- Split View
-
Views
-
Cite
Cite
CHRISTOPHER LITTLEJOHN, DOES SOCIO-ECONOMIC STATUS INFLUENCE THE ACCEPTABILITY OF, ATTENDANCE FOR, AND OUTCOME OF, SCREENING AND BRIEF INTERVENTIONS FOR ALCOHOL MISUSE: A REVIEW, Alcohol and Alcoholism, Volume 41, Issue 5, September/October 2006, Pages 540–545, https://doi.org/10.1093/alcalc/agl053
- Share Icon Share
Abstract
Aims: To determine whether socio-economic status (SES) influences (i) willingness to participate in brief intervention (BI) research, (ii) attendance to receive BI once allocated, and (iii) treatment outcome. Methods: Systematic review of published, randomised controlled trials of BI for non-dependent alcohol misuse in primary health care settings. Results: Eighteen papers met inclusion criteria. There is evidence that once recruited, and following attendance for intervention, participants' SES does not influence treatment outcome. However, the effect of choosing to participate remains unclear, and the generalizability of results to the whole primary care population remains equivocal. Socio-economic status may influence willingness to participate in BI treatment research, and may influence attendance to receive such interventions where allocated. Conclusion: Brief interventions should remain available to all non-dependent hazardous and harmful drinkers in primary care. However, fidelity to research design is suggested to allow for any participation effects to occur. Benefits of such an approach exist for both clinicians and patients. The characteristics of those who participate in BI trials, compared to those who do not, should be studied in detail. Socio-economic variables should be included as potentially important characteristics. The impact of BI on drinking style as well as consumption needs further attention.
(Received 16 January 2006; first review notified 12 June 2006; in revised form 13 June 2006; accepted 13 June 2006)
INTRODUCTION
Brief interventions (BI) have been identified as one of the most effective of all interventions for alcohol misuse (Miller and Wilbourne, 2002). There are multiple meta-analyses demonstrating their efficacy (Bien et al., 1993; Wilk et al., 1997; Moyer et al., 2002). This meta-analytic support has been replicated even when intention-to-treat analysis has been used (Ballesteros et al., 2004), overcoming earlier concerns about the typically high levels of attrition reported in trials of screening and brief intervention (SBI) (Bien et al., 1993; Edwards and Rollnick, 1997). However, does this evidence directly translate to all clinical populations in primary health care? The purpose of this review is to examine whether socio-economic status (SES) is an important variable in willingness to accept SBI, and subsequent outcome following SBI.
SES variables clearly influence alcohol-related harm. Those in lower socio-economic groups suffer disproportionately from alcohol-related mortality and morbidity (Mäkelä, 1999; Bellis et al., 2005). This remains to be fully explained, although there are some pointers to potentially important factors. For example, while increasing alcohol consumption increases oxidative stress as a result of the induction of the CYP 2E1 enzyme pathway, the effects of such increases can be exacerbated or ameliorated by nutritional status (Chase et al., 2005). SES also influences drinking pattern. Pattern of drinking can be as important as overall consumption in the development of alcohol-related harm (Rehm et al., 2003). Population-level research in the UK has demonstrated that the weekly per capita consumption of alcohol is similar across the social spectrum (National Statistics, 2002). If anything, there is a trend towards higher weekly consumption levels amongst professional groups. However, lower socio-economic groups have higher overall levels of non-drinking, and fewer drinking days per week for those who do drink (National Statistics, 2002). However, both groups report similar levels of binge drinking (defined as consuming >8/> 6 units of alcohol for men/women in one day). For weekly per capita consumption levels to be similar across groups, those fewer drinkers amongst the lower SES group must be consuming more, on fewer occasions, than those of higher SES. This increased tendency towards binge drinking in lower SES groups has been previously identified (Mäkelä, 1999; Erens, 2000), and increases the risks of acute, intoxication-related harms, such as trauma and assault.
