Table 6. Parameter estimates and system losses due to Ebola virus disease (EVD) outbreak (June 2014–April 2015) for medroxyprogesterone acetate doses across a census of clinics providing services in Liberia excluding Montserrado County, 2010–2016.
| Time period | Medroxyprogesterone acetate doses | ||
|---|---|---|---|
| β (95% CI)* | p-Value | Mean (95% CI)† | |
| Overall yearly trends pre-EVD | |||
| Monthly change Jan–Dec 2010 | 0.21 (−0.0001, 0.41) | 0.05 | 14.5 (12.0, 17.0) |
| Monthly change Jan–Dec 2011 | 0.42 (0.18, 0.66) | 0.001 | 17.8 (14.6, 21.0) |
| Monthly change Jan–Dec 2012 | 0.32 (0.09, 0.55) | 0.006 | 22.1 (17.9, 26.3) |
| Monthly change Jan–Dec 2013 | 0.27 (0.04, 0.50) | 0.02 | 26.1 (21.5, 30.7) |
| Monthly change Jan–May 2014 | 1.6 (0.96, 2.3) | <0.001 | 31.3 (26.0, 36.5) |
| EVD outbreak (Jun 2014–Apr 2015) | |||
| Monthly change first 4 months (Jun–Sep 2014) | −3.5 (−4.3, −2.7) | <0.001 | 26.1 (22.6, 29.6) |
| Monthly change middle 3 months (Oct–Dec 2014) | 1.1 (0.29, 1.9) | 0.008 | 23.8 (20.0, 27.6) |
| Monthly change last 4 months (Jan–Apr 2015) | 0.74 (−0.24, 1.7) | 0.14 | 24.4 (20.6, 28.2) |
| Post-EVD (May 2015–Dec 2016) | |||
| Monthly change May–Dec 2015 | 0.50 (0.11, 0.88) | 0.01 | 30.5 (24.7, 36.3) |
| Monthly change Jan–Dec 2016 | 0.30 (−0.10, 0.70) | 0.14 | 32.8 (27.7, 37.9) |
*All analyses include fixed-effect control variables for monthly indicator variables and facility-level catchment population.
†Mean represents average monthly number of system outputs across facilities included in a given analysis and across a given time period using a linear mixed model controlling for clustering at the facility level and employing an AR(1) structure to control for autocorrelation in residual errors.