Cost of illness

Since 2001, 73 countries with per-capita gross national incomes in 2003 of no more than 1000 United States dollars (US$) have received Gavi support. Our analysis was based on data for these 73 low- or middle-income countries (Box 1). We estimated the health and economic impact of vaccination against 10 vaccine-preventable diseases: Haemophilus influenzae type b, hepatitis B, human papillomavirus, Japanese encephalitis, measles, Neisseria meningitidis serogroup A, rotavirus, rubella, Streptococcus pneumoniae and yellow fever. Table 1 presents the relevant health outcomes, permanent disabilities and vaccines that we included in our models. While all of our study vaccines are delivered via routine immunization, supplementary immunization activities also occur for Japanese encephalitis, measles, N. meningitidis serogroup A, rubella and yellow fever. Our estimates of the numbers of deaths, cases and DALYs averted as the result of vaccination were developed from health impact models, as previously described.^6,7 We used data on immunization coverages published by the World Health Organization (WHO) and United Nations Children’s Fund to estimate annual vaccine coverages for each study country for the period 2001–2012⁸ and version 9 of Gavi’s strategic demand forecast to estimate the corresponding probable coverages for the period 2013–2020.⁹

Using the cost of illness approach¹⁰ from a societal perspective, we estimated treatment costs and productivity losses averted by vaccination based on the estimated numbers of cases, deaths and disabilities averted.¹¹ For each of the 10 diseases studied, we constructed decision tree models to capture both the short- and long-term averted costs of illness. These costs were broken down into five categories: (i) averted treatment costs; (ii) averted transportation costs for seeking care; (iii) averted reduction in caregivers’ economic output; (iv) averted loss of productivity due to premature death; and (v) averted loss of survivors’ productivity due to disability. All estimates of the averted cost of illness were discounted at 3% and are expressed in 2010 US$. We present separate results for the 20 years following Gavi’s establishment – i.e. 2001–2020 – and the current so-called Decade of Vaccines – i.e. 2011–2020.

In estimating the immunization-attributable averted costs of treatment, transportation and lost caregiver productivity, we used the country-specific estimated proportions of children for whom care was sought¹² as well as data on the duration and rates of hospital admission.^13–18 Country-specific costs of relevant inpatient and outpatient care at hospitals and health centres were primarily obtained from the WHO’s Choosing Interventions that are Cost-Effective (WHO-CHOICE) project.¹⁹ Costs of medications and diagnostics were estimated as proportions of facility costs. We assumed that each inpatient admission or outpatient visit was associated with a fixed transportation cost – i.e. a country-specific estimated mean cost of a return trip to and from a health-care facility.²⁰ We also assumed that caregivers of sick children lost half their daily productivity for an outpatient visit and a full day’s productivity for each day a child was hospitalized. In each study country, a caregiver’s daily productivity was assumed to equal the daily minimum wage.²¹

Lost productivity resulting from convalescence and long-term disability was estimated for cases that could be averted by vaccination. To account for other causes of mortality that may impact the number of survivors entering the workforce, age-specific survival rates were applied to the non-fatal cases. The total number of productive years lost due to disability was estimated using the difference between life expectancy and mean age at disability onset – incorporating relevant disability weights. Estimates of life expectancy were derived from data published by the United Nations’ Population Division²² and disability weights from the 2010 Global Burden of Disease study.²³ For each study country, we estimated lost productivity resulting from disability and premature mortality by multiplying the number of productive life-years lost due to disability or premature death by the projected annual values for the per-capita gross domestic product.²⁴ The values we give for total averted long-term productivity losses represent the projected economic outputs of children whose disability or death are – or will be – prevented through immunization. Children were assumed to begin their economically productive lives when they reached an age of 15 years. Further detail on the key inputs, assumptions and data sources used for our analysis has been published.⁷

Economic and social value

We used a second method to capture the broader economic and social value placed on living longer and healthier lives as a result of vaccination. For this, we applied a value-of-life approach that provides a societal perspective of the full benefits of reduced mortality. The estimated value of a life-year was based on data from two sources: (i) wage risk studies that use data on labour markets to examine the trade-off between wages and risk of mortality while employed; and (ii) stated preference studies in which individuals are asked how much they are willing to pay to avoid certain risks of death.^25,26 Based on earlier work to estimate the annual per-capita value of an increase in life expectancy,^27–30 we assumed that the value of a life-year saved in a particular country was 1.6 times that country’s annual per-capita gross domestic product. The economic and social value of vaccinations was estimated from the number of deaths averted due to vaccines, the difference between life expectancy and mean age of death from each study disease and the relevant per-capita gross domestic product (GDP).³¹

Traditional estimates of the value of a life-year have focused on estimating the full benefits of mortality reduction – with few studies examining the impact of corresponding reductions in morbidity.^28,32 To reflect the benefit of averting morbidity, we estimated the value of a year lived with disability. As in the estimation of years lived with disability and years of life lost – both used to calculate a DALY – disability weights were applied to estimate the impact of various disabling conditions on an individual’s value of life from the age of disease onset to expected age at death.²² Disability weights varied from 0 – representing perfect health – to 1 – representing death. A similar approach was used in a previous estimation of the impact of disability on the value of a life-year.²⁸ Our estimate of the value of a year lived with disability was used to estimate the full economic loss associated with permanent, long-term disability caused by any of the six study diseases that can have permanent sequelae – i.e. H. influenzae type b, Japanese encephalitis, measles, N. meningitidis serogroup A, rubella and S. pneumoniae.

