Applying the netbenefit framework for analyzing and presenting costeffectiveness analysis of a maternal and newborn health intervention.  
Jump to Full Text  
MedLine Citation:

PMID: 22829906 Owner: NLM Status: MEDLINE 
Abstract/OtherAbstract:

BACKGROUND: Coverage of maternal and newborn health (MNH) interventions is often influenced by important determinants and decision makers are often concerned with equity issues. The netbenefit framework developed and applied alongside clinical trials and in pharmacoeconomics offers the potential for exploring how costeffectiveness of MNH interventions varies at the margin by important covariates as well as for handling uncertainties around the ICER estimate. AIM: We applied the netbenefit framework to analyze costeffectiveness of the Skilled Care Initiative and assessed relative advantages over a standard computation of incremental cost effectiveness ratios. METHODS: Household and facility surveys were carried out from January to July 2006 in Ouargaye district (where the Skilled Care Initiative was implemented) and Diapaga (comparison site) district in Burkina Faso. Pregnancyrelated and perinatal mortality were retrospectively assessed and data were collected on place of delivery, education, asset ownership, place, and distance to health facilities, costs borne by households for institutional delivery, and cost of standard provision of maternal care. Descriptive and regression analyses were performed. RESULTS: There was a 30% increase in institutional births in the intervention district compared to 10% increase in comparison district, and a significant reduction of perinatal mortality rates (OR 0.75, CI 0.700.80) in intervention district. The incremental cost for achieving one additional institutional delivery in Ouargaye district compared to Diapaga district was estimated to be 170 international dollars and varied significantly by covariates. However, the joint probability distribution (netbenefit framework) of the effectiveness measure (institutional delivery), the cost data and covariates indicated distance to health facilities as the single most important determinant of the costeffectiveness analysis with implications for policy making. CONCLUSION: The netbenefit framework, the application of which requires householdlevel effects and cost data, has proven more insightful (than traditional ICER) in presenting and interpreting costeffectiveness results of the Skilled Care Initiative. 
Authors:

Sennen Hounton; David Newlands 
Related Documents
:

22778346  The medical education partnership initiative: pepfar's effort to boost health worker ed... 18490786  Estonia's health information system and the digital health record. 8017876  Establishment and use of multiple sclerosis registersedmus. 19858936  How aids funding strengthens health systems: progress in pharmaceutical management. 22373016  A new concept of a multidisciplinary lymphoedema center  established in connection to ... 10187376  Boards of directors under fire: an examination of nonprofit board duties in the health ... 
Publication Detail:

Type: Journal Article; Research Support, NonU.S. Gov't; Research Support, U.S. Gov't, NonP.H.S. Date: 20120719 
Journal Detail:

Title: PloS one Volume: 7 ISSN: 19326203 ISO Abbreviation: PLoS ONE Publication Date: 2012 
Date Detail:

Created Date: 20120725 Completed Date: 20121207 Revised Date: 20130712 
Medline Journal Info:

Nlm Unique ID: 101285081 Medline TA: PLoS One Country: United States 
Other Details:

Languages: eng Pagination: e40995 Citation Subset: IM 
Affiliation:

Centre MURAZ, BoboDioulasso, Burkina Faso. hounton_sennen@yahoo.fr 
Export Citation:

APA/MLA Format Download EndNote Download BibTex 
MeSH Terms  
Descriptor/Qualifier:

CostBenefit Analysis
/
methods* Female Health Services Accessibility / economics Humans Infant Welfare / economics Infant, Newborn Maternal Welfare / economics* Pregnancy 
Comments/Corrections 
Full Text  
Journal Information Journal ID (nlmta): PLoS One Journal ID (isoabbrev): PLoS ONE Journal ID (publisherid): plos Journal ID (pmc): plosone ISSN: 19326203 Publisher: Public Library of Science, San Francisco, USA 
Article Information Download PDF Hounton, Newlands. This is an openaccess article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Received Day: 5 Month: 1 Year: 2012 Accepted Day: 19 Month: 6 Year: 2012 collection publication date: Year: 2012 Electronic publication date: Day: 19 Month: 7 Year: 2012 Volume: 7 Issue: 7 Elocation ID: e40995 ID: 3400570 PubMed Id: 22829906 Publisher Id: PONED1200716 DOI: 10.1371/journal.pone.0040995 
Applying the NetBenefit Framework for Analyzing and Presenting CostEffectiveness Analysis of a Maternal and Newborn Health Intervention Alternate Title:Applying NetBenefit Approach to MNH Interventions  
Sennen Hounton^{1}^{2}*  
David Newlands^{2}  
Talitha L. Feenstraedit1 
Role: Editor 
1Centre MURAZ, BoboDioulasso, Burkina Faso 

