Data and R code for: Adjunctive hypnotic prescribing and complex sleep behaviors in patients continuing zolpidem or eszopiclone

This deposit provides a public participant-level analysis dataset with age and sex withheld, together with R code for the manuscript "Adjunctive hypnotic prescribing and complex sleep behaviors in patients continuing zolpidem or eszopiclone: A secondary cross-sectional analysis." PUBLIC DATA SCOPE AND REPRODUCIBILITY: Participant-level age and sex are withheld for privacy. The manuscript retains its original age- and sex-adjusted analyses; these cannot be fully reproduced from the public dataset. The complete research dataset and original 17-model workflow remain in non-public institutional research records. Public scripts reproduce prescribing counts, crude associations, symptom prevalence and frequency tests, and seven Firth models that do not use the withheld covariates. They do not impute the omitted covariates or substitute unadjusted estimates for the manuscript's adjusted results. Age/sex summary outputs and logs from models using these covariates are also omitted from this archive. The source cohort consists of 157 psychiatric outpatients from Nippon Medical School Chiba Hokusoh Hospital and Nippon Medical School Hospital, Japan. Questionnaires were obtained during routine clinical care in April-June 2022. The present study is a post hoc secondary cross-sectional analysis of this previously reported cohort (Shimoda et al., Journal of Nippon Medical School, 2026;93:153-160; https://doi.org/10.1272/jnms.JNMS.2026_93-209). The primary analysis includes 133 participants continuing zolpidem or eszopiclone and evaluates the association between dual orexin receptor antagonist (DORA) coprescription and additional benzodiazepine hypnotic prescribing. Complex sleep behavior (CSB) presence and preceding-month frequency are secondary outcomes. Nested sensitivity populations include 137 participants after adding zopiclone users and all 157 participants after also adding triazolam users. A separate exploratory comparison of 146 participants examines mutually exclusive prescribing groups across all four index drugs. This subset is obtained from 157 participants by excluding six with ramelteon as the sole adjunct, four with combined DORA and additional benzodiazepine use, and one with combined DORA and ramelteon use. The resulting groups contain 66 participants on the index drug alone, 50 with additional benzodiazepine hypnotics only, and 30 with a DORA as the sole adjunct. These 30 DORA users comprise 26 zolpidem/eszopiclone users shared with the primary cohort and four triazolam users. In contrast, the primary cohort's 30 DORA users comprise the same 26 participants plus three with additional benzodiazepine use and one with ramelteon use. The two 30-person DORA groups are therefore not identical. Zolpidem or eszopiclone preceded adjunctive treatment and remained the principal hypnotic in the primary cohort. Individual adjunct initiation dates, treatment durations, and ordering among adjuncts were unavailable. The dataset supports cross-sectional associations and does not establish preferential second-line selection, causal benzodiazepine reduction, or CSB onset after adjunct initiation. The archive contains a 157-row, 10-variable CSV dataset; a data dictionary; R scripts for the prescribing analyses, supplementary regimen analyses, and figures; results and execution logs for analyses reproducible from the public variables; and a README. Original record identifiers, original row order, and non-analysis clinical fields are omitted. Institutional permission for public release of the de-identified participant-level dataset has been confirmed. CSB-frequency values are structurally blank for participants without CSB. The source-coded maximum-dose category is not a measure of total sedative burden or age-specific prescribing advice. Analyses were verified using R 4.6.1 and logistf 1.26.1. Firth logistic-regression estimates use profile penalized-likelihood confidence intervals and penalized likelihood-ratio tests. Instructions for reproducing the analyses are included in the README. This deposit corresponds to manuscript revision 3 dated 27 September 2026. Funding: This research received no specific funding.