Benefits and Limitations of a Community-Engaged Emergency Referral System in a Remote, Impoverished Setting of Northern Ghana

Sneha Patel; John Koku Awoonor-Williams; Rofina Asuru; Christopher B Boyer; Janet Awopole Yepakeh Tiah; Mallory C Sheff; Margaret L Schmitt; Robert Alirigia; Elizabeth F Jackson; James F Phillips

doi:10.9745/GHSP-D-16-00253

Figures & Tables

Figures

Tables

Download figure
Open in new tab
Download powerpoint
Three-wheeled motorcycles, known as Motorkings, served as emergency transport vehicles in the pilot districts of northern Ghana. Structural modifications were made to enhance patient safety and comfort, including extended reviewmirrors and a tarpaulin to provide privacy and protection.
Download figure
Open in new tab
Download powerpoint
FIGURE 1
Transportation Routes of Patients Using SERC Services, July 2013–June 2015
Abbreviation: SERC, Sustainable Emergency Referral Care.
Download figure
Open in new tab
Download powerpoint
FIGURE 2
Trends in Aggregated Reasons for Referral Reported by 359 Facilities Unexposed and Exposed to the SERC System, July 2013–June 2015
Abbreviation: SERC, Sustainable Emergency Referral Care.
Download figure
Open in new tab
Download powerpoint
FIGURE 3
Trends in the Location of Facility Deliveries, SERC Intervention Areas vs. Comparison Areas, 2009–2015
Abbreviations: CHPS, community-based health planning and services; SERC, Sustainable Emergency Referral Care.
Differences between the baseline and intervention period were statistically significant at P<.001 for health centers and district hospitals.

Tables

Figures

View popup

TABLE 1.

Difference-in-Differences Estimates of the Impact of the SERC Initiative on Hospital-Based Health Measures, Upper East and Upper West Regions, Ghana 2009–2015

	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)	(9)
	Deliveries	Cesarean Delivery Rate	Referrals In	Referrals Out	Pneumonia	Other Upper Respiratory Tract	Septicemia	Accidents	Diarrheal Diseases
Treatment area	−52.93^*	−0.00651	−12.08^**	−6.499	−4.537	−100.5^**	2.555	−15.35	20.96
	(26.18)	(0.0113)	(4.345)	(3.735)	(8.015)	(37.8)	(25.98)	(8.088)	(36.05)
Time period	32.55^***	0.025^**	−3.05^*	−6.80^***	0.687	35.57	23.64^*	−3.24	18.12
	(9.55)	(0.00964)	(1.435)	(1.87)	(12.75)	(47.67)	(11.9)	(3.209)	(19.13)
SERC^a	−4.88	0.0035	12.27^*	1.60	10.99	22.57	35.09	20.52^*	11.71
	(12.76)	(0.015)	(5.18)	(3.52)	(12.92)	(49.82)	(41.91)	(9.90)	(33.54)
Constant	89.73^***	0.12^***	17.45^***	12.18^***	46.96^***	237.40^***	28.38	28.17^***	72.04^***
	(23.81)	(0.02)	(4.17)	(3.04)	(14.13)	(58.81)	(30.88)	(6.84)	(14.96)
Observations	861	795	361	500	787	748	237	796	804
Number of hospitals	14	13	13	14	14	14	10	14	14

Abbreviation: SERC, Sustainable Emergency Referral Care.
Note: Estimates are from multilevel linear regressions of outcomes from monthly hospital records in the Upper East and Upper West Regions of Ghana. Regressions include random facility intercepts to account for clustering at the facility level. Standard errors are calculated assuming an exchangeable correlation structure and are reported in parentheses.
↵* P<.05; ** P<.01; *** P<.001.
↵^a The SERC effect (difference-in-difference) is given by the interaction of treatment area with time period.

View popup

TABLE 2.

Difference-in-Differences Estimates of the Impact of the SERC Initiative on Hospital-Based Maternal Mortality, Upper East and Upper West Regions, Ghana, 2009–2015

	MMR (95% CI)		Differences
	Pre-SERC	Post-SERC	IRD (95% CI)	IRR (95% CI)
Comparison district hospitals (n = 12)	326 (272, 380)	261 (194, 328)	−65 (−140, 10)	0.80^† (0.61, 1.05)
Intervention district hospitals (n = 2)	618 (392, 844)	201 (22, 381)	−417^** (−693, −140)	0.33^* (0.13, 0.83)
Difference-in-differences (SERC effect)			*−352^ (−639, −65)**	0.41^† (0.15, 1.07)

Abbreviations: CI, confidence interval; IRD, incidence rate difference (deaths per 100,000 live births); IRR, incidence rate ratio; MMR, maternal mortality ratio; SERC, Sustainable Emergency Referral Care.
Note: Estimates are from multilevel Poisson regressions of monthly hospital records of births and maternal deaths at 14 facilities in the Upper East and Upper West Regions of Ghana from 2009 to 2015. The hospital MMR is calculated as the number of facility-based deaths per 100,000 live births. The 95% confidence intervals were calculated using robust standard errors accounting for clustering at the facility level.
↵* P<.05; **P<.01; ***P<.001; ‡ P<.10.