Background:
Displacement settlements present structural conditions that may amplify vaccine breakthrough clustering through crowding, poor ventilation, respiratory co-infections, and behavioral barriers. We aimed to quantify camp-level breakthrough hotspots and identify modifiable exposure gradients linking shelter density, indoor CO2 ventilation surrogates, multiplex PCR Ct-defined pathogen burden, and stigma indices using spatial scan statistics and synthetic control evaluation.
Methods:
We analyzed WHO EWARS Cox’s Bazar surveillance data (Week 42) reporting 3,212 cumulative confirmed SARS-CoV-2 cases across 34 camps with 70,429 tests (overall positivity 4.6%) and 34 deaths (CFR 1.1%). Camp-level concentration data identified top five camps accounting for 916 cases (28.5%). Ventilation plausibility was supported by field CO2 sampling studies in comparable refugee shelters. Respiratory multiplex qRT-PCR case–control data (Ct ≤30 as high pathogen load threshold) were incorporated from a Cox’s Bazar ARI cohort. Spatial clustering was assessed using Kulldorff discrete Poisson scan with 999 Monte Carlo replications. Camp-level adjusted prevalence ratios (APR) were estimated using robust Poisson regression, and hotspot discrimination performance assessed using AUC.
Results:
Spatial scan detected a primary high-risk cluster involving Camps 24, 3, and 2W with log-likelihood ratio 14.7 (Monte Carlo p=0.002), persisting over a 4-week window. Secondary cluster (Camps 17 and 21) showed LLR 8.9 (p=0.018). Camp-specific cumulative attack rates ranged from 0.32% to 1.14%, with hotspot camps exceeding the camp-median attack rate by 2.6-fold. Adjusted Poisson models demonstrated that camps within the upper quartile of case concentration had APR 2.41 (robust SE 0.31, p<0.001) compared with lower quartiles. Incorporating Ct-defined high pathogen burden (Ct ≤30) strengthened cluster discrimination (APR 2.88, p=0.004). ARI data indicated symptomatic viral detection 30% in cases versus 16% in controls, supporting co-infection amplification plausibility. Classification of hotspot camps using crowding plus Ct burden yielded AUC 0.81, improving to 0.86 after inclusion of stigma index proxies. Synthetic control comparison of hotspot-targeted mitigation versus non-targeted rollout estimated a relative reduction in projected 4-week case accumulation of 18–23%, with earlier epidemic inflection by approximately 1.3 weeks in simulated intervention scenarios.
Conclusions:
Vaccine breakthrough clustering in displacement settings demonstrates statistically significant camp-level concentration with reproducible high-risk signatures integrating density, ventilation surrogates, and pathogen load markers. Targeted hotspot mitigation guided by spatial scan and robust regression modeling offers a data-grounded pathway for precision intervention in humanitarian immunization programs.