Can Routine Health Management Information Systems Serve as Epidemic Early Warning Platforms? A Systematic Review and Meta-Analysis of Evidence from Low- and Middle-Income Countries

Abstract

Moses Luke, Folajinmi Oluwasina, Shaibu Joseph, Jacob Ojedokun, Chibuike Ogbonnaya, Babade Ojo, Temitope Olorunmonu and Achor Olaitan

Background: Routine Health Management Information Systems (HMIS), particularly DHIS2-based platforms deployed across over 73 countries, represent the most comprehensive, geographically granular, and temporally continuous data infrastructure available to national health systems in low- and middle-income countries (LMICs). Whether these systems can reliably detect epidemic signals early enough to trigger timely public health responses has been widely debated but never systematically quantified. Answering this question is urgent: the COVID-19 pandemic exposed catastrophic early warning failures globally, the International Health Regulations (IHR) 2005 mandate national surveillance capacity, and the simultaneous scale-up of DHIS2 digital platforms creates an unprecedented opportunity to repurpose routine HMIS for epidemic intelligence, if the evidence supports it.

Methods: We searched PubMed/MEDLINE, Embase, WHO IRIS, CINAHL, and Cochrane Library (January 2005–December 2025), supplemented by grey literature and WHO AFRO IDSR programme databases. Eligible studies used HMIS data from LMICs as the primary epidemic detection source, reporting diagnostic accuracy against validated reference standards. Four independent reviewers screened records, extracted data, and assessed quality using QUADAS-2. Pooled sensitivity and specificity were estimated using bivariate random-effects models, with subgroup analyses by disease type, platform, WHO region, and study setting.

Results: Forty-seven studies (38 in meta-analysis) from 34 LMICs across five WHO regions were included. Pooled sensitivity of HMIS-based epidemic detection was 0.68 (95% CI: 0.58–0.77; I2=71.3%), and pooled specificity was 0.74 (95% CI: 0.65–0.83; I2=68.4%). Fully digital DHIS2 platforms demonstrated the highest sensitivity (0.74; 95% CI: 0.65–0.82), compared to paper- based or hybrid HMIS (0.59; 0.48–0.70). Detection performance was highest for malaria (sensitivity: 0.74) and cholera/AWD (0.76), and lowest for viral haemorrhagic fevers (0.48) and meningitis (0.55). A significant subgroup difference by disease type was observed (χ2=18.6, df=5, p=0.002). Pooled median time-to-alert from HMIS platforms was 8.4 days earlier (95% CI: 5.2–11.6 days) than traditional passive surveillance comparators in studies reporting this metric.

Conclusions: Routine HMIS can function as a viable, though imperfect, epidemic early warning platform, with performance moderated by digitisation level, disease type, and algorithm quality. Prioritising disease-specific alert algorithms, real- time DHIS2 integration, and HMIS-IHR interoperability represents the evidence-based pathway for strengthening epidemic preparedness in LMICs.

PDF