2Immpact, University of Aberdeen, Scotland, United Kingdom 

National Institute for Public Health and the Environment, The Netherlands 

Correspondence: * Email: hounton_sennen@yahoo.fr Contributed by footnote: Conceived and designed the experiments: SH. Performed the experiments: SH. Analyzed the data: SH DN. Wrote the paper: SH DN. Conceived the study (as part of his doctoral thesis), led the data collection, constructed the database, performed the statistical analysis and drafted the manuscript: SH. Participated in the design of the study, data collection and data analysis: DN. Read and approved the final manuscript: SH DN. 
Given the importance of determinants affecting access and uptake of maternal and newborn health ^{[1]}, any analysis of effectiveness ^{[2]}, ^{[3]} and costeffectiveness needs to account for the known and unknown covariates. Economic evaluation in general, and costeffectiveness analysis in particular, is an important element of evidencebased policy making in balancing health gains against the costs of interventions ^{[4]}. Cost effectiveness analysis in maternal and newborn health mostly revolves around the use of the standard Incremental Cost Effectiveness Ratio (ICER), which indicates the additional amount of money needed to obtain an extra unit of health gain or to prevent an adverse event compared to alternatives ^{[5]}–^{[6]}. In few situations is there a clearly dominant intervention (the existing intervention is less effective and more costly or a new intervention is more effective and less costly). However, in most maternal and newborn health interventions (complex interventions often building on existing packages of activities and services), new interventions are often more effective and more costly and thus there is a tradeoff of relative advantage or relative cost saving (Figure 1). In these circumstances, there is a need to estimate the maximum a provider (society or the health system) is willing to pay for an additional unit of health gain (averting one extra maternal or newborn death, or achieving one extra institutional delivery, etc.). When using the ICER, it is notoriously difficult to reliably build confidence intervals around the ICER estimates for inferential analysis ^{[7]}, ^{[8]}, and the maximum a provider is willing to pay is often unknown to analysts unless extensive willingness to pay surveys are regularly carried out ^{[10]}. We applied the netbenefit framework ^{[11]}, ^{[12]}, ^{[13]} to an observational maternal and newborn health study to assess the feasibility (including data requirements) and relative advantages in presenting and interpreting results of costeffectiveness analysis. By collecting household costs and by transforming the dataset to have concurrent householdlevel effect and cost data, we could assess the importance of significant determinants on how the costeffectiveness varies at the margin which is important for policy making and scaling up but which is not feasible with the standard ICER statistic.
The data used for this work were derived from the evaluation of the Skilled Care Initiative in Burkina Faso, an observational study. The detailed information about the evaluation methods, the context and components of interventions in Ouargaye district (where the Skilled Care Initiative was developed and implemented), and Diapaga district (comparison area) can be found elsewhere ^{[12]}, ^{[13]}. In summary, Ouargaye and Diapaga are two remote and rural districts in East and Centre East of Burkina Faso in West Africa. As part of the evaluation of the Skilled Care Initiative, an extensive mapping and description of interventions in both districts during the lifetime of the Skilled Care Initiative (2001–2005) was conducted. The most distinctive difference between the mix of interventions in the two districts was the comprehensive community mobilization developed in Ouargaye district and it is on that basis that we estimated and reported the additional cost per extra institutional delivery between the districts and the maternal and perinatal mortality advantage in the intervention district ^{[13]}, ^{[14]}. The unit of analysis was a health district: Ouargaye district (nearly 250,000 inhabitants, where the Skilled Care Initiative project was implemented) is the ‘intervention’ and Diapaga district (nearly 350,000 inhabitants with the standard package of activities in Burkina Faso health districts) is the comparator or ‘status quo’.
Data were collected from January to July 2006. A georeferenced census was conducted in the intervention district (Ouargaye) and a comparison district (Diapaga). Information were collected from all women aged 12–49 with experience of pregnancy in the survey referenced period (2001–2005), on the number of pregnancies, pregnancies lasting more than six months, place of delivery, delivery attendant and newborn survival during the five years prior to the interview.
