Kenya’s emergency-hire nursing programme: A pilot evaluation of health service delivery in two districts

Objective: To assess the feasibility of utilizing a small-scale, low-cost, pilot evaluation in assessing the short-term impact of Kenya’s emergency-hire nursing programme (EHP) on the delivery of health services (outpatient visits and maternal-child health indicators) in two underserved health districts with high HIV/AIDS prevalence.Methods: Six primary outcomes were assessed through the collection of data from facility-level health management forms-total general outpatient visits, vaginal deliveries, caesarean sections, antenatal care (ANC) attendance, ANC clients tested for HIV, and deliveries to HIV-positive women. Data on outcome measures were assessed both pre-and post-emergency-hire nurse placement. Informal discussions were also conducted to obtain supporting qualitative data.Findings: The majority of EHP nurses were placed in Suba (15.5%) and Siaya (13%) districts. At the time of the intervention, we describe an increase in total general outpatient visits, vaginal deliveries and caesarean sections within both districts. Similar significant increases were seen with ANC attendance and deliveries to HIV-positive women. Despite increases in the quantity of health services immediately following nurse placement, these levels were often not sustained. We identify several factors that challenge the long-term sustainability of these staffing enhancements.Conclusions: There are multiple factors beyond increasing the supply of nurses that affect the delivery of health services. We believe this pilot evaluation sets the foundation for future, larger and more comprehensive studies further elaborating on the interface between interventions to alleviate nursing shortages and promote enhanced health service delivery. We also stress the importance of strong national and local relationships in conducting future studies. © 2014 Vindigni et al.; licensee BioMed Central Ltd. This evaluation engaged both qualitative and quantitative methodologies aimed at assessing the feasibility of a low-cost, small-scale, pilot study to assess the impact of the EHP on health service delivery. With fieldwork conducted over two months and data collection from 2004 to 2010, the evaluation focused on the provision of services in 13 health facilities that benefitted from the hiring of EHP nurses. We performed an ecological analysis by comparing the level of health services before and after the assignment of EHP nurses within facilities in two MOH districts. Since most nurses were hired in 2007, this year was used as the referent, or intervention year. The two districts with the greatest quantity of EHP nurses were selected for field-level evaluation—Suba and Siaya Districts, both within Nyanza Province—where HIV prevalence is greatest (Figure 1). The primary investigator travelled with two Kenyan health officers to 13 health facilities comprising a combination of district hospitals, sub-district hospitals, and health centers during May and June 2010. These facilities were selected based on increases in staffing levels, geographic convenience, and to obtain a variety of facility types. Map of Kenya and Nyanza Province. Information on health-care workers, including EHP nurses, was obtained from Kenya’s Human Resource Information System (HRIS), known as the Kenya Health Workforce Information System (KHWIS), which consists of two databases—one housing information on the regulation of nurses and the other containing information on nurse deployment [3]. De-identified deployment data on nurses assigned to each district (MOH and emergency-hire) were obtained using the KHWIS. Each cohort was analysed individually and in total and stratified by province, district, and individual health facility. Nursing data were analysed for demographics, training level, and health facility assignment. Microsoft Access (Microsoft, Redmond, WA, USA) and Stata (StataCorp, College Station, TX, USA) were used for statistical analysis. Provision of health services was determined using Kenya’s HMIS, which collects data at individual health facilities. This information is aggregated and submitted to the MOH’s national office. The HMIS standardized forms document patient services, such as, reproductive health, immunization coverage and other disease surveillance and health outcome parameters (forms in Additional file 1). Health service delivery data was obtained from each health facility visited within Siaya and Suba Districts. Previously collected, monthly-aggregated HMIS data was abstracted dating back to January 2004 through May 2010, although 2004–2005 data was significantly incomplete, therefore, it was not included in this evaluation. A listing of health services is described in Table 1. Abstracted data was analysed in aggregate and stratified by district and facility. Data were averaged based on the quantity of data points available to account for unavailable data. Results of aggregate annual data in Siaya and Suba Districts *significant at 0.05. **significant at 0.01. ***significant at 0.001. ****significant at 0.0001. One key informant interview was also conducted with a convenience sample of medical, nursing, and clinical officer managers at each site (n = 13) to provide further insight into the quantitative trends. Topics for discussion included EHP service delivery benefits, existing challenges regarding service provision, and ideas for increasing the quantity and quality of primary care services. Efforts were made to identify additional causes of service delivery change beyond the deployment of nursing staff (e.g. medical supply availability, medical/public health programmes and campaigns, environmental factors). Face-to-face, 30- to 45-minute discussions were facilitated by the primary author (SV), a male, public health professional with no prior relationship to the discussants. Although the interviews were structured with the same questions at each site (interview guide and questions in Additional file 2), these were intended to be informal discussions without audio-video recording. No staff member refused participation and the only other persons present were two co-authors. To evaluate changes in services in association with EHP implementation, we focused on six indicators—total general outpatients, vaginal deliveries, caesarean sections, antenatal care (ANC) attendance, ANC clients tested for HIV, and deliveries from HIV-positive women. Each of these indicators was collected independently and summed by facility for each year between 2006 and 2009. Annual facility averages were adjusted based on the number of months of data available. Data obtained prior to 2006 were not included since monthly HMIS records were incomplete for those years. Since the EHP was mainly implemented in 2007, the indicators were analysed for changes in sums for each year compared to 2007. A Poisson regression model was used with a year as a categorical predictor, and year 2007 as the referent/intervention category. To account for correlation of data within sites, we specified the working correlation matrix as exchangeable, which assumes the same correlation between any two elements of a cluster and is a reasonable assumption for these data. In this way, we obtained rate ratios and 95% confidence intervals, which assess the rate of the indicator for each year compared to 2007. If the rate ratio is greater than one, this implies the rate of the indicator was greater in that year compared to 2007. If the rate ratio is less than one, it indicates the rate was less in that year compared to 2007. Finally, if the rate ratio is exactly one, the rate was the same compared to 2007.