RCT finds that screening for type 2 diabetes in primary care did not improve clinical outcomes

singled out

If we pick up a disease early, we can treat it early.  So screening tests must be a good thing, right? Not necessarily.

Testing is only beneficial if it enables us to provide better care.  There are many examples from clinical practice that demonstrate how earlier is not necessarily better.  Sometimes “lead-time bias” can mislead us into thinking that testing is beneficial, when in fact, we are just living longer with a disease label.  On other cases, screening can lead to unnecessary treatment without improving survival.

We still don’t know whether population screening for type 2 diabetes is beneficial.  In August, Eyes on Evidence reported on the ADDITION randomised trial which looked at screening for type 2 diabetes in primary care.

Clinical question

In primary care, does screening fror type 2 diabetes lead to reductions on overall mortality?

Testing blood glucose

Screening included capillary blood glucose sampling nad HbA1c as well as OGTT


The trial enrolled non-diabetic participants by calculating a risk score and offering screening to those identified as high risk.  11,737 of the 15,089 patients offered screening took it up.

Randomisation was at the practice level:

  • 15 practices offered screening and subsequently took part in an intensive treatment intervention
  • 13 practices offered screening but then provided usual care.
  • 5 control practices (n=4137) performed the risk calculation but did no screening.

The screening intervention comprised random and fasting blood glucose tests, HbA1c and an oral glucose tolerance test.


The trial found no difference in overall mortality between the patients offered screening and the control group after a median of 9.6 years follow-up.  The sub-group of people who were invited but did not attend screening had higher mortality.


There are a number of limitations with this research.

Were the groups similar at the start of the trial?

It seems likely that there could be important differences between the groups of patients.  Randomising by practice makes this almost certain.  It’s also likely that researchers would have known which group a patient would be assigned to at the time their risk scores were calculated.  Inadequate allocation concealment is associated with the risk of selection bias and overestimates of treatment benefit.

Other confounders

The study contains a number of potential confounders, as noted by the researchers.  To increase statistical power, they combined groups of practices that provided intensive treatment to people identified with diabetes with those that provided usual care.  There was also some indication that the participating patients were different between the control and intervention groups.

Strengths of this research

Large numbers of patients took part in this study over a long period.  The study contains important data about the prevalance of type 2 diabetes and other risk factors for cardiovascular disease.

NHS Evidence comments:

This evidence does not add to the rationale for screening and treatment of diabetes but does demonstrate that people diagnosed with diabetes through screening were at high risk of cardiovascular disease, with high prevalence of modifiable risk factors such as overweight, high blood pressure and high cholesterol levels.


  1. Simmons RK, Echouffo-Tcheugui JB, Sharp SJ, Sargeant LA, Williams KM, Prevost AT, Kinmonth AL, Wareham NJ & Griffin SJ.  Screening for type 2 diabetes and population mortality over 10 years (ADDITION-Cambridge): a cluster-randomised controlled trial.  The Lancet 2012; 380:1741-8.
  2. NICE.  Eyes on Evidence, August 2013.