In addition we costed the standard provision of maternal care in both districts to analyze the main cost structure in both districts, and estimated the incremental cost per delivery based on the assumption that the 30% increase in institutional delivery recorded in Ouargaye district compared to a 10% increase in Diapaga district was mainly attributable to the comprehensive community mobilization of the Skilled Care Initiative ^{[13]}, ^{[14]}.
Data were collected from all women of reproductive age with experience of delivery in the 6 weeks prior to the survey, on costs borne by women and families for institutional delivery, as well as household revenues. A database was subsequently constructed using the household cost survey with geocoordinates, asset ownership quintiles, education of head of household, distance to health facility, household revenues, perinatal death, and family size. The health system in Burkina Faso is very centralized and we assumed the standard maternal care in both remote and rural districts was the same, although differences have been documented among health centers across both districts ^{[14]}. We estimated for each household with an institutional delivery in the intervention district (Ouargaye) the household cost of an institutional delivery and the estimated incremental cost to the health system of an institutional delivery to derive a cost for institutional delivery from a societal perspective. For households with no institutional delivery, only the estimated incremental cost to the health system of an institutional delivery was attributed to the households. Conversely, in the comparison district (Diapaga), we attributed to each household with an institutional delivery the household cost of an institutional delivery but no incremental cost from the health system perspective. This does not mean there was no cost from the health system perspective in Diapaga but that the average standard cost from the health system will cancel out between the intervention and comparison sites.
Assuming we have householdlevel effect and cost data, the traditional equation ΔC/ΔE (ICER) where ΔC is the incremental cost and ΔE the incremental effect can be rearranged by multiplying each arm of the equation by ΔE. The result is ΔC = ΔE * ICER and for any ceiling ratio Ro, ΔC = ΔE * Ro. Thus, a netbenefit statistic can be computed as follows: ΔE*Ro − ΔC = ΔNB. We thus constructed for each observation (household) of the household cost survey an individual netbenefit statistic. The expression of an individual (or household) netbenefit NBi = ΔEi*Ro − ΔCi is similar to a traditional linear regression equation Y = α + δX_{i} + ε_{i} where Y is the dependent variable, α is the intercept, δ the coefficient on an explanatory variable X (continuous variable or dummy variable taking the value 1 for a positive outcome and 0 for a negative outcome for example) and ε_{i} the standard error.
For the evaluation of the Skilled Care Initiative, the household netbenefit was modeled as NB_{i} = α+δSCI_{i} + ε_{i} where NB_{i} is the netbenefit for each household, α is the intercept, δ the coefficient (incremental net benefit ) on the intervention (SCI taking the value zero for households in the intervention district and 1 for households in the comparison district), and ε_{i} the standard error. These statistics are obtained by running an Ordinary Least Squares (OLS) regression. The interpretation is straightforward. When the difference is greater than zero, it means that the incremental cost for one additional unit of effectiveness (in this case institutional delivery) is below the ceiling ratio Ro (the maximum the provider is willing to pay). Then the SCI will be deemed costeffective in Ouargaye district in comparison to Diapaga district. Similarly, if the coefficient is negative, then the incremental cost for one additional unit of effectiveness is above the Ro and the standard health system in Diapaga district will be deemed more costeffective than the SCI intervention.
The basic model, NB_{i} = α+δSCI_{i} + ε_{i}, can then be further improved to include important covariates and therefore allow the examination of the marginal impact of these covariates on incremental cost effectiveness. The final model may look like:
NB_{i} = α + ∑^{P}_{j = 1} β_{j} x_{ij} + δSCI_{i} + ε_{i.}where NB_{i} is the summation of the interaction between the treatment dummy (SCI, coded yes or no) and the covariates, α is the intercept, β_{j,} the parameter which indicates the average change in NB_{i} that is associated with a unit change in X (covariate) whilst controlling for the other explanatory variables, δ the coefficient on the intervention (SCI), and ε_{i} the standard error.
