The Lufwanyama Neonatal Survival Project (“LUNESP”) was a cluster randomized, controlled trial that showed that training traditional birth attendants (TBAs) to perform interventions targeting birth asphyxia, hypothermia, and neonatal sepsis reduced all-cause neonatal mortality by 45%. This companion analysis was undertaken to analyze intervention costs and cost-effectiveness, and factors that might improve cost-effectiveness.
Methods and Findings
We calculated LUNESP's financial and economic costs and the economic cost of implementation for a forecasted ten-year program (2011–2020). In each case, we calculated the incremental cost per death avoided and disability-adjusted life years (DALYs) averted in real 2011 US dollars. The forecasted 10-year program analysis included a base case as well as ‘conservative’ and ‘optimistic’ scenarios. Uncertainty was characterized using one-way sensitivity analyses and a multivariate probabilistic sensitivity analysis. The estimated financial and economic costs of LUNESP were $118,574 and $127,756, respectively, or $49,469 and $53,550 per year. Fixed costs accounted for nearly 90% of total costs. For the 10-year program, discounted total and annual program costs were $256,455 and $26,834 respectively; for the base case, optimistic, and conservative scenarios, the estimated cost per death avoided was $1,866, $591, and $3,024, and cost per DALY averted was $74, $24, and $120, respectively. Outcomes were robust to variations in local costs, but sensitive to variations in intervention effect size, number of births attended by TBAs, and the extent of foreign consultants' participation.
Citation: Sabin LL, Knapp AB, MacLeod WB, Phiri-Mazala G, Kasimba J, Hamer DH, et al. (2012) Costs and Cost-Effectiveness of Training Traditional Birth Attendants to Reduce Neonatal Mortality in the Lufwanyama Neonatal Survival Study (LUNESP). PLoS ONE 7(4): e35560. https://doi.org/10.1371/journal.pone.0035560
Editor: Sten H. Vermund, Vanderbilt University, United States of America
Received: November 29, 2011; Accepted: March 19, 2012; Published: April 24, 2012
Copyright: © 2012 Sabin et al. This is an open-access 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.
Funding: This work was supported by a cooperative agreement between Boston University and the Office of Health and Nutrition of the United States Agency for International Development, GHS-A-00-03-00020-00. NIH/NIAIDS K23 AI 62208 supported Dr. Gill's effort. Additional support for the neonatal resuscitation protocol intervention was provided by a Neonatal Resuscitation Program Grant from the American Academy of Pediatrics, which also contributed toward the costs of Dr. Guerina's travels to Zambia, some of the field office expenditures, and purchase of safe delivery kits. UNICEF also kindly provided safe delivery kits. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests: The authors have declared that no competing interests exist.
In many low-resource countries, neonatal mortality contributes approximately 40% of all under five-year mortality, with birth asphyxia and neonatal sepsis constituting the majority of preventable neonatal deaths , . A major factor in high neonatal mortality is the high proportion of home births. It is estimated that 60 million deliveries occur each year outside of health facilities ; in many low-resource areas over 50% of all births occur in the community, largely in mothers' homes . In many areas, traditional birth attendants (TBAs) are an essential source of basic obstetrical care. TBAs have proven effective in a variety of secondary roles in the community, such as serving as peer educators or breast feeding counselors, but their proximity to the mother/infant pair and their location within the community suggests that TBAs could play a more direct role in reducing neonatal deaths as well.
To test the hypothesis that TBAs could effectively reduce neonatal mortality in a rural African setting, we conducted a randomized controlled effectiveness study in Zambia called the Lufwanyama Neonatal Survival Project (LUNESP) . In LUNESP, TBAs were randomized to receive training and equipment to allow them to perform a set of interventions targeting several of the main causes of neonatal mortality. The interventions included two components: 1) Neonatal Resuscitation Protocol (NRP), which aimed to reduce deaths due to birth asphyxia and hypothermia; and 2) antibiotics with facilitated referral to a health center (AFR), which aimed to reduce sepsis deaths during the first month of life. The primary endpoint was a comparison of mortality rates by day 28 of life among live-born infants. Based on outcomes for 3355 deliveries, we found that infants delivered by intervention TBAs were 45% less likely to die than infants delivered by control TBAs (RR 0.55, 95% CI 0.33 to 0.90). This equated to one death avoided for every 56 deliveries attended by an intervention TBA, for an absolute reduction of 18 deaths per 1000 live births .
Here we report on the economic evaluation component of the LUNESP project. We present our findings from several perspectives. First, we present a financial analysis, based on the actual costs incurred during the intervention. Second, we present an economic analysis, which utilized a societal perspective of costs. Lastly, as evidence for policy-makers, we present a forecasted economic analysis in which the costs and cost-effectiveness of a modified model of the intervention are projected over a ten-year timeframe.
LUNESP was a cluster-randomized trial conducted from 2006–2008 in an impoverished rural district in north-central Zambia among a population with limited access to health care . A total of 120 TBAs were initially randomized to receive the intervention trainings and equipment, or to continue their existing standard of care. All TBAs received one ‘clean delivery’ kit per birth for their regular TBA duties, which included a delivery sheet, cord cutter, cotton cord ties, latex gloves, and soap. Training for intervention TBAs commenced with 4-day sessions for each group of 30 TBAs, followed by 1–2 day refresher trainings approximately every 3–4 months for the duration of the trial. The trainings were conducted collaboratively by a US-based neonatologist and a local master trainer (a Zambian nurse-midwife), assisted by 6–8 Zambian facilitators. To demonstrate competency, intervention TBAs indicated skills retention at each retraining session. Each intervention TBA received one resuscitator mask, a polypropylene bottle with chlorinated water, and a laminated reference card summarizing NRP and trigger conditions for AFR. They also received each of the following per delivery: two flannel receiving blankets, a soft rubber bulb syringe, two 250 mg amoxicillin capsules, one 2-ounce mixing cup and spoon, and a 3 ml oral syringe. A more detailed description of the design, training, and analytic methods used in LUNESP has been published elsewhere .
The study was approved by the ethical committees at Boston University Medical Center (Boston, MA) and the Tropical Diseases Research Centre (Ndola, Zambia). All TBAs and mothers who participated in the LUNESP trial provided written informed consent, using forms in English and the local languages Bemba and Lamba. We did not obtain separate informed consent from participants for the present analysis because we utilized de-identified, aggregated outcome data only from the trial. The LUNESP trial was registered as clinicaltrials.gov NCT00518856.
Analyses were conducted from three perspectives: 1) a financial analysis based on costs incurred during LUNESP; 2) an economic analysis, which factored in societal costs; and 3) a forecasted 10-year economic analysis, which modeled expected intervention economic costs and effectiveness over a future program-appropriate timeframe in order to provide practical information for policy-makers beyond data generated by a short-term research study alone.
The financial analysis was based on the incremental expenditures related to the 33-month intervention. This encompassed 6 months of start-up activities in February–July 2006, including the first training and 27 months of implementation from August 2006-October 2008) (Table 1). All research- and control group-related costs were excluded. We included personnel time for: start-up (US-and Zambia-based collaborators); program monitoring and supervision (Zambian project staff); project coordination (Zambian project staff); and training. We included the costs of travel, food, and accommodation associated with training. Some travel expenses by the US-based neonatologist were defrayed due to cost-sharing with other projects; these savings were incorporated. TBAs were not paid by the government or the study and thus no salaries were included.
Supply costs included training items (resuscitation mannequins, printing, and stationary); single items provided to each TBA at the beginning of the intervention such as a mask, instruction card, and a mixing spoon; and items needed for each birth, including blankets and bulb syringes. Some supplies were purchased locally and others were purchased in the US and brought to Zambia by project staff. We included a one-time customs charge applied to supplies carried into Zambia and a shipping fee for items shipped from the US in 2006. The District Health Management Team (DHMT) was responsible for providing amoxicillin and clean delivery kits, though due to occasional supply issues, the program made several purchases of both items, the costs of which were included. Nominal Zambia-based costs were converted to US dollar values using the average annual exchange rate for the year in which the costs were incurred  and then added to nominal US dollar-based costs. We adjusted total annual US dollar costs by US inflation rates (Consumer Price Index)  and expressed total program costs in real 2006 US dollar figures, categorized as fixed and variable costs over the life of the project. We also estimated the 2006 Present Value (PV) of financial costs (using a discount rate of 3%) , and the 2011 PV for greater comparability with additional analyses.
As recommended by the World Health Organization, we utilized a societal perspective for the economic analysis to better capture the economic resource cost of intervention inputs . First, we included payment to TBAs for their participation. This was based on the supply price of TBAs, estimated by calculating the difference in reported average cash payments per birth made by families to intervention TBAs vs. control TBAs. Second, we eliminated cost-sharing of expenses (i.e., subsidized travel of US-based neonatologist, donations of key supplies). Third, we included the cost of amoxicillin tablets throughout the project, using an estimated international price of US$ 0.1 per 250 mg dose (but assumed consistent delivery kits by the DHMT per their mandate) . Fourth, we deducted the customs tax payment, since taxes are an economic transfer and not an actual resource cost. All US dollar nominal costs were adjusted for US inflation rates , discounted at 3%, and expressed in 2011 US dollar PV terms.
Forecasted 10-year analysis.
We utilized a societal perspective, with a ten-year program lifespan from January 2011 through 2020. We assumed a similar model as that of the trial: 6 months of start-up, 60 participating TBAs, initial intensive 4-day training followed by three 2-day refresher trainings per year, the same supply inputs, and similar levels of local personnel participation in training and monitoring and supervision activities. However, we made several adjustments to the model to localize the intervention and increase its long-term sustainability. We assumed task-shifting by replacing the US-based neonatologist with a local skilled nurse-midwife who would have 5 full-time facilitators for assistance with trainings. We anticipated that TBA turnover would be minimal over time, with training of new TBAs managed by including them in ongoing refresher training rather than holding additional new trainings. US program management was replaced with a team of 3 local staff who would establish the intervention over 30 days. We also estimated compensation for TBAs on the basis of the per birth payments made by families to intervention TBAs during the trial. All nominal costs included expected inflation , ; nominal TBA compensation also incorporated a real annual increase of 1%. We assumed no improvements in or loss of efficiency over time, resulting in similar annual costs after the start-up period. Annual Zambia-based costs in local currency were converted to nominal US dollars using an expected constant real exchange rate equal to the 2011 rate  and added to annual US dollar costs. Total annual costs were deflated by expected inflation and discounted (at 3%) to estimate the 2011 PV.
We estimated the cost effectiveness of the LUNESP interventions using the formula: ICEA = (CI−CC)/(MI−MC), where ICEA is the incremental cost-effectiveness ratio; CI and CC are the total (discounted) costs related to the intervention and control groups, respectively; and MI−MC are the (discounted) mortality figures for intervention and control groups, respectively. Because all costs were incremental, CC = 0. The difference in mortality per year, (MI−MC)t, where t = years 2006, 2007, 2008, was the difference in measured mortality rates per 1,000 live births between intervention and control groups in each project year. Annual lives saved were discounted at 3% per year. We also estimated the number of averted disability-adjusted life years (DALYs), using an approach that incorporates expected years of life lost (not country-specific), with time lived at different ages valued using an exponential function and discounted at 3% .
For the forecasted CEA we conducted three analyses. First, our base case used the estimated economic costs for the 10-year program and assumed that reduced mortality would equal the trial's overall effect size (17.9/1000 births). The number of expected births per TBA was defined as the annual number observed in 2008 (when TBA activity was at its peak) (1.29 births per TBA per month). We estimated both undiscounted and discounted (at 3%) annual deaths avoided, though used the latter to calculate the incremental economic cost per death avoided and DALY averted. In addition, we examined the cost-effectiveness of two alternative ‘optimistic’ and ‘conservative’ scenarios. The former retained the trial's mortality effect and assumed program implementation under more stringent conditions by increasing the number of TBAs per cohort trained from 60 to 80, reducing the refresher training time from 2 to 1 days, reducing monitoring activities from a monthly to bi-monthly timetable, and assuming births based on Zambia's average national density rather than sparsely-populated Lufwanyama (see Table 2). The conservative scenario, relative to the base case, involved reducing the effect size by 25% to 13.4/1000 births, decreasing the mean monthly number of births per TBA to the overall average observed in the trial (1.21), and increasing the number of annual refresher trainings from 3 to 4.
Interpretations regarding ‘cost-effectiveness’ were based on the 2001 recommendation of the Commission on Macroeconomics and Health and adopted by the World Health Organization (WHO), whereby an intervention is ‘highly cost-effective’ if it averts a DALY for less than per capita GDP (Gross Domestic Product) and ‘cost-effective’ if it averts a DALY for less than 3 times per capita GDP .
For the base case, we conducted univariate sensitivity analyses to explore the impact on the costs and cost effectiveness of the 10-year program of varying key input parameters. Certain variables were uncertain—mortality effects, number of births per TBA, and input costs. Others were pre-determined, either wholly or in part, by program or policy staff, including the number of refresher training workshops per year, number of days of each workshop, degree of involvement of the US-based neonatologist in training activities, and number of participating TBAs trained per workshop.
We also conducted multivariate sensitivity analysis with Monte-Carlo simulations using Crystal Ball™ (Oracle Corp, Redwood Shores, CA) to test the model's robustness for each of the three forecasted scenarios given simultaneous changes in key parameters, an approach commonly employed in cost-effectiveness analyses –. Based on the results of the one-way sensitivity analysis for the base case, we selected the four uncertain input parameters (e.g., non-policy determined) whose variations had the greatest impact on the cost-effectiveness outcomes to include in this analysis: intervention effect size, average deliveries/month/TBA, logistic costs of trainings (e.g., room and board), and cost of travel for monitoring and supervision. The probability distributions for input parameters were assumed to be triangular. For the mortality effect, the likeliest value was taken from the trial, with minimum and maximum ranges set at the trial's 95% confidence interval (CI). For the number of births per TBA, we used the trial mean and standard deviation to estimate a 95% CI. For 2011 cost parameters, we used the 2011 estimated costs for likeliest values with minimum and maximum ranges of ±25% (Table 2), an approach frequently employed in similar cost studies .
Intervention costs and cost-effectiveness
In the financial analysis, the program intervention costs in real 2006 US$ were $47,020, $32,599, and $29,282 in 2006, 2007, and 2008, respectively (Table 3). After discounting, the 2006 PV was $106,271. The cost of start-up and supplies, as well as the longer training workshop in June 2006, was primarily responsible for the higher annual costs in 2006 relative to 2007–08. The cost difference between 2007 and 2008 reflected the fact that the study was implemented for only 10 months in 2008. Approximately two-thirds of the cost of the program's first year was US-based, whereas in 2007 and 2008, US-based costs represented less than 30% of total costs. Twenty-seven percent of total financial costs were for the US-based neonatologist.
LUNESP's total economic cost was 8% higher than the total financial cost, mainly due to costs associated with the US-based neonatologist's attendance at all workshops (Table 4). Fixed costs accounted for nearly 90% of economic costs, among which the largest component was training activities (63–65% of total annual costs). Thirty-one percent of all economic costs were for the US-based neonatologist.
In the forecasted analysis, the estimated 2011 PV of the base case program was $256,455, with annual costs of $26,834, far lower than in the financial and economic analyses of the 2006–08 intervention, where annual costs were $49,469 and $53,550, respectively (Table 4). Lower annual cost resulted largely from spreading out first year start-up costs over a longer timeframe. Moreover, in contrast with the cost drivers for the 2006–08 program, training for the 10-year program comprised less than 50% of total costs, rather than nearly two-thirds, whereas monitoring and supervision represented almost one-third of total costs, or about double the proportion of LUNESP's costs. These differences were mainly due to task shifting from the US-based neonatologist to a local trainer (i.e., the lead Zambian trainer, a nurse-midwife). Despite savings from these adjustments, fixed costs were still projected to account for over 80% of total intervention cost. Task-shifting and compensating TBAs had minimal effect on the basic cost structure of the program. Given the low prices of variable cost items—infant receiving blankets, bulb syringes, amoxicillin tablets, and compensation to TBAs—fixed costs still far exceeded variable costs. Overall, the model predicted that the estimated cost per birth would be less than one-half that of LUNESP's economic resource cost (US$29 vs. $68). The variable cost per birth, representing the additional cost of having a trained TBA attend a delivery, was US$5.3, compared to US$8.4 for the 2006–08 trial.
The cost per neonatal death avoided and DALYs averted for both the 2006–08 program and the 2011–2020 program are summarized in Table 5. From the 33 neonatal deaths that were avoided (undiscounted) in 2006–08, we estimated that 158 deaths would be avoided in 2011–2020. For the 2006–08 program, the economic cost per death avoided was $3,900 while the cost per DALY averted was $176 (2011 US dollars). The 10-year forecasted program was much more cost-effective than LUNESP. The base case analysis predicted that the cost per death avoided was $1,866, while the cost per DALY averted was $74. ‘Optimistic’ vs. ‘conservative’ scenarios yielded values of $591 vs. $3,024 and $24 vs. $120 for cost per death avoided and DALY averted, respectively. Given that Zambia's 2011 GDP per capita was estimated to be just under $1400 (in current prices) , this analysis suggested that all scenarios of the intervention, including the conservative scenario, would be highly cost-effective.
The one-way sensitivity analyses showed that the 10-year base case results were most sensitive to variation in two policy-determined variables: the degree of involvement of the US-based neonatologist and number of program TBAs (Table 6). If the neonatologist attended all 3 trainings each year instead of none, the total cost per outcome (death avoided, DALY averted, and birth attended by TBA) would nearly double. Similarly, if the trainings expanded to include 100 TBAs per session rather than 60, cost-effectiveness would increase by approximately 20%. The results were also sensitive to variation in the effect size of the intervention and productivity of a given TBA (i.e., the average monthly number of deliveries attended). If the intervention reduced neonatal mortality by 8.0 (vs. 17.9 in the base case) per 1,000 live births, the cost per death avoided and DALY averted would increase by over 100%. If the intervention was implemented in an area where TBAs helped deliver more babies, e.g., 5 instead of 1–2 per month, cost-effectiveness would improve by more than 50%.
The multivariate sensitivity analyses indicated that the results were robust to simultaneous variation in key input parameters (Table 7). The forecasted median values of cost per death avoided and DALY averted, for all scenarios, were within 5–6% of the point estimates. The 90% probability values were generally less than twice the point estimates. With 90% probability, the cost per DALY averted of the base case scenario would be less than $148, and for the conservative scenario would be less than $276. Thus, even under very conservative policy-determined conditions and extreme outcome conditions, the intervention should still be considered highly cost-effective.
This analysis indicted that training TBAs to perform interventions targeting birth asphyxia, hypothermia, and sepsis was highly cost-effective according to the WHO-accepted definition of cost-effectiveness. Furthermore, modifications to localize investments and spread them out over a 10-year program horizon would significantly enhance the cost effectiveness of the intervention. Some studies have identified even more cost-effective interventions to improve child survival in certain settings such as India ($7 per DALY averted for home-based neonatal care) , and in a number of countries in the same African region as Zambia ($8 per DALY averted for community newborn care) . However, our results compare favorably with the range of interventions presented by the WHO as cost-effective in very low-resource environments, such as delivery by a skilled birth attendant ($37 per DALY averted) and community-based management of serious neonatal infections ($54 per DALY averted) .
Important drivers of our results were the intervention's effect size (reduced mortality) and high fixed costs. The latter was related to the nature of the intervention, which requires substantial training and retraining over time. However, as the sensitivity analysis and 10-year model illustrate, there are ways to task-shift and adapt the training to dramatically reduce some of these costs. By contrast, the former was entirely a function of the local population structure and the low average number of deliveries the TBAs attended—largely due to Lufwanyama's relatively sparse population density (∼6.4 persons/kilometer, roughly one third of Zambia's average ). Thus, the greatest economies are likely to be achieved when TBAs have the opportunity to apply their skills more frequently. In this regard, LUNESP's cost-effectiveness might underestimate the cost-effectiveness that could be achieved in a more densely populated setting.
Strengths and weaknesses
The prospective collection of cost data during the course of an actual field study, rather than estimating costs, is an important strength of the study. Similarly, the intervention's effect size was measured directly from a methodologically-rigorous randomized controlled trial, rather than from assumptions or expert opinion. Both of these enhance the validity of our findings. The forecasted economic analysis takes the analysis one step further, creating several potential scenarios of how an intervention of this kind, conducted in a similar population but over a 10-year time scale, might perform programmatically. In addition, our sensitivity analysis identified several policy- or program-determined variables that could substantially reduce costs and increase sustainability. This provides useful information for policy makers interested in achieving the maximum yield for public health investments.
Several potential limitations of the 10-year projected intervention merit discussion. First, our results are derived from one particular setting. While they may be generalizable to similar contexts characterized by low population density and underdeveloped transportation and health systems where home births are often the only option, the precise degree to which the efficiencies we achieved in the 10-year program can be gained in other settings is uncertain. For example, the LUNESP TBAs were highly motivated, largely because of their participation in a foreign-managed project. Whether TBAs participating in a completely locally-run program would be equally eager and conscientious in applying their skills is impossible to predict. However, it could also be argued that the Lufwanyama context represented a relatively extreme situation in terms of the difficulties of introducing an intervention of this kind. The LUNESP TBAs served a vast geographic area, which made it more difficult for them to access their clients and increased the logistical complexity and costs associated with training workshops and program monitoring. The TBAs' educational levels were also low; 17% had never attended school and only 13% had progressed beyond primary school . Moreover, as mentioned above, LUNESP's TBAs attended relatively few deliveries (an average of 1.2 each month), which had an adverse impact on the intervention's impact and hence lowered cost-effectiveness.
Second, we modified the trial's protocol in the 10-year program, such as the retraining frequency and the number of TBAs per training. In reality, there are limits to the changes in training one can make without decreasing intervention effectiveness. In support of this, Carlo et al, working with an unrelated group of Zambian TBAs, observed significant attrition of skills when TBAs were reassessed six months after primary training . Therefore, making such changes would require a monitoring strategy to ensure that essential skills are acquired and retained. However, it is important to bear in mind that in sparsely-populated areas such as Lufwanyama, where TBAs average 1–2 deliveries monthly, skill retention is likely poorer than in settings where TBAs can attend closer to 5–6 deliveries per month. In more populous areas, the increased utilization of skills could be self-reinforcing and thus permit fewer refresher workshops. This emphasizes the need to assess TBAs' skills regularly and systematically in order to optimize retraining efforts.
Third, our results assume that task shifting from expensive and experienced US-based experts to local experts paid at local wages will not compromise the intervention's effectiveness. We consider this a reasonable assumption: neonatal resuscitation is not technically complex to teach or to perform, and should not require the very high expertise of tertiary care center neonatologist. In other settings, training skills have been transferred successfully to local trained staff such as mid-wives .
Fourth, we lacked the data to incorporate the treatment of conditions for children who survived due to the work of the TBAs (additional costs) as well as the treatment of perinatal complications which were averted by the involvement of the TBAs (cost savings). The incorporation of such additional potential costs and savings would inevitably strengthen our analyses.
Finally, in our forecasted model, we used a simple, linear progression of costs over time after making initial baseline cost assumptions for year 1 (2011), rather than employing more sophisticated techniques, including non-linear cost projections. However, handling the inherent uncertainty of future values with sensitivity analysis encompassing a range of scenarios and a Monte Carlo simulation analysis is a rigorous approach that is intuitive and transparent. In addition, a rigorous cost-effectiveness analysis of a nationwide version of the intervention might provide even more relevant information for policy-makers. Yet given the variation in so many key parameters that such an analysis would need to encompass, and our lack of reliable data on these varying parameters, we do not believe such an analysis would produce reliable results. We hope instead that our rigorous analysis of a modestly-sized program, based as it is on trial-generated data with transparent modifications and a careful examination of uncertain variables, can be used for programmatic purposes in settings similar to that of LUNESP.
Conclusion and policy implications
This analysis indicates that training TBAs in practical skills targeting the key causes of neonatal mortality is highly cost-effective. Moreover, it shows that the cost-effectiveness of this strategy can likely be improved if implemented programmatically over a longer time horizon and with deliberate cost-saving measures. These results further strengthen the rationale for implementing programs similar to LUNESP in other disadvantaged communities with extremely limited access to health care.
We gratefully acknowledge the many contributions from members of the LUNESP study team. We also appreciate the strong support of CGHD faculty and staff, including Jonathon Simon, Deirdre Pierotti, and Donald Thea. We thank Joseph Tham and Yesim Tozan for their comments on the analysis.
Conceived and designed the experiments: LLS ABK WBM GP-M JK DHH CJG. Performed the experiments: LLS CJG. Analyzed the data: LLS. Contributed reagents/materials/analysis tools: LLS. Wrote the paper: LLS ABK WBM GP-M JK DHH CJG. Collected and organized additional data for the analysis: LLS ABK.
- 1. Black RE, Cousens S, Johnson HL, Lawn JE, Rudan I, et al. Global, regional, and national causes of child mortality in 2008: a systematic analysis. Lancet 375: 1969–1987.RE BlackS. CousensHL JohnsonJE LawnI. RudanGlobal, regional, and national causes of child mortality in 2008: a systematic analysis.Lancet37519691987
- 2. Lawn JE, Cousens S, Zupan J (2005) 4 million neonatal deaths: when? Where? Why? Lancet 365: 891–900.JE LawnS. CousensJ. Zupan20054 million neonatal deaths: when? Where? Why?Lancet365891900
- 3. Darmstadt GL, Lee AC, Cousens S, Sibley L, Bhutta ZA, et al. (2009) 60 Million non-facility births: who can deliver in community settings to reduce intrapartum-related deaths? Int J Gynaecol Obstet 107: Suppl 1S89–112.GL DarmstadtAC LeeS. CousensL. SibleyZA Bhutta200960 Million non-facility births: who can deliver in community settings to reduce intrapartum-related deaths?Int J Gynaecol Obstet107Suppl 1S89112
- 4. Montagu D, Yamey G, Visconti A, Harding A, Yoong J (2011) Where Do Poor Women in Developing Countries Give Birth? A Multi-Country Analysis of Demographic and Health Survey Data. PLoS One 6: e17155.D. MontaguG. YameyA. ViscontiA. HardingJ. Yoong2011Where Do Poor Women in Developing Countries Give Birth? A Multi-Country Analysis of Demographic and Health Survey Data.PLoS One6e17155
- 5. Gill CJ, Phiri-Mazala G, Guerina NG, Kasimba J, Mulenga C, et al. Effect of training traditional birth attendants on neonatal mortality (Lufwanyama Neonatal Survival Project): randomised controlled study. BMJ 342: d346.CJ GillG. Phiri-MazalaNG GuerinaJ. KasimbaC. MulengaEffect of training traditional birth attendants on neonatal mortality (Lufwanyama Neonatal Survival Project): randomised controlled study.BMJ342d346
- 6. OANDA website. Historical Exchange Rates. OANDA website. Historical Exchange Rates.Available: http://www.oanda.com/currency/historical-rates/. Accessed 2012 March 10. Available: http://www.oanda.com/currency/historical-rates/. Accessed 2012 March 10.
- 7. United States Department of Labor: Bureau of Labor Statistics. Consumer Price Index: All Urban Consumers. United States Department of Labor: Bureau of Labor Statistics. Consumer Price Index: All Urban Consumers.Available: ftp://ftp.bls.gov/pub/special.requests/cpi/cpiai.txt. Accessed 2012 March 10. Available: ftp://ftp.bls.gov/pub/special.requests/cpi/cpiai.txt. Accessed 2012 March 10.
- 8. Gold MR, Siegel JE, Russell LB, Weinstein MC, editors. (1996) Cost Effectiveness in Health and Medicine. New York: Oxford University Press. MR GoldJE SiegelLB RussellMC Weinstein1996Cost Effectiveness in Health and MedicineNew YorkOxford University Press
- 9. Edejer T, Baltussen RM, Adam T, Hutubessy R, Acharya A, et al. (2003) Making choices in health: WHO guide to cost-effectiveness analysis. Geneva 312.T. EdejerRM BaltussenT. AdamR. HutubessyA. Acharya2003Making choices in health: WHO guide to cost-effectiveness analysis.Geneva312
- 10. Management Sciences for Health. International Drug Price Indicator Guide: 2010. Management Sciences for Health. International Drug Price Indicator Guide: 2010Available: http://erc.msh.org/dmpguide/pdf/DrugPriceGuide_2010_en.pdf. Accessed 2011 September 5. Available: http://erc.msh.org/dmpguide/pdf/DrugPriceGuide_2010_en.pdf. Accessed 2011 September 5.
- 11. International Monetary Fund. World Economic Outlook Database April 2011: Report for Selected Countries and Subjects. International Monetary Fund. World Economic Outlook Database April 2011: Report for Selected Countries and Subjects.Available: http://www.imf.org/external/pubs/ft/weo/2011/01/weodata/weorept.aspx?sy=2000&ey=2012&scsm=1&ssd=1&sort=country&ds=.&br=1&c=754&s=PCPI%2CPCPIPCH&grp=0&a=&pr.x=42&pr.y=7. Accessed 2011 June 29. Available: http://www.imf.org/external/pubs/ft/weo/2011/01/weodata/weorept.aspx?sy=2000&ey=2012&scsm=1&ssd=1&sort=country&ds=.&br=1&c=754&s=PCPI%2CPCPIPCH&grp=0&a=&pr.x=42&pr.y=7. Accessed 2011 June 29.
- 12. Murray CJL (1994) Quantifying the burden of disease: the technical basis for disability-adjusted life years. Bulletin of the World Health Organization 72: 429–445.CJL Murray1994Quantifying the burden of disease: the technical basis for disability-adjusted life years.Bulletin of the World Health Organization72429445
- 13. World Health Organization (2001) Macroeconomics and health: investing in health for economic development. Geneva. World Health Organization2001Macroeconomics and health: investing in health for economic development.Geneva
- 14. Sicuri E, Bardaji A, Nhampossa T, Maixenchs M, Nhacolo A, et al. Cost-effectiveness of intermittent preventive treatment of malaria in pregnancy in southern Mozambique. PLoS One 5: e13407.E. SicuriA. BardajiT. NhampossaM. MaixenchsA. NhacoloCost-effectiveness of intermittent preventive treatment of malaria in pregnancy in southern Mozambique.PLoS One5e13407
- 15. Halton KA, Cook D, Paterson DL, Safdar N, Graves NCost-effectiveness of a central venous catheter care bundle. PLoS One 5: KA HaltonD. CookDL PatersonN. SafdarN. GravesCost-effectiveness of a central venous catheter care bundle.PLoS One5
- 16. McCabe CJ, Goldie SJ, Fisman DNThe cost-effectiveness of directly observed highly-active antiretroviral therapy in the third trimester in HIV-infected pregnant women. PLoS One 5: e10154.CJ McCabeSJ GoldieDN FismanThe cost-effectiveness of directly observed highly-active antiretroviral therapy in the third trimester in HIV-infected pregnant women.PLoS One5e10154
- 17. Wilford R, Golden K, Walker DG (2011) Cost-effectiveness of community-based management of acute malnutrition in Malawi. Health Policy Plan Advance Access published March 4 2011: 1–11.R. WilfordK. GoldenDG Walker2011Cost-effectiveness of community-based management of acute malnutrition in Malawi.Health Policy Plan Advance Access published March 42011111
- 18. International Monetary Fund. World Economic Outlook Database. International Monetary Fund. World Economic Outlook Database.Available: http://www.imf.org/external/pubs/ft/weo/2010/02/weodata/weorept.aspx?sy=2008&ey=2015&scsm=1&ssd=1&sort=country&ds=.&br=1&c=754&s=NGDPDPC&grp=0&a=&pr1.x=71&pr1.y=11. Accessed 2011 November 8. Available: http://www.imf.org/external/pubs/ft/weo/2010/02/weodata/weorept.aspx?sy=2008&ey=2015&scsm=1&ssd=1&sort=country&ds=.&br=1&c=754&s=NGDPDPC&grp=0&a=&pr1.x=71&pr1.y=11. Accessed 2011 November 8.
- 19. Bang AT, Bang RA, Reddy HM (2005) Home-based neonatal care: summary and applications of the field trial in rural Gadchiroli, India (1993 to 2003). J Perinatol 25: Suppl 1S108–122.AT BangRA BangHM Reddy2005Home-based neonatal care: summary and applications of the field trial in rural Gadchiroli, India (1993 to 2003).J Perinatol25Suppl 1S108122
- 20. WHO-CHOICE website. CHOosing Interventions that are Cost-Effective (analysis results for Africa-E region). WHO-CHOICE website. CHOosing Interventions that are Cost-Effective (analysis results for Africa-E region).Available: http://www.who.int/choice/results/mnh_afroe/en/index.html. Accessed 2011 November 8. Available: http://www.who.int/choice/results/mnh_afroe/en/index.html. Accessed 2011 November 8.
- 21. Trading Economics website. Population density (people per sq. km) in Zambia. Trading Economics website. Population density (people per sq. km) in Zambia.Available: http://www.tradingeconomics.com/zambia/population-density-people-per-sq-km-wb-data.html. Accessed 2011 November 23. Available: http://www.tradingeconomics.com/zambia/population-density-people-per-sq-km-wb-data.html. Accessed 2011 November 23.
- 22. Carlo WA, Wright LL, Chomba E, McClure EM, Carlo ME, et al. (2009) Educational impact of the neonatal resuscitation program in low-risk delivery centers in a developing country. J Pediatr 154: 504–508 e505.WA CarloLL WrightE. ChombaEM McClureME Carlo2009Educational impact of the neonatal resuscitation program in low-risk delivery centers in a developing country.J Pediatr154504508 e505
- 23. Bang AT, Bang RA, Baitule SB, Reddy MH, Deshmukh MD (1999) Effect of home-based neonatal care and management of sepsis on neonatal mortality: field trial in rural India. Lancet 354: 1955–1961.AT BangRA BangSB BaituleMH ReddyMD Deshmukh1999Effect of home-based neonatal care and management of sepsis on neonatal mortality: field trial in rural India.Lancet35419551961
- 24. Hutton G, Schellenberg D, Tediosi F, Macete E, Kahigwa E, et al. (2009) Cost-effectiveness of malaria intermittent preventive treatment in infants (IPTi) in Mozambique and the United Republic of Tanzania. Bull World Health Organ 87: 123–129.G. HuttonD. SchellenbergF. TediosiE. MaceteE. Kahigwa2009Cost-effectiveness of malaria intermittent preventive treatment in infants (IPTi) in Mozambique and the United Republic of Tanzania.Bull World Health Organ87123129
- 25. Beigi RH, Wiringa AE, Bailey RR, Assi TM, Lee BY (2009) Economic value of seasonal and pandemic influenza vaccination during pregnancy. Clin Infect Dis 49: 1784–1792.RH BeigiAE WiringaRR BaileyTM AssiBY Lee2009Economic value of seasonal and pandemic influenza vaccination during pregnancy.Clin Infect Dis4917841792
- 26. Nannan N, Norman R, Hendricks M, Dhansay MA, Bradshaw D (2007) Estimating the burden of disease attributable to childhood and maternal undernutrition in South Africa in 2000. S Afr Med J 97: 733–739.N. NannanR. NormanM. HendricksMA DhansayD. Bradshaw2007Estimating the burden of disease attributable to childhood and maternal undernutrition in South Africa in 2000.S Afr Med J97733739