Stratified Wilson and Newcombe Confidence Intervals for Multiple Binomial Proportions

Abstract

This article proposes the stratified Wilson confidence interval for multiple binomial proportions and the stratified Newcombe confidence intervals for multiple binomial proportion differences. Both confidence intervals are presented in closed forms to facilitate easy calculations. The confidence levels of the proposed intervals are theoretically justified and demonstrated through extensive simulations. The coverage rates are found to be rather satisfactory. When the Wilson and Newcombe methods are used in unstratified analysis, the proposed methods may serve as the counterparts for stratified analysis. The proposed methods are applied to a vaccine trial to compute the stratified sero-conversion rate and rate difference over multiple clinical centers.

Ceftolozane/Tazobactam Plus Metronidazole for Complicated Intra-abdominal Infections in an Era of Multidrug Resistance: Results From a Randomized, Double-Blind, Phase 3 Trial (ASPECT-cIAI)

“…The noninferiority hypothesis was tested through a 2-sided 95% confidence interval (CI) approach. The weighted difference in cure rates and the 95% CI around the difference in cure rates between study treatments were calculated using a stratified Newcombe CI with minimum risk weights [ 25 , 26 ]. If the lower bound of the 95% CI for the difference (ceftolozane/tazobactam plus metronidazole minus meropenem) was above −10 percentage points, noninferiority was claimed.…”

Second-Generation Hydrogel Coils for the Endovascular Treatment of Intracranial Aneurysms

“…For binary outcomes, the absolute difference of the proportion of outcome events between the 2 arms, expressed as percentages, was calculated along with a 2-sided Newcombe 95% confidence interval (CI) and P value with Cochran–Mantel–Haenszel weights, stratified by rupture status. 12 A preplanned sensitivity analysis of the primary end point explored the worst-case scenario in the analysis population where all missing outcomes for patients randomized to the hydrogel arm were evaluated as unfavorable and all those in the bare platinum arm as favorable. For post hoc analyses, we calculated Newcombe CI for the absolute difference in the proportion of unfavorable outcomes between treatments within subgroups, and we examined odds ratios (±the interaction with treatment) by Wald tests from logistic regression.…”

Seroprevalence of Heartland Virus Antibodies in Blood Donors, Northwestern Missouri, USA

“…Because there were differences in the rates of blood donors per population in the included counties, we computed the estimate of the seroprevalence within the region by using a stratum-weighted estimate and 95% CI ( 10 ). We estimated a seroprevalence of 0.9% (95% CI 0.4‒4.2%) in blood donors >16 years of age in the 10-county region.…”