PU520 Unit 4 Case-Control Study Design Critique Example

What an Odds Ratio of 1.8 Can Establish: A Study Design Critique Reading the Methods of a Composite Hospital-Based Case-Control Study of Sleep Medication and Hip Fracture in Older Adults, Section by Section, From Control Selection to Recalled Exposure

[Student Name]

Purdue University Global

PU520: Principles of Epidemiology

Unit 4 Study Design Critique

[Instructor Name]

[Date]

The study critiqued is composite and its figures are bracketed for the writer's own data; the reporting guideline and published research cited are real.

Summary

Question: is recent sleep medication use associated with hip fracture in adults aged 65 and older? Design: case-control, at one hospital. Cases: [180] patients admitted with a first hip fracture. Controls: [360] older patients booked for elective joint replacement on the same orthopedic service, matched on age and sex. Exposure: use of any prescription sleep medication in the month before admission, by interview. Result: odds ratio [1.8], 95 percent confidence interval [1.2] to [2.7], adjusted for age, sex and number of chronic conditions.

Selection of Controls

Controls should represent the population that produced the cases, so that their exposure reflects what cases' exposure would have been without the disease (Wacholder et al., 1992). Patients awaiting elective joint replacement are not that population. Many have chronic joint pain, which disturbs sleep, so they may use sleep medication more than other older adults. If so, control exposure is inflated and the odds ratio is biased toward one, meaning the true association could be stronger. Alternatively, elective surgery patients may be healthier and more mobile than the general older population, with less use of sedating drugs, which would bias the odds ratio upward. The direction cannot be known without data on the source population. A control group chosen for convenience can push the odds ratio up or down, and the paper gives no way to tell which.

Measurement of Exposure

Exposure was recalled in an interview after admission. Patients who have just broken a hip may search their memories for causes and recall medication use more fully than controls, inflating the odds ratio. Some cases, in pain and on strong analgesics, may recall less accurately. Pharmacy dispensing records would measure exposure independently of the outcome. The paper also does not separate long-acting from short-acting drugs, though a large earlier case-control study found increased hip fracture risk with long-acting hypnotics and anxiolytics but not short-acting ones, with risk rising with dose (Ray et al., 1987). Grouping all sleep medications together may hide the drugs responsible.

Confounding

Adjustment for age, sex and number of chronic conditions leaves important confounders unmeasured: dementia, frailty, poor vision, prior falls, alcohol use and other sedating medications, such as antidepressants. Each is linked to both sleep medication use and falls. A count of chronic conditions treats a mild condition and severe dementia as equivalent.

Alternative Explanations

Reverse timing: people beginning to decline may sleep poorly and receive a sleep medication shortly before a fall that was already more likely. A dose-response analysis and a check of timing, such as use for at least three months before the fracture, would help. Chance: the confidence interval excludes one, so chance is less likely, but the study is modest in size.

Statistical Analysis

The study reports a single adjusted odds ratio from conditional logistic regression, appropriate for matched sets, but does not show the unadjusted estimate or how much each adjustment changed it. Readers therefore cannot see whether confounders moved the estimate. It also reports no analysis by drug type, dose or duration, which would test whether the association behaves as a causal effect should. With [180] cases, subgroup analyses would be imprecise, but presenting them with their intervals would still be informative.

Strengths

The study has clear case definitions, confirmed by imaging, and an outcome that is unambiguous and recorded at admission. Matching on age and sex addresses two strong determinants of both fracture and medication use. Its question is clinically important, and its main finding is in the same direction as earlier research. A critique that lists only weaknesses misjudges what the study offers.

Reporting Gaps

The paper does not state how many eligible cases and controls declined, how interviewers were trained or whether they knew which patients were cases. These items are part of the reporting guideline for observational studies and matter for judging bias (von Elm et al., 2007).

What the Design Permits

The study shows an association between recalled recent sleep medication use and hip fracture among hospitalized older adults. Given its control group, recall-based exposure and limited adjustment, it cannot establish that sleep medication causes fractures or estimate the size of any effect. Its finding is consistent with earlier, larger studies, which raises plausibility but does not repair its own weaknesses.

How a Better Study Would Differ

Population-based controls from the same region; exposure from pharmacy records, separated by drug half-life and dose; adjustment for dementia, frailty, other sedating drugs and prior falls; and timing analyses. A cohort using pharmacy and hospital records would avoid recall altogether.

Implications for Practice

The finding, together with prior research, supports caution in prescribing long-acting sleep medication to older adults, but the study itself should not be cited as evidence of a specific risk.

References

Ray, W. A., Griffin, M. R., Schaffner, W., Baugh, D. K., & Melton, L. J. (1987). Psychotropic drug use and the risk of hip fracture. New England Journal of Medicine, 316(7), 363-369. https://doi.org/10.1056/NEJM198702123160702

von Elm, E., Altman, D. G., Egger, M., Pocock, S. J., Gøtzsche, P. C., & Vandenbroucke, J. P. (2007). The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement: Guidelines for reporting observational studies. The Lancet, 370(9596), 1453-1457. https://doi.org/10.1016/S0140-6736(07)61602-X

Wacholder, S., McLaughlin, J. K., Silverman, D. T., & Mandel, J. S. (1992). Selection of controls in case-control studies: I. Principles. American Journal of Epidemiology, 135(9), 1019-1028. https://doi.org/10.1093/oxfordjournals.aje.a116396