Bulletin of the World Health Organization

Global cost of child survival: estimates from country-level validation

Liselore van Ekdom a, Karin Stenberg b, Robert W Scherpbier c, Louis W Niessen d & on behalf of the ad hoc Study Group for Child Health Cost Validation

a. House of Performance, Utrecht, Netherlands.
b. World Health Organization, Department of Health Systems Financing, Avenue Appia 20, 1211 Geneva 27, Switzerland.
c. World Health Organization, Beijing, China.
d. Department of International Health, Johns Hopkins Bloomberg School of Public Health, Baltimore, United States of America.

Correspondence to Karin Stenberg (e-mail: stenbergk@who.int).

(Submitted: 04 July 2010 – Revised version received: 18 January 2011 – Accepted: 20 January 2011 – Published online: 11 February 2011.)

Bulletin of the World Health Organization 2011;89:267-277. doi: 10.2471/BLT.10.081059

Introduction

In keeping with the fourth Millennium Development Goal (MDG4), nations have pledged to reduce child mortality by two-thirds between 1990 and 2015. This calls for a scale-up of child survival interventions, whose global cost the World Health Organization (WHO) recently estimated for 75 countries that have a high burden of mortality among children aged less than five years.1 In an effort to support the countries in greatest need, WHO’s Department of Child and Adolescent Health and Development (CAH) has identified 33 countries (Appendix A, available at: http://www.who.int/choice/publications/p_2011_cost_validation_webannexes.pdf) that together contribute 78% of all deaths among children under 5 years of age.

Updated and improved global data on child health have recently become available. Also available now are updated estimates on the prevalence of malnutrition; new intervention coverage estimates made available by Countdown-to-20152; updated price estimates from the WHO-CHOICE project3 and updated estimates of the resources needed to scale up immunization,4 malaria5 and prevention of mother-to-child transmission (PMTCT) of human immunodeficiency virus (HIV) programmes.6

Although global price estimates for the scale-up of packages of selected child health interventions are regularly published,7,8 to our knowledge none has been empirically validated against existing country data.9 Empirical validation differs from the type of conceptual validation by experts that we first conducted.1 This paper reports on the results of this country-level empirical validation process, for which we use country feedback data in conjunction with new, published global data on epidemiology and prices. It presents a validated and revised global price tag for the scale-up of child survival interventions required to attain MDG4 and serves as an investment guide for governments and their development partners.

Methods

We used the original cost projection developed by WHO to derive the 2007 global price tag for child survival. The validation comprised only the key interventions included in the CAH model, including the management of pneumonia, diarrhoea and severe malnutrition, as well as nutrition counselling (complete list available in Appendix B, available at: http://www.who.int/choice/publications/p_2011_cost_validation_webannexes.pdf).

We identified major cost drivers by country and then sent a validation questionnaire covering these cost drivers to 32 countries with high child mortality. Each data input category for the costing model (disease incidence, intervention coverage and prices and input volumes for individual-level and programme-level activities) was validated against local data. Original assumptions were replaced by local data or by recent updated estimates on incidence, coverage and price. Additional costs were calculated through nine model scenarios that assessed the effects of: (i) the validation and of updated epidemiologic and coverage data; (ii) population change and scale-up sensitivity ranges; and (iii) updated unit price estimates. These cross-validated results were subsequently extrapolated to 75 countries. Lastly, these final estimates were combined with published estimates of the resources needed to scale up immunization, malaria interventions and PMTCT to compute a revised total global price tag.

Cost drivers

We retained the components of the original price tag, which included interventions addressing the major causes of death among children aged less than 5 years:10,11 newborn sepsis, pneumonia, diarrhoea, malaria, PMTCT and malnutrition. As in our original study,12 we based cost estimates on an ingredients approach, and we focused on validating input assumptions using patient-level and programme-level costs.

Patient costs refer to the costs of an intervention at the point where services (e.g. drugs, supplies) are delivered to a client. Patient costs vary depending on disease burden, current intervention coverage, differences in case management protocols and visit prices. Programme costs are expenses incurred at the district, province or country administrative level (e.g. in-service training for health workers or information, education and communication activities).13 Costs vary across countries depending on the choice of delivery mechanisms and level of investment in areas such as health worker training, health-care infrastructure and mechanisms for generating service demand in the community.

It was not feasible to validate every single assumption in the original model, so our questions referred to cost drivers, defined as follows. Interventions were identified as cost drivers when the relative proportion of an intervention contributed to 10% or more of overall patient costs in the specific country. A programme cost category was identified as a cost driver if it fulfilled the following two criteria: (i) the cost category contributed to 10% or more of the overall programme cost estimated for the specific country, and (ii) the particular component accounted for at least 70% of the costs of that category. More details are provided in Appendix C, available at: http://www.who.int/choice/publications/p_2011_cost_validation_webannexes.pdf

Survey instrument and analysis

A self-administered questionnaire was developed and sent to WHO experts based in 32 countries with high child mortality in all 6 WHO regions (Appendix A). These experts, responsible for child health in each WHO country office and usually having a medical background, were the first point of contact. The respondents liaised with ministry of health officials responsible for child health to jointly review the costing assumptions and input data used by WHO to derive cost estimates. Involvement of child health staff at the ministries was intended to ensure that the replies obtained were nationally representative and in line with national policies. The data collection verification process lasted from 2 to 4 months in each country.

The survey contained two types of questions: generic questions applicable to all countries and country-specific questions covering each country’s cost drivers (Appendix C and Box 1). The survey showed the country-specific values used for the original calculations and prompted respondents to review each one and to explicitly indicate if they agreed or disagreed with these estimates. Respondents who disagreed with the original assumption were asked to provide alternative data and their sources (Appendix A).

Box 1. Information solicited in questionnaire sent to 32 countries with high child mortality

Questions on pneumonia, diarrhoea and malnutrition were part of the generic questions section, as these conditions are among the leading causes of child mortality in the countries in our study.14 Generic questions on community health workers were also included for all countries.

In relation to the calculation of patient costs, the questionnaire solicited information on: (i) the epidemiology/incidence of major conditions (pneumonia and diarrhoea) and the coverage of interventions addressing these problems (breastfeeding counselling, antibiotics for pneumonia, oral rehydration salts for diarrhoea); (ii) differences in local standard case management protocols compared with generic World Health Organization standards (quantities of medicines and supplies); (iii) prices and quantities of medicines and supplies; (iv) costs of outpatient visits and inpatient management (Appendix C).

In relation to programme costs, generic questions were included about the number of community health workers needed and their remuneration. In addition, the survey included country-specific questions on information, education and communication activities, as well as on the organization of health worker training, infrastructure and equipment.

For data quality control, we asked respondents to: (i) clearly state their reference source and (ii) indicate if health ministry officials agreed with the new data.

Country responses were classified into four categories and acted upon as follows:

  • valid agreement: the original assumption was considered valid by the respondent and remained unchanged;
  • valid disagreement: the newly validated information was considered to be adequately supported and was used in place of the original assumption;
  • invalid disagreement: the newly validated information was considered to be inadequately supported and the original assumption was maintained;
  • information not provided: the original estimate remained unchanged.

Price adjustments

Cost data were provided by countries in their local currencies and were converted to 2004 United States dollars (US$). When an input price was reported in US$ for a tradable good (such as medicines), we used the US$ deflation rate. When respondents provided input prices in local currency units, we applied average annual inflation rates obtained from the International Monetary Fund. When input prices for non-tradable goods such as staff salaries were provided in US$, they were converted into local currency units for the equivalent year, deflated to the 2004 level using the country-specific local currency deflation rate, and then converted to 2004 US$ equivalents.

Stepwise validation and estimate updates

Table 1 lists nine successive model estimates grouped into three categories: (i) validation (V), (ii) sensitivity analyses (S), and (iii) information updates (U). In each successive model estimate new information was added in a stepwise fashion to a preceding estimate. When country-validated data had been provided, they were maintained and not replaced in subsequent models. The stepwise presentation shows the net effect of each step (Table 1 and Appendix D, available at: http://www.who.int/choice/publications/p_2011_cost_validation_webannexes.pdf).

Model V estimates show the effect of adding new and validated country data to the original 2007 estimates. The U1–U4 model estimates show the effect of successively updating the data on population, disease incidence15,16 and intervention coverage,2 while Models S1–S3 show the effect of changes in demographic figures and in the assumed progress of the scale-up process. Model U5 estimates present the effect of recent unit price updates by WHO-CHOICE, which were performed by using new input values for the independent variables in the CHOICE regression analysis.3

Extrapolation to 75 countries

WHO’s Commission on Macroeconomics and Health (CMH) has developed the only index available to date for classifying countries’ health systems into four different levels of strength based on constraints other than lack of funds (e.g. constraints related to demand and care-seeking, health sector policy and broader economic and political factors).17 We estimated the average percentage increase in the projected cost of child survival interventions and related activities, by CMH category, for the 26 countries included in Model U5. The original costs for the remaining 49 countries were then increased in accordance with this adjustment factor. In this manner we extrapolated the results to all 75 countries that together accounted for 94% of all child deaths in the world (the 32 countries included in the validation accounting for 78% of such deaths). Finally, we incorporated recently updated price tags for immunization and malaria programmes and for PMTCT.

Results

Validating and adapting global cost assumptions

More than 80% (26/32) of the countries with high child mortality rates responded. This included countries from all regions, the majority (12) of them in the African Region, as outlined in Appendix A. The level of agreement with the generic assumptions used for the original global cost estimates varied substantially among country respondents, as shown in Fig. 1. It ranged from 80% agreement to 100% disagreement, although for specific areas little information was available.

Fig. 1. Classification of country responses as a percentage of total feedback provided, by input category, for 26 respondent countries
Fig. 1. Classification of country responses as a percentage of total feedback provided, by input category, for 26 respondent countries
BF, breastfeeding; CHW, community health worker; IMCI, Integrated Management of Childhood Illness; ORS, oral rehydration salts; PD, per diem; VC, visit costs.a Combined coverage and combined incidence refer to aggregate results for pneumonia and diarrhoea.b Combined diarrhoea prices and quantities refers to ORS and zinc;c Combined pneumonia prices and quantities refers to paracetamol, salbutamol (oral), amoxicillin (oral), ampicillin (injectable), gentamycin (oral).d Number of responses: CHW, 10; IMCI training, 1.

Country-level information on the incidence of common childhood illnesses, especially pneumonia and diarrhoea, was particularly scarce. When available, the data provided by the countries did not conform to our quality criteria. Of the 26 respondent countries, 17 (65%) provided updated information on the coverage of pneumonia and diarrhoea management interventions. Demographic and Health Surveys were the source of information for all countries (Appendix A). About 65% of the respondent countries provided no data on the coverage of breast feeding counselling.

Not surprisingly, agreement on drug prescription quantities was strong; 77% of the respondents agreed with the global default estimates for drugs used to treat pneumonia and diarrhoea, based on standardized WHO treatment guidelines. Most disagreement had to do with the country’s choice of first-line treatments (e.g. co-trimoxazole versus amoxicillin for pneumonia).

A total of 11 countries reported updated prices for oral rehydration salt sachets. Nine countries reported a price that was, on average, 50% higher than the original global median price. Less information was provided about the average cost of outpatient visits and hospital admissions, a reflection of the scarcity of cost studies at the country level and/or the limited use of their results for national planning.

Upon request, 13 countries provided updated per diem costs for Integrated Management of Childhood Illness in-service training participants. On average, the new per diem amounts were more than double the original estimates. Similarly, the original figures for community health worker (CHW) pay were reported as inaccurate by 16 of the 26 respondent countries. Nine countries presented higher figures, four reported a lower amount and three reported that CHWs received no remuneration.

Consensus was limited on the CHW density needed to support family care practices and community-based care. Of 10 countries that provided data, 5 reported a density higher than the model assumption of 1 CHW per 1000 rural residents and 1 CHW per 1500 urban residents, whereas 5 reported a lower density. The updated numbers of existing CHWs that were provided by respondent countries also differed from the estimates available in the WHO database,14 most likely owing to differences in definition. For the 16 countries that reported using CHWs, the new data resulted in a median increase of 53% in CHW remuneration levels, which emerged as a major cost component.

Validating outcomes for 26 countries

Table 2 shows the new cost estimates based on validated country inputs. New country-level data (Model V) resulted in a 53% increase in total costs for the 26 countries surveyed. When new population data were provided (U1), costs fell by 4%. With updated incidence calculations for severe malnutrition (U2), costs increased further, to 77% more than the original estimate. Updating intervention coverage lowered total costs somewhat, to 61% of the original estimate (U3). Combining these updates (U4) yielded a total increase of 63%.

The validation survey (V) showed relatively robust patient cost estimates (an increase of 20%). Programme cost estimates, however, more than doubled (+140%) in Model V. CHW expenditures contributed the most to this increase. Such expenditures increased by 280% (range: 0–1146%), on average, for the 19 countries that provided feedback, mainly because of higher updated figures for CHW remuneration (rather than number of CHWs). The second most influential adjustment was an average doubling of infrastructure costs (+103%), mostly comprised of the equipment needed to upgrade existing hospitals. As a result of the validation process, the cost of training and of information, education and communication activities decreased to 91% and 78% of the original cost estimates, respectively (Appendix D).

The sensitivity analysis (S2, S3) showed that population size was not a major cost driver; low and high population estimates caused a change of only 2% to 3% in total costs. Changing the model to a linear scale-up (S1) lowered total costs by 11%. Model U5 results in a 16% increase in overall costs when updated WHO-CHOICE prices for the year 2005 were used.3

Overall our analysis showed that country feedback affected costs the most. Additional substantial effects resulted from updated WHO-CHOICE prices (U5) and from improved estimates of the incidence of severe malnutrition (U2). The design of the scale-up curve (S1) had an important effect (−11%), along with updated programme coverage figures (+5%). Population projections (U1) had a small effect; the shift from 2002 to 2008 United Nations projections caused a 3% cost decrease and the use of low versus high projections had a moderate effect (range: +2 to −3%). We are referring to gross changes; the importance of these factors may vary across countries.

As shown in Fig. 2, the original price tag substantially underestimated total costs. After the validation, the median per capita investment needed in 2015 increased by 24% for the 26 countries surveyed (range: −31% to +498%) (Appendix E, available at: http://www.who.int/choice/publications/p_2011_cost_validation_webannexes.pdf).

Fig. 2. Scale-up curves comparing original costs to revised cost estimates for intervention package,a for 26 respondent countries
Fig. 2. Scale-up curves comparing original costs to revised cost estimates for intervention package,<sup>a</sup> for 26 respondent countries
S, sensitivity analyses; U, information updates; V, validation.a Of the World Health Organization’s Department of Child and Adolescent Health and Development.b In billions of 2004 United States dollars.

We extrapolated the new findings to estimate the costs (in 2005 US$ for consistency with recent publications from the High Level Taskforce on International Innovative Financing for Health Systems and others) for the 75 countries having high child mortality rates, based on the percentage change, per CMH category, observed among the countries in our survey. Our analysis, based on revised estimates and on new published cost estimates for PMTCT,9 immunization18 and malaria programmes,19 has shown the need to invest an additional US$ 74 billion from 2010 to 2015, starting with US$ 10 billion in 2010 and reaching US$ 14 billion by 2015 (Table 3). This represents an increase of US$ 33 billion (+80%) over previous estimates for the span of 6 years. The planned introduction of new vaccines resulted in increased immunization costs. On the other hand, the incremental resources needed for PMTCT decreased due to recent progress in scaling up and to new, lower estimates of the incidence of HIV infection.

The newly derived costs correspond to a per capita increase of US$ 2.6 in total health expenditure in 2015 relative to 2007. As shown in Fig. 3, countries with weak health systems (CMH1) need to make the greatest investment; in such countries, the additional resources needed per capita are two to three times greater than in countries belonging to CMH categories 3 and 4. Countries with high child mortality rates in the Eastern Mediterranean Region were found to require the greatest investment (Appendix F, available at http://www.who.int/choice/publications/p_2011_cost_validation_webannexes.pdf, presents the additional funds needed per capita by WHO region).

Fig. 3. Additional investment needed per capita in 2015 to scale up entire child survival package,a by Commission on Macroeconomics and Health (CMH) category, for 75 countries
Fig. 3. Additional investment needed per capita in 2015 to scale up entire child survival package,<sup>a</sup> by Commission on Macroeconomics and Health (CMH) category, for 75 countries
CAH, Department of Child and Adolescent Health and Development (World Health Organization); PMTCT, prevention of mother-to-child transmission.a Of the World Health Organization’s Department of Child and Adolescent Health and Development.b In 2005 United States dollars.c Country typology by constraint quartiles: 1, most constrained; 4, least constrained.17 Countries included: CMH1, 7; CMH2, 8; CMH3, 6; CMH4, 5.

In response to the United Nations Millennium Declaration, governments have committed themselves to scaling up health services to reach the MDGs.20 Fig. 4 shows that for all 26 survey countries combined, the additional investment needed would represent a median increase of approximately 12% (range: 1–55%) over the total health expenditure for 2007. Countries in CMH categories 1 and 2 would require a median increase of 18% above 2007 levels, compared with 6% for countries in CMH categories 3 and 4.

Fig. 4. Increase over the total health expenditure (THE) for 2007 required to scale up entire child survival package,a by Commission on Macroeconomics and Health (CMH) category, for 26 respondent countries, 2010–2015
Fig. 4. Increase over the total health expenditure (THE) for 2007 required to scale up entire child survival package,<sup>a</sup> by Commission on Macroeconomics and Health (CMH) category, for 26 respondent countries, 2010–2015
a Of the World Health Organization’s Department of Child and Adolescent Health and Development.b Cost in 2015 compared with 2006 expenditure.c Country typology by constraint quartiles: 1, most constrained; 4, least constrained.17 Countries included: CMH1, 7; CMH2, 8; CMH3, 6; CMH4, 5.

Discussion

Our revised estimate of the global price tag for selected child survival packages for the period 2010–2015 is 74 billion in 2005 US$. Incorporating the country information received resulted in an 80% increase in our 2007 global estimates. Patient cost estimates were relatively robust. Country respondents reported agreement with the treatment protocols recommended by WHO, although some countries reported prices higher than the original estimates. The wide variation in drug prices among countries has important policy implications for child health programme financing.21,22 Epidemiologic cost drivers are also important: a change in the estimated incidence of severe malnutrition alone increased original costs by 15%. The original price tag, relying on generic model defaults, resulted in considerable underestimation. Programme cost estimates increased substantially (+140%). A significant share of the total costs corresponded to health systems resources related to service delivery.

While our empirical validation reduced some measurement uncertainties, others pertaining to the scale-up process and specific implementation strategies remain.2325 The debate surrounding facility-based versus community-based care, for example, is likely to continue. While facility-based care may prevent the more severely ill children from dying,26 access to those facilities is limited. Timely community-based care can prevent deaths by increasing access and preventing children from becoming severely ill.27 With improved standardization of care delivery options, systematic costing research can indicate the resource implications for each option, thus informing government programme strategies about the most cost-efficient alternatives.

Our study has also shown that essential country-level epidemiological data (e.g. incidence rates or cause-specific mortality figures) are rarely available.28 Data on the provision of essential child survival services are rare; 65% of the survey respondents could not provide data on breastfeeding counselling. Countries depend extensively on data that are synthesized and disseminated internationally. Existing information, particularly regarding current epidemiological trends, needs to be more widely disseminated. Needs-based planning requires more routine data collection at the local level and more economic research. Internationally comparable data on community healthy workers are lacking. This points to the need to develop comparable human resource estimates, an area in which human resources for health observatories can play an important role.29

Recent analyses have highlighted the resource needs associated with different scale-up strategies.30 National-level costing is needed for national planning involving local stakeholders and to determine the actual resources required to scale up strategies in a manner that accounts for local economies of scale. Child health planning should be conducted in conjunction with broader health sector planning. Given resource constraints, countries will need to prioritize health activities within their national resource envelopes.

The relative lack of information on country-level spending on child health complicates priority setting.31 Better instruments and processes, such as sub-accounts for child health32 and the monitoring of current (under-)spending, strengthen the case for investing in child survival and help direct resources towards effective uses. Both external and in-country financing can play a key role, especially in countries with weak health systems. The recent launch of a Global Plan for Maternal, Newborn and Child Health may increase partnerships around country plans to fund activities in this area.33 While donors are pledging increased resources for maternal and child health,34 developing country governments need to raise even more funds. Our empirical revision revealed an 80% increase in the global price tag for scaling up child survival interventions, which means that countries must reassess their budget allocation towards child health if they are to reach MDG4. When possible, options should be explored to increase national health expenditures: the estimated resource needs are equivalent to a gradual increase of 12% over 2007 levels.


Acknowledgements

At the time this study was conducted Liselore van Ekdom, Health Economist, was affiliated with the Erasmus School of Economics and Institute of Health Policy and Management, Erasmus University, Rotterdam, Netherlands, and Robert W Scherpbier was a Medical Officer with the Department of Child and Adolescent Health and Development, World Health Organization, Geneva, Switzerland.

The authors thank the following for their collaboration during this study: members of the Ad Hoc Study Group for Child Health Cost Validation who supported this research; Faduyile Adetomowo, Carlos Avila-Figueroa, Geoffrey Bisoborwa, Richard Cibulskis, Teshome Desta, Chris Drasbek, Dainah Fajardo, Suzane Farhoud, Abuova Gaukhar, Phanuel Habimana, Job Hawap, Raymond Hutubessy, Theopista John, Abbas Uddin Khan, Brigitte Kini, Mehmet Kontas, Sudhansh Malhotra, Emmalita Manalac, Elisabeth Mason, Martha Mejía, Khounphet Mongkhongkham, Assumpta Muriithi, Juliet Nabyonga, Godwin Ntadom, Josephine Obel, Mikael Ostergren, Sergio Pieche, Mona Rakha, Severoni Santino, Tessa Tan-Torres Edejer, Tinu Taylor, Basak Tezel, Marianna Trias, Daisy Trovoada, Victoria de Urioste Blanco, Severin von Xylander and Edwin Ylagan.

Competing interests:

None declared.

References

Share