A costeffectiveness acceptability curve graphically represents the levels of certainty around the costeffectiveness ratio of two interventions by plotting hypothetical estimates of ceiling ratios to the probability that an intervention is costeffective ^{[15]}, ^{[16]}, ^{[17]}, ^{[18]}. The pvalues obtained from the net benefit regression are twosided but only onesided values are needed to test whether the incremental netbenefit is positive (the new intervention is preferred) or negative (the standard intervention is preferred) so the regression pvalues were divided by two. For negative incremental netbenefits the probability that the new intervention is preferred equals the onesided pvalue, and for positive incremental netbenefits, the probability that the new intervention is preferred equals 1 minus the one sided pvalue. The costeffectiveness acceptability curves are obtained by plotting a graphical representation of hypothetical ceiling ratios against the probability that the intervention is costeffective and can be used to construct confidence intervals around the ICER (Figure 2).
The study was approved by the Burkina Faso Health Sciences Ethical Review Board. Informed consent from all participants (head of household or adult member of the household) involved in our study was obtained verbally prior to the interviews. Written consent was not deemed necessary by the ethics committees as the survey is a classic Demographic and Health Survey (DHS) type survey and data were to be analyzed anonymously and no individual data analysis was proposed. The investigation team relied on trained supervisors to ensure consent was secured and the ethics committees approved this consent procedure.
Table 1 presents the sample statistics of the costeffectiveness analysis of the Skilled Care Initiative. The ICER value represents the additional resources required per additional delivery in a health facility. Summary estimates are presented using international dollars. International dollars are adjusted for differential purchasing power across countries. In Burkina Faso at the time of the intervention one international dollar was equivalent to 167 CFA (West African Francs). As we can observe, the ICER estimate varies significantly across subsets of the population by covariates. For example, the incremental cost for one extra institutional delivery among people with no education was I$168 compared to I$84 among people with some level of education (at least primary school level). The incremental cost of one extra institutional delivery among people within 5 km of the closest health facility was I$303 compared to I$383 among people at least 5 km away from the closest heath facility. These results confirm the existence of important subgroups, and thus the importance of assessing how these covariates affect the overall costeffectiveness of the Skilled Care Initiative. This could only be achieved through a joint probability distribution with all covariates which is possible with the netbenefit approach but not with stratified analysis of the traditional ICER.
Table 2 presents the coefficients of the netbenefit estimates, obtained from a classic (OLS) regression procedure, with netbenefit as outcome and the intervention dummy (SCI) as independent variable. The coefficient for the intervention (in this case the Skilled Care Initiative) corresponds to the incremental netbenefit and is equivalent to the standard ICER.
We used different ceiling ratios around the value of the computed ICER (approximately I$170) and as the ceiling ratios increase moving from zero to infinity, the coefficients (incremental netbenefits) obtained from the OLS regression on the intervention and covariates increase (Tables 3 and Table 4 in Appendix S1).
Tables 3 and 4 in Appendix S1 illustrate how the coefficient changes from a negative value (negative incremental netbenefit, the intervention not reaching an acceptable level of costeffectiveness) to a positive value (positive incremental netbenefit, the intervention deemed costeffective). In Table 3, for example, the probability that the intervention is costeffective varies from zero to 98% in a narrow interval between the values I$150 and I$180 (international dollars). Figure 2 highlights one of the major advantages of the netbenefit framework, that we can see clearly how different the ICER values are when adjusted or not to the covariates. It is therefore important to adjust the ICER to important subgroups. One can observe in Table 5 in Appendix S1 and Figure 3 that the probability of the intervention being costeffective when the ceiling ratio is 60,000 CFA (approximately I$360 international dollars) is only 24% for households living 5km or less from a health facility, whilst the corresponding probability for households living 5 km or more from a health facility is nearly 99%. This is a major difference to the standard ICER approach where it is not possible to indicate the probability that the intervention is costeffective adjusting for other covariates.
