MRC/CSO Social and Public Health Sciences Unit - Informing Healthy Public Policy programme

Using instrumental variables to evaluate public health interventions

Supervisors: Dr Peter Craig; Dr Michele Hilton Boon; Dr Vittal Katikireddi; Dr Hilary Thomson
Interest is growing in the use of natural experiments to evaluate policies and interventions that are not amenable to randomised controlled trials. In the absence of randomisation researchers must find other ways to guard against selective exposure to the intervention. One is to use an instrumental variable to model exposure. Such approaches have advantages over other methods, but rely on assumptions that can be difficult to test. This project will review the use of instrumental variables in public health, evaluate their contribution to the evidence base, and identify how (if necessary) practice could be improved.

How effective is the ‘public health approach’ to preventing violence?

Supervisors: Dr Peter Craig; Dr Ruth Dundas; Dr Frank Popham; Dr Vittal Katikireddi
The ‘public health approach’ to preventing violence, pioneered by the Violence Reduction Unit set up Strathclyde Police in 2005, has become a model strategy for dealing with knife crime. Research using recorded crime and victimisation survey data suggests that, while changes in the pattern of violent crime are consistent with the focus of the new policing strategy, trends in violent crime in Glasgow since 2005 largely follow trends elsewhere in Scotland. The project will take a natural experimental perspective, using hospitalisation data alongside other sources to identify the effect of the public health approach to preventing violence.

Evidence for ‘health in all policies’

Supervisors: Dr Hilary Thomson; Dr Michele Hilton Boon; Dr Vittal Katikireddi
There is a diverse range of outcomes which may be considered within a ‘health in all policies’ approach. This can result in conflict about the overall benefits and harms, and can make decision making difficult. For example, investment in road infrastructure may impact positively or negatively on health determinants, such as economic opportunities, air quality, accidents and community networks. This project will use a range of evidence synthesis and primary research methods to consider whether and how diverse outcomes can be systematically considered by those pursuing a health in all policies approach to policy-making.

What are policy recommendations in public health research papers based on?

Supervisors: Dr Peter Craig; Dr Hilary Thomson; Dr Michele Hilton Boon
Research quality assessment increasingly takes account of actual or potential impact. Researchers face powerful incentives to emphasise the relevance of their work for policy and practice, yet reporting guidelines provide little guidance on how to make recommendations. This project will investigate how far recommendations for policy and practice made in research papers are supported by the data reported, and how recommendations from individual studies compare with recommendations from related systematic reviews. It should appeal to candidates interested in public health policy and evidence synthesis methods.