This review addresses three questions. First, does SES influence the willingness of hazardous and harmful (i.e. non-dependent) drinkers to participate in BI research? Second, does SES influence subsequent attendance to receive BI in participants so allocated? Third, does SES influence BI treatment outcomes? The hypothesis developed for testing is that hazardous and harmful drinkers of lower SES will be less willing to participate in BI research, less likely to attend to receive BI, and will show poorer treatment outcomes when compared to those of higher SES.
METHODS
‘Cinahl’, ‘Embase’, ‘Medline’, ‘PsycInfo’, ‘the Cochrane Library’, and ‘Index to Theses’ (www.theses.com) were searched using the terms ‘brief intervention’ and ‘minimal intervention’ as keywords. Attempts were made to focus the searches by combining these with ‘alcohol’ or ‘primary care’. Reference lists of selected papers, and of published systematic reviews and meta-analyses (Bien et al., 1993; Wilk et al., 1997; Poikolainen, 1999; Moyer et al., 2002; Beich et al., 2003; Ballesteros et al., 2004; Bertholet et al., 2005) were also searched. Full papers were reviewed based upon the content of published abstracts with potentially relevant titles. Papers were included that reported on randomized controlled studies of BI for non-dependent, hazardous, or harmful alcohol drinking in primary care settings. Only papers in English were considered. Papers describing secondary analysis of previously published BI trial data were excluded. BI was defined as an intervention providing feedback and advice to change to non-dependent, non-treatment seeking alcohol drinkers, where the intervention was provided by generalist medical or nursing staff. Trials that reported solely on specialist treatments for alcohol use disorders were excluded. Specialist treatment was defined as any intervention delivered by a practitioner or service whose specialist focus was substance misuse or addiction.
RESULTS
The literature searches produced 1281 citations, reduced to 606 when combined with ‘alcohol’ or ‘primary care’. Many citations were duplicated across databases. The search of the Cochrane Library produced 106 citations, none of which were included. The search of ‘Index to Theses’ (www.theses.com) produced no relevant citations.
Forty-four primary papers were found that involved BIs for alcohol misuse in primary care settings. Of these 26 were excluded. Eight papers were sub-analyses of previously published data (Adams et al., 1998; Fleming et al., 2000, 2002; Freeborn et al., 2000; Gordon et al., 2003; Grossberg et al., 2004; Manwell et al., 2000; Mundt et al., 2005). Four papers were long-term follow-ups of previously published trials (Kristenson et al., 2002; Nilssen, 2004; Reiff-Hekking et al., 2005; Wutzke et al., 2002). Ten papers were excluded because they either reported a high proportion of alcohol dependent participants or did not actively attempt to screen out those with alcohol dependence (Kristenson et al., 1983; Wallace et al., 1988; Romelsjö et al., 1989; Seppä, 1992; Israel et al., 1996; Burge et al., 1997; McIntosh et al., 1997; Maisto et al., 2001; Copeland et al., 2003; Saitz et al., 2003). One paper described a BI delivered by a specialist psychiatrist (Chang et al., 1997). One paper described a non-randomised design (Scott, 2000). One paper addressed multiple health behaviours, confounding attempts to focus upon alcohol consumption alone (Burton et al., 1995). One paper recruited from but did not report on the data gathered from primary care (Saunders et al., 1992).
Of the remaining 18 papers (Table 1), 2 papers were included despite also recruiting outwith primary care (Rollnick et al., 1992; Skutle, 1992), as the primary care data appeared relevant to the review topic. One paper was included despite the single BI being delivered by research staff trained in motivational interviewing (Senft et al., 1997). This was because addiction specialists did not provide the intervention, and the intervention described appeared similar to that as would be provided by trained generalist primary care clinicians. One paper was included despite including those with symptoms of alcohol dependence, as the proportion was reported to be very small (2%) (Ockene et al., 1999). One paper was included despite being a sub-analysis of previously published data (Aalto and Sillanaukee, 2000), because the sub-analysis directly related to the topic of this review.
