Inverse probability weighted Cox model in multi-site studies without sharing individual-level data

Abstract

The inverse probability weighted Cox proportional hazards model can be used to estimate the marginal hazard ratio. In multi-site studies, it may be infeasible to pool individual-level datasets due to privacy and other considerations. We propose three methods for making inference on hazard ratios without the need for pooling individual-level datasets across sites. The first method requires a summary-level eight-column risk-set table to produce the same hazard ratio estimate and robust sandwich variance estimate as those from the corresponding pooled individual-level data analysis (reference analysis). The second and third methods, which are based on two bootstrap re-sampling strategies, require a summary-level four-column risk-set table and bootstrap-based risk-set tables from each site to produce the same hazard ratio and bootstrap variance estimates as those from their reference analyses. All three methods require only one file transfer between the data-contributing sites and the analysis center. We justify these methods theoretically, illustrate their use, and demonstrate their statistical performance using both simulated and real-world data.

Estimating the marginal hazard ratio by simultaneously using a set of propensity score models: A multiply robust approach

“…We also extended our method to multisite settings so that each participating site may postulate multiple site‐specific propensity score models. It can be done in a privacy‐protecting way using data‐sharing methods of Shu et al 47 Specifically, each site first estimates the empirical likelihood weights for its members using multiple propensity score models, and then obtains its risk‐set table using the resultant empirical likelihood weights. Finally, instead of sharing individual‐level data across sites, it suffices for sites to share their summary‐level risk‐set tables to the analysis center to estimate the marginal hazard ratio 47 …”

“…For example, the Sentinel System is a national electronic system funded by the US Food and Drug Administration to monitor the safety of approved medical products using data from more than a dozen health plans and delivery systems 44 . The IPW Cox model stratified on data‐contributing site provides one approach to estimating marginal hazard ratios in multisite studies, where each site fits a site‐specific propensity score model 45‐47 . In this section, we extend the proposed multiply robust method in Section 4 to enable each participating site to postulate multiple site‐specific propensity score models.…”

Adjuvant Treatment of Early Ovarian Clear Cell Carcinoma: A Population-Based Study of Whole Abdominal Versus Pelvic Nodal Radiotherapy

“…IPW-adjusted log-rank tests and Cox multivariable regressions were applied to compare the HR for FFS and OS between the 2 chemoRT regimens. [23][24][25] Additionally, flexible parametric models for time-to-event data were applied to calculate the treatment effect on FFS and OS adjusted for covariables and time (supplemental eAppendix 2 and eTables 2-5). 26 The competing risk models of Fine and Gray 27 were applied to determine the cumulative incidence of treatment failure and cancer-specific mortality (CSM) among the treatment groups.…”

“…Statistical analyses were performed using R statistical version 3.5.2 (R Foundation for Statistical Computing). [22][23][24][25][26][27][28]…”

Methodological frontiers in vaccine safety: qualifying available evidence for rare events, use of distributed data networks to monitor vaccine safety issues, and monitoring the safety of pregnancy interventions

“…47 More recently, simulation as well as empirical studies have shown that privacy-protecting methods, such as sharing of aggregated data sets, perform similarly to analysis in individual-level pooled data. [48][49][50]…”