Putting the results in Figure 3 into practice, one can say that if policy makers in rural Burkina Faso were prepared to invest I$360 to achieve an extra institutional delivery the results will vary significantly for households whether they are situated closer (less than 5 km) or further away (more than 5 km) from a health facility. This is very important information for the scaling up process.
Given that education, distance to the closest health facility and asset ownership are major determinants of the effects of the Skilled Care Initiative, it is critical to be able to assess the effects on the marginal costeffectiveness of adjusting for these important covariates.
When running the OLS regression with netbenefit as the dependent variable, and education, distance to closest health facility, and asset quintiles as independent variables, the variability of the costeffectiveness acceptability curves (CEAC) by covariates (Figures 2 and 3) highlights the importance of adjusting the costeffectiveness analysis by covariates. We can explore for each covariate the difference in the probability that the Skilled Care Initiative (SCI) is costeffective when adjusted or not for the selected covariate but only a joint probability distribution with all covariates is appropriate. As for any regression analysis, we can assess the variability accounted for by the covariates and their probability of correlation with the intervention. It is important to point out that although education, distance to health facilities, and asset ownership are well documented determinants of health outcomes, these three factors appear to explain only a small fraction of the costeffectiveness of the Skilled Care Initiative in rural Burkina Faso (adjusted R^{2} around 13%).
Table 6 in Appendix S1 presents the results of the regression analysis with interaction terms. When running an OLS regression using the netbenefit framework with the interaction terms, the main interest is in the coefficients of the intervention dummy (SCI) which corresponds to the incremental netbenefit. The magnitude and significance of the coefficients on the interaction terms (interaction between the covariates and the intervention dummy) indicate how costeffectiveness of the Skilled Care Initiative is expected to vary at the margin. A large and statistically significant coefficient on an interaction will usually point to an important population subgroup. From Table 6 in Appendix S1, we may infer that there is an interaction between the intervention and the education dummy (first interaction term) and between the intervention dummy and asset ownership dummy (third interaction term) but no interaction between the intervention dummy and distance to health facility (second interaction term). However, the observed interactions may be due to collinearity between covariates and from this point we may decide to include or omit a covariate based on prior knowledge or practical implication, based on model diagnosis. The largest and most significant coefficients for collinearity are observed between distance and asset ownership, and between the intervention dummy and asset ownership quintiles (data not shown). We conclude that distance was the most important covariate among the three analyzed.
By applying the netbenefit framework, we were able to show that the probability (adjusting for the covariates) of the Skilled Care Intervention to achieve one extra institutional delivery in rural Burkina Faso when the ceiling ratio is approximately I$360 is only 24% for households living 5km or less from a health facility, whilst the corresponding probability for households living 5 km or more from a health facility is nearly 99%. This major piece of information (variability in probability in costeffectiveness by important subgroups) together with the possibility of providing some level of certainty (confidence intervals) around the ICER estimate would not have been possible just by using the traditional ICER method for this observational study. As pointed out by Hosh JS et al ^{[2]} the existence of important subgroups (which is the case for maternal and newborn health) affects how the costeffectiveness varies at the margin and need to be accounted for when analyzing and interpreting costeffectiveness results.
