Katherine Baicker and Amitabh Chandra (B & C) have an essay in the New England Journal of Medicine clarifying what it means to say that a health policy is evidence-based. If you are a student of health policy, take this to heart. In this post, I want to argue for a complementary goal: that we should strive to be not just evidence-based but also normatively-literate in advocating health policies. I’ll summarize B & C’s points and then explain what I mean by normatively-literate.
B & C argue that evidence-based health policies [EBHPs] must (1): be specified in sufficient detail so that it is clear what they would mean if they were implemented. (2) Policies should be based on credible empirical estimates of the magnitudes of the policies’ effects. Above all, we need (3) to distinguish between policies and goals, and not mistake a goal-based slogan for an EBHP. Therefore, a slogan such as
“Target population health” doesn’t qualify as a policy, let alone EBHP, because myriad policies fall under the population health banner, including influenza vaccination, smoking cessation, [etc.]. Slogans like “population health,” “single payer,” or “malpractice reform” may be an effective way to signify a political position or rally support (after all, who’s against population health?), but in avoiding specificity, they sidestep the hard work of assessing the relative effectiveness and implementation details of the policies included under their umbrella.
Yes to all the above.
However, precise specification and empirical study of policies need to be complemented by careful reasoning about goals. My curmudgeonly view is that reasoning about goals is if anything rarer in health policy than evidence-based reasoning. Robin Hanson says that
we seem to have a meta-norm that norm application should be automatic and obvious. We are to just know easily and surely which actions violate norms, without needing to reflect on or discuss the matter. We are to presume that framing effects are unimportant and that everyone agrees on the relevant norms and how they are to be applied.
Hanson is talking about norms about what we shouldn’t do, but the same naïveté prevails in normative thinking about what we should do, that is, our goals. The “norms are automatic” meta-norm states (falsely) that goals are simple, easily specified, and likely to be incontrovertible. I see it differently: policy goals are complicated, hard to express clearly, and the attempt to clarify them frequently uncovers deep controversies.
How does this affect policy studies?
First, goals shape data, often in ways we don’t see. It isn’t just that wanting to justify a particular policy outcome leads to sloppy methods. Instead, our goals shape what we measure, often at a subtle level of detail.
Consider the goal of “population health”.
- Who counts as a member of the population? Suppose that I am interested in US population health. But should I be? If people are moral equals, why is the US border relevant? Suppose I answer that membership in the nation creates duties to care for fellow citizens. If so, does the population include non-citizen residents?
- How do you measure ‘health’? Does health = longevity, or does the quality of life matter? If quality matters, does quality refer to our subjective appraisal of how well our life is going, or to objective measures of functioning? Is health the absence of disease or the presence of well-being? If it is the absence of disease which, if any, mental health disorders count?
- What function maps the distribution of health in the population to a number? Population health quantifies the health outcomes of a group. Since most people hold the modern norm that we are all moral equals, it’s intuitive to quantify population health as average health. However, consider two countries, A and B. Both are 80% white and 20% black. In A, whites have a life expectancy of 80 years and blacks have a life expectancy of 75 years. In B, both races have a life expectancy of 79 years. Is one population healthier than the other? They both have the same average life expectancy (79 years). Most workers in population health, however, view inequalities in health as problems and would judge B to have better population health. If we grant this, how should we measure inequality? By the size of the difference between group means (5 years)? If so which groups? Should inequality be measured using a Gini coefficient? Suppose we use an average and an inequality measure to summarize population health. If we use two numbers, can we still compare the health of populations? Conversely, suppose we combine the average and the inequality measure into a weighted composite and call that population health. Then how should we derive and justify those weights?
Defining population health will require us to make normative commitments on all of these issues. The results of every empirical study using that measure will be conditional on those commitments.
Second, policy choices involve disputes about both facts and goals. B & C ask “who, after all, is against population health?” It’s true, I’ve never heard anyone argue for lowering population health. But many people give population health little weight in policy choices, and not necessarily because they are evil. They may have normative arguments for other views.
- How important is population health relative to other goods or principles? For the sake of argument, let’s stipulate that Americans would be healthier if they smoked less and owned fewer handguns. Let’s stipulate further that the most effective policies in these areas would restrict what we can consume or own. If so, how much do we value liberty relative to population health?
- Does it matter how people got sick? Population health requires even-handed concern for everyone’s health. Measures of population health make no distinction about what causes poor health. But does even-handed commitment fit well with our other moral judgments? Many people believe that we have a duty to care for those who, through no fault of their own, are victims of disease or accident. However, they feel that we are less responsible for those who knowingly engage in behaviours that expose them to disease or trauma. Are we responsible for the care of a mountain climber paralyzed from a fall, or a motorcycle rider from a crash, or someone who was shot while he was committing a violent crime? Should we have to pay for the cancer care of a smoker who was repeatedly counselled to quit? For the anti-virals of someone who shared injection needles? Who engaged in risky sex? (Perhaps your attitudes vary across these cases. If so, why?) For many, appeals to the goal of population health conflict with their view that the duty to care is conditional on the victim’s responsibility.
The population health example shows that policy goals can be complicated, hard to clarify, and controversial on many points. It follows that health policy students and advocates need to be both evidence-based and normatively-literate. By ‘normatively-literate’, I mean that you should appreciate the normative presuppositions and implications of your work and the reasons why others disagree with your commitments. I say ‘normatively-literate’ rather than ‘norm-based’ to acknowledge that there are often better prospects for closure on what the evidence says than about whether the goals of a policy are good or right.
Unfortunately, the ‘ideal student’ who is both a strong empiricist and a deep student of philosophy may not exist (unless your name is Amartya Sen). Everyone needs to specialize. But the health policy community needs an appreciation for both poles of the goals/policies distinction. We need more philosophers who can follow the evidence and more empiricists who can defend their goals ‘in the seminar’.