Bias in case-control studies of screening effectiveness

Effect of population breast screening on breast cancer mortality to 2005 in England and Wales: A nested case-control study within a cohort of one million women

Objectives

To evaluate the effectiveness of the NHS breast screening programme (NHSBSP) on breast cancer mortality in England and Wales and to compare findings with a cohort analysis of the same underlying population.

Methods

A nested case-control study within a cohort of 959,738 women in England and Wales aged 49–64 who were eligible for routine NHSBSP screening during 1991–2005. Cases who died from breast cancer in 1991–2005 were matched to controls without breast cancer at the case diagnosis date and alive when the case died. Risk of breast cancer mortality associated with intention to screen (ITS) (7047 cases/28,188 controls) and screening attendance (4707 cases/9413 controls) was examined. Bias was minimised in accordance with currently advocated best practice. Odds ratios (ORs) were calculated using conditional logistic regression. Results were compared with findings from an incidence-based breast cancer mortality cohort analysis.

Results

ITS was associated with a 21% breast cancer mortality reduction (OR = 0.79, 95% confidence interval [CI]: 0.71–0.88, P < 0.001). Attendance ≤5 years before diagnosis was associated with a 47% reduction in breast cancer mortality after self-selection correction (OR = 0.53, 95% CI: 0.46–0.62, P < 0.001). Breast cancer mortality reduction associated with ITS was 21% in both the case-control and cohort analyses, but the impact of attendance was marginally greater in the case-control analysis (36% vs. 32%).

Conclusions

Case-control studies designed and analysed according to current best practice guidelines offer an effective means of evaluating population breast screening.

Outcomes of surveillance mammography after treatment of primary breast cancer: a population-based case series

GOAL

To ascertain outcomes of surveillance mammography (SM) following treatment of early stage unilateral primary breast cancer (PBC) in a population based case series.

METHODS

Random samples from all 12,279 women having breast surgery within 4 months after diagnosis of PBC, between July 1991 and December 1993 in Ontario, were drawn from a database created by deterministic linkage of PBC files from the Ontario Cancer Registry (OCR) with episodes of breast surgery extracted from the hospital Discharge Abstract Database (DAD), and mammography from the Ontario physician billings database (OHIP). Among women having >or=1 episode(s) of breast surgery subsequent (SBS) to the date of diagnosis up to December 2000, a sample of 1,200/5,064 (23.7%) was drawn, and among women with no SBS, a sample of 400/7,215 (5.5%). Among these two samples, operative, pathology, and mammography reports were abstracted from original charts. Treatments were abstracted and categorized. Women with complete data for Stages 1 and 2 unilateral PBC were included. From the subsequent surgery sample, 609/1,200 (50.8%) were excluded because of simultaneous or sequential bilateral breast cancers or mastectomies within 6 months, missing stage information, Stage 3 or 4 PBC, or missing primary charts. From the no subsequent surgery sample, 90/400 (22.5%) were excluded by the same criteria. Episodes of bilateral 2-view X-ray mammography, beginning >or=6 months after the diagnosis of unilateral PBC, and if multiple, at least 11 months apart, and not prompted by a clinical concern or symptom, were classified as SM. We confirmed episodes of cancer recurrence within the ipsilateral conserved breast (CRICB) and metachronous contralateral primary breast cancer (CPBC) >or=6 months after the diagnosis of the unilateral PBC from original operative and pathology reports. We used Cox models to describe the association of exposure to >or=1 episode(s) of SM with the risk of death from breast cancer among the study population, and separately among women experiencing CRICB or CPBC.

RESULTS

Eligible women comprising 591/1,200 and 310/400 produced a combined case series of 901/1,600 (56.3%). Women with >or=1 episode(s) of SM numbered 721/901 (80.0%). We confirmed 84 CRICB events among 584 women initially treated by lumpectomy (14.4%), and 49 CPBC events among all 901 women in the study population (5.4%). Among women having >or=1 episode(s), the 25th percentile of observed follow up was 1,631 days, the 50th, 4,287 days, and the 75th 5,011 days. Among women without any SM, the 25th percentile of observed follow-up was 440 days, the 50th, 891 days, and the 75th, 1,849 days. Hazard ratio (HR) for death due to breast cancer associated with >or=1 episode of SM was 0.28 (95% CI 0.22-0.37), adjusted for age, stage, type of surgery, adjuvant chemotherapy, and tamoxifen. Among 84/584 women with CRICB, unadjusted HR = 0.36 (95%CI 0.13, 1.00) and among 49/901 women with CPBC, unadjusted HR = 0.86 (0.20-3.77).

