“The US Centers for Disease Control and Prevention has recognised public health programmes that reduced deaths from coronary heart disease and stroke and prevented deaths from smoking as two of the greatest public health achievements of the 20th century, both championed by robust evidence connecting these risks to reductions in healthy life expectancy.
[..] To our knowledge, no study links a comprehensive set of modifiable risk factors to health-care spending by condition.
[..] To address this research gap, we estimated US health-care spending attributable to 84 modifiable risk factors in 2016, with the latest data available. We included behavioural risks, such as tobacco use and dietary risks; metabolic risks, such as high body-mass index (BMI) and high blood pressure; and environmental risks, such as air pollution and occupational carcinogens. Knowledge of health-care spending attributable to modifiable risk factors can inform choices and priorities for the design of public and private health promotion and prevention programmes, both in the USA and elsewhere.
[..] We estimated health-care spending attributable to modifiable risk factors (henceforth referred to as attributable spending) using a simple two-step process using data from two existing studies. The first dataset was from the Institute for Health Metrics and Evaluation’s Disease Expenditure Project, from which we extracted estimates of how much was spent on health care for 154 mutually exhaustive health conditions by age group and sex in 2016 in the USA. The second dataset was from the Global Burden of Diseases, Injuries, and Risk Factors Study (GBD) 2017, from which we extracted health condition-specific estimates of health burden—number of deaths, years lived with disability (YLDs), and disability-adjusted life-years (DALYs)—and the estimated population attributable fractions for 84 modifiable risk factors, for each health condition, age, and sex group in 2016. Population attributable fractions measure the portion of health burden for each health condition that is attributable to each risk factor based on the relative risks associated with risk exposures and actual risk exposure in the population.
[..] We extracted population attributable fractions from GBD 2017, which assessed health burden for 195 countries from 1990 to 2017. In addition, GBD 2017 measured the portion of the health burden for each health condition that is attributable to 84 modifiable risk factors, for each age and sex group. [..] More than 463 data sources were used to make these estimates.
[..] In 2016, we estimated that US$730·4 billion (95% UI 694·6–768·5), or 27·0% (95% UI 25·7–28·4) of total US health-care spending (included in the Disease Expenditure Project), was attributable to the modifiable risk factors included in this study. Attributable spending is largely associated with older ages (ie, ≥45 years) and chronic conditions such as cardiovascular diseases; diabetes and urogenital, blood, and endocrine diseases; and management of metabolic risks, which includes treatment of hypertension and hyperlipidaemia. [..] High BMI had the most attributable spending in 2016, at $238·5 billion (95% UI 178·2–291·6). 33·2% (95% UI 29·5–37·2) of this attributable spending was on diabetes and urogenital, blood, and endocrine diseases and 26·0% (23·4–28·5) was on cardiovascular diseases [see Figure below].
Health-care spending attributable to risk factor categories by aggregated health condition (A) and age group (B), 2016 (Due to risk interaction and mediation, attributable spending by risk category does not sum up to total attributable spending.)
[..] At an aggregated risk level (level 1), metabolic risks, which included high systolic blood pressure, high fasting plasma glucose, and high BMI, were associated with the greatest attributable spending, at $508·0 billion (95% UI 468·9–549·1) in 2016. Behavioural risks, including dietary risks and tobacco smoke, had the second-highest attributable spending, at $349·4 billion (329·1–369·6). Finally, $75·5 billion (62·0–89·5) of spending was attributed to environmental risks.
[..] The majority of total attributable spending was at older ages, with 86·7% (95% UI 86·1–87·4) occurring in people aged 45 years and older. This is illustrated by the two risk factors with the highest attributable spending: for high BMI, 45·9% (43·0– 49·1) of the total attributable spending was at ages 45–64 years and 42·3% (38·0–46·1) at ages 65 years and older, whereas for high systolic blood pressure, 37·7% (35·6–39·8) of attributable spending was at ages 45–64 years and 56·0% (53·6–58·3) at ages 65 years and older.
