The team behind the heavily-criticised GP Survey has issued a robust defence in a new paper on BMJ.com.

When originally published in July, the survey was attacked by doctors and by the BMA for its low response rate and failure to show that those patients who failed to respond – around 60 per cent – held the same opinions as those who did.
 
Because two questions in the survey were used to determine the distribution of £68 million to GPs, and some claimed to have lost thousands of pounds as a result, the issue rankled. Professor Martin Roland of Cambridge and colleagues attempt to lay these doubts to rest.
 
They do this by comparing the age, sex, and deprivation profile of those who responded against those who were sent the questionnaire. This shows that the responses under-represented the young, and over-represented the old: not surprising, as older people are more likely to attend GP surgeries and have an opinion to offer, as well as time to fill in a questionnaire. The responses also under-represented some ethnic minorities, especially those of Asian or Asian/British origin.
 
Practices where responses were high tended to score well on the two questions on GP access that were worth money. However, the team says that this can be accounted for by the demography of these practices and does not imply non-response bias. There was no evidence that a low response to these questions disadvantaged the practice the respondents belonged to (or a high one, for that matter).
 
A major reason why some practices suffered was a change in the way the scores were linked to GPs’ pay. The thresholds for getting any money at all  were increased, and practices that just passed the threshold got nothing – in the previous year, just passing the threshold would have given you 50 per cent of the total payment on offer.
 
This change alone would have cost most practices money, but was counterbalanced by there being more money in the pot. That meant that some practices lost money and others gained, while their performance remained unchanged.
 
This may seem an odd way to incentivise GPs, but the message from Professor Roland is: “Don’t blame the survey”.