Study . | No. screened (N) . | Positive screened (N) . | Declined to participate in research interview to establish eligibility N (%) . | Trial participants (N) . | Randomised to BI (N) . | Number attended to receive BI (N) . | Analysis of non-attenders reported? . | SES variables considered re outcome? . | |||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
Heather et al., 1987 | Not reported | Not reported | Not reported | 104 | 32 to simple advice, 34 to multiple sessions | Not reported | No | No | |||||||
Scott and Anderson, 1990 | 11 521 | 384 (32 randomised to no-interview control group) | 187 (53) | 72 | 33 | Not reported | No | Yes | |||||||
Nilssen, 1991 | 21 647 | 338 | Not reported | 338 | 113 | Not reported | No | No | |||||||
Anderson and Scott, 1992 | 8483 | 803 (105 randomised to no-interview control group) | 337 (48) | 154 | 80 | Not reported | No | Yes | |||||||
Rollnick et al., 1992 | 3467 in Primary Care | Not reported | Not reported | 110 [total participants recruited in both Primary Care and Hospital] | 26 for simple advice; 32 for brief counselling [total participants recruited in both Primary Care and Hospital] | 50% of Primary Care eligible patients attended interview during which randomisation and BI (if indicated) provided | No | No | |||||||
Skutle, 1992 | 1797 | 167 (includes unknown number recruited from work-site location) | 81 declined, and further 12 agreed to participate but did not attend (55.7) | 55 [total participants recruited in both Primary Care and work-site] | 5 for simple advice; 9 for brief counselling [total participants recruited in both Primary Care and work-site] | 82.1% | No | No | |||||||
Richmond et al., 1995 | 13 017 | 713 | 246 (34.5) | 378 | 96 randomised to 5 sessions; 96 randomised to single session | 47 (49%) returned for at least first of 5 sessions; all single session participants received the intervention | Yes | No | |||||||
Fleming et al., 1997 | 17,695 | 2,450 | 1,705 (69.6) | 774 | 392 | 307 (78.3%) | Yes | Yes | |||||||
Senft et al., 1997 | 8017 | 620 | 104 (17) | 516 | 260 | 205 (79%) | Yes (re. Follow-up) | No | |||||||
Córdoba et al., 1998 | Not reported | Not reported | Not reported | 546 | 104 | Not reported | No | No | |||||||
Tomson et al., 1998 | 2338 | 222 | 18 (8.1) did not attend (and further 15 excluded) | 82 | 39 | 30 (76.9%) | No | No | |||||||
Fleming et al., 1999 | 6073 (further 514 refused) | 656 | 260 (39.6) | 158 | 87 | 84 (96.6%), 92.4% followed up at 12 months | No | Yes | |||||||
Ockene et al., 1999 | 9772 | 1760 | 275 (16%) | 545 | 274 | Not reported | No | No | |||||||
Aalto et al., 2000 | 7539 | 353 | Note: directly invited by GP to participate: 216 (61) declined | 137 | All 137 randomised to 1, 3 or 7 sessions | 188 (86.1%) attended | No | No | |||||||
Aalto et al., 2001 | 4258 | 658 | Note: directly invited by GP to participate: 308 (47) declined | 350 | All 350 randomised to receive 1, 3 or 7 sessions | 296 (84.6%) attended | No | No | |||||||
Aalto and Sillanaukee, 2000 | Sub-analysis of compliance rates for Aalto (2000,2001) | ||||||||||||||
Curry et al., 2003 | 4793 | 380 | 28 (7) | 333 | 166 | 151 (91%) | Yes | No | |||||||
Fleming et al., 2004 | Not reported | Not reported | Not reported | 151 | 81 | Not reported | No | No |
Study . | No. screened (N) . | Positive screened (N) . | Declined to participate in research interview to establish eligibility N (%) . | Trial participants (N) . | Randomised to BI (N) . | Number attended to receive BI (N) . | Analysis of non-attenders reported? . | SES variables considered re outcome? . | |||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
