Work

Among rural older adults, an apparent physical-activity contrast in nonblank medication-field presence was markedly attenuated when chronic-disease management registration was incorporated into the analysis

Year: 2026

Type: article

Abstract: Background: Medication fields created for chronic-disease management may mislead when reused as medication-exposure measures in broader older-adult populations. Using routine exercise-frequency as routine-recorded physical activity (PA), we compared PA + records (daily, at least weekly, or occasional exercise) with PA- records (no exercise) and examined how nonblank medication-field presence changed after management-registration context was incorporated. Methods: This cross-sectional study used 2,270 examination records from a 2025 rural primary-care program in northeastern China; PA was available for 2,226 records. The outcome was any nonblank text in the main-medication field. Modified Poisson regression estimated prevalence ratios (PRs) with small-sample-adjusted cluster-robust inference across 20 sites. Sequential models incorporated demographics, site, and four hypertension/diabetes management-registration categories. Logistic regression estimated standardized proportions and absolute differences. Medication text underwent secondary audit, including 200 randomly sampled retained entries independently reviewed by two pharmacists. Results: Nonblank medication fields occurred in 1,114/1,157 (96.3%) records with hypertension and/or diabetes management registration versus 27/1,113 (2.4%) with neither registration. Field presence was low without registration and near-complete with hypertension-only or dual registration; the diabetes-only stratum was small (n = 37). Overall field presence was 56.4% in PA+ and 42.8% in PA- records (PR 1.317, 95% CI 1.076-1.612). After registration context was incorporated, the PR was 1.001 (95% CI 0.960-1.043); full adjustment yielded 1.002 (95% CI 0.959-1.046). Standardized proportions were 51.26% and 51.14%, a difference of 0.12 percentage points (95% CI -2.07 to 2.30). Among 1,368 retained entries, archived labels classified 75.9% as antihypertensive, 14.5% as glucose-lowering, 0.5% as other cardiovascular, 1.7% as other medications, and 7.4% as unresolved. Inter-reviewer agreement for broad therapeutic class was 99.5% (κ = 0.982), and parser-manual-reference concordance was 99.0%; these assessments did not verify medication exposure. Conclusion: Nonblank main-medication-field recording was strongly patterned by chronic-disease management registration, and the apparent overall PA-related contrast was markedly attenuated after this context was incorporated. Blank fields cannot establish medication nonuse, and nonblank or consistently parsed text cannot establish complete, current medication exposure.

Source: Frontiers in Pharmacology

DOI: 10.3389/fphar.2026.1965382