New paper out in Health Affairs:
Public health encompasses a broad array of programs designed to prevent the occurrence of disease and injury within communities. But policy makers have little evidence to draw on when determining the value of investments in these program activities, which currently account for less than 5 percent of US health spending. We examine whether changes in spending by local public health agencies over a thirteen-year period contributed to changes in rates of community mortality from preventable causes of death, including infant mortality and deaths due to cardiovascular disease, diabetes, and cancer. We found that mortality rates fell between 1.1 percent and 6.9 percent for each 10 percent increase in local public health spending. These results suggest that increased public health investments can produce measurable improvements in health, especially in low-resource communities. However, more money by itself is unlikely to generate significant and sustainable health gains; improvements in public health practices are needed as well.
Doing these kind of analyses are not easy. As far as I can tell, the methods look pretty robust:
Multivariate regression models for panel data were used to estimate the effect of local public health spending on each mortality measure while controlling for the effects of other community characteristics that influence population health. Time-trend variables were used in the models to account for general downward trends in mortality that occurred independently of spending.
An important methodological complication arises in this analysis because local public health spending levels may be influenced by unobserved community characteristics that simultaneously influence mortality. For example, deteriorating local economic conditions may cause public health spending to decline and mortality risks to rise, resulting in incorrect inferences about how spending influences mortality.
To address this possible source of bias, we used instrumental variables methods to distinguish the effects of spending on mortality from the effects of unmeasured characteristics that simultaneously influence spending and mortality. This methodology requires the identification of variables that influence local public health spending but have no direct effects on community mortality rates. We used measures of local public health decision-making structures for this purpose, including whether the agency is governed by a local board of health with policy making authority, and whether the agency operates under the centralized administrative control of state government. Theory and previous studies indicate that these characteristics influence the ability and inclination of local public health agencies to secure external funding sources for their work.
Specification tests confirm that the structural characteristics meet the requirements for instrumental variables (see the Appendix). Using a two-step process, we first estimated the effect of the instrumental variables on spending levels, and then used the natural variation in spending produced by these variables to estimate how spending affects mortality.
And the results are pretty dramatic. For every 10% increase in public health spending, infant mortality went down almost 7 per 1000 births. Deaths by heart disease went down more than 3 per 100,000 population. Deaths from diabetes and cancer both went down more than 1 per 100,000 population.
It’s important to note that this study can’t prove causation. But it provides some compelling evidence that public health spending does work. It’s equally important to note that this study looked at mortality. There are likely a lot of benefits, unmeasured here, that fall short of preventing death.
Public health spending already comprises a paltry amount of our overall health care spending. In the highest year measured (2005), it was $40.84 per person. I have no doubt it will likely be on the chopping block, though, as we try and reduce spending, especially at the state and local level. That’s a pity, because public health spending, even in a small amount, matters.
