The British Government and its gambling regulator, the Gambling Commission, have a new way of assessing the adult prevalence of problem gambling. Previously it was funding regular dedicated national gambling surveys, the British Gambling Prevalence Surveys, of which three had been carried out, in 2000, 2007 and 2010. We were led to believe that, as part of fulfilling their responsibility to protect the public, similar surveys would be carried out every three years. Britain was a leader internationally in that respect. Unfortunately the funding was withdrawn and a cheaper way of trying to assess prevalence was chosen which involves adding questions about gambling on to general health surveys. We now have the first results from that exercise in the form of chapters about gambling in the reports of the Eighth Scottish Health Survey (SHeS) 2012 and the Health Survey for England (HSE) 2012 (although the fieldwork for the latter was carried out in March 2013). Some of the disadvantages of this new way of doing things are immediately apparent. Although the leading social survey organisation, NatCen, which carried out the earlier British Gambling Prevalence Surveys, was involved in both the Scottish and English health surveys, the reports are separate and some of the analyses carried out are different. No similar exercise has been carried out in Wales.

The same two sets of problem gambling screening questions as were used in the British Gambling Prevalence Surveys were used again in the two health surveys (the DSM and the PGSI scales). The overall problem gambling prevalence found in the SHeS was 7 per 1000 according to both scales, and in the HSE was about 5 per 1000 according to DSM and 4 per 1000 according to PGSI (in the report the figures are given separately for men and women and are not combined). The English figures are lower than those found in the 2010 British Gambling Prevalence Survey, a fact highlighted in the press release issued by the Gambling Commission and not surprisingly made much of by the gambling industry. If these new figures are a true reflection of rates of problem gambling then a reduction is to be welcomed. There is a need for caution however. The methods used in nearly 200 gambling surveys from around the world were reviewed in a 2012 report by the Ontario Problem Gambling Research Centre and the Ontario Ministry of Health and Long-Term Care. One of a number of conclusions in that report was that estimated rates of problem gambling tend to be lower when questions about gambling are included in general health surveys and higher when gambling is the main focus of the whole survey. There was another difference between the methods used in the 2010 British Gambling Prevalence Survey and those used in the recent health surveys. The former used 'computer-assisted interviewing' whereas the latter used paper and pencil self-completion which makes it much easier for respondents to miss out questions or even whole sections, which was made even more likely by the much greater length of the health surveys. As many as 19% of those who were eligible to complete the problem gambling screening questions, declined to do so. Because there is no way of knowing whether those with gambling problems were over-represented amongst those who declined, this adds uncertainty to the prevalence estimate which is already somewhat uncertain due to the only moderate response rate achieved by such surveys these days (taking into account both failure to make contact at a proportion of randomly chosen addresses and a failure to obtain responses from all adults at those addresses where contact is made, the overall response rate in both health surveys was estimated at about 56%). It is worth noting that 'computer-assisted interviewing' was introduced into the British Gambling Prevalence Surveys for the first time in 2010, which could have been a factor in the apparent 40 to 50% increase in the estimated prevalence of problem gambling between 2007 and 2010.

The bare overall prevalence figure masks substantial differences between different socio-economic groups within a country's population. For a start, gender differences remain strong, particularly it seems in Scotland where the problem gambling prevalence amongst men who had done any gambling at all, even if only once, in the last year, was 19 per 1000 compared to only 1 to 3 per 1000 amongst women. The comparable figures for England were 9 to 11 per 1000 for men and 2 to 3 per 1000 for women. There was also evidence from both health surveys that problem gambling was significantly higher in relatively deprived areas. In Scotland, prevalence was estimated to be nearly 7 times as high amongst that third of the respondents who lived in the more deprived areas according to a standard Index of Multiple Deprivation compared to the prevalence amongst the third who lived in the least deprived areas. In England the rate was nearly twice as high amongst those respondents who lived in Spearhead Primary Care Trust areas compared to other respondents (just over a third of the sample lived in Spearhead PCT areas, the most health deprived areas of England according to indicators such as life expectancy, cancer and cardiovascular disease mortality and an Index of Multiple Deprivation). There was also strong evidence from both surveys that problem gambling is related to poorer mental health: those with scores on the General Health Questionnaire (GHQ) which indicated a clinically significant level of mental health problem were 4 to 5 times as likely to have a gambling problem as those who scored zero on the GHQ.

The point should also be made that, when trying to interpret such findings, the cut-off point on a screening scale for defining a particular respondent as having a gambling problem, although following an internationally agreed convention, is somewhat arbitrary. It is generally agreed that gambling problems are not a clear-cut matter of some people having such a problem and others not. Of the two problem gambling screening scales, the Canadian Problem Gambling Severity Index (PGSI) recognises that problem gambling lies on a continuum. For example in the HSE, according to the PGSI, to the 6 per 1000 men defined as 'problem gamblers' could be added another 17 per 1000 whose gambling was putting them at 'moderate risk' and to the 2 per 1000 women 'problem gamblers' could be added another 4 per 1000 at 'moderate risk'; in the SHeS, the comparable figures for men were 7 per 1000 'problem gamblers' plus 21 per 1000 at 'moderate risk' (in Scotland the numbers of women at 'moderate risk' were too low to put a figure on it).

The final point to make is that this new method of carrying out national problem gambling surveys leaves a lot to be desired in other ways. Compared to the questions that could be asked in the gambling-dedicated British Gambling Prevalence Surveys, the numbers of questions that could be asked about gambling in the general health surveys was comparatively very small. There were no questions for example about the frequency of people's gambling and no attempt was made to ask any questions about how much people spent or lost gambling. That means that it is impossible to explore further what it is about people's gambling that is creating problems or why prevalence might be higher in certain groups compared to others. There was no room for questions about people's motives for gambling which was something that the series of British Gambling Prevalence Surveys started to explore in the 2010 survey and which could be developed further. Nor was there any scope for asking about people's attitudes towards gambling, a subject which is of vital importance if we want to understand what the British public thinks about gambling and gambling policy and how attitudes might or might not be changing, a subject which had already been taken some considerable way in the 2007 and 2010 British Gambling Prevalence Surveys.