Methods to Account for Competing Events With Unknown Duration in Health Policy Evaluations: The Case of COVID ‐19

OBJECTIVE: To compare methods for estimating policy effects in the presence of competing events such as COVID-19. STUDY SETTING AND DESIGN: We created a simulation study to evaluate how single interrupted time series (ITS) and comparative interrupted time series (CITS) estimators performed when policy implementation occurred near a competing event (COVID-19). The onset of the competing event was known (March 2020) but the duration was unknown. ITS approaches included assuming different COVID-19 durations (ITS-Assume) or searching for the COVID-19 duration that provided the best fit to the data (ITS-Grid Search). Outcomes included percent bias, power, and Type 1 error. DATA SOURCES AND ANALYTIC SAMPLE: We varied the duration of COVID-19 effects, the time between the policy implementation and COVID-19 onset, and the magnitude and sign of the policy effects relative to COVID-19 (2080 scenarios). Each simulated time series had 99 monthly observations (October 2015-December 2023). PRINCIPAL FINDINGS: CITS had the lowest bias and Type 1 error and had high power across scenarios, except for intercept changes and when assumptions about shared COVID-19 effects between treatment and control groups were violated. ITS-Grid Search had lower bias and Type 1 error than ITS-Assume, which produced biased parameter estimates when the assumptions were incorrect. Estimates were more accurate when the policy was implemented prior to the competing event. CONCLUSIONS: Using estimation approaches that leverage control groups, such as CITS, can mitigate the potential bias of competing events in ITS frameworks, but these approaches are biased when competing event effects differ between treatment and control groups. Searching for the best-fitting competing event duration does not require a control group and outperforms methods that assume the duration of competing event effects a priori. More generally, researchers should be cautious about bias, specification error, and low power when using ITS methods with competing events.

Authors

Institutions

Publication Details

Journal
Health Services Research
Published
2026-11-17
DOI
https://doi.org/10.1111/1475-6773.70180
Primary Topic
Advanced Causal Inference Techniques
Type
article
Field-Weighted Citation Impact
0.00
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
article

Methods to Account for Competing Events With Unknown Duration in Health Policy Evaluations: The Case of COVID ‐19

Marisa Elena Domino, Caleb Easterly, Mark Holmes, Justin G. Trogdon
Health Services Research
Advanced Causal Inference Techniques
article

Methods to Account for Competing Events With Unknown Duration in Health Policy Evaluations: The Case of COVID ‐19

Marisa Elena Domino, Caleb Easterly, Mark Holmes, Justin G. Trogdon
article en

Abstract

OBJECTIVE: To compare methods for estimating policy effects in the presence of competing events such as COVID-19. STUDY SETTING AND DESIGN: We created a simulation study to evaluate how single interrupted time series (ITS) and comparative interrupted time series (CITS) estimators performed when policy implementation occurred near a competing event (COVID-19). The onset of the competing event was known (March 2020) but the duration was unknown. ITS approaches included assuming different COVID-19 durations (ITS-Assume) or searching for the COVID-19 duration that provided the best fit to the data (ITS-Grid Search). Outcomes included percent bias, power, and Type 1 error. DATA SOURCES AND ANALYTIC SAMPLE: We varied the duration of COVID-19 effects, the time between the policy implementation and COVID-19 onset, and the magnitude and sign of the policy effects relative to COVID-19 (2080 scenarios). Each simulated time series had 99 monthly observations (October 2015-December 2023). PRINCIPAL FINDINGS: CITS had the lowest bias and Type 1 error and had high power across scenarios, except for intercept changes and when assumptions about shared COVID-19 effects between treatment and control groups were violated. ITS-Grid Search had lower bias and Type 1 error than ITS-Assume, which produced biased parameter estimates when the assumptions were incorrect. Estimates were more accurate when the policy was implemented prior to the competing event. CONCLUSIONS: Using estimation approaches that leverage control groups, such as CITS, can mitigate the potential bias of competing events in ITS frameworks, but these approaches are biased when competing event effects differ between treatment and control groups. Searching for the best-fitting competing event duration does not require a control group and outperforms methods that assume the duration of competing event effects a priori. More generally, researchers should be cautious about bias, specification error, and low power when using ITS methods with competing events.

Health Services ResearchVol. 61(6)
University of North Carolina at Chapel Hill (US), University of North Carolina Health Care (US)
Advanced Causal Inference Techniques
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon