Newborn screening and prophylactic interventions for sickle cell disease in 47 countries in sub-Saharan Africa: a cost-effectiveness analysis

Sickle cell disease (SCD) is a hereditary disorder of hemoglobin (Hb) characterized by inheritance of two abnormal Hb genes, at least one of which is responsible for the production of HbS. The most common clinical phenotype is the homozygote (Hb SS), also known as sickle cell anemia. It is commonly characterized by chronic hemolytic anemia and recurrent vaso-occlusion which is responsible for the painful crises that characterize the disease. In Africa and Asia, SCD occurs in areas where malaria is endemic [1]. Malaria and bacterial infections such as pneumococcal infections result in morbidity and mortality of SCD patients in Sub-Saharan Africa (SSA) [2, 3]. Malaria contributes not only to mortality, but also to anemia and SCD crises [2].

In SSA, SCD is a disease of public health concern. In 2010, the region had nearly 80 % of the projected 306,000 newborns with SCD worldwide [4]. Mortality estimates among SCD children in SSA range from 50–90 % before age 5 [5]. In contrast, life-expectancy for SCD patients has improved in western countries and most patients live to their 40s and 50s [6]. In those countries, newborn screening programs were introduced incorporating testing newborns for SCD plus interventions (penicillin prophylaxis and pneumococcal vaccination) to prevent bacterial infections among diagnosed cases. Although newborn screening was shown to be cost-effective among African American infants [7], this approach has not been widely adopted in SSA. In SSA, malaria is endemic and newborn screening programs require an additional component of malaria prevention. Evidence from prospective studies in Benin and Angola indicate that a package of prophylactic interventions can reduce the infant and under-5 mortality among newborns with SCD to levels similar to those of the general population [8, 9].

Despite the high SCD burden in SSA, newborn screening and prophylactic intervention (NSPI) programs have not been widely implemented due to concerns of affordability, costs, operational challenges and competing priorities. This analysis assesses the cost-effectiveness of a NSPI program in SSA.

A decision analytic model was developed to compare NSPI for SCD relative to no screening, from the perspective of the national health systems in 47 countries in SSA. The model structure was adapted from a previous cost-effectiveness analysis of newborn screening for SCD in the United States [7], and based on a lifetime Markov state transition cohort with annual cycle lengths and the standard half-cycle correction. A graphical representation of the model structure is displayed in Fig. 1. Infants identified to have SCD could either remain alive with SCD, or progress to the terminal steady state of death. In each annual cycle, infants accrued life-years, adjusted for disability. Infants in the NSPI arm of the model consumed health care resources in each cycle. The resource use associated with no screening was assumed to be zero. The common model structure was populated with country-specific model inputs, in order to yield 47 model results.

In the model, every infant was tested at birth for SCD using isoelectric focusing (IEF). Infants with a positive confirmatory test result would initiate an array of prophylactic therapies to reduce the risk of malaria and other infections. Unscreened infants were assumed to remain undiagnosed and untreated. For these infants, the annual probability of mortality was stratified between infants born in rural versus urban areas because prior studies have suggested that infection rates and access to healthcare services could differ between rural and urban areas [5]. A third option of delaying screening upon the onset of symptoms was not included, because SCD-associated mortality may occur before the onset of symptoms, particularly in rural areas. In the intervention arm, the annual probability of mortality among treated infants was stratified for four separate age categories: less than 5 years, 6 years to 20 years, 21 years to 30 years and 31 years and older. For the age categories below 30 years, mortality rates were obtained from a Tanzanian hospital-based cohort and generalized to the other 46 countries included in the model [10]. Beyond 30 years of age, a multiplier was applied to the annual probability of mortality in the general population. All future costs and benefits were discounted at 3 % annually [11]. Calculations were performed using TreeAge Pro 2014, R1.0 (Treeage Software Inc., Williamstown, MA, USA).

Formal ethics approval and consent to participate was not required in this retrospective analysis of observational data.

The annual number of infants born with SCD in the 47 countries in SSA is estimated at 228,169. The annual incidence is unevenly distributed, with estimates ranging from 0 cases in Djibouti, Lesotho, and Mauritius, to 38,217 and 85,186 in Democratic Republic of Congo (DR Congo) and Nigeria, respectively.

Mean life expectancy of unscreened newborns in rural and urban areas across SSA was estimated at 1.7 years and 6.7 years, respectively. All country-specific base case model outputs are displayed in Table 3. The life expectancy of screened and treated infants ranged from 24.2 years in the Central African Republic to 32.4 years in Cape Verde (mean 27.0 years). The increase in life expectancy associated with NSPI relative to no treatment translates into 2,414,612 disability-adjusted life years (DALYs) that could be averted across all countries. Importantly, DALY estimates from DR Congo and Nigeria alone account for more than 50 % of DALYs that could be averted in this analysis.