In this study, the small degree of variability explained by the three determinants (household education level, distance, asset ownership) may be explained by the existence of a more important unknown variable and/or the content validity of our constructed metrics of the covariates. It is well documented that the mix of assets affects the constructed wealth index and the constructed index may not reflect consumption and socioeconomic status well. Furthermore, the remoteness of the study context and similarity in household characteristics across the districts may result in a very low variability of the asset ownership quintiles ^{[19]}.
Our results point first to the importance of covariates (Figure 2). There is a clear difference between the probability that the intervention is costeffective at a given ceiling ratio when adjusted or not adjusted to the covariates. Distance appears to be the most important covariate in our study context. Cost effectiveness of the Skilled Care Initiative varies significantly by the average distance of households to the nearest health facility (after adjusting for all covariates).
Unlike some developed countries where extensive studies have been conducted to generate appropriate ceiling ratios for health outcomes and where there are evaluation bodies, such as NICE in the United Kingdom and CADTH in Canada, that continually update these estimates, researchers and policy makers in the developing world have almost no information about their contextual ceiling ratios. In these settings, the identification of an ICER point estimate where there is no information on uncertainties may be of little use as policy makers will have no basis to judge whether or not the ICER is good value for money. It is thus important to revise the current practice of the presentation and interpretation of costeffectiveness analysis results of maternal and newborn interventions to improve evidencebased decision making.
There are increasing opportunities through household surveys (Demographic and Health Surveys, Multiple Indicators Cluster Surveys, Living Standard Measurement Surveys) to construct personlevel or userlevel (household, individual) effect and cost (willingness to pay, outofpocket expenditures) data, necessary for application of the netbenefit framework.
A costeffectiveness analysis requires the comparison of like with like, which is not often the case with observational studies. However, in spite of inherent differences between any two populations, comparability of settings on specific characteristics within or between subsets of populations is important before the implementation of an intervention can be evaluated ^{[20]}. The analyst will need to identify and adjust for significant covariates or important subgroups within a population and descriptive statistics such mortality ratios and ICER estimates only do not account for these important covariates ^{[3]}. Lastly, constructing the database from an observational study to apply the net benefit framework requires careful planning. Typically, a clinical trial database will have patientlevel effect and cost data. In observational studies, only effect data are usually collected and additional (often separate) costing exercise conducted, sometimes through modeling, sometimes with field data collection. We were able to demonstrate that by carefully designing an observational study to collect concurrent patientlevel (household or individual) effect and cost data we can apply the netbenefit framework in maternal and newborn health.
We have demonstrated that the netbenefit framework is applicable to observational studies common in maternal and newborn health and that the cost effectiveness of a maternal and newborn health intervention varies by important covariates (sub groups). By adjusting the intervention costeffectiveness results to the covariates, we were able to identify distance to health facilities as an important determinant of the costeffectiveness analysis. We could not have reached the same conclusion employing the traditional ICER approach. These advantages in the presentation and interpretation of costeffectiveness analysis results, and especially in providing information on the marginal costeffectiveness of important covariates, necessitate that we revise the traditional methods and tools of maternal and newborn health household surveys to include household outofpocket expenditure for health outcomes. The recent development of household cost surveys alongside Demographic and Health Surveys (such as outofpocket expenditure data from National Health Accounts) offers broader opportunities for the construction of concurrent household level effect and cost datasets and application of the netbenefit framework for assessing the costeffectiveness of maternal and newborn health interventions.
