Estimating Vaccine Coverage in Conflict Settings Using Geospatial Methods: A Case Study in Borno State, Nigeria

"Globally comparable estimates across space and time are critical for understanding geographies lagging in progress to avoid leaving vulnerable children behind."

Ongoing conflict, terrorism, and political instability persisting in northern and northeastern portions of Nigeria, particularly Borno, present barriers to reaching children with vaccination programmes. The Equity Reference Group for Immunization emphasises the importance of understanding vaccination coverage in conflict settings, while acknowledging that assessing immunisation status in these settings is extremely challenging. Leveraging complex relationships of outcomes and covariates across space and time, model-based geostatistical (MBG) methods can provide alternative vaccine coverage estimates for unsafe and insecure areas where, for example, there is an inability to sample, as well as increased uncertainty in underlying population estimates. This paper assesses the use of geostatistical models to estimate vaccination coverage in conflict settings and considers the use of model-based estimates alongside survey estimates as an alternative tool for assessing coverage in such settings.

The researchers estimated first- and third-dose diphtheria-tetanus-pertussis (DTP) vaccine coverage in Borno state, Nigeria, using a spatiotemporal MBG modelling approach. In brief, this model produces spatiotemporal estimates of dose-specific DTP coverage under the assumption that coverage is similar in location-years that have similar covariate patterns and that are close together in space and time. They then compared these data to estimates from recent conflict-affected, household-based surveys. They compared sampling cluster locations from recent household-based surveys to geolocated data on conflict locations and modelled spatial coverage estimates while also investigating the importance of reliable population estimates when assessing coverage in conflict settings.

In brief, geospatial population estimates suggest that a substantial proportion of children under 5 years of age in Borno state live in local government areas (LGAs) that were not represented in the 2016-17 Multiple Indicator Cluster Survey (MICS/NICS) and 2018 Demographic and Health Survey (DHS). Modelled estimates in Borno state suggest substantially lower coverage for both DTP1 and DTP3 than reported by the most recent available surveys. This pattern was more consistently observed in Borno state than in other states with representative sampling. In short, while no model of coverage in conflict settings is a substitute for reliable data, "Model-based geostatistical estimates infer coverage in unsampled areas using historical trends, nearby data, and spatial covariates, which provide an additional tool when investigating coverage in conflict-affected areas."

In discussing the findings, the researchers point to past studies documenting substantial conflict-associated service disruptions in Borno state, "suggesting that the lower estimates from geospatial modelling - driven by low estimated coverage in unsampled LGAs - are plausible, particularly as conflict, low coverage, and sampling limitations are likely to be highly correlated....For example, ...conflict-affected areas are likely to experience delays or difficulties in accessing care, destruction of health care infrastructure, higher case fatality rates, particular challenges during outbreak control activities, and higher levels of poverty....Besides conflict, other factors likely contribute to low coverage..., including...lack of availability of immunisation services at facilities and increasing vaccine hesitancy as a result of misconceptions and low health literacy in the region..."

In conclusion: "Understanding vaccination coverage in conflict settings is an equity priority. To reach every child with immunisation services, reliable estimates of vaccination coverage in conflict locations are needed. Innovative model-based coverage estimates, the expansion of new survey methodologies, and reliable population estimates will all be critical tools to ensure that vulnerable children living in conflict-affected areas will be protected against preventable disease and death."

Full list of authors, with institutional affiliations: Alyssa N. Sbarra, University of Washington; Sam Rolfe, University of Washington; Emily Haeuser, University of Washington; Jason Q. Nguyen, University of Washington; Aishatu Adamu, Bayero University Kano and London School of Hygiene and Tropical Medicine; Daniel Adeyinka, University of Saskatchewan and Federal Ministry of Health (Abuja, Nigeria); Olufemi Ajumobi, University of Nevada Reno and Federal Ministry of Health (Abuja, Nigeria); Chisom Akunna, Federal Ministry of Health (Abuja, Nigeria) and the Intercountry Centre for Oral Health (ICOH) for Africa; Ganiyu Amusa, Jos University Teaching Hospital and University of Jos; Tukur Dahiru, Ahmadu Bello University; Michael Ekholuenetale, University of Ibadan; Christopher Esezobor, University of Lagos and Lagos University Teaching Hospital; Kayode Fowobaje, University of Ibadan and Centre for African Newborn Health and Nutrition; Simon I. Hay, University of Washington; Charles Ibeneme, Abia State Ministry of Health and African Field Epidemiology Network; Segun Emmanuel Ibitoye, University of Ibadan; Olayinka Ilesanmi, University of Ibadan and University College Hospital; Gbenga Kayode, Institute of Human Virology Nigeria and Utrecht University; Kris Krohn, University of Washington; Stephen S. Lim, University of Washington; Lyla E. Medeiros, University of Washington; Shafiu Mohammed, Ahmadu Bello University and Technical University of Berlin; Vincent Nwatah, National Hospital Abuja and University of Liverpool; Anselm Okoro, Society for Family Health; Andrew T. Olagunju, McMaster University and University of Lagos; Bolajoko O. Olusanya, Centre for Healthy Start Initiative; Osayomwanbo Osarenotor, University of Benin; Mayowa Owolabi, University of Ibadan and University College Hospital; Brandon Pickering, University of Washington; Mu'awiyyah Babale Sufiyan, Ahmadu Bello University; Benjamin Uzochukwu, University of Nigeria Nsukka; Ally Walker, University of Washington; and Jonathan F. Mosser, University of Washington