This dissertation contains no tests, graphs or tables copied and pasted from the Internet, unless specifically acknowledged and the source is detailed in the dissertation and in the references sections. The health outcomes achieved did not justify the additional costs of frequent surveillance in the intervention, but the focus should be on the quality and consistency of surveillance.
INTRODUCTION
- Task Shifting and the Health Human Resources Challenge
- Community Health Workers
- The Nompilo Project RCT
- Chapter Outline
CRH-UKZN therefore decided to investigate the effectiveness of a CQI in the management and monitoring of CCGs. The (CQI) methodology that CRH-UKZN implemented in the intervention arm of the RCT introduces a quality mentor (QM) into the monitoring of CCGs.
BACKGROUND LITERATURE
- Primary Health Care in South Africa
- Maternal Neonatal Child and Women’s Health (MNCWH)
- General Characteristics of Community Health Worker Programmes
- Community Health Workers in South Africa
- Community Case Management
Maternal health care services such as antenatal and postnatal care are also provided by the clinics. The training undertaken by CHWs is much shorter and has a limited curriculum compared to health care providers such as nurses.
THEORETICAL FRAMEWORK
- Agency Theory and the Effect of Supervision
- Supervision and the “Crowding Out” of Motivation
- Supervision as an Incentive for Shirking
- Intrinsic Motivation and Signalling in Agency Theory
- Legitimacy of Control
- Supervision and Heterogeneity in Reciprocity Preferences
- The Effect of Supervision on the Conformists Agent
- Summary of Supervision and Agency Theory
- Supervision Studies of CHW Interventions
- Cost-effectiveness of CHWs in Promotion of MNCWH Interventions in Rural
- Costing of the Promise EBF Trial in South Africa
- Costing of the Promise EBF in Rural Uganda
- Conclusion
According to Frey's (1993) “crowding out” theory, increased control creates an agent's distrust of the principal. Falk and Kosfeld (2006, p. 1612) argue that in the context of mutual agents, enhanced monitoring signals that the principal has low expectations of the agent's level of work effort.
THE NOMPILO PROJECT STUDY DESIGN
The Nompilo Project Study Design
The study population included CCG supervisors, CCGs and mothers of infants born in the 12 months prior to undertaking the baseline survey. In the study, CCG supervisors were randomized into an intervention and control arm, with each arm comprising 15 CCG supervisors respectively. In intervention arms, enhanced CCM training was received and teams were supported through continuous quality improvement (CQI) supervision.
The CQI methodology used in the intervention arm was based on identifying measurable goals at the beginning of the intervention and then developing tests of changes that could lead to improvements in the uptake of MNCWH service packages by participating mothers. In the control arm of the RCT, CCGs received standard provincial CCM training, which was not specifically tailored for use in areas with a high prevalence of HIV/AIDS. Control CCGs and their supervisors continued to receive routine training or support provided by KZN DH in the district.
CCGs in the control arm relied on healthcare supervisors who were registered nurses in the local hospital (CRH-UKZN, 2014). In the post-intervention period, HIV-adapted CCM training and CQI supervision appear to be most effective compared to standard provincial DoH training.
METHODS
Cost Analysis
Terminology related to resource costs appears to be controversial, with different meanings of the same terms (Gold et al, 1996; Drummond, 2005). According to Gold et al (ibid), resource costs generally fall into three categories, namely direct costs, indirect costs and intangible costs. This study took the contractor's perspective, so direct costs are the most important.
Direct costs are defined as "the value of all goods, services and other resources consumed in the delivery of an intervention or in the management of the side effects or other current and future consequences associated with it" (Gold et al, 1996, p . 179). Thus, direct costs include the costs required to carry out the intervention that would normally be contained in a health facility. Gold et al (ibid) further divide direct costs into direct health costs and direct non-health costs.
According to Meunnig (2008, p. 7), indirect costs do not include the use of goods and services and are related to costs of productivity, morbidity and mortality. Morbidity and mortality costs, also known as intangible costs, attempt to assign a monetary value to an individual's pain, suffering, stigma, or life years and can be highly subjective (Meuning, 2008).
Cost-Effectiveness Analysis
The technical efficiency of the competing intervention is observed through the cost-effectiveness ratio, which reflects the cost of obtaining one unit of health outcome – for example, the cost per child vaccinated (Gold et al, 1996, p.3). The intervention that produces the lowest cost per unit of health outcome would be considered the “most efficient way to improve health” as it indicates that maximum results are produced within a given budget (Gold et al, 1996, p.4; Shemilt et al., 2002, p.196; Fox-Rushby and Cairns, 2005, p.13). When comparing two competing complex interventions that may have different intensities, such as the duration of training of health professionals, an incremental cost-effectiveness ratio (ICER) would be the most appropriate tool to use.
The ICER provides a summary measure of the incremental cost per unit of health gained in adopting one medical intervention over another, which in most cases tends to be the current existing intervention. Drummond et al (2005, p. 8) caution against the use of an ICER, noting an overarching weakness with most cost-effectiveness analyses: the existing treatment or practice to which the intervention is being compared may no longer be the important thing. cost-effective treatment in the first place. The implication is that while an intervention may indeed be effective, it may not necessarily be what society wants (Elliot and Payne, 2005, p.16).
The study is a retrospective CEA based on a randomized control trial conducted by CRH-UKZN in the Ugu Health District, KwaZulu-Natal Province between May 2012 and November 2013. This cost-effectiveness analysis compares training and supervision of a CCG that provided by Provincial DoH KwaZulu-Natal, with respect to HIV-tailored CCM training and CQI management and oversight of CCGs by CRH-UKZN.
RESEARCH METHODOLOGY
- Overview
- Data Collection
- Framework for Data Analysis
- Comparator
- Time Horizon
- Choice of the Discount Rate
- Health Outcomes
- Estimating Resources and Costs
- Classification of Resource Inputs
- The Line Item Method of Classification
- Non-recurrent Inputs
- Non-recurrent Inputs – Once-Off HIV-adapted CCM and CQI Training and
- Non-recurrent Inputs – Equipment
- Non-recurrent Inputs – Building Space
- Non-recurrent Costs Office Furniture
- Recurrent Inputs
- Recurrent Inputs – Personnel
- Recurrent Inputs – Fortnightly Mentoring Sessions
- Recurrent Inputs – Quarterly Learning Sessions
- Recurrent Inputs – Telephone, Internet Data, Stationery
- Recurrent Inputs – Utilities
- Incremental Cost-Effectiveness Analysis
- One Way Sensitivity Analysis
- Limitations of the One Way Sensitivity Analysis
- Conclusion
In addition, the costs of the resource inputs used in the intervention are clearly visible from the program budget. Staff identified in the operation of the Nompilo RCT project include three quality mentors, 60 CCGs and 15 supervisors and one administrator. Telephone expenditure costs attributed to the Nompilo Project RCT are not reflective of program costs and therefore had to be adjusted to reflect use by quality mentors.
This paper performs a stepwise cost-effectiveness analysis which allows for a comparative analysis between the control and intervention arms of the Nompilo project. A major limitation of the Nompilo Project RCT is resource use, activities and costs in the control arm, where it is not tracked, as the primary focus has been the effectiveness of the intervention relative to the control arm. CCGs in the control arm did not undergo additional HIV-tailored CCM and CQI training, therefore the incremental costs are the full costs of the training.
Calculating the financial cost of the control arm allows its cost-effectiveness ratio (CER) to be calculated. Difference between the costs of the recurring meetings in the respective departments of the RCT (100% of the costs).
RESULTS
- Nompilo Project Programme Unit Costs
- Line Item Classification of the Nompilo Project Costs
- Recurrent Costs
- Non-Recurrent Costs
- Analysis of Personnel Costs
- Activity Based Costing of the Nompilo Project RCT
- Estimating the Control Programme Costs
- Effectiveness
- Cost-effectiveness
- Sensitivity Analysis
- Conclusion
There are three activities that have been identified in relation to the intervention aspect of the Nompilo RCT project. We estimated the program costs of the control by calculating the incremental costs of the intervention and subtracting them from the intervention arm. In the RCT of the Nompilo project, additional activities included the introduction of quality mentors, quarterly teaching hours and fortnightly mentoring sessions, as shown in Table 9.
CCG supplies, office telephone and data costs are similar between both arms of the Nompilo Project RCT resulting in zero additional cost. Program costs of the control arm are approximately 40% of the total costs of the intervention arm. The incremental cost-effectiveness ratio was calculated by subtracting the CER control arm from that of the CER intervention arm as illustrated in Table 11 below.
The results show that the model developed to determine the cost-effectiveness ratio of the RCT of the Nompilo project is robust to changes in the discount rate. The main drivers of total program costs are staff costs and bi-weekly mentoring sessions.
DISCUSSION
- Choice between the Intervention and the Control Arm
- Employment of Full Time Quality Mentors
- A Relook at Supportive Supervision
- The Intrinsic Motivation of CCGs
- Limitations
- Ignoring Team Size Effect
- Limitations of a Retrospective Cohort Analysis
- Duration of the Intervention
- The Limitations of Relying on a CEA
- Conclusion
Consideration of the MNCWH interventions, where there was no statistically significant difference in outcomes between the control and intervention arms, implies that the standard provincial training and supervision of CCGs is as effective as the HIV-tailored training and CQI supervision. Based on CER alone, it is recommended that the KZN-DoH training and supervision of CCGs be chosen as it is the more cost-effective of the two programmes. The result of the cost-effectiveness analysis suggests that the HIV-adapted CCM training and CQI supervision is less efficient in its use of resource inputs compared to standard provincial training and supervision.
An implicit assumption in the above argument is that the marginal labor productivity of the Quality Mentor is a function of the intrinsic motivation of the CHW. CRH-UKZN's main motivation when implementing the RCT was to determine the effectiveness of the intervention relative to the status quo. It is argued that the 15-month time horizon of the CRT was not sufficient to determine the cost-effectiveness of the intervention.
CCGs reported an increase in confidence and knowledge as a result of the support and training they received in the intervention. The limitations of the intervention arm do not increase the likelihood that the HIV-tailored CCM training and CQI surveillance of CCGs will become cost-effective compared with the provincial CCM training and surveillance of CCGs.
CONCLUSION
However, this seems to be consistent with the hidden cost theories of supervision presented by Frey (1993), Benabou and Tirole (2003); Falk and Fischbacher (2006). Rather than focusing on frequency, it is the quality and consistency of supervision that has the potential to increase the performance of CCGs. Cost-effectiveness analysis of the use of community health workers in promoting maternal health services in Butere District, Rural Western Kenya.
Improving PMTCT Program Coverage through a participatory quality improvement intervention in South Africa. Gilson, L., Walt G., Heggenhougen K, Owour-Omindi L, Perera M, Ross D., Salazar L (1989) National community health worker programs: How they can be strengthened. A brief summary of the South African mother, newborn, child and women (MNCWH) and nutrition strategic plan 2012-2016.
Tebenderana J, Kirkwood B and Meek S, (2014) Community health worker supervision in low-income countries - a review of impact and implementation issues. Vol 8, No 2, Paper and Proceedings of the One Hundred and Twenty Annual Meeting of the American Economic Review pp.201 – 205.
