column should read 99th percentile and not the 90th percentile.DisclosuresNone.1 K/DOQI Workgroup. K/DOQI clinical practice guidelines for cardiovascular disease in dialysis patients. Am J Kidney Dis. 2005; 45 (4 Suppl 3): S1–S153.Google Scholar2 Khan NA, Hemmelgarn BR, Tonelli M, Thompson CR, Levin A. Prognostic value of troponin T and I among asymptomatic patients with end-stage renal disease: a meta-analysis. Circulation. 2005; 112: 3088–3096.LinkGoogle Scholar3 Easterbrook PJ, Berlin J, Gopalan R, Matthews DR. Publication bias in clinical research. Lancet. 1991; 337: 867–872.CrossrefMedlineGoogle Scholar4 Panteghini M, Pagani F, Yeo KT, Apple FS, Christenson RH, Dati F, Mair J, Ravkilde J, Wu AH; Committee on Standardization of Markers of Cardiac Damage of the IFCC. Evaluation of imprecision for cardiac troponin assays at low-range concentrations. Clin Chem. 2004; 50: 327–332.CrossrefMedlineGoogle Scholar5 Apple FS, Murakami MM. Serum 99th percentile reference cutoffs for seven cardiac troponin assays. Clin Chem. 2004; 50: 1477–1479.CrossrefMedlineGoogle Scholar Previous Back to top Next FiguresReferencesRelatedDetailsCited By Vircoulon M and Combe C (2012) Conséquences cardiaques de l’insuffisance rénale chronique, EMC - Néphrologie, 10.1016/S1762-0945(12)40514-9, 9:1, (1-14), Online publication date: 1-Jan-2012. May 23, 2006Vol 113, Issue 20 Advertisement Article InformationMetrics https://doi.org/10.1161/CIRCULATIONAHA.106.619643 Originally publishedMay 23, 2006 PDF download Advertisement SubjectsChronic Ischemic Heart DiseaseDiagnostic Testing

What are the implications of optimism bias in clinical research?

What are the implications of optimism bias in clinical research?

Commentary: unbiased divination, unbiased evidence, and the patulin clinical trial.

Histories of the randomized clinical trial (RCT) are constantly unearthing dramatic precursors for concurrent controls, random assignment to comparison group, blind assessment, placebo controls, and statistical inference. 1‐4 All these forerunners seem to indicate a constantly budding but, somehow, stumbling and aborted development toward the modern methodology. While most of the methods (at least in primitive forms) seem known in numerous early sources, it is not clear why medicine as a profession refused to commit itself to safeguards against bias until after World War II. The 60th anniversary of the under-appreciated and critical placebo controlled clinical trial of patulin for the common cold is a timely moment to offer some historical reflections on the origin and final acceptance of the RCT. This commentary will hypothesize that the modern methods to reduce bias in clinical research—especially randomization, concurrent controls, and blinding—in at least rudimentary forms—were well known to educated physicians and, in fact, most literate European and Middle Easterners for thousands of years. Most of these assurances against bias or chicanery have their origin in religious divination rituals. From the 16th century onwards until the 1950s, many physicians sought to adopt these methods for revealing secular and unbiased medical truths. Despite valiant efforts by medical reformers, the bulk of the medical community ignored or rejected the introduction of unbiased comparison of clinical outcomes. Until 1950, most of the medical profession was not attracted to any form of clinical experimentation that challenged the personal judgement of physicians or seemingly threatened to treat patients as less than unique individuals. ‘Science’ was thought best confined to the laboratory where variability could be contained and physicians were best suited to delicately apply this knowledge within the context of ‘the art of healing’. The 1944 patulin trial and its more famous younger sibling, the 1948 streptomycin trial for tuberculosis, were both deeply embedded in the British biometry statistical tradition. These trials were not so important for their methodological innovations, as much as they came to represent potent exemplary models for an innovative team of statisticians and epidemiologists who were disseminating already known methodologies and convincing colleagues of the need and the possibility for a ‘clinical science’. This re-construction of medicine’s self-identity was as great a revolution as any previous transformation in the history of medicine. The patulin trial was a key component of this revolution as well as an important illustrative example of the very process itself.

Selection Bias in Clinical Research When Subjects Are Excluded Because of Failure to Estimate Left Ventricular Mass by Echocardiography

Quantitative echocardiographic imaging often is used in clinical research, but the method is not successful in all patients. The goals of this study were as follows: (1) to determine the frequency with which left ventricular mass (LVM) could be estimated by echocardiography, (2) to characterize persons in whom LVM could not be estimated from echocardiography, and (3) to determine whether failure to estimate LVM was independently related to any common cardiovascular risk factors, thereby suggesting a potential for selection bias. This research was part of a community-based case-control study of risk factors in hypertension. Five hundred hypertensive subjects and 506 normotensive age- and sex-matched control subjects were studied prospectively. The mean age was 61 years (range: 39 to 76 years), the mean weight was 74 kg (range: 37 to 127 kg), 46% of the subjects were men, and 56% had a history of smoking. Two-dimensionally guided M-mode echocardiography was attempted in all subjects for determination of LVM (Penn convention). LVM was estimated when the endocardial borders could be accurately identified from a true radial cavity diameter. LVM was estimated in 803 subjects (79.8%). Subjects in whom estimates of LVM could not be obtained were older and had the following characteristics: higher systolic blood pressure; larger chest, waist, and abdominal circumferences; higher waist/hip ratio; and larger body mass index (all P < .02). In multivariate analysis the only independent predictors of failure to estimate LVM by echocardiography were advanced age ( P = .0001), larger chest size ( P = .0001), and history of smoking ( P = .004). The inability to estimate LVM in 20% of subjects has important implications to our understanding the strength of association between potential risk factors and LVM, an independent measure of cardiovascular risk. LVM is more likely to be missing in older subjects who have a large chest and a history of smoking. The effects of such a selection bias should be considered in any analysis, and a characterization of the excluded subjects should be developed. (J Am Soc Echocardiogr 1998;11:1050-5.)