Contains Table 4 (Using the netbenefit regression results to create costeffectiveness acceptability curves), Table 5 (Costeffectiveness acceptability curves from the netbenefit regression results with distance as covariate) and Table 6 (Covariates adjusted netbenefit regression estimates with different ceiling ratios, interactions).
(DOCX)
Click here for additional data file (pone.0040995.s001.docx)
Notes
Competing Interests: The authors have declared that no competing interests exist.
Funding: This research is part of a larger maternal health research programme, Immpact (Initiative for Maternal Mortality Programme Assessment), funded by the Bill & Melinda Gates Foundation, the UK Department for International Development, the European Commission and USAID. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
This work was undertaken as part of an international research programme – Immpact. A special thanks to Nicolas Meda, Peter Byass, and Bocar Kouyate for technical support to Sennen Hounton’s doctoral research. The views expressed herein are solely those of the authors.
Sennen Hounton is a medical epidemiologist with expertise in maternal and newborn health, health systems and economic evaluation. Sennen Hounton is an MD (Benin), MPH in Epidemiology (University of Oklahoma, USA) and a PhD in Public Health from University of Aberdeen, (Scotland, UK). He was a Senior Research Fellow with Immpact (Initiative for Maternal Mortality Program Assessment). He is currently Maternal Health Technical Adviser at the United Nations Population Fund (New–York), and served as Scientific and Technical Advisor on the WHO Alliance for Health Policy and System Research Scientific and Technical Advisory Committee.
David Newlands is a Senior Lecturer in Economics, previously Team Leader of the Economic Outcomes of an international research programme – Immpact. He is currently a Senior Lecturer at the Business School of University of Aberdeen, Scotland, UK.
References
1.  Ronsmans C,Graham WJ. Year: 2006Lancet Maternal Survival Series steering group: Maternal mortality: who, when, where, and why.Lancet, 368(9542)1189200 
2.  Hounton S,Chapman G,Menten J,De Brouwere V,Ensor T,et al. Year: 2008Accessibility and utilisation of maternity care within a Skilled Care Initiative in rural Burkina Faso.Tropical Medicine and International Health, 13: Suppl 1, p 44–52 
3.  Hounton S,Byass P,Bassane B. Year: 2009Towards maternal and perinatal mortality reduction in rural Burkina Faso: community are not empty vessels.Global Health Action72 
4.  Drummond MF,Sclpher MJ,Torrance GW,O’Brien BJ,Stoddart GL. Year: 2005Methods for the economic evaluation of health care programmes.Oxford University Press, 3rd Ed, New York 396 p 
5.  Adam T,Lim SS,Mehta S,Bhutta ZA,Fogstad H,et al. Year: 2005Cost effectiveness analysis of strategies for maternal and neonatal health in developing countries.BMJ, 12331(7525)1107 
6.  Edejer TT,Aikins M,Black R,Wolfson L,Hutubessy R,et al. Year: 2005Cost effectiveness analysis of strategies for child health in developing countries.BMJ, 331(7526)1177 
7.  Hogan DR,Baltussen R,Hayashi C,Lauer JA,Salomon JA. Year: 2005Cost effectiveness of strategies to combat HIV / AIDS in developing countries.BMJ331(7530)1431716282380 
8.  Briggs A,Fenn P. Year: 1998Confidence intervals or surfaces? Uncertainty of the costeffectiveness plane.Health Econ77237409890333 
9.  Briggs AH,Gray AM. Year: 1999Handling uncertainty when performing economic evaluation of healthcare interventions.Health Technol Assess, 3 
10.  Ternent L,McNamee P,Newlands D,Belemsaga D,Gbangou A,et al. Year: 2010Willingness to pay for maternal health outcomes: are women willing to pay more than men?Appl Health Econ Health Policy, 8(2)99109 
11.  Stinnett AA,Mullahy J. Year: 1998Net health benefits: a new framework for the analysis of uncertainty in costeffectiveness analysis.Med Decis Making, 18(2 suppl)S6880 
12.  Hoch JS,Briggs AH,William AR. Year: 2002Something old, something new, something borrowed, something blue: a framework for the marriage of health econometrics and costeffectiveness analysis.Health Econ1141543012112491 
13.  Hoch JS,Rockx MA,Krahn AD. Year: 2006Using netbenefit regression framework to construct costeffectiveness acceptability curves: an example using data from a trial of external loop recorders versus Holter monitoring for ambulatory monitoring of “community acquired” syncope.BMC Health Services Research66816756680 
14.  Newnlands D,YugbareBelemsaga D,Ternent L,Hounton S,Chapman G. Year: 2008Assessing the costs and costeffectiveness of a Skilled Care Initiative in rural Burkina Faso.Tropical Medicine and International Health13(1)616718578813 