Sensitivity analysis

Multivariate Monte Carlo simulations, with 10 000 replications, were performed to assess the impact on cost estimates of uncertainty in the values of several key parameters: labour-force participation, per-capita GDP, the numbers of cases and deaths averted, the multiplier – otherwise set at 1.6 – used in the estimation of the value of a life-year, transportation costs and WHO-CHOICE treatment costs. Cost values were sampled from γ distributions to represent the right skew of observed costing data while non-cost values were sampled from β distributions. Lower and upper ranges of distributions were derived from health impact models or published literature. Results from the sensitivity analysis were used to construct 90% uncertainty ranges around point estimates and generate a tornado diagram illustrating the degree to which individual parameters influenced the final results. The analysis was performed using version 6 of the @RISK software package (Palisade Corporation, Ithaca, United States of America).

Cost of illness

Table 2 presents estimates of the health and economic impact, for the period of 2001–2020 and the decade of 2011–2020, of averting 10 vaccine-preventable diseases in the 73 countries. According to our analyses, use of life-saving vaccines will avert an estimated 20 million deaths, 500 million cases of illness, 9 million cases of long-term disability and 960 million DALYs between 2001 and 2020. During the Decade of Vaccines, introduction and/or increased coverage of the modelled vaccines are projected to avert over 14 million deaths, 350 million cases of illness, 8 million cases of long-term disability and 700 million DALYs.

By 2020, immunizations since 2001 will have averted an estimated US$ 350 billion (uncertainty range: 260–460 billion) in total costs due to illness. Most of these costs – about US$ 250 billion (uncertainty range: 190–330 billion) – will have been averted since 2011, of which about US$ 240 billion (uncertainty range: 180–320 billion) represents averted productivity loss caused by premature death. Between 2011 and 2020, US$ 4 billion (uncertainty range: 3–4 billion) in treatment costs, US$ 350 million (uncertainty range: 240–490 million) in transportation costs and US$ 730 million (uncertainty range: 650–790 million) in caregiver productivity losses could be averted. When we estimated the total cost of illness averted by vaccination against each of our study diseases, it appeared that greater costs could be averted, per vaccinated individual, through protection against H. influenzae type b, S. pneumoniae, human papillomavirus and measles than protection against any of the other six study diseases (Table 3).

Projected introductions of new vaccines and supplementary immunization activities account for US$ 170 billion – about 66% – of the estimated economic benefits of the Decade of Vaccines, with the remainder attributable to the scale-up in coverage of vaccines introduced before 2011. Supplementary immunization activities against measles represented the largest contributor to our estimates of the overall averted costs of illness – representing approximately US$ 130 billion (uncertainty range: 70–220 billion), or about 37% of total averted costs, and US$ 76 billion (uncertainty range: 42–133 billion), or about 30% of total averted costs, for the periods 2001–2020 and 2011–2020, respectively. The second, third and fourth largest drivers of the cost of illness appeared to be H. influenzae type b, S. pneumoniae and hepatitis B, respectively. Fig. 1 shows the disease-specific costs of illness averted by vaccination, for the periods 2001–2020 and 2011–2020. Most of these averted costs were represented by the productivity saved as the result of reduced mortality. Of the 73 countries included in the analysis, the five with the largest birth cohorts – i.e. Bangladesh, India, Indonesia, Nigeria and Pakistan – together accounted for more than half of the estimated total cost of illness averted between 2011 and 2020 (available from the corresponding author). In terms of the estimated total averted costs of illness per vaccinated individual (available from the corresponding author), the study countries in the WHO African Region came highest, at US$ 71, followed by those in the South-East Asia (US$ 58), Western Pacific (US$ 56), Eastern Mediterranean (US$ 45) and European (US$ 33) Regions and then the Region of the Americas (US$ 20).

Sensitivity analysis indicated that our estimates of the averted costs of illness were most sensitive to variation in estimates of the numbers of deaths averted by vaccination, particularly against diseases associated with a high disease burden and young age at onset – e.g. H. influenzae type b, measles, rotavirus and S. pneumoniae (available from the corresponding author). Cost parameters such as per-capita GDP proved to be less influential.

Economic and social value

In terms of their overall broader economic and social impact, we estimated the vaccinations we investigated to be worth approximately US$ 820 billion (uncertainty range: 560–1200 billion) and US$ 600 billion (uncertainty range: 420–870 billion) over the periods 2001–2020 and 2011–2020, respectively. About 97% of each of these values was represented by the value of averted mortality. Averted morbidity contributed much less – about US$ 19 billion (uncertainty range: 16–22 billion) and US$ 16 billion (uncertainty range: 14–19 billion) over the periods 2001–2020 and 2011–2020, respectively. Over half of the estimated economic and social value of vaccination in 2001–2020 is attributable to vaccinations against H. influenzae type b, hepatitis B and S. pneumoniae. Fig. 2 presents a year-on-year comparison of the estimated economic and social values, costs of illness averted and deaths averted between 2001 and 2020. The annual fluctuations shown are largely a result of supplementary immunization activities against measles, which are scheduled to peak every other year.

Similar to our estimates of the averted costs of illness, our broader estimates of the economic and social value of vaccinations were most sensitive to variation in the estimated health impact of vaccinations against childhood illnesses with a high disease burden – e.g. H. influenzae type b, measles, rotavirus and S. pneumoniae.