Endocrine Abstracts (2011) 27 P54

Confounding factors and variations in HbA1c collection methods have not shown different HbA1c results as compared to the National Paediatric Diabetes Audit Results

Thomas Reed & Birgit E van Meijgaarden


Basildon and Thurrock Hospitals, Essex, UK.


The National Diabetes Paediatric Audit Report (NDPAR) 2009–10 was published earlier this year. A total of 155 units submitted data to the report, representing an overall increase of 31 units since the 2008–9 audit. Whilst non-participation has been largely attributed to lack of resources and technical infrastructure, there are some concerns over the method of data collection and interpretation.

Our unit entered the data by using the automatic extraction function of the Twinkle database. With this method only the latest single HbA1c entrance is used which gave our unit a median value of 9.1% (range=6.0–15.0%, n=214). We reviewed the data of the same 235 patients that we submitted to the 2009–10 NDPAR and used different criteria to analyse the HbA1c. We hypothesised that audit results could be confounded by not taking an average HbA1C per patient over the whole audit period, the HbA1c within 6 or 12 months of diagnosis as well as patients that came to us from out of area with difficult to manage diabetes.

In our analysis we calculated the mean HbA1c value for each patient over the whole period giving a median of 9.13% (n=216), representing no difference to the NDPAR.

With regards to patients with newly diagnosed diabetes during the audit period, exclusion gave a median of 9.13% (n=195). When we excluded new diagnoses up to a year before the audit period we found a median of 9.23% (n=174), showing a slight increase of the median.

Finally, we excluded the patients that were from out of area which resulted in a median of 9.13% (n=207).

Excluding both new diagnoses up to a year before the audit period and our out of area patients resulted in an unexpected increase of the median to 9.25% (n=165).

Overall the results of our review show no, or only minimal, differences in HbA1c values for the different confounding factors as compared to the NDPAR, suggesting that the current data collection is a valid method.

Article tools

My recent searches

No recent searches.