The uses of spatial analysis in medical geography: a review

Soc Sci Med. 1986;23(10):963-73. doi: 10.1016/0277-9536(86)90253-4.


This paper is a review of how geographers and others have used spatial analysis to study disease and health care delivery patterns. Point, line, area and surface patterns, as well as map comparisons and relative spaces are discussed. Problems encountered in applying spatial analytic techniques in medical geography are pointed out. The paper is intended to stimulate discussion about where medical geography can and should go in this area of study. Point pattern techniques include standard distance, standard deviational ellipses, gradient analysis and space and space-time clustering. Line methods include random walks, vectors and graph theory or network analysis. Under areas, location quotients, standardized mortality ratios, Poisson probabilities, space and space-time clustering, autocorrelation measures and hierarchical clustering are discussed. Surface techniques mentioned are isolines and trend surfaces. For map comparisons, Lorenz curves, coefficients of areal correspondence and correlation coefficients have been used. Case-control matching, acquaintance networks, multidimensional scaling and cluster analysis are examples of methods that are based on relative or non-metric spaces. The review gives rise to the discussion of several general points: problems encountered in spatial analysis, theory building and verification, the appropriate role of technique and computer use. Some suggestions are made for further use of spatial analytic techniques in medical geography: Monte Carlo simulation of point patterns, network analysis to study referral systems and health care for pastoralists, geographic information systems to assess environmental risk, difference mapping for disease and risk factor map comparisons and multidimensional scaling to measure social distance.

MeSH terms

  • Cross-Sectional Studies
  • Delivery of Health Care / trends*
  • Humans
  • Monte Carlo Method
  • Morbidity*
  • Mortality*
  • Risk
  • Space-Time Clustering