A practical measure of health facility efficiency: an innovation in the application of routine health information to determine health worker productivity in Ethiopia

Table 2 Determinants of the productivity estimated by multilevel linear mixed-effects models

Covariates	Unadjusted model	Null model	Model 1	Model 2	Model 3
Covariates	Est	Est	Est	Est	Est
Fix effects
Intrinsic contextual covariate: provider-mix
HO equivalent clinical staff^a	− 0.44^***		− 0.25^***	− 0.26^***	− 0.25^***
Para-clinical staff^a	− 0.16^***		− 0.04	− 0.04	− 0.04
Admin staff^a	− 0.26^***		− 0.23^***	− 0.22^***	− 0.22^***
Intrinsic contextual covariate: facility size
Number of beds in health centers^b	1.003			1.004	1.005^**
Extrinsic contextual covariate
Woreda population (per 10,000)^b	1.004^**				1.004^***
Region (Ref: Oromiya)^b
Addis Ababa	0.67^***				0.98
SNNPR	0.79^***				0.85^***
Constant		0.85^***	1.32^***	1.29^***	1.28^***
Random effects
Zone-level residual variance		0.02	0.02	0.02	0.008
Zone-level ICC		0.28	0.25	0.25	0.16
Woreda-level variance		0.02	0.01	0.01	0.01
Woreda-level ICC		0.51	0.44	0.44	0.36
Observations	1128	1128	1128	1128	1128
AIC		55.78	− 232.26	− 233.63	− 255.90
Log-likelihood		− 23.89	123.13	124.81	138.95

The analytical sample is consisting of 1128 health centers, from 369 nested within 39 zones
The dependent variable was log transformed to achieve the normality as it was highly skewed to the right
ICC intraclass correlation; Est effect estimates
^aVariables were log transformed to achieve the normality as they are highly skewed to the right. Their associated effect estimates present the elasticity of provider-mix
^bEffect estimates of the variables were exponentiated for ease of interpretation
***p < 0.001, **p < 0.01, *p < 0.05

ISSN: 1478-4491