Answers are grounded in this paper and cite their sources.
Long-term subnational tuberculosis (TB) surveillance can be biased when administrative boundaries and population denominators change over time. We examined temporal, geographic and clinical patterns of TB in Khuzestan, Iran, using a harmonised framework designed to preserve numerator-denominator comparability across a 16-year surveillance period. This retrospective registry-based observational study included 9,090 TB registration episodes accumulated over 16 surveillance years (2005–06 to 2020–21), not verified unique individuals. Temporal and spatial analyses used 6,047 residence-based new-and-relapse episodes in a fixed 10-unit domain; treatment-outcome and diagnostic-delay analyses included 8,718 and 9,058 episodes, respectively. Negative-binomial population-offset models estimated notification trends, and pooled grid analyses assessed spatial autocorrelation. Modified Poisson and median regression assessed episode-level outcomes and delay. The model-standardised age-adjusted notification rate declined by 2.67% per year (95% CI − 3.55 to − 1.77). Pooled crude notification rates across the 20 historical county groups ranged from 1.77 to 22.64 per 100,000 person-years. Global spatial autocorrelation in the primary 10-km grid was modest (Moran’s I = 0.0648; permutation P = 0.0344), and none of 432 local tests remained significant after false-discovery-rate correction. Adverse outcomes occurred in 944/8,718 eligible episodes (10.8%). Median diagnostic delay among 9,058 nonnegative intervals was 55 days (IQR 30–94), with a 95th percentile of 322 days. Notifications declined within the historically comparable 10-unit domain, alongside geographic heterogeneity and substantial diagnostic delay. Explicit geographic and cohort harmonisation links complementary surveillance indicators while preserving their different denominators. No local cluster remained supported after false-discovery-rate correction. The findings may inform programme review and hypothesis generation but do not establish province-wide incidence trends or causal effects.
No takes yet. Share an insight, caveat, or question.
Zadeh et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: