This study relied on secondary data and is constrained by the limited extent of those data. In particular, there were very few historic data that we were able to access, and we were generally unable to compare the situation of the health workforce before and after financing-policy change. Secondary data are also affected by well-known quality concerns that in the cases of individual datasets are difficult to assess. We benefited, however, from recent initiatives to strengthen HRH databases in several of the countries. Despite constraints in the data collection process, we believe the data we have used were the most up-to-date at the time of collection (2011) and those believed the best quality available in each country.
Of the four case-study countries that have removed or introduced exemptions for user fees for RMNH (in the fifth case-study country, Zimbabwe, no discrete policy change was introduced), only in Nepal is there clear evidence of positive impact on utilization without significant exception. In Ghana, better evidence is available in relation to the earlier maternal health exemption programme than the more recent inclusion of free maternal health services in the NHIS, although an evaluation of the NHIS exemption for pregnant women was due in 2012, according to Ministry of Health sources. It appears clear that utilization increased where free care was effectively available, but implementation difficulties, most notably under-funding of the programme, implied that effective free care (at least as judged by users) was not sustained, with the implication that higher rates of utilization also could not be sustained. In Zambia, fee removal was not specifically targeted at Maternal and Newborn Child Health (MNCH) services and there is conflicting evidence of the impact of fee removal on utilization. In Sierra Leone, data suggest an initial increase in outpatient visits for children under five years in the first year of the policy, but a gradual decline since then, and an overall fall in immunization levels, which may have been caused by factors external to the policy. These findings illustrate the importance of attending to the supply side, including human resource constraints, when seeking to support access to effective health care through financing policy change.
In Nepal and Zambia, there is some evidence that user-fee removal has particularly enhanced the utilization of poorer groups (Nepal) or areas (Zambia), although we have expressed some doubt about the Zambian analysis on this point. In other countries, it has not been possible to break down utilization change in this way.
The HRH situation in case-study countries is more variable than might have been expected. At national level, shortages of HR relative to the needs of RMNH services are not universal. However, in general, there are local shortages relative to need, either because of overall national level shortages, which are acute in Sierra Leone and more marginal in Zimbabwe, or because maldistribution creates local shortage where there is national sufficiency. The relative contribution of health workers in the private sector is difficult to measure. Although such workers represent capacity to deal with RMNH needs, they may be under-used to the extent that people are unable to access those health workers due to the financial barrier. In Nepal where the proportion of health staff in the private sector is highest, this issue is more important than it yet is in the African countries. However, economic and private sector growth in these countries implies that questions of access to private-sector health staff and their influence on the overall balance of need and HRH capacity will require a more sophisticated analysis.
Low salaries are not the general situation of health workers in the case-study countries, with salary levels for doctors in Nepal, Sierra Leone and Zambia, implying that they must be located at least in the top 2% of the income distribution, and in Ghana, the top 3 to 4%j. Other cadres, other than nurses in Nepal, are not quite so well paid. The situation in Sierra Leone for non-doctor health workers and for all health workers in Zimbabwe is more moderate, with pay levels at 3- to 9-fold per capita GNI/GDP.
The relatively high salary levels for at least some health workers suggest that their market position or collective bargaining power is strong. One explanation of this is the greatly increased level of international migration since the 1990s. This implies a global market for scarce medical skills in which some countries seem positioned to compete, although the sustainability of that level of competition is questionable both in the medium and long terms. Benchmarks are not available and the expectations of well-educated Africans and Asians, whose economies are characterized by high degrees of inequity in income distribution, are likely to be relatively high in comparison to national incomes per capita than in countries where education is less scarce. Given that only 2% of Zambians (for example, Zambia Living Conditions Monitoring Survey, 2004) are educated to degree level or above, it may be a reasonable expectation of those who are, that their incomes should locate them in the same elite.
Another key issue is the extent to which competence in skilled birth attendance is difficult to assess across the case-study countries. The research has relied on rules of thumb about who counts or does not count as an SBA. There are particular difficulties in this assessment in Zambia, where no separate category of midwife exists and where nurses are not all trained to an adequate level of competence in skilled birth attendance; and in Sierra Leone where Maternal Child Health (MCH) aides do not meet the international definition of SBA but are locally expected to play this role. Even health workers who have initially been provided with sufficient training but who are not highly motivated, have not subsequently practised in the role of SBA, or have not received sufficient in-service training since, will not in practice have the requisite level of skill. Hence, the capacity to scale up to 95% coverage of RMNH services is probably more limited than it appears.
