Geospatial estimation of reproductive, maternal, newborn and child health indicators: a systematic review of methodological aspects of studies based on household surveys

Table 1 Summary of the characteristics for the selected studies

	All studies	Study outcomes^a
	All studies	Malaria	Child mortality	Malnutrition	Vaccination	Other outcomes
Number of studies	82	34	14	11	8	19
Covariates^a
Agriculture and livestock	17	3	3	5	4	5
Climate	43	28	5	5	4	5
Health-related interventions/outcomes	24	9	5	5	1	7
Remoteness	43	19	7	5	5	11
Satellite indices	19	4	5	3	4	7
Sociodemographic	53	17	6	11	6	17
Topography/land cover	59	30	7	10	4	12
No covariates	15	4	7	1	2	2
Geographic coverage
Single country	50	26	6	7	2	9
Multi-country	32	8	8	5	6	10
Temporal component
No	46	19	2	8	7	11
Yes	36	15	12	4	1	8
Spatial resolution^a
Less than 5x5km	23	18	0	1	3	2
5x5 to 10x10km	20	6	5	5	3	4
Lower admin. level	30	2	8	6	1	14
Not reported	12	10	1	1	0	0
Uncertainty^a
Standard deviation map	14	6	1	1	5	2
Interval map/table	28	12	8	2	0	10
Relative map	7	0	2	3	0	2
Other metrics	13	9	2	0	1	1
Not reported	22	7	3	6	2	4
Modeling technique^a
Bayesian–MCMC	35	24	3	3	3	2
Bayesian–INLA	28	4	7	6	3	12
Classical GLM	17	5	2	2	2	6
Spatial interpolation	2	0	1	1	0	0
Ensemble models	12	1	5	4	1	5
Out-of-sample pred.
Cross-validation	22	3	7	5	4	6
Hold-out	24	18	2	1	0	4
Not reported	36	13	5	6	4	9
Model fit metrics^a
Bias	34	12	7	6	4	9
RMSE/MSE	30	3	7	6	6	12
Coverage	24	8	6	4	4	5
DIC/AIC	19	6	3	3	1	6
MAE	16	7	2	3	2	3
Correlation	15	11	0	2	1	2
Other metrics	31	15	4	3	1	9
None reported	11	5	2	3	1	1

^aThese characteristics allow studies to be classified in more than one subgroup
MCMC Markov Chain Monte Carlo, INLA Integrated Nested Laplace Approximation, GLM Generalized Linear Models, RMSE Root Mean Squared Error, MSE Mean Squared Error, DIC Deviance Information Criterion, AIC Akaike Information Criterion, MAE Mean Absolute Error

ISSN: 1476-072X