• Housing
• policy analysis SI

We all believe that policies outside the traditional health sector have major impacts on health. But empirically demonstrating such social causation is another matter. Housing is one such example. There is a strong theoretical case for improved housing (eg, insulation, mould reduction) improving health (eg, respiratory and cardiovascular diseases).1 2 In some cases, one can conduct randomised trials. For example, we have previously reported that people randomised to homes that were then insulated had substantial and significant improvements (ie, about 50% reductions and highly statistically significant) on multiple self-reported health measures and visits to general practitioners. On the objective measure of respiratory hospitalisations, we found a rate ratio of 0.53 (95% CI 0.22 to 1.29) for randomised subjects versus controls.3

Demonstrating that actual housing policies impact on health—especially on more objective measures of health—using ‘real-life’ data from observational studies is extremely challenging. Classic epidemiological threats to internal validity abound: people taking up, or offered, the housing intervention may be very different from those not (confounding); robust …