Heather et al., 1987 | Not reported | Not reported | Not reported | 104 | 32 to simple advice, 34 to multiple sessions | Not reported | No | No | |||||||
Scott and Anderson, 1990 | 11 521 | 384 (32 randomised to no-interview control group) | 187 (53) | 72 | 33 | Not reported | No | Yes | |||||||
Nilssen, 1991 | 21 647 | 338 | Not reported | 338 | 113 | Not reported | No | No | |||||||
Anderson and Scott, 1992 | 8483 | 803 (105 randomised to no-interview control group) | 337 (48) | 154 | 80 | Not reported | No | Yes | |||||||
Rollnick et al., 1992 | 3467 in Primary Care | Not reported | Not reported | 110 [total participants recruited in both Primary Care and Hospital] | 26 for simple advice; 32 for brief counselling [total participants recruited in both Primary Care and Hospital] | 50% of Primary Care eligible patients attended interview during which randomisation and BI (if indicated) provided | No | No | |||||||
Skutle, 1992 | 1797 | 167 (includes unknown number recruited from work-site location) | 81 declined, and further 12 agreed to participate but did not attend (55.7) | 55 [total participants recruited in both Primary Care and work-site] | 5 for simple advice; 9 for brief counselling [total participants recruited in both Primary Care and work-site] | 82.1% | No | No | |||||||
Richmond et al., 1995 | 13 017 | 713 | 246 (34.5) | 378 | 96 randomised to 5 sessions; 96 randomised to single session | 47 (49%) returned for at least first of 5 sessions; all single session participants received the intervention | Yes | No | |||||||
Fleming et al., 1997 | 17,695 | 2,450 | 1,705 (69.6) | 774 | 392 | 307 (78.3%) | Yes | Yes | |||||||
Senft et al., 1997 | 8017 | 620 | 104 (17) | 516 | 260 | 205 (79%) | Yes (re. Follow-up) | No | |||||||
Córdoba et al., 1998 | Not reported | Not reported | Not reported | 546 | 104 | Not reported | No | No | |||||||
Tomson et al., 1998 | 2338 | 222 | 18 (8.1) did not attend (and further 15 excluded) | 82 | 39 | 30 (76.9%) | No | No | |||||||
Fleming et al., 1999 | 6073 (further 514 refused) | 656 | 260 (39.6) | 158 | 87 | 84 (96.6%), 92.4% followed up at 12 months | No | Yes | |||||||
Ockene et al., 1999 | 9772 | 1760 | 275 (16%) | 545 | 274 | Not reported | No | No | |||||||
Aalto et al., 2000 | 7539 | 353 | Note: directly invited by GP to participate: 216 (61) declined | 137 | All 137 randomised to 1, 3 or 7 sessions | 188 (86.1%) attended | No | No | |||||||
Aalto et al., 2001 | 4258 | 658 | Note: directly invited by GP to participate: 308 (47) declined | 350 | All 350 randomised to receive 1, 3 or 7 sessions | 296 (84.6%) attended | No | No | |||||||
Aalto and Sillanaukee, 2000 | Sub-analysis of compliance rates for Aalto (2000,2001) | ||||||||||||||
Curry et al., 2003 | 4793 | 380 | 28 (7) | 333 | 166 | 151 (91%) | Yes | No | |||||||
Fleming et al., 2004 | Not reported | Not reported | Not reported | 151 | 81 | Not reported | No | No |
Study . | No. screened (N) . | Positive screened (N) . | Declined to participate in research interview to establish eligibility N (%) . | Trial participants (N) . | Randomised to BI (N) . | Number attended to receive BI (N) . | Analysis of non-attenders reported? . | SES variables considered re outcome? . | |||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
Heather et al., 1987 | Not reported | Not reported | Not reported | 104 | 32 to simple advice, 34 to multiple sessions | Not reported | No | No | |||||||
Scott and Anderson, 1990 | 11 521 | 384 (32 randomised to no-interview control group) | 187 (53) | 72 | 33 | Not reported | No | Yes | |||||||
