Study Designs in Epidemiology

a. Cohort study
   • Scheme:

   • Selection of comparison groups (“sampling”) by exposure
   • By conduct:
     – Prospective (concurrent) cohort
     – Retrospective cohort
   • Main analytic approach:
     – Comparison of disease incidence by exposure status
   • Limitations:
     – effect of confounding
     – bias (selection bias, information bias)
  •Example:
     A total of 211 581 postmenopausal women who completed a baseline questionnaire in 1982 and had no history  of cancer, hysterectomy, or ovarian surgery at enrollment. MAIN OUTCOME MEASURE: Ovarian cancer mortality, compared among never users, users at baseline, and former users as well as by total years of use of estrogen replacement therapy (ERT).

b. Case-control study
   • Scheme:

   • Sampling by outcome; exposure information collected retrospectively
   •Main analytic approach: Comparison of exposure prevalence by outcome status
   •Limitations
     - Fatalities lost for study
     - Selection of controls
  •Exposure information bias
  •Example:
      OBJECTIVE: to examine the association between residential status and risk of hip fracture in older people. DESIGN: population-based case-control study. SETTING: Auckland, New Zealand. SUBJECTS: a random sample of all individuals > or = 60 years, hospitalized with a fracture of the proximal femur between July 1991 and February 1994. Controls were age and gender frequency-matched to the cases, randomly selected from a random sample of general practitioners. MAIN OUTCOME MEASURES: radiographically-confirmed fracture of the proximal femur. Fractures sustained as a result of major trauma, such as in a motor vehicle crash, and those associated with pre-existing pathological conditions were excluded. RESULTS: individuals living in institutions were almost four times more likely to sustain a hip fracture [age- and gender-adjusted odds ratio (OR)=3.8; 95% confidence interval (CI): 3.0-4.8] than those living in private homes. After adjustment for potential confounding factors, the risk of hip fracture associated with living in an institution remained significantly increased (P< 0.0001), although the magnitude of the risk was somewhat diminished (OR=2.2; 95% (CI: 1.5-3.5). CONCLUSIONS: living in an institution is associated with an increased risk of hip fracture in older people. Specific factors that place these individuals at increased risk need to be identified, in order to develop intervention strategies.

c. Experimental studies
   • Scheme:

   • Sampling by controlled exposure; follow-up of outcome
   • Main analytic approach: Comparison of outcome incidence by exposure status
   • Types
     - Randomized studies (RCT)
     - Non-randomized studies
  • Purpose of randomization
    - equal distribution of confounding variables
  • Limitations
    - Selected populations, efficacy studies, expensive, ethical challenges
 • Example :
      BACKGROUND: Oral leukotriene receptor antagonists have been shown to have efficacy in chronic asthma. OBJECTIVE: To compare the clinical benefit of montelukast, a once-daily oral leukotriene receptor antagonist; placebo; and inhaled beclomethasone. DESIGN: Randomized, double-blind, double-dummy, placebo-controlled, parallel-group, 12-week study. SETTING: 36 sites worldwide. PATIENTS: 895 patients 15 to 85 years of age with chronic asthma and an FEV1 50% to 85% of predicted. INTERVENTIONS: Montelukast, 10 mg once daily at bedtime; inhaled beclomethasone, 200 microg twice daily, administered with a spacer device; or placebo. MEASUREMENTS: Primary end points were daytime asthma symptom score and FEV1. Secondary end points were peak expiratory flow rates in the morning and evening, as-needed beta-agonist use, nocturnal awakenings, asthma-specific quality of life, and worsening asthma episodes.