CONCLUSION

SM was associated with a significant reduction in the hazard for breast cancer death. Among women who experienced CRICB, the reduction was of borderline significance, and the reduction was not significant among women who experienced CPBC.

Population-based case-control study of PSA and DRE screening on prostate cancer mortality

OBJECTIVES

The efficacy of screening for prostate cancer (PCa) with digital rectal examination (DRE) and prostate-specific antigen (PSA) measurement has not been proved in randomized clinical trials. In an earlier case-control study, we found that DRE might reduce PCa mortality. The present case-control study assessed the association between PSA and DRE testing and PCa mortality.

METHODS

The case subjects included 74 Olmsted County residents who had died from 1992 to 2005 with PCa as the underlying cause of death. From 1 to 3 community control subjects (alive at time of case subject's death) were matched to each case subject. The medical records were reviewed to identify DREs and PSA determinations performed 0 to 5 years before the date the case was diagnosed (index date). Tests performed in the absence of symptoms were considered to be "screening tests." Conditional logistic regression analysis was used to estimate the odds ratios and 95% confidence intervals for the association of screening (defined in multiple ways) and PCa mortality.

RESULTS

From 1 to 5 years before the index date, control subjects were more likely than case subjects to have undergone a previous screening PSA test or DRE (81.3% versus 60.8%, P = 0.0005). The unadjusted odds ratio associated with a previous screening PSA test or DRE was 0.34 (95% confidence interval 0.18 to 0.63), and the odds ratio adjusted for potential confounders was 0.35 (95% confidence interval 0.17 to 0.71). PSA testing was frequently done in conjunction with DRE, making evaluation of the individual effects difficult.

CONCLUSIONS

The results of this case-control study suggest a potential benefit of screening by PSA testing and/or DRE on PCa mortality.

Protection of mammography screening against death from breast cancer in women aged 40-64 years

OBJECTIVE

This study assessed the efficacy of community-based screening mammography in protecting against breast cancer death, asking whether age differences in efficacy persisted in the 1990s.

METHODS

In a case-control study with follow-up, odds ratios (OR) were used to estimate the relative mortality rates from invasive breast cancer among women with at least one screening mammogram in the two years prior to a baseline reference date compared to non-screened women, adjusting for potential confounding. The multicenter population-based study included 553 black and white women diagnosed during 1994-1998 who died in the following five years, and 4016 controls without breast cancer.

RESULTS

Efficacy for reducing the rate of breast cancer death within five years after diagnosis was greater at ages 50-64 years (OR = 0.47, 95% confidence interval (CI) 0.35-0.63) than at ages 40-49 (OR = 0.89, 95% CI 0.65-1.23), and greater among postmenopausal (OR = 0.45, 95% CI 0.33-0.62) than premenopausal women (OR = 0.74, 95% CI 0.53-1.04). Estimates of efficacy were conservative, as shown by sensitivity analyses addressing whether cancer was discovered by a screening mammogram, age at which screening was received, the length of the screening observation window, and years of follow-up after diagnosis.

CONCLUSIONS

Despite the persistence of age differences in efficacy of mammography screening, with greater observed benefit for women aged 50-64 years, these findings support current screening recommendations for women 40-64 years old.

Lower risk of death from gastric cancer among participants of gastric cancer screening in Japan: a population-based cohort study

OBJECTIVE

This study aimed to investigate the association between gastric cancer screening and mortality from gastric cancer.

METHODS

In 1990, 47,605 Japanese subjects were recruited and completed a questionnaire about participation in gastric cancer screening and life-style. We followed up their vital status through December 2001. In this cohort, 41,394 subjects without a history of cancer were allocated to the screened group or the unscreened group according to their response to the question about gastric cancer screening. We estimated the relative risk (RR) of death from gastric cancer, death from any cause except gastric cancer, and incidence of gastric cancer with adjustment for potential confounding variables.