Nilssen, 1991 | 21 647 | 338 | Not reported | 338 | 113 | Not reported | No | No | |||||||
Anderson and Scott, 1992 | 8483 | 803 (105 randomised to no-interview control group) | 337 (48) | 154 | 80 | Not reported | No | Yes | |||||||
Rollnick et al., 1992 | 3467 in Primary Care | Not reported | Not reported | 110 [total participants recruited in both Primary Care and Hospital] | 26 for simple advice; 32 for brief counselling [total participants recruited in both Primary Care and Hospital] | 50% of Primary Care eligible patients attended interview during which randomisation and BI (if indicated) provided | No | No | |||||||
Skutle, 1992 | 1797 | 167 (includes unknown number recruited from work-site location) | 81 declined, and further 12 agreed to participate but did not attend (55.7) | 55 [total participants recruited in both Primary Care and work-site] | 5 for simple advice; 9 for brief counselling [total participants recruited in both Primary Care and work-site] | 82.1% | No | No | |||||||
Richmond et al., 1995 | 13 017 | 713 | 246 (34.5) | 378 | 96 randomised to 5 sessions; 96 randomised to single session | 47 (49%) returned for at least first of 5 sessions; all single session participants received the intervention | Yes | No | |||||||
Fleming et al., 1997 | 17,695 | 2,450 | 1,705 (69.6) | 774 | 392 | 307 (78.3%) | Yes | Yes | |||||||
Senft et al., 1997 | 8017 | 620 | 104 (17) | 516 | 260 | 205 (79%) | Yes (re. Follow-up) | No | |||||||
Córdoba et al., 1998 | Not reported | Not reported | Not reported | 546 | 104 | Not reported | No | No | |||||||
Tomson et al., 1998 | 2338 | 222 | 18 (8.1) did not attend (and further 15 excluded) | 82 | 39 | 30 (76.9%) | No | No | |||||||
Fleming et al., 1999 | 6073 (further 514 refused) | 656 | 260 (39.6) | 158 | 87 | 84 (96.6%), 92.4% followed up at 12 months | No | Yes | |||||||
Ockene et al., 1999 | 9772 | 1760 | 275 (16%) | 545 | 274 | Not reported | No | No | |||||||
Aalto et al., 2000 | 7539 | 353 | Note: directly invited by GP to participate: 216 (61) declined | 137 | All 137 randomised to 1, 3 or 7 sessions | 188 (86.1%) attended | No | No | |||||||
Aalto et al., 2001 | 4258 | 658 | Note: directly invited by GP to participate: 308 (47) declined | 350 | All 350 randomised to receive 1, 3 or 7 sessions | 296 (84.6%) attended | No | No | |||||||
Aalto and Sillanaukee, 2000 | Sub-analysis of compliance rates for Aalto (2000,2001) | ||||||||||||||
Curry et al., 2003 | 4793 | 380 | 28 (7) | 333 | 166 | 151 (91%) | Yes | No | |||||||
Fleming et al., 2004 | Not reported | Not reported | Not reported | 151 | 81 | Not reported | No | No |
Study . | No. screened (N) . | Positive screened (N) . | Declined to participate in research interview to establish eligibility N (%) . | Trial participants (N) . | Randomised to BI (N) . | Number attended to receive BI (N) . | Analysis of non-attenders reported? . | SES variables considered re outcome? . | |||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
Heather et al., 1987 | Not reported | Not reported | Not reported | 104 | 32 to simple advice, 34 to multiple sessions | Not reported | No | No | |||||||
Scott and Anderson, 1990 | 11 521 | 384 (32 randomised to no-interview control group) | 187 (53) | 72 | 33 | Not reported | No | Yes | |||||||
Nilssen, 1991 | 21 647 | 338 | Not reported | 338 | 113 | Not reported | No | No | |||||||
Anderson and Scott, 1992 | 8483 | 803 (105 randomised to no-interview control group) | 337 (48) | 154 | 80 | Not reported | No | Yes | |||||||
