AI Summary & Key Takeaways
AI InsightBecause daily outpatient record counts may reflect healthcare utilization, healthcare-seeking behavior, and service availability in addition to potential variation in disease activity, these findings should be interpreted as population level associations with record-based healthcare utilization rather than as direct measures of individual disease severity, individual visit frequency, or causal effects.
📌 3 Key Takeaways
Abstract
Aims: This study aimed to evaluate the association between seborrheic dermatitis (SD) outpatient record counts and meteorological factors using 16 years of daily-resolution data and to investigate the non-linear and lagged association between temperature and daily outpatient record counts using a distributed lag non-linear model (DLNM). Methods: A total of 68.518 International Classification of Diseases, Tenth Revision (ICD-10) L21-coded outpatient record entries from a tertiary dermatology center between January 2010 and December 2025 were analyzed using a retrospective ecological time-series design. The unit of analysis was the calendar day. Weekend days were excluded from the primary analysis because of limited routine outpatient service availability. The weekday time series comprised 4.174 calendar days, including 216 zero-count days. The primary fully adjusted negative binomial regression analysis included 3.866 weekday observations with complete meteorological data, including 202 zero-count days. Negative binomial regression was used to account for overdispersion, and heteroskedasticity-and autocorrelation-consistent (HAC) standard errors were applied to provide robust statistical inference in the presence of residual temporal autocorrelation. Models were adjusted for day of the week, long-term temporal trend, and meteorological variables. DLNM analysis was performed on 3.819 weekday observations to evaluate the cumulative effects of temperature over a 0-7-day lag period. The exposure-response and lag-response dimensions were modeled using B-spline functions with four degrees of freedom each. Results: Across all eligible weekday calendar days, the median daily SD outpatient record count was 13 [interquartile range (IQR), 8-19]. Daily outpatient record counts showed marked seasonal variation, with a mean of 17.3±9.0 in winter and 10.4±6.6 in summer. In the fully adjusted negative binomial model, each 1°C increase in mean daily temperature was associated with a 1.7% decrease in daily outpatient record counts (IRR, 0.983; 95% CI, 0.979-0.987; p<0.001). In the DLNM analysis, compared with the median temperature (15.1°C), low temperatures (10th percentile, 2.1°C) were associated with a cumulative relative risk (RR) of 1.285 (95% CI, 1.170-1.411) over lag days 0-7, whereas high temperatures (90th percentile, 28.9°C) were associated with a cumulative RR of 0.722 (95% CI, 0.650-0.803). Conclusion: SD outpatient record counts exhibited a distinct seasonal pattern. Lower temperatures were associated with increased daily outpatient record counts, whereas higher temperatures were associated with decreased record counts in a non linear and lagged manner. Because daily outpatient record counts may reflect healthcare utilization, healthcare-seeking behavior, and service availability in addition to potential variation in disease activity, these findings should be interpreted as population level associations with record-based healthcare utilization rather than as direct measures of individual disease severity, individual visit frequency, or causal effects.