Background

Estimates of effect heterogeneity (i.e. the extent to which the causal effect of one exposure varies across strata of a second exposure) can be biased if the exposure-outcome relationship is subject to uncontrolled confounding whose severity differs across strata of the second exposure.

Methods

We propose methods, analogous to the E-value for total effects, that help to assess the sensitivity of effect heterogeneity estimates to possible uncontrolled confounding. These E-value analogues characterize the severity of uncontrolled confounding strengths that would be required, hypothetically, to 'explain away' an estimate of multiplicative or additive effect heterogeneity in the sense that appropriately controlling for those confounder(s) would have shifted the effect heterogeneity estimate to the null, or alternatively would have shifted its confidence interval to include the null. One can also consider shifting the estimate or confidence interval to an arbitrary non-null value. All of these E-values can be obtained using the R package EValue.

Results

We illustrate applying the proposed E-value analogues to studies on: (i) effect heterogeneity by sex of the effect of educational attainment on dementia incidence and (ii) effect heterogeneity by age on the effect of obesity on all-cause mortality.

Conclusion

Reporting these proposed E-values could help characterize the robustness of effect heterogeneity estimates to potential uncontrolled confounding.

Background

We aimed to investigate mortality trends in hospitalized patients with septic shock in the US. To achieve this objective, we tested hypothesis that mortality decreased in patients identified by the code of septic shock while mortality did not change in those with septic shock identified by vasopressor use.

Methods

We conducted a serial cross-sectional analysis using Nationwide Inpatient Sample database from 2005 through 2011. First, we identified all adult patients aged ≥18 years hospitalized for septic shock by the following criteria: 1) primary ICD-9 diagnosis of infection plus procedure code for vasopressor use, 2) primary ICD-9 diagnosis of infection plus septic shock in non-primary field, and 3) primary ICD-9 diagnosis of septic shock. Second, we stratified all identified patients by record of vasopressor use. The outcome of interest was year-to-year changes in the in-hospital all-cause mortality.

Results

From 2005 to 2011, we identified 109,812 weighted hospitalizations with septic shock. Overall, there was a significant downward trend in in-hospital mortality (from 46 % in 2005 to 42 % in 2011; P trend  = 0.003); the adjusted mortality also decreased significantly (OR for comparison of 2005 with 2011, 0.98; 95 % CI, 0.96-1.00; P < 0.001). In stratified analysis, the mortality trend was not significant in the subgroup with vasopressor use (from 42 % in 2005 to 40 % in 2011; P trend =0.57); similarly, the adjusted mortality did not change significantly (OR, 1.01; 95 % CI, 0.97-1.05; P =0.62). By contrast, there was a downward trend in mortality in the subgroup without vasopressor use (from 47 % in 2005 to 43 % in 2011; P trend =0.002); likewise, the adjusted mortality decreased significantly (OR, 0.97; 95 % CI, 0.95-0.99; P =0.002) CONCLUSIONS: From 2005 to 2011, we found a modest decrease in in-hospital mortality among patients identified with septic shock. However, in the subgroup with vasopressor use, we found no significant change in mortality. Our data challenge the conventional wisdom that mortality in this population has improved during the last decade.