Original Articles
A Study in Contrasts: Eligibility Criteria in a Twenty-Year Sample of NSABP and POG Clinical Trials

https://doi.org/10.1016/S0895-4356(97)00240-0Get rights and content

Abstract

We studied changes in eligibility criteria—the largest impediment to patient accrual—in two samples of clinical trials. Trials from the NSABP (National Surgical Adjuvant Breast and Bowel Program) and POG (Pediatric Oncology Group) were analyzed. After eliminating duplications, the criteria in each protocol were enumerated and classified according to a novel schema. NSABP trials contained significantly more criteria than POG trials, and added precision criteria (making study populations homogeneous) at a faster rate than POG studies. The difference between NSABP studies (explanatory trials) and POG studies (pragmatic trials) suggest that large numbers of eligibility criteria are not necessary for quality studies. We recommend that: (1) the inclusion/exclusion criteria distinction be abandoned; (2) eligibility criteria be explicitly justified; (3) the need for each criterion be assessed when new trials are planned; (4) criteria in phase III trials restricting patient accrual be minimized; and (5) further research be done to assess the impact of criteria on generalizability.

Introduction

Clinical trials are the keystone in the development and evaluation of new treatments in oncology. Despite the pivotal role of clinical trials, only a small proportion of cancer patients are actually treated in trials. Friedman and colleagues report that only 1.6% of U.S. cancer patients are enrolled in NCI-funded phase II and III clinical trials [1]. Low accrual rates have been identified repeatedly as a critical problem affecting U.S. cancer trials 2, 3.

Why are so few patients enrolled in clinical trials? Many cancer patients never have the opportunity to enroll in a research study either because they are treated in a hospital that doesn't participate in multicenter clinical trials or because no trials are available for their type and stage of disease. The NCI's Community Clinical Oncology Program (CCOP) aims toward—and has been successful in—increasing the involvement of community hospitals in cancer clinical trials 4, 5, 6, 7. But even within institutions that actively participate in clinical trials, only a minority of cancer patients for whom a study is available are treated in studies [8].

Much of what has been written on barriers to clinical trial enrollment in oncology has focussed on physician 9, 10, 11, 12, 13, 14and patient 15, 16, 17, 18factors influencing accrual to research studies. In fact, however, criteria for clinical trial eligibility are the largest barrier to trial accrual. McCusker and colleagues describe the enrollment experience of a cohort of 454 patients in a single medical oncology clinic: of the 342 patients for whom a trial was available for their type and stage of disease, 43% were ineligible, 4% were not enrolled due to physician refusal, 2% refused to give consent, 1% were not enrolled for “other” reasons, and only 25% were actually enrolled in a research study [19]. Kotwall and colleagues describe the enrollment experience of a cohort of 592 women with stage I or II breast cancer (a protocol was available for all of the women): 46% were ineligible, 5% were not enrolled due to physician refusal, 25% refused to provide consent, 6% were not enrolled for “other” reasons, and only 18% were actually enrolled in a research study [20]. Other studies of accrual of oncology patients to trials have reported similar results 21, 22.

The exclusion of patients from cancer research is an important issue for a number of reasons. Patients who are barred from study participation may be deprived of benefits possibly associated with trial treatment [23]. A number of studies have observed that cancer patients treated in clinical trials have better outcomes than those who are treated outside of trials 24, 25, 26. It is uncertain whether observed differences were due to selection or treatment factors. Weijer and colleagues examined treatment differences between women with early stage breast cancer treated within and outside clinical research studies at a single institution and found evidence of two specific differences between the groups [27]. When age and stage of disease were controlled for in the analysis, women in trials received higher doses of chemotherapy and more frequent blood tests than other patients.