Rollnick et al., 1992 | 3467 in Primary Care | Not reported | Not reported | 110 [total participants recruited in both Primary Care and Hospital] | 26 for simple advice; 32 for brief counselling [total participants recruited in both Primary Care and Hospital] | 50% of Primary Care eligible patients attended interview during which randomisation and BI (if indicated) provided | No | No | |||||||
Skutle, 1992 | 1797 | 167 (includes unknown number recruited from work-site location) | 81 declined, and further 12 agreed to participate but did not attend (55.7) | 55 [total participants recruited in both Primary Care and work-site] | 5 for simple advice; 9 for brief counselling [total participants recruited in both Primary Care and work-site] | 82.1% | No | No | |||||||
Richmond et al., 1995 | 13 017 | 713 | 246 (34.5) | 378 | 96 randomised to 5 sessions; 96 randomised to single session | 47 (49%) returned for at least first of 5 sessions; all single session participants received the intervention | Yes | No | |||||||
Fleming et al., 1997 | 17,695 | 2,450 | 1,705 (69.6) | 774 | 392 | 307 (78.3%) | Yes | Yes | |||||||
Senft et al., 1997 | 8017 | 620 | 104 (17) | 516 | 260 | 205 (79%) | Yes (re. Follow-up) | No | |||||||
Córdoba et al., 1998 | Not reported | Not reported | Not reported | 546 | 104 | Not reported | No | No | |||||||
Tomson et al., 1998 | 2338 | 222 | 18 (8.1) did not attend (and further 15 excluded) | 82 | 39 | 30 (76.9%) | No | No | |||||||
Fleming et al., 1999 | 6073 (further 514 refused) | 656 | 260 (39.6) | 158 | 87 | 84 (96.6%), 92.4% followed up at 12 months | No | Yes | |||||||
Ockene et al., 1999 | 9772 | 1760 | 275 (16%) | 545 | 274 | Not reported | No | No | |||||||
Aalto et al., 2000 | 7539 | 353 | Note: directly invited by GP to participate: 216 (61) declined | 137 | All 137 randomised to 1, 3 or 7 sessions | 188 (86.1%) attended | No | No | |||||||
Aalto et al., 2001 | 4258 | 658 | Note: directly invited by GP to participate: 308 (47) declined | 350 | All 350 randomised to receive 1, 3 or 7 sessions | 296 (84.6%) attended | No | No | |||||||
Aalto and Sillanaukee, 2000 | Sub-analysis of compliance rates for Aalto (2000,2001) | ||||||||||||||
Curry et al., 2003 | 4793 | 380 | 28 (7) | 333 | 166 | 151 (91%) | Yes | No | |||||||
Fleming et al., 2004 | Not reported | Not reported | Not reported | 151 | 81 | Not reported | No | No |
Does SES influence willingness to participate in BI research trials?
The high level of non-participation among potential participants has been previously identified (Edwards and Rollnick, 1997). Of the 18 papers comprising this review, 12 provided data on numbers of positively screened potential participants who declined to participate in the research. On average 38% (SD 21.41, 95% CI 25.89–50.15) of potential participants declined to participate following screening. Only two papers (11.1%) compared the characteristics of potential participants who consented against those who declined (Aalto and Sillanaukee, 2000; Senft et al., 1997). Unlike many other studies, both these papers informed positively screened, potential participants of the specific alcohol-focus of the research. Aalto and Sillanaukee (2000) reported that it was older, heavier drinking men who were more likely to participate in the study. They also found that differences in employment status or educational status did not influence likelihood of participation. Senft et al. (1997) also reported that educational status did not influence participate rates. These variables were operationalized differently by different authors (Table 2).