[..] Among adults (aged ≥20 years), five preventable risk factors—high BMI, high systolic blood pressure, high fasting plasma glucose, dietary risks, and tobacco smoke—accounted for a substantial proportion of attributable spending. The health conditions with the most attributable spending were cardiovascular diseases and diabetes. Attributable spending was heavily skewed toward older ages. Younger adults (ie, 20–44 years) accounted for a small portion of the total, although other research shows that it is at early ages when first exposure to many risks occur and when many patterns of healthy behaviours are formed.
[..] We found that the proportion of total health burden attributed to the modifiable risk factors considered in this study was substantially higher than the proportion of health-care spending attributed to those same risk factors—46·2% versus 27·0%. This discrepancy is due to two key factors. First, health-care spending is not perfectly associated with health burden, and for some health conditions, a large amount of spending occurs independent of health burden. This is especially true for pharmaceutical spending and health-care spending on prevention and disease management. Second, several of the health conditions that have the most attributable health burden actually have relatively little health-care spending. Examples with little total spending but a large fraction of the attributable health burden were lung cancer ($7 billion in total health-care spending), drug use disorders ($13 billion), and alcohol use disorders ($8 billion). Conversely, some health conditions with relatively large amounts of health-care spending have relatively small population attributable fractions, such as uncomplicated labour and delivery ($71 billion) and preventive dental care ($61 billion), which both have population attributable fractions of zero, and musculoskeletal disorders, which have over $380 billion in health-care spending but a population attributable fraction of only 22·3%. Ultimately, health-care spending and health burden are not well aligned, and many of the health conditions with the most spending are less attributable to the risks considered in this study.
[..] Globally, nearly 50% of health burden was attributable to modifiable risk factors in 2017. While the USA spends more per person on health than any other country, the fact that a large portion of spending can be attributed to key risk factors is unlikely to be an anomaly. Lessons learned about which risk factors contribute the most to health-care spending and the methods employed in this study could be expanded to consider attributable spending in other countries.
[..] There were six main limitations to this research. First, this study measures the attribution of health-care spending to modifiable risk factors. We use attribution in the epidemiological, relative sense, addressing the question of what proportion of an outcome (in this case, health-care spending) can be tied to risk exposure. In this interpretation, attribution is measured at a single moment, such that reducing existing risk exposure might not necessarily lead to proportional reductions in attributable health-care spending in the long term. This is due to the complex relationships among health burden, life span, and health-care resource use, and is a limitation associated with nearly all attribution studies. Because average health-care spending per person and disease incidence tend to increase with age, improvements in health that lead to reductions in spending on one health condition might be replaced by spending on another. For these reasons, it would be a misinterpretation of these results to directly infer how spending would decrease based on reductions in risk exposure. Instead, this study highlights how much of the health-care spending that occurred in 2016 was attributable to these risk factors.
[..] Second, population attributable fractions directly linking risk factors and health-care spending do not exist. Instead, this study relied on population attributable fractions reflecting the relationship between risk factors and health outcomes, adjusted downwards to reflect how health-care spending is associated with health burden. Third, attributable spending estimates are based on reducing risk factors to their theoretical minimums, which in some cases is not a feasible public health goal. No estimates were made for partial reductions and it is unclear without further research how partial reductions in risk exposure might affect spending. Fourth, several modifiable risk factors were not included in this study, such as poor medication adherence, vaccine compliance, inadequate sleep, elevated stress levels, as well as more distal but crucially important drivers such as socioeconomic status. [..] Fifth, as described above, the underlying input data used for this study are not measured with certainty, and estimates rely on necessary assumption. [..] Sixth, despite illustrating a novel method and producing a unique comprehensive set of attributable spending estimates, these estimates are from 2016 and are 4 years out of date upon publication.”
Full article, Bolnick HJ, Bui AL, Bulchis A et al. Lancet 2020.10.1