15.  Hoch JS,Dewa CS. Year: 2007Lessons from trialbased costeffectiveness analysis of mental health interventions. Why uncertainty about outcomes, estimates and willingnesstopay matters.Pharmacoecnomics, 25(10)807816 
16.  Hoch JS,Blume JD. Year: 2008Measuring and illustrating statistical evidence in a costeffectiveness analysis.Journal of Health Economics2747649518179834 
17.  Claxton K. Year: 1999The irrelevance of inference: a decision making framework to the stochastic evaluation of health care technologies.J Health Econom18341364 
18.  Fenwick E,Claxton K,Sculpher M. Year: 2001Representing uncertainty: the role of costeffectiveness acceptability curves.Health Econ1077978711747057 
19.  Filmer D,Pritchett LH. Year: 2001Estimating wealth effects without expenditure data–or tears: an application to educational enrollments in states of India.Demography38(1)1153211227840 
20.  Victora CG,Black RE,Boerma JT,Bryce J. Year: 2011Measuring impact in the Millennium Development Goal era and beyond: a new framework to largescale effectiveness evaluations.Lancet, 1377(9759)8595 
Figures
Tables
Table 1 Sample statistics from costeffectiveness analysis of the SCI, Ouargaye and Diapaga districts, Burkina Faso.
Group variables  Mean  SD  SE 
Overall analysis  
Comparison district (N = 48 272)  
Cost^{*}  1042  –  – 
Effect^{**} (%)  31.5  0.464  0.002 
Intervention district (N = 40 469)  
Cost^{*}  4576  –  – 
Effect^{**} (%)  44  0.496  0.002 
Cost increment  3534  –  – 
Effect increment (%)  12.4  0.003  
Sample ICER^{***}  28430 (I$170) 
*Average cost (heath system perspective) for maternal health provision in West Africa Francs.
**Institutional delivery.
***(Ratio of incremental cost to incremental effect).
Summary estimates are presented in the manuscript using international dollars (international dollars are ones adjusted for differential purchaising power. In Burkina Faso at the time of the intervention it was around 167, WHO, 2005) to enable easy understanding of international readers.
Table 2 Simple netbenefit regression estimates with different ceiling ratios, Skilled Care Initiative, Burkina Faso.
Group variables  Mean  SD  SE  
Overall analysis  
Cost increment  3534  –  –  
Effect increment (%)  12.4  0.003  
Sample ICER***  28430 (I$170)  
Stratified analysis by education  
None  
Cost difference  3534  
Effect difference (%)  13  –  0.003  
Sample ICER  28 115 (I$168)  
Some  
Cost difference  3534  
Effect difference (%)  25  –  0.015  
Sample ICER  13 950 (I$84)  
Stratified analysis by distance  
Less than 5 km  
Cost difference  3534  
Effect difference (%)  7  0.004  
Sample ICER  50 558 (I$303)  
More than 5 km  
Cost difference  3534  
Effect difference (%)  5.5  0.005  
Sample ICER  63 906 (I$383)  
Incremental netbenefit  Net Monetary Benefit coefficients (SE)  
Ceiling ratios Overall  Edu = None  Edu = Some  Dist:>5 km  Dist: ≤5 km  
Ro = 0–3534 (0.0)  −3534 (0.0)  −3534 (0.0)  −3534 (0.0)  −3534 (0.0)  
R = 15 000–1670 (48)  265 (226)  −1648 (50)  −2705 (76)  −2485 (59)  
R = 25 000–426 (80)  2798 (377)  −391 (83)  −2152 (126)  −1786 (98)  
R = 35 000 817 (113)  5331 (528)  867 (116)  −1599 (176)  −1086 (137) 
Edu: Education, Dist: Distance to health facility.
Summary estimates are presented in the manuscript using international dollars (international dollars are ones adjusted for differential purchaising power. In Burkina Faso at the time of the intervention it was around 167) to enable easy understanding of international readers.
Table 3 Simple netbenefit regression estimates with different ceiling ratios, Skilled Care Initiative, Burkina Faso.
Explanatory Variables  NMB  NMB  NMB  NMB  NMB  NMB 
With Ro = 0^{a}  With R 15000  With R 25000  With R 35000  With R 45000  With R 55000  
[SE]  [SE]  [SE]  [SE]  [SE]  [SE]  
(pvalue)  (pvalue)  (pvalue)  (pvalue)  (pvalue)  (pvalue)  
Constant term  −1042  3682  6832  9981  13130  16280 
[0]  [33]  [55]  [76]  [98]  [120]  
(0.000)  (0.000)  (0.000)  (0.000)  (0.000)  (0.000)  
Intervention strategy (SCI)  −3534  −1670  −426  816  2060  3203 
[0]  [48]  [80]  [113]  [145]  [178]  
(0.000)  (0.000)  (0.000)  (0.000)  (0.000)  (0.000)  
R^{2} (adjusted)  1  0.013  0.000  0.001  0.002  0.004 
F (1, 88741)  –  1185  27.8  52  200  345 
Prob > F  –  <0.000  <0.000  <0.000  <0.000  <0.000 
aWhen Ro = 0, NMB = −Cost.
Summary estimates are presented in the manuscript using international dollars (international dollars are ones adjusted for differential purchaising power. In Burkina Faso at the time of the intervention it was around 167) to enable easy understanding of international readers.
Article Categories:

Previous Document: Temporal patterns of medications dispensed to children and adolescents in a national insured populat...
Next Document: Heart failure and preserved left ventricular function: long term clinical outcome.