Epidemiological Trends, Statistical Correlations, and Forecasting of Notifiable Infectious Diseases in South Korea,...

Notifiable infectious diseases impose a substantial, evolving burden on public health systems. This study characterized long-term epidemiological trends, inter-class associations, and age-stratified patterns of notifiable infectious diseases in South Korea and projected future burden through 2028.

We analyzed aggregate national surveillance data (2001-2024) from the Korea Disease Control and Prevention Agency (KDCA) via the Korean Statistical Information Service (KOSIS). Because the statutory classification changed in 2020 from a four-group (Class 1-4, gun) to a four-grade (geup) system, all inferential trend, correlation, and forecasting analyses were restricted to the internally consistent 2001-2019 group-system period, and 2020-2024 grade-system data were used descriptively. Secular trends were estimated with ordinary least squares and, for count-specific inference, negative-binomial log-linear models (annual percent change, APC); residual autocorrelation was assessed (Durbin-Watson, Ljung-Box). Structural breaks were estimated objectively (Bai-Perron least-squares). Inter-class associations were tested with Pearson/Spearman correlations and re-evaluated with partial correlation (controlling for calendar year) and first-difference detrending to guard against spurious co-trending. Forecasts used a random-walk ARIMA (identified by ADF stationarity testing, ACF/PACF, and AICc) and a damped Holt-Winters model on the H1N1-adjusted series, combined into an ensemble; performance was validated by rolling-origin cross-validation (RMSE, MAE, MAPE).

After handling the 2009 H1N1 pandemic outlier, total notifications rose significantly over 2001-2019 (slope = 8344 cases/year; 95% CI 6603-10,084; R² = 0.866; p < 0.001). Class 2 (respiratory/vaccine-preventable) diseases showed the strongest trend (APC = 17.2%/year; 95% CI 8.0-27.3), with an objectively estimated structural break at 2004 (supF = 10.5) separating an early variable phase from sustained growth (2005-2019: +7418 cases/year; R² = 0.94). Raw Class 1-Class 2 correlation (Pearson r = 0.569, p = 0.011) did not survive temporal adjustment (partial r = 0.13, p = 0.59; detrended r = -0.31, p = 0.21), indicating shared secular co-trending rather than a direct epidemiological association; only Class 2-Class 3 remained associated after detrending (r = 0.52, p = 0.03). Young working-age adults (30-49 years) showed the steepest Class 1 increases. Ensemble projections for the pre-COVID trajectory were 238,000 (2025) to 269,000 (2028) annual notifications, with wide prediction intervals reflecting substantial uncertainty.

South Korean notifiable-disease epidemiology was dominated by a genuine upward Class 2 respiratory trend; several previously reported inter-class correlations were attributable to common temporal trends rather than shared transmission. Post-COVID normalization monitoring and respiratory-disease surveillance investment are priority actions.