Another problem with exclusions is that trials that prevent substantial proportions of patients from participating may yield results that are not widely applicable (i.e., not generalizable) to the broader patient population of interest. Optimally, clinical trials should mirror the patient population in clinical practice; such trials will maximize accrual rates and foster the widespread applicability of results essential to maximize a trial's impact on medical practice [28]. Advocates of more restrictive trials assert that narrow trials are more efficient because patient heterogeneity (i.e., variance) is reduced [29]. In oncology, however, we know too little about prognostic factors to be able to define a truly homogenous population of patients [30]. Furthermore, both Buyse and George have independently argued that including patients of differing prognoses can allow a trial to accrue patients more quickly, and thus answer the question of interest more efficiently 31, 32.

How many eligibility criteria do typical phase III cancer trials have? Are all the criteria necessary? In an attempt to answer these questions, Begg and Engstrom studied a synchronous sample of trials investigating chemotherapy in the treatment of stage II, node positive breast cancer [33]. They found that each of the studies contained a large number of criteria defining the eligible patient population (the average number of criteria per study was 23). They also found that, among these otherwise comparable trials,substantial variation existed in eligibility criteria. Begg and Engstrom conclude that [33]:

[t]he rationale for these exclusions is not clearly understood in many cases and may to some extent be due to the unchallenged perpetration of conventions that are more applicable to laboratory experiments in which the experimenter is able to exercise much more control of the conduct of the study.

The magnitude and nature of the exclusion criteria were such as to “cast doubt on the generalizability of the results from the clinical trials program.”

Crucial questions remain regarding eligibility criteria in clinical trials that aim to guide clinical practice: How have the numbers of criteria in comparable trials changed over time? What kinds of criteria have been added or dropped? How do these changes affect the generalizability of study results? To begin to address these problems, we undertook a study of two diachronous samples: clinical trials of the National Surgical Adjuvant Breast and Bowel Program (NSABP) and the Pediatric Oncology Group (POG).

Section snippets

Methods

Clinical trials from the NSABP and POG were selected because they each constitute a single cooperative group that has focussed a series of clinical trials on a single disease: in the case of the NSABP, early stage breast cancer; for POG, acute lymphocytic leukemia. Furthermore, both cooperative groups are the source of distinguished and influential research results and recommendations. Finally, both groups of studies cover a similar time span: 1972 to 1992. Trials were chosen up to the

Results

Of the 22 NSABP breast cancer trials between 1972 and 1992, 11 examined the role of chemotherapy in the treatment of stage II, node positive breast cancer (NSABP protocols B-05, B-07, B-08, B-09, B-10, B-11, B-12, B-15, B-16, B-22, and B-25). The number of criteria for all studies is shown in Fig. 1. The number of eligibility criteria in the larger set of 22 studies increased from 21 to 44 over the 20-year period (coefficient of correlation, τ = 0.54, p = 0.0006). In the subset of 11 trials

Discussion

Until relatively recently, the scope of ethical enquiry into clinical research was limited to a few discrete areas. Following the final report of the National Commission for the Protection of Human Subjects of Biomedical and Behavioral Research, the Belmont Report, research ethicists have focussed their attention primarily on informed consent and the assessment of risk–benefit ratio in clinical research [35]. Freedman and Shapiro recently reviewed the cumulative index (1979–1990) of the main

Recommendations

Based on our findings, we have the following recommendations for design of clinical trials.

First, the distinction between inclusion and exclusion criteria ought to be abandoned. Clearly, nothing is gained by distinguishing between inclusion and exclusion criteria: the two categories of criteria are interchangeable by merely adding a “not” to any given criterion. As we have seen, when the distinction is made, duplications occur. Such duplications make IRBs and perhaps even

Acknowledgements

Funding for this research was provided by a joint grant from the Medical Research Council of Canada (M.R.C.) and the Social Sciences and Humanities Research Council of Canada. Dr. Weijer's research was supported by a fellowship from M.R.C.

The authors are grateful to Dr. Richard Margolese and Ms. Riva Kaplan of the S.M.B.D. Jewish General Hospital for providing copies of the NSABP protocols, and to Ms. Susan Devine of the Montreal Children's Hospital for providing copies of POG protocols. Dr.