. | Variable as defined by study . | . | |
---|---|---|---|
Author . | Employment . | Education . | |
Aalto and Sillanaukee (2000) | Working/studying or unemployed or retired | Comprehensive school or vocational school or college or university | |
Fleming et al. (1997) | Professional or technical-mechanical or student or labour-machine or retired or farm or sales-service or homemaker or unemployed | High school or less or some college or college degree or more | |
Fleming et al. (1999) | — | High school or less or some college or college degree or more | |
Richmond et al. (1995) | Employed % | With further education % | |
Scott and Anderson (1990), Anderson and Scott (1992) | Social class | — | |
Senft et al. (1997) | — | Some college or more education % |
. | Variable as defined by study . | . | |
---|---|---|---|
Author . | Employment . | Education . | |
Aalto and Sillanaukee (2000) | Working/studying or unemployed or retired | Comprehensive school or vocational school or college or university | |
Fleming et al. (1997) | Professional or technical-mechanical or student or labour-machine or retired or farm or sales-service or homemaker or unemployed | High school or less or some college or college degree or more | |
Fleming et al. (1999) | — | High school or less or some college or college degree or more | |
Richmond et al. (1995) | Employed % | With further education % | |
Scott and Anderson (1990), Anderson and Scott (1992) | Social class | — | |
Senft et al. (1997) | — | Some college or more education % |
. | Variable as defined by study . | . | |
---|---|---|---|
Author . | Employment . | Education . | |
Aalto and Sillanaukee (2000) | Working/studying or unemployed or retired | Comprehensive school or vocational school or college or university | |
Fleming et al. (1997) | Professional or technical-mechanical or student or labour-machine or retired or farm or sales-service or homemaker or unemployed | High school or less or some college or college degree or more | |
Fleming et al. (1999) | — | High school or less or some college or college degree or more | |
Richmond et al. (1995) | Employed % | With further education % | |
Scott and Anderson (1990), Anderson and Scott (1992) | Social class | — | |
Senft et al. (1997) | — | Some college or more education % |
. | Variable as defined by study . | . | |
---|---|---|---|
Author . | Employment . | Education . | |
Aalto and Sillanaukee (2000) | Working/studying or unemployed or retired | Comprehensive school or vocational school or college or university | |
Fleming et al. (1997) | Professional or technical-mechanical or student or labour-machine or retired or farm or sales-service or homemaker or unemployed | High school or less or some college or college degree or more | |
Fleming et al. (1999) | — | High school or less or some college or college degree or more | |
Richmond et al. (1995) | Employed % | With further education % | |
Scott and Anderson (1990), Anderson and Scott (1992) | Social class | — | |
Senft et al. (1997) | — | Some college or more education % |
Neither paper identified employment-related socio-economic classification. Only Aalto and Sillanaukee (2000) differentiated between graduates and non-graduates, but did not identify the SES of participants within these groups.
Does SES influence attendance for BI once enrolled in a treatment trial?
Of the 18 papers, 12 (67%) reported on the attendance rates of those randomly allocated to the intervention condition. Reported attendance rates ranged from 49% (Richmond et al., 1995) through 99% (Ockene et al., 1999). Four papers compared the characteristics of attendees compared to defaulters. Fleming et al. (1997) reported that variables such as employment (stratified by occupational level; see Table 2) or educational status (college degree or not) did not account for whether participants attended or defaulted. Richmond et al. (1995), however, found that higher SES (operationalization detailed in Table 2) was associated with higher attendance. In their study, younger, less educated, heavier drinkers were most likely to default (Richmond et al., 1995).
Two papers also examined whether socio-demographic variables influenced follow-up retention rates. Senft et al. (1997) found that higher educational status increased the likelihood of follow-up attendance, while Curry et al. (2003) did not.
Does SES influence the outcome following BI for alcohol misuse?
