The Critical Role of Supervision in Retaining Staff in Obstetric Services: A Three Country Study

Millennium Development Goal (MDG) 5 commits us to reducing maternal mortality rates by three quarters and MDG 4 commits us to reducing child mortality by two-thirds between 1990 and 2015. In order to reach these goals, greater access to basic emergency obstetric care (EmOC) as well as comprehensive EmOC which includes safe Caesarean section, is needed. The limited capacity of health systems to meet demand for obstetric services has led several countries to utilize mid-level cadres as a substitute to more extensively trained and more internationally mobile healthcare workers. Although this does provide greater capacity for service delivery, concern about the performance and motivation of these workers is emerging. We propose that poor leadership characterized by inadequate and unstructured supervision underlies much of the dissatisfaction and turnover that has been shown to exist amongst these mid-level healthcare workers and indeed health workers more generally. To investigate this, we conducted a large-scale survey of 1,561 mid-level cadre healthcare workers (health workers trained for shorter periods to perform specific tasks e.g. clinical officers) delivering obstetric care in Malawi, Tanzania, and Mozambique. Participants indicated the primary supervision method used in their facility and we assessed their job satisfaction and intentions to leave their current workplace. In all three countries we found robust evidence indicating that a formal supervision process predicted high levels of job satisfaction and low intentions to leave. We find no evidence that facility level factors modify the link between supervisory methods and key outcomes. We interpret this evidence as strongly supporting the need to strengthen leadership and implement a framework and mechanism for systematic supportive supervision. This will promote better job satisfaction and improve the retention and performance of obstetric care workers, something which has the potential to improve maternal and neonatal outcomes in the countdown to 2015. © 2013 McAuliffe et al.