RESULTS

The risk of death from gastric cancer among the screened group was significantly lower than that among the unscreened group. The multivariate RR of death from gastric cancer for screened individuals compared with those not screened was 0.54 (95% confidence interval [CI]: 0.38, 0.77). The RR of death from any cause except gastric cancer was 0.83 (95% CI: 0.77, 0.90), and the RR of incidence of gastric cancer was 0.94 (95% CI: 0.79, 1.13).

CONCLUSION

Our data suggest that gastric cancer screening or factors associated with it may be associated with lower mortality from gastric cancer.

Benefit of screening mammography in reducing the rate of late-stage breast cancer diagnoses (United States)

OBJECTIVE

We studied the benefit of modern mammography screening in community settings, evaluating age-related differences in rates of late-stage breast cancer detection.

METHODS

Our multicenter population-based case-control study included 931 black and white women with incident breast cancer (American Joint Commission on Cancer Stage IIB or higher) diagnosed 1994-1998 and 4,016 randomly sampled controls never diagnosed with breast cancer. Adjusted odds ratios (ORs) estimated the relative rate of late-stage diagnosis in screened and non-screened women.

RESULTS

Women aged 50-64 at diagnosis with at least one screening mammogram in the previous 2 years were significantly less likely to have late-stage diagnosis (OR = 0.41, 95% CI 0.33-0.52). Results for women aged 40-49 were consistent with a screening benefit, although the confidence interval marginally overlapped the null (OR = 0.81, 95% CI 0.64-1.02). Mammography screening was associated with lower rates of late-stage breast cancer among both premenopausal (OR = 0.64, 95% CI 0.50-0.81) and postmenopausal (OR = 0.44, 95% CI 0.35-0.56) women.

CONCLUSIONS

With modern mammography in the community, rates of late-stage breast cancer diagnoses are lower in screened compared to non-screened women ages 40 and older, but age-related differences persist.

Measuring screening intensity in case-control studies of the efficacy of mammography

Of great interest in studies of screening for breast cancer is the relative efficacy of different screening frequencies (intensities). Prior work has suggested that estimates of the association between screening intensity and outcome in case-control studies would not produce valid results and that only binary indicators (no screens vs. one or more) of exposure can be used. Using case-control studies drawn from simulated cohorts of 30,000-40,000 women, the authors found that biases demonstrated in prior studies can be explained by 1) misclassification of true exposure groups by observed screening history, and 2) differential exposure misclassification of cases and controls. Binary as well as ordered categorical and interval measures can be biased unless they account for misclassification. By combining measurements of screening history from multiple periods of observation of varying lengths and using repeated-measures logistic regression models, the effect of screening intensity can be estimated in the presence of misclassification. Assessing the effect of screening intensity in case-control studies of mammography is possible if principles and methods for misclassification and measurement error guide the analysis.

Review in depth and meta-analysis of controlled trials on colorectal cancer screening by faecal occult blood test

BACKGROUND

Several randomized studies have shown that colorectal cancer (CRC) screening by faecal occult blood test (FOBT) reduces CRC mortality. These trials have different designs, especially concerning FOBT frequency and duration, as well as the length of follow-up after stopping FOBT campaigns.

AIMS

To review the effectiveness of screening for CRC with FOBT, to consider the reduction in mortality during or after screening or to identify factors associated with a significant mortality reduction.

METHODS

A systematic review of trials of FOBT screening with a meta-analysis of four controlled trials selected for their biennial and population-based design. The main outcome measurements were mortality relative risk (RR) and 95% confidence interval (CI) of biennial FOBT during short (10 years, i.e. five or six rounds) or long-term (six or more rounds) screening periods, as well as after stopping screening and follow-up during 5-7 years. The meta-analysis used the Mantel-Haenszel method with fixed effects when the heterogeneity test was not significant, and used 'intent to screen' results.

RESULTS

Although the quality of the four trials was high, only three were randomized, and one used rehydrated biennial FOBT associated with a high colonoscopy rate (28%). A meta-analysis of mortality results showed that subjects allocated to screening had a reduction of CRC mortality during a 10-year period (RR 0.86; CI 0.79-0.94) although CRC mortality was not decreased during the 5-7 years after the 10-year (six rounds) screening period, nor in the last phase (8-16 years after the onset of screening) of a long-term (16 years or nine rounds) biennial screening. Whatever the design of the period of ongoing FOBT, CRC incidence neither decreased nor increased, although it was reduced for 5-7 years after the 10-year screening period. Neither the design nor the clinical or demographic parameters of these trials were independently associated with CRC mortality reduction.

