What risk models work best for predicting whether a person will get type 2 diabetes?
There is some evidence that we can prevent diabetes from developing if we can reliably identify people who are at risk. This systematic review set out to find and evaluate risk models that have previously been validated and published in any language.
The reviewers found 48 studies in all. They considered them too heterogenous to combine their data into an overall meta-analysis.
Of the risk models reported in these studies, seven were considered practical in clinical setting.
The key findings were:
- Studies need to be clearer about how to implement the risk models they were validating.
- There were very few follow-up studies reporting the effectiveness of implementing the risk models.
- Adding genetic markers to clinical and sociodemographic risk factors did not improve the performance of the models.
Usefully, the review recommends some criteria to guide selection of a relevant clinical score for diabetes risk.
Users of this research should consider:
- Whether this review addresses a sufficiently well-defined question.
- Did the reviewers find all of the relevant studies? They did not search for grey literature, for example.
- The reviewers’ reasons for excluding particular risk models.
- To what extent are your patients or population similar to those included in the scope of the review? The reviewers state their intent is to focus on populations with no previous risk factors. Is there a valid sub-group analysis or subset of paper/s that can help you particularise the evidence to your patients?
The full text of this review is available from the BMJ at the link below.
Noble D, Mathur R, Dent T, Meads C, Greenhalgh T. Risk models and scores for type 2 diabetes: systematic review. BMJ. 2011 Nov 28;343:d7163. doi: 10.1136/bmj.d7163.