Statistical Analysis Plan (Preregistration): Do Factors Reported as Risk Factors for the Development of Post-COVID-19 Condition Predict Symptom-Specific Duration?
This record contains the preregistered statistical analysis plan (Version 1.1, November 12, 2026) and its Supplement S1 for a study developing and internally validating models that predict the symptom-specific duration of post-COVID-19 condition. Study population. The survey targeted 7,312 patients diagnosed with COVID-19 between March 2020 and June 2024 at designated medical institutions for Class II infectious diseases in Hiroshima Prefecture, Japan. The study population comprises the 2,689 respondents (36.8%; 1,524 adults and 1,165 children) to self-administered follow-up questionnaire surveys, linked with medical record information from the acute phase. Questions. The primary question is whether predictors prespecified on the basis of systematic reviews and meta-analyses of risk factors for the development of post-COVID-19 condition predict the duration of each of 13 symptoms, and whether their predictive performance differs across symptoms. Secondary questions address differences between adults and children, the added value of acute-phase laboratory values, and the incremental value of the established risk factors. The models will be implemented in a publicly available web tool. Methods. Predictors were prespecified through a structured PubMed search (Supplement S1); no data-driven variable selection will be performed. Symptom duration, recorded in intervals, is modeled with a discrete-time survival model with a complementary log-log link, combined with a logistic model for symptom occurrence. Performance will be assessed by bootstrap optimism correction. No external validation is performed. Information examined before registration. The survey data were collected before this plan was registered. Before the plan was finalized, only the marginal distributions of the predictors and the number of outcome events were examined (October 5, November 9, and November 12, 2026; details in Section 4 of the plan). The joint distribution of predictors and outcomes was not examined. Files. (1) Statistical analysis plan; (2) Supplement S1: procedure for identifying the literature used to prespecify the predictors, and factor tables.
Authors
- Aya Sugiyama (ORCID: https://orcid.org/0000-0002-8087-4319)
Publication Details
- Journal
- Zenodo (CERN European Organization for Nuclear Research)
- Published
- 2026-11-12
- DOI
- https://doi.org/10.5281/zenodo.24152419
- Primary Topic
- Long-Term Effects of COVID-19
- Type
- article
- Field-Weighted Citation Impact
- 0.00