CONCLUSION

Biennial FOBT decreased CRC mortality by 14% when performed over 10 years, without evidence-based benefit on CRC mortality when performed over a longer period. No independent predictors of CRC mortality reduction have been identified in order to allow a CRC screening programme in any subgroups of subjects at risk.

Prospective study of screening for stomach cancer in Japan

Although screening for stomach cancer is a widespread community service in Japan, the benefits of the screening program remain unclear. Our study investigated prospectively the relation between participation in stomach-cancer screening during the past 12 months and subsequent deaths. Data was derived from the Japan Collaborative Cohort Study, in which 480 stomach-cancer deaths were identified during an 8-year follow-up period. Cox proportional hazard regression was used to estimate the relative risk of death from stomach cancer and that from other causes while adjusting for potential confounding factors. In men, screening participation was associated significantly with a reduced risk of death from stomach cancer (relative risk [RR] = 0.54; 95% confidence interval [CI] = 0.41-0.70). The extent of the risk reduction was greater than potential health selection (for deaths other than stomach, RR = 0.71). The adjustment for potential confounding variables, however, attenuated the difference in risk of death (stomach cancer, RR = 0.65; other causes, RR = 0.71). In women, the magnitude of the association between screening participation and death from stomach cancer (RR = 0.74; 95% CI = 0.52-1.07) was equal to that for deaths from non-stomach cancers (RR = 0.74). Subgroup analysis showed that women with a parental history of stomach cancer had a reduced risk of death from stomach cancer associated with screening (RR = 0.32; 95% CI = 0.12-0.87). The present results underline the potential for selection bias in observational studies, and thus it remains an open question whether screening for stomach cancer is effective.

Mammographic screening: case-control studies

BACKGROUND

The case-control design can be used to evaluate the benefit of cancer screening programmes.

MATERIALS AND METHODS

This paper outlines the main methodological features of the case-control design in this context, and indicates some potential biases. It also reviews the existing case-control literature on mammographic screening.

RESULTS

Case-control studies consistently indicate a reduction of approximately 50% in breast cancer mortality associated with mammography. This result indicates greater benefit than shown in randomised trials; however, one should recognise that trials indicate effectiveness whereas case-control studies indicate efficacy. The two types of evidence are broadly compatible when one allows for screening non-compliance and contamination in the randomised trials.

CONCLUSIONS

The case-control evidence supports and is consistent with the findings of randomised trials of mammography. Effectiveness estimates from trials indicate the benefit of screening to the population as a whole, and are pertinent to the public policy debate as to the value of offering screening. In contrast, case-control studies indicate benefit to actual screening participants. As such, case-control estimates of efficacy are appropriate for individual decision-making by women about their use of mammography when it is potentially available to them.

The efficacy of lung cancer screening conducted in 1990s: four case-control studies in Japan

The efficacy of lung cancer screening is still controversial. In order to evaluate efficacy of mass screening for lung cancer in 1990s, the Japanese Ministry of Health and Welfare planned to conduct four independent case-control studies in four different regions; Miyagi, Gunma, Niigata, and Okayama Prefecture. The study design of all the four studies was a matched case-control study in which the decedents from lung cancer were defined as cases. In Gunma Prefecture, a screening examination is annual miniature chest X-ray only, whereas sputum cytology is added for high-risk screenees in others. Matching conditions were gender, year of birth, smoking histories (except Okayama), and municipality. Smoking adjusted odds ratio (OR) of dying from lung cancer for those screened within 12 months before case diagnosis compared with those not screened ranged 0.40-0.68. Three of four studies revealed statistically significant reduction of the risk for lung cancer death. OR of pooled analysis, where all sets were combined and analyzed, was 0.56 (95% confidence interval: 0.48-0.65). Recent mass screening program for lung cancer in Japan could reduce the risk for lung cancer death. However, the possibility exists that some confounding factors affected the results. In order to elucidate whether the results can be applied to Western countries, further studies will be required.

