Evidence and reducing health inequalities

A recent paper in Critical Public Health argued that there is a danger of going down “rabbit holes” in health inequalities research if we focus too much on causal inference. It is written by people I respect and have worked with, so this response is written in a spirit of friendly discussion. First I think there is a conflation, the authors argue

…. there remains scepticism of the evidence on the causal relation between social inequalities and health inequalities, and the social policy solutions to address them.

These are two separate issues. That social inequalities in health vary over time and place implies that policy can reduce them. I haven’t seen any convincing evidence that we can’t reduce health inequalities.

However, I think there is more debate about which policies are most effective and I support rigorous efforts to find out what they are. The authors argue

Health inequalities researchers accustomed to assigning a causal interpretation only to the results of randomized controlled trials or experimental evaluations of policy implementation, may share the fallacious belief that anything short of this ‘gold standard’ raises concerns. However, researchers from disciplines concerned with structural, political economy or lifecourse determinants of health inequalities employ distinct approaches to establish causality.

The argument that a causal framework is alien to those in the social sciences seems strange as in work across the social sciences (including economics, political science, and sociology), the potential outcomes (aka counterfactual) approach to causality is well established. It covers RCTs, natural experiments, and adjustment based regression methods. As a health inequalities researcher whose work covers lifecourse, and policy approaches it is the framework I try to utilise. A strength of such a framework is that its assumptions are clear.

The authors highlight complexity as a framework, and I have argued that such a framework seems similar to potential outcomes but its assumptions are not always clear. The authors cite Greenhalgh and as I have previously written her paper actually made the case for methods in the potential outcomes family.

The authors suggest that causal methods may hinder policy action on health inequalities. There is no reason it should but it is important that policy should be evidence informed and evaluated. There is much research and discussion about the impact of the English health inequalities strategy under the New Labour government. There seems some agreement it could have been more effective. To study and raise questions about the strategy’s effectiveness doesn’t stop future macro level policy in this area. For example, the authors highlight the possible role of income inequality in health inequality. Under New Labour income inequality remained high although its growth was largely halted. Hence, there remains much potential to reduce income inequality to possibly reduce health inequalities, and this can be studied.

Groups opposed to a particular policy direction will often choose to use a particular scientific framework to block policy action. Just look at how industry uses complexity. This doesn’t negate the scientific value of the framework.