The cost of universal newborn screening using IEF in all 47 countries is estimated at US$312 million per year. Lifetime treatment cost of the prevention package (mean, range) was US$876 (US$828-US$930). The sum cost of a universal NSPI program and the net present value of lifetime treatment costs were US$514 million per year. Most costs would be borne by Nigeria (approximately US$136 million) and DR Congo (approximately US$62 million), but for 34 other countries, the budget impact would still exceed US$1 million per year.

The incremental cost-effectiveness ratios and corresponding 95 % confidence intervals of NSPI in the 47 countries are displayed in Table 3 and, for the subset of 34 countries where screening is cost-effective in the base case, in Fig. 2. The cost per DALY averted ranges from US$144 (95 % CI: $98-$323) in Equatorial Guinea to an ICER approaching infinity in the three countries that report zero burden of SCD and where no DALYs are therefore expected to be averted. The population-weighted average cost per DALY averted for all 47 countries is US$213. Among 34 countries with a cost per DALY averted below one time per capita gross domestic product (GDP), the cost per DALY averted is US$190.

To our knowledge, this is the first study to systematically evaluate the cost-effectiveness of NSPI programs across most countries in SSA. Based on the commonly accepted threshold of less than one time per capita GDP to assess a highly cost-effective intervention [11], we conclude that NSPI provides excellent value per dollar spent in the settings of many, but not all countries in SSA. We propose to classify countries into three distinct categories: In 24 countries, NSPI is almost certainly cost-effective; in 13 countries, NSPI is either unlikely to be, or is certainly not cost-effective; and in 10 countries, NSPI is probably cost-effective, but the findings are subject to varying degrees of uncertainty. The countries where NSPI is not cost-effective tend to be located in areas with a historically low malaria burden. In Djibouti, Lesotho, and Mauritius, the incidence of SCD is essentially zero. In Botswana, Eritrea, Ethiopia, Somalia, South Africa, and Swaziland, the incidence rate of SCD is approximately 0.01 % or less, or 10 cases out of 100,000 live births, such that the corresponding cost per DALY averted exceeds the World Health Organization (WHO) cost-effectiveness threshold, meaning that significant resources would generate relatively small improvements in population health. However, results from our scenario analysis also suggest that NSPI would be cost-effective in some of these countries if the actual incidence rate in these two populations exceeded that published rates by a factor of 2. In contrast, the results are uncertain in countries with a higher, but still relatively low annual incidence of SCD, around 0.15–0.80 %, or 150–800 cases out of 100,000 live births, such as around Central African Republic, Gambia, Guinea-Bissau, Liberia, Madagascar, Namibia, Niger, Rwanda, Sao Tome and Principe, and Zimbabwe. For these countries, the element of uncertainty is only introduced once multiple parameters are varied simultaneously in probabilistic sensitivity analyses.

In the remaining 24 countries, our model suggests NSPI would provide a very attractive return for investments in healthcare and could lead to meaningful improvements in population health. The average cost per DALY averted for these 24 countries included in the model is US$184, which is not only cost-effective according to the WHO definition, but also compares favorably relative to other healthcare interventions in SSA. For example, the cost-effectiveness of first line antiretroviral therapy for HIV was reported at US$620 per year of life gained in Cote d’Ivoire [20], and is likely to be even higher for second-line antiretroviral therapy where annual treatment costs alone have been estimated at US$1,037 in South Africa [21]. Similarly, the cost-effectiveness of antiretroviral therapy for prevention among serodiscordant couples has been estimated at $590 per life year saved [22]. For vaccines, a review of 44 studies found that the cost per DALY averted was below US$100 in about half of all studies reported and almost a fourth of all vaccine studies reported an estimate exceeding US$500 [23].

Our results support the WHO’s call urging member countries where SCD is a public health problem to establish national health programs, operate specialized centers for SCD and facilitate access to care and treatment including NSPI [24, 25]. Providing access to NSPI in those states will improve childhood survival thereby contributing towards the attainment of Millennium Development Goal 4. While this analysis focuses on currently available interventions for SCD in SSA, in future, technological and medical advances may alter the clinical course of SCD as well as the cost of managing the disease in these countries.

Using IEF to screen all newborns for SCD plus administration of prophylactic interventions to affected children is cost-effective in the majority of countries in SSA.