Case-control study designs

Bradbury et al. (2004), however, recently reanalyzed the relation between tamoxifen and cataracts and described it as a null association, They used a nested, matched, case-control study design and data collected in the General Practice Research Database. They identified all women 30-79 years old who were diagnosed with breast cancer and treated with tamoxifen within 6 months, or with bladder cancer, colorectal cancer, or nonmelanoma skin cancer between January 1991 and December i999. From this population they identified all newly diagnosed cases of cataract and matched four female controls to each case on age, index date, and study entry data. They assessed the risk of cataracts for current, past, and sometime users of tamoxifen... 

The cohort study offers two additional strategies for analysis — case-cohort and nested case-control study designs the case-cohort comprises all individuals in the cohort with a particular disease and a random sample of unaffected individuals from the cohort the nested case-control design includes cohort members with a particular disease (cases) and unaffected individuals (controls) matched on relevant study covariates ... 

The biomarker is applied to a population etiological study focused on investigating the role of HCA exposure to cancer. Due to concerns of biases introduced by cancer status ( reverse causation ), a nested case-control study design is preferred. 

Schlesselman, J. J. Case-Control Studies, Design, Conduct, Analysis, Oxford University Press, New York, 1982. 

Figure 7.1 The design of a cohort study and its relationship with case-control study designs. Individuals who died during follow-up in the shaded area are eligible for inclusion in a case-control study including incident cases between t and t2- The controls in the case-control study are individuals who did not have the disease of interest at the time of involvement. Arrows indicate end of follow-up, either because of (a) loss to follow-up, (b) mortality due to cause of interest, or (c) other causes of death. The risk ratio, or relative risk (RR) = ( died cause of interest in exposed/E person years)/( died cause of interest controls/E person years)...

In a prospective, case-control study, designed to determine causes of INRs over 6.0 in an outpatient anticoagulant unit, there was a clear dose-dependent association between the use of paracetamol (acetaminophen) and having an INR greater than 6.0 (212). The authors studied 93 patients with INRs over 6.0 (cases) and 196 patients with INRs of 1.7-3.3 (controls) during warfarin therapy. The likelihood of an INR greater than 6.0 increased from an odds ratio of 3.5 for doses of 2275 549 mg per week, to 6.9 for doses of 4550-9099 mg per week, to a 10-fold increase at a dose of over 9100 mg per week. 

Rose, S. and M.J. van der Laan. Why match investigating matched case-control study designs with causal effect estimation. Int Biostat, 5, Article 1, 2009, http / / www.bepress.com / ijb / vol5 / issl /1 /. 

Example of risk data from a case-control study design... 

Mexico City pregnant women followed from first trimester (N=668) PbB levels in nested case—control study design Spontaneous abortion rate Odds ratio of 1.8/ 5 pg/dl PbB increase dose—response seen Borja-Aburto etal. (1999)... 

In a case-control study design, the point of departine is an at-risk popnlation that is defined, not as one that is known to have been exposed, but as one that is suspected of having an elevated disease frequency (the case). The disease frequency is compared with that of a sample drawn from a second population that is thought not to be at risk (the eontrol). Elevation of the disease frequency in the study group drawn from the case population compared with the study group drawn from the control ... 

NEWMAN T B, BROWNER s w, CUMMINGS s R, HULLEY s B (1988) Designing a new study 11. Cross-sectional and case-control studies, in Hulley S B and Cummings S R, Designing Clinical Research, Baltimore, Williams Wilkins, 75-86. 

Although earlier work had showed a positive association between coffee consumption and pancreatic cancer rates across countries,2 it was the much-publicized case-control study of MacMahon et al.3 in 1981 that attracted widespread attention to the question of a possible link. In that study, which was designed primarily to investigate the role of smoking and alcohol in pancreatic cancer, 369 pancreatic cancer patients prior to diagnosis and 644 hospital controls reported their typical daily coffee and tea consumption. Unexpectedly, the authors found a significantly increased risk of pancreatic cancer associated with coffee consumption (overall rela-... 

The Canadian Task Force categorized the quality of evidence based on the type of research study. The quality of evidence was organized into three classes Class I evidence comes from procedures having at least one randomized controlled study to support them. Class II is divided into three subclasses, where II-l involves a well-designed controlled study without randomization. Class II-2 evidence comes from well-designed cohort or case-control studies, preferably carried out at more than one research setting. Class II-3 involves uncontrolled research with dramatic results (e.g., penicillin trials in the 1940s). Class III evidence includes the opinions of experts and authorities in the field based on clinical... 

Epidemiological studies have different strengths and weaknesses associated with their design (Table 4) shows some of the strengths and weaknesses of the two main methods of prospective cohort studies and retrospective case-control studies. 

Breslow, N.E. Day, N.E. (1980) Statistical Methods in Cancer Research, Vol. 1, The Analysis of Case-Control Studies (lARC Scientific Publications No. 32), Lyon, ARCPress Breslow, N.E. Day, N.E. (1987) Statistical Methods in Cancer Research, Vol. 2, The Design and Analysis of Cohort Studies (lARC Seientific Publications No. 82), Lyon, IP RCPress Cohen, S.M. Ellwein, L.B. (1990) Cell proliferation in carcinogenesis. Science, 249, 1007-1011... 

---