By the end of this module, you should be able to:

• define the following study designs and understand the situations in which they are most appropriate:
  • prospective cohort study
  • retrospective cohort study
  • case-control study
• create a 2X2 table for a cohort study and calculate and interpret a relative risk from it
• create a 2X2 table for a case-control study and calculate and interpret an odds ratio from it

References

Users’ guides to the medical literature : a manual for evidence-based clinical practice / The Evidence-Based Medicine Working Group ; edited by Gordon Guyatt, Drummond Rennie. Chicago, IL : AMA Press, c2002.

Fletcher, Robert H. Clinical epidemiology : the essentials / Robert H. Fletcher, Suzanne W. Fletcher. 4th ed. Baltimore : Lippincott Williams & Wilkins, c2005.

You are a medical student working in an infectious disease clinic. You are taking blood from an HIV positive patient. He moves suddenly as you access the vein and the needle pierces your glove and pricks your finger. Suddenly, you are very interested in the risk of HIV transmission from an occupational needlestick exposure and wonder about the effectiveness of anti-retroviral medication (post-exposure prophylaxis or PEP) in preventing HIV transmission.

The original case-control of HIV seroconversion was done with a population of hospital workers who had a needlestick injury. “Cases” were workers who subsequently became HIV positive (seroconverted) and “controls” were workers who did not seroconvert. The case-control study compared the features of exposure and post-exposure treatment among cases versus controls.

Back to your occupational needlestick injury...

You search through the literature and find a (mythical) retrospective cohort studying exactly this question. 1,000 nurses working in AIDS clinics in Africa experienced a needlestick exposure in 1998. Only half of the nurses could afford to purchase anti-retroviral medications. Using medical records, the 1,000 nurses were “followed” for 6 months and their HIV tests at the end of that period were reported. Of the 500 who did not take “PEP” (post-exposure prophylaxis), 50 (10%) were HIV positive at six months. Among the 500 who took “PEP” medications, 10 (2%) were HIV positive.

The information from this cohort study can be displayed in a new type of 2x2 table. As we have previously discussed, “the truth always rises to the top”, and here the truth is the outcome of interest – seroconversion. Insert the numbers from the mythical cohort study in Africa into the following 2x2 table on a piece of paper before moving to the next page. Label the exposure and outcome categories.

Unlike 2x2 tables constructed for diagnostic tests, we do not calculate sensitivity and specificity for cohort studies. Rather, we calculate a “measure of association” or “measure of effect” called a relative risk. A relative risk is simply a ratio of rates. Here it would be a ratio of the risk of seroconversion in the exposed group (10/500 = 2%) relative to the risk of seroconversion in the unexposed group (50/500 = 10%).

Risk for seroconversion in the nurses exposed to PEP was 0.2 times the risk among nurses who were not exposed. PEP protected those nurses from becoming HIV positive after needlestick injury.

In classical epidemiology, cohort studies were designed to establish causality (e.g. cholera epidemics were associated with exposure to contaminated water, and cohort studies were used to establish this association). In these studies, the exposure is a risk factor. However, in clinical epidemiology, the “exposure” is often a treatment or intervention which may prove protective.

Among the 1000 nurses exposed to HIV positive blood, 200 of them had a volume of exposure that was large (> 3ml). The remaining 800 nurses had a lower volume of exposure. Of the 60 nurses who seroconverted, 50 of them were in the large volume group. Fill in the following 2x2 table on a piece of paper before moving to the next screen. Calculate and interpret the relative risk of seroconversion among nursing groups defined by the volume of exposure.

RR = (50/200) / (10/800)

= 25% / 1.25%

= 20

We will use this same clinical example to illustrate what a case-control study looks like. Remember, case-control studies involve the identification of patients with disease outcomes (cases) and patients without disease outcomes (controls). Cases and controls are compared with respect to their past exposures.

Imagine that you have designed a study to look at the depth of needle penetration as a risk factors for HIV seroconversion. You have identified 100 people who became HIV positive following needlestick injury (cases), and 1000 people who remained HIV negative after needlestick injury (controls). You collect information about the location and estimated depth of the needlestick injury penetration. You divide your “exposure” into superficial (skin depth only) or “deep” (intramuscular or intravenous penetration).

Among the 100 cases, 80/100 (80%) report a deep needlestick injury. Among the 1000 controls, 150/1000 (15%) report a deep needlestick injury.

Before moving to the next screen, fill out the above 2x2 table on a piece of paper for a basic, case-control study.

In case-control studies, a relative risk has no meaning because you have decided how many cases and controls there will be in your study, and this in turn affects the relative risk calculation. Therefore, relative risks are not used in case-control studies. Instead, there is a new measure of effect called an odds ratio.

Therefore, "deep" exposure is a very strong risk factor for HIV seroconversion.

Important Trivia: Mathematically, when a disease is “rare” (say, <5% of the population under study), an odds ratio is an excellent approximation of a relative risk. This means that you can interpret an odds ratio the same way as a relative risk. This is called the “rare disease assumption”, commonly discussed in case-control studies.

Clinical Scenario

You are a medical student working in an infectious disease clinic. You are taking blood from an HIV positive patient. He moves suddenly as you access the vein and the needle pierces your glove and pricks your finger.

Your research has shown you that although your injury was superficial, and involved a low volume of blood, HIV PEP is has been shown to be protective against seroconversion. You decide to discuss the possibility of post-exposure prophylaxis with your clinical supervisor.

You are pleased with your new expertise in epidemiological study design and associated measures of effect.

Resources

CCFP Recommendations for PEP

CMAJ review on PEP

You should now be able to:

• define the following study designs and understand the situations in which they are most appropriate:
  • prospective cohort study
  • retrospective cohort study
  • case-control study
• create a 2X2 table for a cohort study and calculate and interpret a relative risk from it
• create a 2X2 table for a case-control study and calculate and interpret an odds ratio from it

Credits

Credits