Nonselection and nonresponse bias in clinical research

Nonselection and nonresponse bias in clinical research

Medical Editors Trial Amnesty

Reports of properly conducted randomised controlled trials are the foundation of safe and effective health care. However, a substantial proportion of all controlled trials never contribute to this knowledge base because they are not submitted for publication. 1 Dickersin K Min YI NIH clinical trials and publication bias. On-line J Curr Clin Trials[serial on-line]. 1993; (Apr 28: Doc No 50) Google Scholar , 2 Easterbrook PJ Berlin JA Gopalan R Matthews DR Publication bias in clinical research. Lancet. 1991; 337: 867-872 Summary PubMed Scopus (2262) Google Scholar This fact has several important implications for patient care. First, underreporting of trials reduces the power of systematic reviews to detect moderate but clinically important treatment effects. As a result, patients may be denied effective forms of health care. Second, since trials that show more promising treatment effects are more likely to be submitted for publication, research syntheses based on published studies can give misleading conclusions about treatment effectiveness, thus exposing patients to useless or even harmful therapies. 3 Egger M Davey Smith G Misleading meta-analysis. BMJ. 1995; 310: 752-754 Crossref PubMed Scopus (317) Google Scholar Finally, patients may be asked to participate in new research studies designed to address questions that have, in truth, already been answered. 4 Savulescu J Chalmers I Blunt J Are research ethics committees behaving unethically? Some suggestions for improving performance and accountability. BMJ. 1996; 313: 1390-1393 Crossref PubMed Scopus (151) Google Scholar

Women's health--missing from U.S. medicine

Acknowledgments Introduction Part One: Critiques of the Androcentric Focus in Clinical Research and Practice 1. Androcentric Bias in Clinical Research 2. Underdiagnosis and Inadequate Treatment: AIDS and Women 3. Androcentric Bias in Psychiatric Diagnosis 4. Separate, but Equal? Women and Obstetrics/Gynecology Part Two: Ignoring Diversity among Women in Clinical Research and Practice 5. Elderly Women 6. Women of Color 7. Lesbians Part Three: Including Women in Medical Education 8. Feminist Methodologies and Clinical Research 9. More Than OAdd Women and StirO: Integrating Disease and Health of Women into the Medical Curriculum 10. A Chilly Climate for Women in Medicine 11. WomenOs Ways of Knowing: Changing Teaching Techniques and Evaluation Conclusion Women and Health: A Selected Bibliography complies by Faye A. Chadwell Index

Publication bias in clinical research

In a retrospective survey, 487 research projects approved by the Central Oxford Research Ethics Committee between 1984 and 1987, were studied for evidence of publication bias. As of May, 1990, 285 of the studies had been analysed by the investigators, and 52% of these had been published. Studies with statistically significant results were more likely to be published than those finding no difference between the study groups (adjusted odds ratio [OR] 2·32; 95% confidence interval [Cl] 1·25-4·28). Studies with significant results were also more likely to lead to a greater number of publications and presentations and to be published in journals with a high citation impact factor. An increased likelihood of publication was also associated with a high rating by the investigator of the importance of the study results, and with increasing sample size. The tendency towards publication bias was greater with observational and laboratory-based experimental studies (OR=3·79; 95% CI=1·47-9·76) than with randomised clinical trials (OR=0·84; 95% CI=0·34-2·09). We have confirmed the presence of publication bias in a cohort of clinical research studies. These findings suggest that conclusions based only on a review of published data should be interpreted cautiously, especially for observational studies. Improved strategies are needed to identify the results of unpublished as well as published studies.

Informed consent as a source of bias in clinical research

Clinical observation suggested that schizophrenic patients more often refused to participate in experiments. In two studies psychiatric patient volunteers and nonvolunteers were compared on discharge diagnoses (psychiatric and medical), medication given, demographic measures, and hospitalization characteristics. In each study, patients with paranoid schizophrenia were more likely to refuse participation. Also, in each study, volunteers received more benzodiazepines and more nonpsychiatric medication. The latter difference could not be explained on the bases of rated severity of or probability of drug use in diagnosed medical illnesses, or number of medical illnesses. Characteristics of schizophrenic patients such as suspiciousness, negativism, and ambivalence were suggested as possible sources of their refusal to participate. Nonvolunteers were suggested to be less agreeable and compliant in general, and perhaps less likable, possibly explaining their receiving fewer benodiazepines and nonpsychiatric medications. Attempts to characterize and correct the volunteer bias should be made, including offering tangible reward.