Privacy-protecting estimation of adjusted risk ratios using modified Poisson regression in multi-center studies

Shu, Di; Young, Jessica; Toh, Sengwee

doi:10.1186/s12874-019-0878-6

Cited by 7 publications

(9 citation statements)

References 27 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Modified Poisson regression was used to determine the association between receipt of FP counselling and current use of modern contraception, prevalence ratio (PR) was used as the measure of association. Modified Poisson regression directly estimates prevalence ratios and produces confidence intervals with the correct nominal coverage when individual-level data are available [22,23]. We used sample weights calculated based on the multistage sampling design in all analyses.…”

Section: Discussionmentioning

confidence: 99%

Missed opportunity for family planning counselling along the continuum of care in Arusha region, Tanzania

et al. 2021

View full text Add to dashboard Cite

Introduction Adequate sexual and reproductive health information is vital to women of reproductive age (WRA) 15 to 49 years, for making informed choices on their reproductive health including family planning (FP). However, many women who interact with the health system continue to miss out this vital service. The study aimed to identify the extent of provision of FP counselling at service delivery points and associated behavioral factors among women of reproductive age in two districts of Arusha region. It also determined the association between receipt of FP counselling and contraceptive usage. Methods Data were drawn from a cross-sectional survey of 5,208 WRA residing in two districts of Arusha region in Tanzania; conducted between January and May 2018. Multistage sampling technique was employed to select the WRA for the face-to-face interviews. FP counseling was defined as receipt of FP information by a woman during any visit at the health facility for antenatal care (ANC), or for post-natal care (PNC). Analyses on receipt of FP counseling were done on 3,116 WRA, aged 16–44 years who were in contact with health facilities in the past two years. A modified Poisson regression model was used to determine the Prevalence Ratio (PR) as a measure of association between receipt of any FP counseling and current use of modern contraception, controlling for potential confounders. Results Among the women that visited the health facility for any health-related visit in the past two years, 1,256 (40%) reported that they received FP counselling. Among the women who had had births in the last 30 months; 1,389 and 1,409 women had contact with the service delivery points for ANC and PNC visits respectively. Of these 31% and 26% had a missed FP counseling at ANC and PNC visit respectively. Women who were not formally employed were more likely to receive FP counselling during facility visit than others. WRA who received any FP counseling at PNC were significantly more likely to report current use of modern contraception than those who did not (adjusted PR [adj. PR] = 1.28; 95% Confidence Interval [CI]: 1.09, 1.49). Conclusion Overall, only 40% women reported that they received any form of FP counseling when they interfaced with the healthcare system in the past two years. Informally employed women were more likely to receive FP counselling, and women who received FP counselling during PNC visits were significantly more likely to use contraceptive in comparison to the women who did not receive FP counselling. This presents a missed opportunity for prevention of unintended pregnancies and suggests a need for further integration of FP counseling into the ANC and PNC visits.

show abstract

Section: Discussionmentioning

confidence: 99%

Missed opportunity for family planning counselling along the continuum of care in Arusha region, Tanzania

et al. 2021

View full text Add to dashboard Cite

show abstract

“…The unrestricted benchmarks are the estimates of 𝛽 w in (21), denoted as βw,bm , and its estimated variance, denoted as V𝛽 w ,bm , from the combined data.…”

Section: Unrestricted Benchmarksmentioning

confidence: 99%

“…Specifically, early studies in meta-analysis and meta-regression analysis provide inference, but largely center around combining randomized controlled trials, and a typically used pooling approach is IVW. 16,17 Recently, a growing number of studies develop privacy-preserving methods to provide inference by pooling aggregate data across multiple studies: most of them are tailored to specific parametric models, including linear models, 18,19 logit models, 20 Poisson models, 21 Cox models, 22,23 and generalized linear models 24 , while Jordan et al 25 and Duan et al 26 consider the efficient pooling of the more general MLE. Among these studies, only Toh et al 18 and Shu et al 22 account for nonrandom treatment assignments by using propensity scores, though the asymptotic theory is lacking.…”

Section: Introductionmentioning

confidence: 99%

Federated causal inference in heterogeneous observational data

Xiong

Koenecke

Powell

et al. 2023

Statistics in Medicine

View full text Add to dashboard Cite

We are interested in estimating the effect of a treatment applied to individuals at multiple sites, where data is stored locally for each site. Due to privacy constraints, individual‐level data cannot be shared across sites; the sites may also have heterogeneous populations and treatment assignment mechanisms. Motivated by these considerations, we develop federated methods to draw inferences on the average treatment effects of combined data across sites. Our methods first compute summary statistics locally using propensity scores and then aggregate these statistics across sites to obtain point and variance estimators of average treatment effects. We show that these estimators are consistent and asymptotically normal. To achieve these asymptotic properties, we find that the aggregation schemes need to account for the heterogeneity in treatment assignments and in outcomes across sites. We demonstrate the validity of our federated methods through a comparative study of two large medical claims databases.

show abstract

“…There is often only a single parameter of interest in meta-analysis and a linear functional form imposed in meta-regression analysis. Recent studies develop privacy-preserving methods to pool the summarylevel information across multiple studies for broader classes of models such as linear models (Toh et al, 2018, Li et al, 2019, logistic models (Li et al, 2016), Poisson models (Shu et al, 2019), and GLM (Wolfson et al, 2010)); only a few of these studies provide pooling methods for confidence intervals, but they are restricted to specific models and lack asymptotic theory for the pooling methods (e.g., Poisson models (Shu et al, 2019) or Cox models (Shu et al, 2020)).…”

Section: Introductionmentioning

confidence: 99%

“…For example, the federated estimators in logistic models(Fienberg et al, 2006, Slavkovic et al, 2007, Li et al, 2016, Poisson models(Shu et al, 2019), and MLE(Blatt and Hero, 2004, Karr et al, 2007, Zhao and Nehorai, 2007, Lin and Karr, 2010, Snoke et al, 2018.4 An iterative approach can provide estimators that are closer to those from the pooled individual-level data. However, we show that, asymptotically, the difference between iterative and non-iterative approaches is a higher order term that can be neglected.…”

mentioning

confidence: 99%

Federated Causal Inference in Heterogeneous Observational Data

Xiong¹,

Koenecke²,

Powell³

et al. 2021

Preprint

View full text Add to dashboard Cite

Analyzing observational data from multiple sources can be useful for increasing statistical power to detect a treatment effect; however, practical constraints such as privacy considerations may restrict individual-level information sharing across data sets. This paper develops federated methods that only utilize summary-level information from heterogeneous data sets. Our federated methods provide doubly-robust point estimates of treatment effects as well as variance estimates. We derive the asymptotic distributions of our federated estimators, which are shown to be asymptotically equivalent to the corresponding estimators from the combined, individual-level data. We show that to achieve these properties, federated methods should be adjusted based on conditions such as whether models are correctly specified and stable across heterogeneous data sets.

show abstract

Privacy-protecting estimation of adjusted risk ratios using modified Poisson regression in multi-center studies

Cited by 7 publications

References 27 publications

Missed opportunity for family planning counselling along the continuum of care in Arusha region, Tanzania

Missed opportunity for family planning counselling along the continuum of care in Arusha region, Tanzania

Federated causal inference in heterogeneous observational data

Federated Causal Inference in Heterogeneous Observational Data

Contact Info

Product

Resources

About