This study is a cross-sectional descriptive study of healthcare facilities and healthcare providers in obstetric care in Malawi, Tanzania and Mozambique. The study was approved by the Institutional Review Board of Columbia University, New York; Global Health Ethics Committee Trinity College, Dublin; and the Institutional review boards of College of Medicine, Malawi, Eduardo Mondlane University, Mozambique and Ifakara Health Institute, Tanzania. Data were collected from healthcare providers and healthcare facilities between October and December 2008. Providers who indicated they had performed basic or emergency obstetric care tasks in the prior three months were eligible for participation. In Malawi, a near-national sample of facilities (N = 84) intended to provide EmOC services was identified and included central, district, rural and CHAM (faith-based organisations) –operated hospitals and a randomly sampled urban and recently upgraded health centres designated to provide EmOC. A few districts/facilities were excluded in Malawi due to their recent participation in another human resources study in which similar data had been collected from health workers. In Tanzania, due to the size of the country, cluster sampling was employed. One region was randomly selected in each of the eight geographic zones and all districts within those eight regions were then included in the sampling frame. The primary hospital serving the district was identified for inclusion; either the government-run district hospital or voluntary agency-run (VA) designated district hospital (DDH). In some districts that also contain the regional headquarters, the regional hospital was included in the sample when there was no district hospital serving the community. One health centre (HC) was randomly selected in each district, thus there were two facilities from each district in the study (N = 90). In Mozambique, a near national sample of general, district and rural hospitals was included to maximise the potential participation of the NPC cadre tecnico de cirurgia. In addition, two to three health centres (type 1 and type 2) providing maternity care, and therefore at least some basic EmOC functions, were randomly selected in each district for inclusion in the study (N = 138). Facilities were sampled from all rural regions outside Maputo City, as mid-level cadres such as surgical technicians are concentrated primarily in health facilities in rural regions with obstetricians and nurse midwives being concentrated in the Maputo City area.. Selected facilities were similar within and across the three countries and therefore the different selection approaches are unlikely to have influenced the results. Eligible providers were given detailed information about the study and its requirements and signed a consent form if they wished to participate. The actual response was limited by the numbers of eligible staff actually available in the facilities at the time the facility was visited and the data collector’s efforts to ensure minimum disruption to health service delivery. The facility survey was completed by the data collection team who compiled information on key facility metrics such as the number of beds in the facility and the availability of equipment and other resources. The team was assisted in this process by the facility and maternity in-charge and specific members of staff with expertise or access to records in the relevant area. The availability and functionality of equipment was confirmed through visual inspection. Data from the detailed facility level survey for Mozambique was not available at the time of writing and facility level analyses thus utilize the Malawi and the Tanzania data. Participants indicated which of five methods of supervision best described the supervision experienced at their healthcare facility-“Formal supervision process with regular pre-arranged supervision meetings”, “Supervision is available if I request it from my line manager”, “Supervision consists of negative feedback when performance is poor”, “ I never receive any supervision or feedback on my performance”, or “other” form of supervision. These categorisations were derived from informal discussions with ministry and district/council level staff. Job satisfaction was assessed using 5-items derived from a previously validated 7-item scale [17], [18]. Two items from the scale were dropped as they assessed satisfaction with supervision and were likely to inflate any estimates of the relationship between supervision methods and job satisfaction. The remaining items were summated (e.g. “In general, I am satisfied with this job”, “I am satisfied with my pay compared to similar jobs in other organizations”) and compiled scores on the augmented job satisfaction scale ranged from 5 (low job satisfaction) to 25 (high job satisfaction). Three items were used to assess the likelihood that participants would leave their current position-“would consider working for another hospital/clinic”, “seriously thought about leaving this hospital/clinic”, and “actively seeking other employment”. On a 5-point Likert scale total scores ranged from 3 (low intention to leave) to 15 (high intention to leave). It is possible that certain demographic and occupational factors like age, gender, and cadre may impact on independent variables such as supervision methods and dependent variables like job satisfaction. Under this rationale we thus include age, gender, and cadre as covariates in all analyses. Even after adjustment for demographic and occupational characteristics it is possible that facility level factors may confound relationships between supervision methods and healthcare worker outcomes. In the HSSE study, comprehensive facility level information was collected from all facilities sampled in Malawi and Tanzania. Although the aim was to collect similar information in Mozambique some errors occurred in collection and coding that prevented the matching of individual and facility level data and thus it has been excluded from this analysis. We include metrics of hospital size, geographic isolation and the availability of resources in all multilevel analyses conducted using data from Malawi and Tanzania. Hospital size was estimated from the number of beds recorded in the facility. Geographic isolation was indexed using the distance to the nearest referral hospital. Finally, the presence of ten key resources was recorded for each facility in order to gauge the adequacy of the facilities and resources available (e.g. availability of: electricity, clean water, staff room, meals for staff, staff toilet facilities, allowances for overtime work). The outcome variables, job satisfaction and intention to leave, were treated as continuous variables and predicted using linear multilevel modeling. Due to the hierarchical structure of the data with healthcare workers nested within facilities multilevel random coefficient modeling was deemed to be the most appropriate technique to answer most of the study questions [19]. This analytic method allows for uneven number of assessments per facility and estimates random variation in both the sampling of facilities and the sampling of workers within those facilities. The analytic strategy for the multilevel analyses was as follows: firstly we estimate two separate random intercepts models using supervision methods to predict intentions to leave and job-satisfaction adjusting for background characteristics and facility level factors. These analyses aim to clearly specify a link between supervision methods and outcome measures with adjustments for potentially confounding factors. We contrast each supervision method (e.g. negative feedback) with formal supervision. The predictive model is common across the three models and adjusts for demographic and occupational characteristics at Level 1 and facility level intercept at Level 2, as shown in Model 1 below. Standard nomenclature is used where i represents the healthcare worker, and j represents the facility. To identify if facility level characteristics influence the relationship between supervision methods and the two dependent variables of interest (i.e. job satisfaction and intentions to leave) we estimate a series of multilevel random intercepts and random slopes models. This set of analyses firstly involves identifying if the slope or relationship between supervision and the key outcomes varies between facilities (see Model 2 below). For example, in the case of job satisfaction Model 2 captures the extent that the facility-level slope of the relationship between the presence of formal supervision and job satisfaction varies (u 4j) from the overall average slope in this relation across all facilities (γ40). If significant variation in slopes between facilities is identified (e.g. the link between formal supervision and job satisfaction is substantially stronger in some facilities than in others) our aim is to then estimate the degree to which facility level factors may explain the variance component between the facilities. The key terms which are added to Model 1 and Model 2 to specify the random slopes and their determinants are detailed in Model 3. Initial model: Level 1: Job satisfaction/Intentions to leaveij = β 0j+eij Level 2: β 0j = γ00+u 0j Job satisfaction/Intentions to leaveij = γ00+u 0j+eij Specification of the level 1 and level 2 random intercept model: Level 1: Job satisfaction/Intentions to leaveij = β 0j+β 1×Ageij+β 2×Genderij+β 3×Occupationij+β 4×Supervision methodsij+eij Through substitution: Level 1: Job satisfaction/Intentions to leaveij = γ00+γ10×Ageij+γ20×Genderij+γ30×Occupationij+γ40×Supervision methods ij+u 0j+eij Adding level 2 predictors where β 0j = γ00+γ01Wj+u 0j Through substitution: Model 1: Job satisfaction/Intentions to leaveij = γ00+γ10×Ageij+γ20×Genderij+γ30×Occupationij+γ40×Supervision methodsij+γ01×Number of bedsj+γ02×Facility resourcesj+γ03×Distance to referral hospitalj+u 0j+eij Addition of random slope to Model 1 and explaining variability in the random slope: Model 2: β 4×Supervision methodsj = γ40×Supervisory methodsj+u 4j Adding level 2 predictors of the link between supervision methods and job satisfaction: Model 3: β 4×Supervisory methodsj = γ40×Supervisory methodsj+γ41(Supervision methods * Number of beds)j+γ42(Supervision methods * Facility resources)j+γ43(Supervision methods * Distance to referral hospital)j+u 4j

N/A

• Dima Health