IHP 515 Module 4 Study Design Paper Example

Two Ways to Ask One Question: Cohort and Case-Control Designs for Overdose Risk After Jail Release

[Student Name]

Southern New Hampshire University

IHP 515: Population-Based Epidemiology

Module Four Paper

[Instructor Name]

[Date]

Two Ways to Ask One Question: Cohort and Case-Control Designs for Overdose Risk After Jail Release

The descriptive work on overdose deaths in Harlan County found that nearly a quarter of those who died had left jail or prison in the previous year, and few had received medication for opioid use disorder. That raises an analytic question: does release from jail without medication treatment increase the risk of overdose death, compared with release with treatment? This paper designs a cohort study and a case-control study to answer it, calculates illustrative measures of association and compares the designs' strengths, biases and practicality.

The Logic of Cohort Studies

Grimes and Schulz (2002a) explain that cohort studies begin with people who do not yet have the outcome, classify them by exposure and follow them over time to see who develops the outcome. Because exposure is measured before the outcome occurs, the sequence of events is clear. Cohort studies can calculate incidence in exposed and unexposed groups and compare them as a relative risk. They can be prospective, following people into the future, or retrospective, using existing records to reconstruct a cohort from the past. Their weaknesses include cost and time, inefficiency for rare outcomes and loss of participants to follow-up, which can bias results.

A Retrospective Cohort Design

A retrospective cohort could link county jail records for 2019 to 2022 with pharmacy, treatment and death records. The cohort would include all people with opioid use disorder documented at booking who were released during those years. Those who received methadone or buprenorphine before or at release form the treated group; those who did not form the untreated group. Each person would be followed for twelve months after release for overdose death. Suppose 1,200 people meet criteria: 300 treated and 900 untreated. Over twelve months, 4 treated and 36 untreated people die of overdose. Incidence is 4 divided by 300, or 1.3%, in the treated group and 36 divided by 900, or 4.0%, in the untreated group. The relative risk is 4.0 divided by 1.3, or about 3.0: untreated people had three times the risk of death.

The Logic of Case-Control Studies

Schulz and Grimes (2002) describe case-control studies as research in reverse. They begin with people who have the outcome, the cases, and a comparison group without it, the controls, and look backward to compare how often each group was exposed. Because they start from the outcome, case-control studies are efficient for rare outcomes and can examine several exposures at once. They cannot directly measure incidence; instead they estimate the odds ratio, which approximates the relative risk when the outcome is rare. Their main vulnerabilities are selection bias, if controls do not represent the population that produced the cases, and recall or information bias, if exposure is measured differently for cases and controls.

A Case-Control Design

This version would begin by identifying all 60 overdose deaths in the county from 2019 to 2023 among people released from jail in the prior year, the cases, and select 240 controls from the same release records who were alive twelve months after release, matched on release year. Exposure, lack of medication treatment at release, would be determined from the same pharmacy and treatment records for both groups, reducing information bias. Suppose 52 of 60 cases and 150 of 240 controls were untreated. The odds of being untreated among cases are 52 to 8, or 6.5; among controls, 150 to 90, or 1.67. The odds ratio is 6.5 divided by 1.67, or about 3.9.

Table 1. Illustrative Case-Control Results

Note. Odds ratio = (52 x 90) / (8 x 150) = 3.9. Figures are illustrative.

A Classic Example

Doll and Hill (1950) used a case-control design to study smoking and lung cancer, comparing patients with lung cancer in London hospitals with patients admitted for other conditions. They found that lung cancer patients were far more likely to be heavy smokers and that risk rose with the amount smoked. Their study showed how a case-control design could detect a strong association quickly and cheaply, prompting cohort studies that later confirmed the link. The same logic applies here: a case-control study can provide timely evidence while a larger cohort is assembled.

Comparing the Designs

Both illustrative studies point in the same direction, but they differ in practicality and vulnerability. The cohort study directly measures risk and establishes that exposure preceded death, but it requires linking records for everyone released, and people who move out of the county may be lost to follow-up, possibly biasing results if those who leave differ in risk. The case-control study needs fewer records and less time and is well suited to a relatively rare outcome, but its validity depends on choosing controls from the same population as cases and measuring exposure the same way for both.

Confounding

Both designs face confounding. People who receive medication treatment in jail may not resemble untreated people on factors that also shape overdose risk, such as severity of addiction, housing after release or motivation. If sicker people are less likely to be treated, the untreated group's higher risk might partly reflect severity rather than lack of treatment. Grimes and Schulz (2002b) note that confounding can be addressed through matching, stratification or statistical adjustment; this study would adjust for age, sex, prior overdose and housing status where records allow.

Recommendation

For Harlan County, a case-control study is the faster first step, using existing records to produce evidence within months. A retrospective cohort, linking records for all releases, should follow as data systems allow, providing direct estimates of risk that decision makers find easier to interpret.

Both studies will need approval from the health department's data governance committee and agreements with the jail and treatment providers, which should begin now so that data linkage does not delay the work.

Conclusion

Cohort and case-control studies approach the same question from opposite directions, one following exposure to outcome and the other tracing outcome back to exposure. Understanding their logic, measures and biases lets epidemiologists choose the right design for the question and resources at hand.

References

Doll, R., & Hill, A. B. (1950). Smoking and carcinoma of the lung: Preliminary report. BMJ, 2(4682), 739-748. https://doi.org/10.1136/bmj.2.4682.739

Grimes, D. A., & Schulz, K. F. (2002a). Cohort studies: Marching towards outcomes. The Lancet, 359(9303), 341-345. https://doi.org/10.1016/S0140-6736(02)07500-1

Grimes, D. A., & Schulz, K. F. (2002b). Bias and causal associations in observational research. The Lancet, 359(9302), 248-252. https://doi.org/10.1016/S0140-6736(02)07451-2

Schulz, K. F., & Grimes, D. A. (2002). Case-control studies: Research in reverse. The Lancet, 359(9304), 431-434. https://doi.org/10.1016/S0140-6736(02)07605-5