This research highlights gaps in systematic and well planned coordination between financing policy and HR policy. In our case-study countries, there have been laudable attempts to plan for the impact of fee removal or reduction, and sometimes concomitant supportive change, even if not specifically responding to the needs of financing-policy change. The global literature review suggests that poor coordination is widespread. In some cases, such as in Niger and Zambia, measures were taken after problems associated with the removal of fees became evident. In the case of Zambia, of which we know more, the measures of compensation appeared to come too little and too late, sometimes not at all.
A number of countries that removed fees also increased health worker pay to some extent at around the same time, although it is not clear that this was directly in compensation of changes brought about by fees in all cases. In Sierra Leone, the two policy debates have been clearly linked and salaries were increased in preparation for the FHCP
. Such explicit linkage is not apparent in the other countries. In Zambia it is claimed that the user-fee removal policy came with no plan or budget to recruit and deploy health workers
Rather like pay, in some cases additional recruitment was undertaken concurrent to fee reform, but it is not clear in Zambia or Ghana that this was carefully planned as part of a package of complementary policies. In contrast, Sierra Leone did plan increased recruitment as an element of the FHCP and this had been 'partially achieved’ at the time of a review in June 2010 in the sense that it was seen as contingent on the salary uplift.
Rather it appears that pay reform, recruitment activity and user-fee reform are among a plethora of interventions that are being introduced concurrently but with insufficient coordination. The literature on user fees is now quite clear on the need for associated measures, and the implications of the neglect of these are clear. In the first place, failure to coordinate undermines the impact of user-fee reform through what appear as implementation problems and result in the failure of policies to secure expected results or to sustain them. In the second place, user-fee reform may be exacerbating HR problems. The clearest case of this is the Zambian one. Ironically, the focus of user-fee removal on rural districts, intended to target access improvements in rural areas, has had a disproportionate effect on workloads in rural areas, which were already significantly greater than in urban areas. Worsening the relative conditions in rural areas is likely, other things being equal, to worsen the maldistribution of HRH and may result in rural access deteriorating. In Sierra Leone, loss of user-fee income has resulted in the loss of volunteer workers, who in practice depended on user fees for an income rather than constituting volunteers in the strict sense. This may explain the declining rate of immunization, as this appears to depend to some extent on such workers.
The difficulties of policy coordination are well-known and are not confined to low- and middle-income settings
. The specific set of policy process issues involved in user-fee removal have been analysed by Meessen et al.
, who find that what they describe as 'good practice’ has more often than not been absent in the six African countries they review. One common feature they identify is a 'top-down’ and in many cases sudden and surprise move to remove charges that planners and policy makers at lower levels then struggle to adjust to. This may partly explain some of the problems in our case-study countries too.
The HRH situation also affects user-fee reform in the sense that there is some evidence among our case studies that staff who feel aggrieved because of a sense of overwork, underpay or deterioration in conditions, are more likely to undermine user-fee reform in the interests of maintaining the status quo. In all countries, there was evidence that services intended to be free were not always experienced as such by users, although to different extents. This problem was seen to be small in Sierra Leone, and to have reduced in Nepal, but indicates a clear link between the two areas of policy in this direction. At the extreme, informal fees can simply replace formal ones.
Linkage also operates in both directions through the medium of quality of care. User-fee removal can only be counted as successful to the extent that users recognize a better option in the reformed service, comparing both quality and price variables. Consideration of utilization as an indicator of the effectiveness of policy reform measures the direct and desired outcome, improved access, but also indicates the extent of users’ preferences for the reformed service
. The observation that initial increases in utilization are not sustained (of our case studies, most likely in Zambia) implies that neither measure of success is long lived. Health workers make perhaps the most critical contribution to quality of care and to whether any utilization gains following fee removal are sustained. Interpersonal aspects of quality of care - whether users are treated with dignity and respect and given the attention their problem requires - always rank highly in studies of the attributes of quality that matter to users, and are mainly under the control of health workers. Health workers also have influence on whether drugs and other supplies are available when required: they can conceal available stocks, and can use initiative to replace drugs that are out of stock at the time, for example. Aggrieved health workers who do not support user-fee removal because they have not been adequately compensated for the lost income and increased workload, are least likely to support the maintenance of quality in any of its dimensions.
Among the associated measures well-recognized in the existing literature is the need to ensure replacement of user-fee income where it is important at the local level. User-fee income has typically been used to provide bonuses to staff, employ additional contract staff and to support drug supply. All the case studies of user-fee removal or exemption except Nepal identified problems in either the failure to replace user-fee income or the inadequacy of the replacement in form or amount.