@article{PAR00015229,
  title = {{I}mpact of socioeconomic inequalities on geographic disparities in cancer incidence : comparison of methods for spatial disease mapping},
  author = {{G}oungounga, {J}. {A}. and {G}audart, {J}. and {C}olonna, {M}. and {G}iorgi, {R}och},
  editor = {},
  language = {{ENG}},
  abstract = {{B}ackground: {T}he reliability of spatial statistics is often put into question because real spatial variations may not be found, especially in heterogeneous areas. {O}ur objective was to compare empirically different cluster detection methods. {W}e assessed their ability to find spatial clusters of cancer cases and evaluated the impact of the socioeconomic status (e.g., the {T}ownsend index) on cancer incidence. {M}ethods: {M}oran's {I}, the empirical {B}ayes index ({EBI}), and {P}otthoff-{W}hittinghill test were used to investigate the general clustering. {T}he local cluster detection methods were: i) the spatial oblique decision tree ({S}p{ODT}); ii) the spatial scan statistic of {K}ulldorff ({S}a{TS}can); and, iii) the hierarchical {B}ayesian spatial modeling ({HBSM}) in a univariate and multivariate setting. {T}hese methods were used with and without introducing the {T}ownsend index of socioeconomic deprivation known to be related to the distribution of cancer incidence. {I}ncidence data stemmed from the {C}ancer {R}egistry of {I}sere and were limited to prostate, lung, colon-rectum, and bladder cancers diagnosed between 1999 and 2007 in men only. {R}esults: {T}he study found a spatial heterogeneity (p < 0.01) and an autocorrelation for prostate ({EBI} = 0.02; p = 0.001), lung ({EBI} = 0.01; p = 0.019) and bladder ({EBI} = 0.007; p = 0.05) cancers. {A}fter introduction of the {T}ownsend index, {S}a{TS}can failed in finding cancers clusters. {T}his introduction changed the results obtained with the other methods. {S}p{ODT} identified five spatial classes (p < 0.05): four in the {W}estern and one in the {N}orthern parts of the study area (standardized incidence ratios: 1.68, 1.39, 1.14, 1.12, and 1.16, respectively). {I}n the univariate setting, the {B}ayesian smoothing method found the same clusters as the two other methods ({RR} > 1.2). {T}he multivariate {HBSM} found a spatial correlation between lung and bladder cancers (r = 0.6). {C}onclusions: {I}n spatial analysis of cancer incidence, {S}p{ODT} and {HBSM} may be used not only for cluster detection but also for searching for confounding or etiological factors in small areas. {M}oreover, the multivariate {HBSM} offers a flexible and meaningful modeling of spatial variations; it shows plausible previously unknown associations between various cancers.},
  keywords = {{S}patial analysis ; {C}luster detection ; {C}ancer ; {O}blique decision tree},
  booktitle = {},
  journal = {{BMC} {M}edical {R}esearch {M}ethodology},
  volume = {16},
  numero = {},
  pages = {art. 136},
  ISSN = {1471-2288},
  year = {2016},
  DOI = {10.1186/s12874-016-0228-x},
  URL = {https://www.documentation.ird.fr/hor/{PAR}00015229},
}