A multicenter observational study is conducted to evaluate the efficacy of a novel glucago… — USMLE Step 3

A multicenter observational study is conducted to evaluate the efficacy of a novel glucagon-like peptide-1 (GLP-1) receptor agonist compared to a dipeptidyl peptidase-4 (DPP-4) inhibitor in preventing the progression of non-alcoholic steatohepatitis (NASH) in patients with type 2 diabetes. The primary outcome is the 3-year incidence of biopsy-proven NASH progression.

The initial analysis yields a crude hazard ratio (HR) of $0.70$ (95% CI: $0.55-0.88$ ), favoring the GLP-1 receptor agonist.

To address potential methodological concerns, the investigators perform two separate stratified analyses based on concurrent high-intensity statin use and baseline body mass index (BMI). The results are as follows:

Stratification Variable	Stratum	Hazard Ratio (95% CI)
Statin Use	Concurrent high-intensity statin	$0.71$ ( $0.50-0.95$ )
No high-intensity statin	$0.69$ ( $0.48-0.92$ )
Baseline BMI	$\text{BMI} > 35 \text{ kg/m}^2$	$0.45$ ( $0.35-0.58$ )
$\text{BMI} \leq 35 \text{ kg/m}^2$	$0.95$ ( $0.75-1.20$ )

Based on these epidemiological findings, which of the following conclusions are correct? (Select all that apply)

Concurrent high-intensity statin use acts neither as a significant confounder nor as an effect modifier in this association.Answer
Baseline BMI demonstrates effect modification, indicating that the clinical benefit of the GLP-1 agonist differs based on patient weight.Answer
C
Concurrent high-intensity statin use acts as a significant confounder, requiring the calculation of a pooled Mantel-Haenszel hazard ratio for accurate reporting.
D
Baseline BMI acts primarily as a confounder, indicating that its distribution is unequal between the GLP-1 agonist and DPP-4 inhibitor groups.
E
The investigators should utilize a multivariable regression model to adjust for baseline BMI in order to report a single, accurate hazard ratio for the entire cohort.

Answer

The correct conclusions are that statin use is neither a confounder nor an effect modifier, while baseline BMI demonstrates effect modification.

The correct statements correctly distinguish between confounding and effect modification based on stratified data. A variable is neither a confounder nor an effect modifier if the stratum-specific risk estimates equal each other and equal the crude estimate (as seen with statin use:

0.71 \approx 0.69 \approx 0.70

). A variable is an effect modifier if the stratum-specific risk estimates differ significantly from one another (as seen with BMI:

0.45 \neq 0.95

Step-by-Step Solution

Evaluate the statin use stratified analysis.

The stratum-specific hazard ratios (

0.71

and

0.69

) are nearly identical to each other and to the crude hazard ratio (

0.70

Similarity between strata rules out effect modification, while similarity to the crude estimate rules out confounding.

Evaluate the baseline BMI stratified analysis.

The hazard ratio for

\text{BMI} > 35

(

0.45

) is substantially different from the hazard ratio for

\text{BMI} \leq 35

(

0.95

, which crosses

1.0

A significant difference between stratum-specific estimates is the hallmark of effect modification, meaning the treatment's efficacy depends on the patient's BMI.

Determine the appropriate reporting method for the BMI findings.

The findings should be reported separately for each BMI stratum.

Effect modifiers represent true biological differences in treatment response and should never be "adjusted away" in a multivariable model or pooled into a single estimate.

Key Concept

Distinguishing confounding from effect modification using stratified analysis

Estimated Time:2m 0s

Question

Answer

Step-by-Step Solution

Key Concept

Common Mistakes

Topics

More questions from this topic