Approved Research

Three natural experiments using the UK Biobank

Lay summary

Aims:

To answer, using methods that go beyond simple correlations to isolate causal effects, the following questions

Research Question 1: Does health feedback affect health outcomes and healthcare use?

Research Question 2:  Does family environment affect long term health?

Research Question 3:  Did changes in the school leaving age in Scotland in 1963  affect longer term health?

Scientific rationale

Each of these questions are typically difficult to answer because there are usually other factors that complicate the relationships we are interested in. For example people typically receive adverse health feedback because they are in poor health, and so receiving adverse health feedback is correlated with worse health, but it does not necessarily cause it. The same is true for factors in the family environment and the level of education one receives. The novelty of this research is that we plan to exploit features of the UK Biobank dataset to answer the above questions causally, and go beyond the typical criticism of much health research, that 'correlation does not equal causation'.

Project duration: 36 Months

Public health impact:

Research Question 1: Information based interventions such as health feedback are increasingly popular in public health policy. This is for two main reasons: i) they are typically much lower cost compared to clinical interventions, and; ii) greater use of health apps on smartphones and devices such as fitbit means that there is greater potential to deliver personalised public health messages.

As such this research will be of interest to health policymakers. There is currently a government review of the NHS health check and other personalised health interventions and it is envisaged that this research will feed directly into this.

Research Question 2: Causal mechanisms in public health are difficult to identify, the findings under this research question will help to provide evidence of how differing health outcomes might occur through different family environment. Through providing evidence on these mechanisms, the findings may be suggestive of interventions that may improve health.

Research Question 3: Understanding how public policy affects health outcomes has great value in informing how public policy is designed. In this case, providing evidence of the potentially beneficial health effects of keeping people in education may influence policy decisions in this area.

A wider aim is that the research will promote the use of quasi-experimental methods on observational data in health research.