A case-control follow-up study for disease-specific mortality

Case-control studies often rely on subjects to report their own screening or exposure information: this information is often obtained from cases after the event of interest has occurred. This is problematic for mortality outcomes, because dead subjects cannot report the desired information. To avoid this problem, Weiss and Lazovich (1996, American Journal of Epidemiology 143, 319-322) proposed obtaining exposure or screening information from potential cases, i.e., subjects diagnosed with disease, at the time of disease diagnosis, and also from a referent series. The design is best viewed as a new scheme for sampling from a cohort. I review estimation of the effects of time-varying screening or exposure in cohort studies, using a new factorization. I then show how this factorization, together with ignorability assumptions, allows valid estimation from these new designs. Even when the sampling fraction of nondiseased subjects is unknown, causal risk ratios are estimable if diagnosis is rare in the cohort. I illustrate and compare conventional and new methods with data from the Health Insurance Plan study.

Fecal occult blood screening for colorectal cancer: open issues

Data from seven case-control and--mainly--three randomized clinical trials consistently indicate that biennial fecal occult blood screening (FOBT) can reduce colorectal cancer (CRC) mortality by approximately 20% after 10-18 years. The reduction may be greater in compliant subjects. In the long-term, incidence also appears to be reduced. There are suggestions that the effect of annual screening may be greater, although data are inadequate to quantify the potential advantages of annual versus biennial screening. The issue of the effectiveness of FOBT in the general population and, more important, of comparative cost-effectiveness with other possible screening tests for CRC, however, remain open to discussion.

An evaluation of screening for lung cancer in Niigata Prefecture, Japan: a population-based case-control study

Although an annual screening programme for lung cancer has been carried out widely in Japan since 1987, there is insufficient evidence to confirm its efficacy in terms of reducing mortality. In order to evaluate the efficacy of the lung cancer screening which has been widely carried out in Japan since 1987, a case-control study was conducted in Niigata Prefecture, Japan. In the study area, chest X-ray examinations for all participants and sputum cytology for high-risk participants were offered annually. Case subjects, who had died from lung cancer (174), and control subjects matched by sex, year of birth, residence and smoking status (801), who had been alive at the time of diagnosis of the corresponding case, were selected from the National Health Insurance holders. Screening histories of the subjects were compared between cases and matched controls for the identical calendar period before the time of diagnosis of the cases. The odds ratio of death from lung cancer for those screened within 12 months vs those not screened was 0.401 (95% CI: 0.272-0.591) with adjustment by smoking index. Our results suggest that annual lung cancer screening might reduce mortality from lung cancer by approximately 60%.

A case-control study for evaluating the efficacy of mass screening program for lung cancer in Miyagi Prefecture, Japan

BACKGROUND

In Miyagi Prefecture, Japan, a mass screening program for lung cancer has been conducted since 1982 (miniature chest X-ray for all screenees and sputum cytology for those with a smoking index > or = 600) [smoking index 600 = 30 pack years, the average number of cigarettes smoked per day multiplied by the number of years of regular smoking]. Over 1500 lung carcinomas, including 250 roentgenographically occult lung tumors, were detected and treated up to 1999. In the current study, a nested case-control study was conducted in the population that was screened in 1989 to evaluate the efficacy of the screening program for lung cancer.

METHODS

To reduce self-selection bias, the source population was defined as screenees with negative results in 1989 (284,226 individuals). In the population, 474 individuals died of lung carcinoma during 1992-1994. After exclusion, 328 patients who died of primary lung carcinoma at between ages 40 years and 79 years were defined as the cases. Six controls were supposed to be selected in the source population for each case and matched by gender, year of birth, municipality, and smoking habits. Controls who had died or moved before the matched case was diagnosed were excluded. Finally, 328 cases and 1886 controls were selected. Screening histories were compared, and odds ratios were calculated using conditional logistic regression analysis.

RESULTS

Within the 12 months before diagnosis, 241 of 328 cases (73.5%) had attended the screening compared with 1557 of 1886 controls (82.6%). The smoking-adjusted odds ratio was 0.54 (95% confidence interval, 0.41-0.73).

CONCLUSIONS

The mass screening program for lung cancer in Miyagi Prefecture was capable of reducing by 46% the risk of death from carcinoma of the lung.

