Project topics and materials
Get Complete Project Material File(s) Now! »
Relevant Literature
There is an extensive literature relating health financing policy changes to health service utilisation (Lagarde, Barroy and Palmer 2012; Masiye et al., 2008; Nabyonga et al., 2005; Ridde, 2003; Wilkinson et al., 2001) however, much of that evidence could be biased (Lagarde and Palmer, 2008). Although a few studies have accounted for specific time-series properties and problems (Lagarde, Barroy and Palmer 2012; Wilkinson et al., 2001), most of the analysis has focussed on a simple comparison of average utilisation before and after a policy change (Ridde, 2003; Nabyonga et al., 2005; Masiye et al., 2008). Even if the analysis has gone beyond a simple comparison before and after the change, which is not always the case, the impact of policy on curative and preventative care has been the predominant theme, rather than maternal health services. Evidence from this literature suggests that removing user fees increases access to curative services for the vulnerable groups (Deininger and Mpuga, 2005; Lagarde, Barroy and Palmer, 2012; Lagarde and Palmer, 2008; Masiye et al., 2008; Wilkinson et al., 2001), leads to provider choice substitution (Koch, 2012) but may negatively affect service quality (Lagarde and Palmer, 2008) and utilisation amongst non-targeted groups (Lagarde, Barroy and Palmer, 2012). Also, curative care utilisation increases could have been at the expense of preventative services (Wilkinson et al., 2001). In Ridde and Morestin’s (2011) review of 20 articles, they note that the abolition of user fees has generally had positive effects on the utilisation of health services. With regard to maternal health services, as would be expected, user fees have had negative effects on utilisation (Nanda, 2002). Utilisation of antenatal care (ANC) services in Zimbabwe and Tanzania declined with the introduction of user fees. In Ghana, with the introduction of the fee exemption policy on deliveries, the proportion of institutional deliveries increased in the Central and Volta region, and, encouragingly, the increase was higher for women facing the greatest financial barrier to health care and were at the greatest risk of maternal mortality (Penfold et al., 2007). Asante et al. (2007) provide further evidence of equity improvements; fee exemption policy reduced the overall costs of delivery by 8% to 22%, depending on the type of delivery.
Although user fees do matter, there are other factors affecting institutional deliveries. Gabrysch and Campbell (2009) identify 20 determinants, based on a review of 80 articles. They group determinants into four broad themes: (1) socio-cultural factors, (2) perceived benefit/need of skilled attendance, (3) economic accessibility and (4) physical accessibility. The identified factors influence decision-making at the individual and household level; they also include measures affecting the ability to pay and the role of distance as access obstacles. They suggest that other factors, such as the quality of care, are not easily captured in household surveys, although they are reported as being essential in qualitative studies. Thus, there is a need to examine the effect of supply-side factors, which is done here. In addition to the factors mentioned by Gabrysch and Campell (2009), the use of ANC services positively affects the utilisation of institutional deliveries and skilled attendance (Gage, 2007), as does previous delivery at a health facility (Bell et al., 2003; Stephenson et al., 2006). Essentially, experiences with the health system, especially positive ones, gained through ANC visits or previous deliveries can affect delivery. Similarly, ANC provides opportunities for health workers to recommend a place of delivery, based on pregnancy risk assessments, although women with lower risks may be encouraged to deliver without a skilled assistant. Moreover, ANC attendance breeds familiarity with the health system and health facility; thus, women who seek ANC are more likely to use the same facility for delivery. However, the positive relationship observed between seeking ANC and delivering at a health facility could result from other confounding factors, such as the availability and access to services (Breen and Ensor, 2011; Gabrysch and Campbell, 2009); the same has been suggested for previous deliveries (Gabrysch and Campbell, 2009; Stephenson et al., 2006). For instance, the use of ANC or delivery services may indicate the presence of a nearby health facility offering delivery services. In many developing countries, it should be noted, ANC services are provided through outreach services, mobile clinics and small facilities, many of which do not offer delivery services. To address these problems, we include factors to proxy for the availability of health services. As implied by the previous discussion, quality of care, which covers both the perceived quality and the medical quality of care, is an important factor influencing the choice to deliver at a health facility. Although this implication has been confirmed in Hadley’s (2011) qualitative study, few quantitative studies manage to capture the quality of care. Therefore, including such factors, when available, as is the case here, is a necessary addition to the literature. Finally, alternative delivery options should also be considered, as they are likely to impact on institutional delivery and skilled birth attendance. In the African context, the primary alternative is a traditional birth attendant (TBA), an alternative that may or may not be an appropriate substitute. TBAs may not provide satisfactory