Skip to main content

Advertisement

Table 2 Characteristics of the included papers

From: Spatial analysis of cluster randomised trials: a systematic review of analysis methods

Author and year Location Intervention Outcome Map Spatial method type Spatial method
Binka 1998 [22] Ghana Permethrin-impregnated bed net All-cause child mortality Yes Straight line distance Distance to discordant observation and points of interest. Standardised mortality rates calculated at several distances
Alexander 2003 [27] Papua New Guinea Diethylcarbamazine (DEC) plus ivermectin versus DEC alone Spatial distribution of Wuchereria bancrofti and microfilaraie Yes Spatially structured random effect Negative binomial model with a distance parameter in the covariance structure of a random effect. Measures half distance of spatial correlation
Gimnig 2003 [19] Kenya Permethrin-treated bed net Spatial distribution of mosquitoes Yes Straight line distance Distance to discordant observation and to points of interest. Poisson regression model with a random effect for cluster
Hawley 2003 [18] Kenya Permethrin-treated bed net All-cause child mortality, anaemia, and density of mosquitoes Yes Straight line distance Distance to discordant observation and to points of interest. Cox regression model with a random effect for adjusted for cluster
Miguel and Kremer 2004 [23] Kenya Deworming Helminth infection No Density Total number and proportion of treated students within 6 km of a school. Included in primary analysis random effects model adjusted for cluster (School)
Kroeger 2006 [20] (1) Mexico (2) Venezuela Insecticide-treated curtains and water container covers Reduction in entomological indices for Dengue No Straight line distance Distance to nearest participant with outcome at the beginning of the study Odds ratio of outcome for nearby houses compared to houses further away
Ali 2007 [21] Vietnam Vaccine campaign Vaccine uptake Yes Straight line distance and density Distance to points of interest and density. Random effects mode including typhoid prevalence and private practitioner density
Lenhart 2008 [24] Haiti Insecticide-treated bed nets Reduction in entomological indices for Dengue Yes Density Total number of bed net households within 100 m. Spearman’s correlation of number of bed net houses within 100 m compared with change in entomological measures
Silcocks 2010 [29] UK (3 trials) (1) Sun exposure, (2) home safety intervention, (3) intervention to reduce baby walker use (1) Lip cancer, (2) number of injuries per individual, (3) ownership of baby walker No Spatially structured random effect Spatial weights matrix in covariance of random effect. A multiple membership spatial random effects model with fixed North/South or East/West gradient covariate effect for gradient
Chao 2015 [25] India Typhoid vaccine Vaccine effectiveness Yes Density The sum of the risk of those within 100 m of a participant called the potential exposure. Included in a model with a random effect for cluster