References (48)

  • CB Begg et al.

    Participation of community hospitals in clinical trialsAnalysis of five years of experience in the Eastern Cooperative Oncology Group

    N Engl J Med

    (1982)
  • KM Taylor et al.

    Physician's reasons for not entering eligible patients in a randomized clinical trial of surgery for breast cancer

    N Engl J Med

    (1984)
  • EG Mansour

    Barriers to clinical trials. Part IIIKnowledge and attitudes of health care providers

    Cancer

    (1994)
  • AB Benson et al.

    Oncologists' reluctance to accrue patients onto clinical trialsAn Illinois Cancer Center study

    J Clin Oncol

    (1991)
  • KM Taylor et al.

    Fundamental dilemmas of the randomized clinical trial processResults of a survey of the 1,737 Eastern Cooperative Oncology Group investigators

    J Clin Oncol

    (1994)
  • GR Langley et al.

    Why are (or are not) patients given the option to enter clinical trials?

    Control Clin Trials

    (1987)
  • GR Morrow et al.

    Behavioral aspects of clinical trialsAn integrated framework from behavior theory

    Cancer

    (1994)
  • WS Schain

    Barriers to clinical trials. Part IIKnowledge and attitudes of potential participants

    Cancer

    (1994)
  • HA Llewellyn-Thomas et al.

    Cancer patients' decision making and trial-entry preferences

    Med Decision Making

    (1995)
  • BR Cassileth et al.

    Attitudes towards clinical trials among patients and the public

    JAMA

    (1982)
  • J McCusker et al.

    Cancer patient accessions into clinical trialsA pilot investigation into some patient and physician determinants of entry

    Am J Clin Oncol

    (1982)
  • CA Kotwall et al.

    Reasons for nonentry in randomized clinical trials for breast cancerA single institution study

    J Surg Oncol

    (1992)
  • CB Begg et al.

    Cooperative groups and community hospitalsMeasurement of impact in the community hospitals

    Cancer

    (1983)
  • JY Lee et al.

    Accrual of radiotherapy patients to clinical trials

    Cancer

    (1983)
  • Cited by (71)

    • Under-served groups remain underserved as eligibility criteria routinely exclude them from breast cancer trials

      2022, Journal of Clinical Epidemiology
      Citation Excerpt :

      Patient recruitment continues to be a problem in clinical trials globally [1], and engaging underserved groups in trial recruitment poses additional challenges, sometimes due to the small size of these target populations but also other recognized barriers [3,5]. In breast cancer research, the disparity in clinical trial participation of underserved groups and how this might affect trial results and the conclusions that can be drawn from them is not clear [6–8]. Overall, breast cancer mortality has decreased over the last 10 years due to progress in prevention, early detection, and treatments [9].

    • Time and Effort Required for Tissue Acquisition and Submission in Lung Cancer Clinical Trials

      2017, Clinical Lung Cancer
      Citation Excerpt :

      While predictive biomarkers provide the foundation of personalized or precision medicine, they have added to the complexities and costs of clinical research.2 Independent of these considerations, clinical trial protocols have become more lengthy, and eligibility criteria more stringent.3-6 Even before tissue requirements were routinely incorporated into study protocols, less than 5% of adults with cancer in the United States participated in clinical trials.7-9

    • Thoracic Oncology Clinical Trial Eligibility Criteria and Requirements Continue to Increase in Number and Complexity

      2017, Journal of Thoracic Oncology
      Citation Excerpt :

      Although our results are not directly generalizable to clinical trials in other cancer types, they are likely representative. Increases in eligibility criteria have been reported in other settings, including breast, gastrointestinal, and gynecologic cancers.19,22 Small numbers of radiation therapy and surgery protocols may obscure significant changes in eligibility criteria, although the observed values suggests that this is unlikely.

    View all citing articles on Scopus
    View full text