assistance, due to low levels of literacy, non-existent to poor training and limited obstetric skills, all of which negatively affect the delivery process, especially when there are complications (Garces et al., 2012; Singh et al., 2012). On the other hand, TBAs could be better than nothing, especially if they are properly trained. Although maternal mortality rose after TBAs were banned in Malawi, they then fell, once TBAs were trained and reinstated (Ana, 2011). There is further evidence that trained TBAs reduce neonatal mortality (Gill et al., 2012), and are a feasible and affordable option in countries with limited medical skills capital; however, they need an appropriate support network to work effectively (Stekelenburg et al., 2004). The evaluation of user fee abolition and institutional deliveries in Zambia, discussed below, highlights the importance of the aforementioned supply-side factors, such as quality of care, and demand-side factors, such as user fees and alternatives, in influencing the demand for institutional deliveries. A natural experiment in Zambia underpins the identification strategy in the empirical analysis, although additional practical realities must also be considered. The policy change in Zambia, a country consisting of 72 districts in 9 provinces, was implemented under different conditions. For instance, drugs and financing that were to be provided to some districts were not provided successfully (Carraso et al., 2010), which could have compromised the quality of care in those districts, leading to an exodus of unsatisfied clients into other districts or provinces. Thus, policy implementation potentially plays an important role in the analysis, as there could be provincial or district interdependencies in the outcomes. Therefore, the experimental setting analysed is further complemented with tests for cross-sectional dependence, following Pesaran’s (2004) CD test, as well as Breush and Pagan’s (1980) Lagrange Multiplier (LM) test, and corrections for the identified dependencies. These tests and corrections address a major critique of panel data analysis; cross-sections are unlikely to be independent. Furthermore, quality of care, one of the important supply-side factors, is likely to differ by province and/or district, influencing effect size in these regions. Zellner’s (1962) seemingly unrelated regression (SUR), which controls for spatial dependence, is used to estimate the potentially different effect sizes across regions.
Preliminary analysis
In Panel I of Table 4.2 we compare the means of the selected variables in the period prior to the abolition of user fees (2003q1-2006q1) to the means in the post-abolition period (2006q2- 2008q4). A test of mean differences before and after the policy change finds statistically significant difference for antenatal visits, drug availability and health centre client contact, although these estimates do not control for any trends, prior to the abolition of user fees. Moreover, the standard errors do not account for any within-group correlation. Differences in means for institutional deliveries at the provincial level are presented in Appendix C3. Panel II of Table 4.2 shows the correlation between the variables in the system. Institutional deliveries are statistically significantly correlated with the first lag, suggesting persistence and justifying the dynamic specification in the analysis. It is also evident that many of the previously described relationships from the literature hold in this data, at least at the level of correlations. ID is positively correlated with ANC, but negatively correlated with DA and TBAs. However, correlation is not a result that can stand on its own.
1 Introduction
1.1 Overview
1.2 Background on the health system and maternal health care in Zambia
1.3 Research questions and methods used
2 The utilisation of antenatal care in Zambia: a micro-econometric approach
2.1 Introduction
2.2 Background and related literature
2.3 Econometric methods
2.4 Data and variable specification
2.5 Results
2.6 Conclusion
3 Utilisation of focused antenatal care in Zambia: examining individual and community level factors using a multilevel analysis
3.1 Introduction
3.2 Methods
3.3 Results
3.4 Discussion
3.5 Conclusion
4 Assessing regional variations in the effect of the removal of user fees on institutional deliveries in rural Zambia.
4.1 Introduction
4.2 Relevant Literature
4.3 The Data
4.4 Empirical Methodology
4.5 Estimation Results and Discussion
4.6 Conclusion
5 Evaluating the impact of the removal of user fees on facility based deliveries in rural Zambia: a difference- in- differences approach
5.1 Introduction
5.2 Conceptual framework
5.3 Empirical strategy
5.4 Data and description of variables
5.5 Results
5.6 Discussion
5.7 Conclusion
6 Summary and Conclusion
References
A Appendix for Chapter 2
A.1: Summary statistics for the explanatory variables, 1996 (N=4425)
A.2: Summary statistics for the explanatory variables, 2001/2 (N=3846)
A.3: Summary statistics for the explanatory variables, 2007 (N=3618)
B Appendix for Chapter 3
B.1: Description of two-level multilevel models
C Appendix for Chapter 4
C.1: List of 54 districts where user fees were removed for the entire district on 1st April 2006.
C.2: Means of institutional deliveries at provincial level
C.3 Graphical analysis of the abolition of user fees
C.3.5: Trend in Population at National and Provincial level, 2003q1-2008q4
C.4: Map of Zambia showing provinces
D Appendix for Chapter 5
D.1: Full results-Odds ratios
D.2: Specification test-without selective migration
D.3: Specification tests-including interaction terms
D.4: Specification tests-placebo