A novel form of ascertainment bias in case-control studies of cancer screening

In case-control studies of cancer screening, some have generally admonished investigators against case definitions based on diagnosis dates because of lead-time bias. However, perhaps partly due to vagueness, the admonitions have been frequently ignored. A recurrence-time model simulates case ascertainment when diagnosis must occur within a specific calendar period. The model depends on screening test sensitivity and rate, age-specific preclinical incidence rates, and preclinical duration time and survival time distributions. For one study of sigmoidoscopic screening for colorectal cancer, when the true odds ratio is 1, its estimate is 0.50 to 0.75 under plausible assumptions. This bias can affect any observational study wherein case definition depends on diagnosis times (e.g., health-plan enrollment data). To avoid bias in observational investigations of cancer screening wherein the case definition depends on the diagnosis date, one must ensure that both screening and preclinical incidence do not occur before the case definition period.

Reducing colorectal cancer mortality by repeated faecal occult blood test: a nested case-control study

Randomised trials have shown the efficacy of faecal occult blood testing (FOBT) in reducing colorectal cancer mortality, but observational studies are needed to monitor such efficacy in population programmes. We conducted a nested case-control study on a cohort of 21,879 subjects who participated in a colorectal screening programme from 1978 to 1995, undergoing at least one FOBT test. 95 fatal cases of colorectal cancer were eligible for the study. For each fatal case, 5 non-fatal matched controls were randomly selected from the cohort. FOBT screening history was less common among cases than controls. The odds ratio of colorectal cancer mortality among 'attenders' (defined as those who underwent a second FOBT within 2 years of study entry) with respect to 'non-attenders' was 0.64 (95% confidence interval 0.36-1.15). We also computed odds ratios defining exposure as one or more tests in the detectable preclinical period, hypothesising various lengths for the latter, which, however, yielded an efficacy estimate biased towards the null. A strong inverse relationship was observed between mortality and the number of tests, but this phenomenon is interpretable as 'healthy screenee bias'. The results suggest that the potential efficacy in preventing colorectal cancer mortality through annual FOBT screening may be of the order of one third.

Seventeen-year evaluation of breast cancer screening: the DOM project, The Netherlands. Diagnostisch Onderzoek (investigation) Mammacarcinoom

The DOM project is a non-randomized population-based breast cancer screening programme in Utrecht which started in 1974-75. The 17-year effect has been evaluated by a case-control study of breast cancer deaths during the period 1975-92 in women living in the city of Utrecht, born between 1911 and 1925, whose breast cancers were diagnosed after the initiation of the DOM project. Controls (three for each case) were defined as women having the same year of birth as the case, living in the city of Utrecht at the time the case died, and having had the opportunity of screening in the DOM project. Screening in the period 1975-92 indicated a breast cancer mortality reduction of 46% (odds ratio of 0.54, 95% confidence interval 0.37-0.79). The strongest protective effect was found at a screening interval of 2 years or less (mortality reduction of 62%, odds ratio of 0.38), and for the highest number of screens (mortality reduction of 68%, odds ratio of 0.32 for more than four screens). Exclusion of breast cancer deaths that occurred within 1 year of diagnosis, to allow for 'lead-time' bias, gave an odds ratio of 0.61. Early diagnosis of breast cancer by screening reduces breast cancer mortality in the long term. Bias due to the study design may slightly overestimate the protective effect. A screening programme with a 2-yearly, or smaller, interval between successive screens will improve the protection of screening.

Case-control studies of cancer screening: theory and practice

This review summarizes methodologic theories for the design of cancer screening case-control studies and examines the methods applied in studies published in English from 1980 through 1996. In addition to summarizing state-of-the-art methodologic approaches, we identify areas where obvious gaps exist between theory and practice, and we recommend potential areas where theory and methodology may need further development. In particular, we focus on three major areas: 1) the selection of case and control subjects, 2) the definition of exposure (i.e., exposure to the screening test), and 3) bias. Each area is considered carefully by summarizing current theory, reviewing cancer screening applications, and linking recommended methodologic approaches to those used in practice to identify areas where inconsistencies exist. In general, we found methodologic theory and practice in this field of research to be consistent. However, discrepancies were identified in the area of exposure definition, including the use of screening frequency and the use of a detectable, curable preclinical phase for case subjects as the exposure measures. Even when recommended methods were followed, a number of difficulties arose in practice. Specific concerns included the ability to carry out the following: identifying all case subjects within a source population, defining eligibility criteria to ensure that case and control subjects had equal access to screening during the exposure period, distinguishing between symptomatic and diagnostic tests, and controlling for self-selection bias. Careful scrutiny is warranted in all aspects of the design of cancer screening case-control studies, and caution is advised in the interpretation of study results.