Table 1

Summary of geospatial analysis tools and related objectives and data used as part of the study’s methodology

Geospatial toolsObjective and detailsData used
1.Inverse distance weighted (IDW) interpolationEstimate the geographical distribution of paediatric surgical conditions stratified by age and time of delay to receive surgical care.40 Manual breaks were used.SOSAS survey: prevalence of paediatric surgical conditions (n=221) among 196 children.
2.Hotspot analysisIdentify statistically significant (90%, 95% and 99%) clusters of untreated surgical conditions (also called ‘unmet surgical care’). Areas of most vulnerability and highest priority for future interventions regarding age and time of delay were depicted as red bullets.SOSAS survey: prevalence of untreated surgical conditions (n=168) among 196 children.
3.Service area—Network AnalystEvaluate geographical accessibility to surgical care based on the time needed to reach surgically capable and bellwether hospitals. Optimal accessibility was defined as a catchment area within a 2-hour distance, as suggested by the LCoGS.3 Distances calculated for travelling by public transportation and by foot were based on the national road network instead of a straight-line distance (Euclidean distance). Distances were broken down into six categories, including ≤2 hours (optimal), 2–6 hours, 6–12 hours, 12–14 hours, 1–2 days, and ≥2 days. We assumed a constant speed limit of 30 km/hour for public transportation and 5 km/hour for travel on foot. The child population inside a 2 hour distance (called ‘catchment population’) was estimated in a two-step process. First, the child population density was calculated at a regional level. Second, the regional child population density was multiplied by the area (km2) corresponding to ≤2 hours in the service area map. Estimates at the national level were calculated by using the national population density.Hospital assessment for 15 healthcare facilities.
Road network41
4.Voronoi diagramsEvaluate the crude area of coverage (km2) for surgically capable and bellwether hospitals. We assumed that patients travel to the closest hospital. If no hospital was available within the region, then we assumed that patients would travel to the closest hospital from a neighbouring region. The hospital area of coverage (km2) at the regional level was calculated as the average of the polygons within each region. A polygon was counted for a specific region only if its related hospital point was inside that region (online supplemental material 4).Hospital assessment for 15 healthcare facilities.
  • All analyses were performed by using ArcMap packages.

  • LCoGS, Lancet Commission on Global Surgery; SOSAS, Surgeons OverSeas Assessment of Surgical Need.