Nine of the 18 papers reported a significant outcome of intervention compared to control. Of these, four papers reported on SES influence on outcome. Anderson and Scott (1992) collected data on the social class of participants. They did not report the proportion of the intervention group who attended for the subsequent BI appointment, however they did report their findings based upon intent-to-treat analysis. Fleming et al. (1997) reported that 78% of the intervention group received a BI, and analysed the results in relation to employment status stratified by occupational status. Fleming et al. (1999) reported that 96.6% of the intervention group attended at least one appointment, and analysed the results in relation to educational status. Each of these papers reported significant reductions in the intervention groups' alcohol consumption. However, Anderson and Scott (1992) and Fleming et al. (1997) also found that binge drinking was reduced in the intervention group relative to controls. None of these papers found that SES influenced changes to alcohol consumption at follow-up. In contrast, Ockene et al. (1999) found that BI did not alter binge drinking in those who drank in that pattern alone; in those who mixed binge drinking with an overall heavy consumption, BI did reduce binging as well as overall consumption. Their analysis of whether level of outcome was influenced by educational level was non-significant; other socio-economic variable were not analysed.
In addition, there were two papers that did not report on SES influence on outcome but did report on binge drinking outcomes. Curry et al. (2003), while finding significant reductions in overall alcohol consumption amongst the intervention group compared to control, found that binge drinking reduced in both intervention and control group with no significant difference between them. Saitz et al. (2003) also reported significant reductions in alcohol consumption in the intervention group, but no reduction in binge drinking.
DISCUSSION
SBI programmes offer the opportunity to provide some level of treatment to large numbers of people who might otherwise go untreated. This has been shown to produce significant reductions in harmful drinking amongst those people. However, it can be difficult to recruit hazardous and harmful drinkers to participate in BI trials, and once recruited not all participants randomised to the intervention group will attend to receive this. Amongst those who are recruited and do attend, it appears that SES does not necessarily influence treatment outcome, contrary to the hypothesis being tested (Scott and Anderson, 1990; Anderson and Scott, 1992; Fleming et al., 1997). However, questions remain over the representativeness of those who do participate in such trials (Edwards and Rollnick, 1997). Many patients decline to be screened initially. Of those who do screen positive significant numbers can refuse to participate further. Few studies have examined the potential differences between those who do and those who do not participate further. The evidence suggests that educational and employment status does not influence participation rates (Senft et al., 1997; Aalto and Sillanaukee, 2000); however, the crude way these variables were defined could easily hide SES differences. Stronger evidence exists that once recruited and randomized to intervention, that variables such as occupational stratification do not influence subsequent attendance to receive treatment (Fleming et al., 1997). This is not unequivocal however as others have found the opposite: that higher SES does predict higher attendance (Richmond et al., 1995). In relation to the follow-up data that are available for analysis, some have found that higher educational status increases follow-up attendance (Senft et al., 1997) while others have not (Curry et al., 2003).
The possibility exists that results are influenced by unknown, differential characteristics of those who participate compared to those who do not. Without specific attention to these characteristics it cannot be confidently stated that BI research populations are representative of the general primary care population. While SES may not be an important variable in those who do choose to participate in BI research, this is not to say that SES does not influence that choice itself.
In addition, the effect of BI on binge drinking also warrants further investigation. Given the reported links between binge drinking and SES, the influence of the latter should be part of such investigations.
It is concluded that BIs should remain available to all hazardous and harmful drinkers in primary care regardless of SES. However, fidelity to the research studies should be observed. While negotiating the clinical agenda with the patient, alcohol should be identified as relevant by the clinician as indicated, and the patient's permission to discuss this further should be sought. For those who withhold permission, the fact that the clinician has raised the subject may give pause for reflection, and the possibility of future discussion may be enhanced by the display of respect for self-governance that a request for permission displays. For clinicians, this prevents the demoralizing experience of delivering an unwanted intervention to a resistant patient, and targets BI at a population where efficacy has been demonstrated.
Future research is needed to clarify what, if any, role SES might play in moderating the impact of SBI programmes. Public health risk factors tend to cluster together in those living in the context of socio-economic deprivation. If ‘standard’ SBI is more difficult to deliver to populations with high deprivation, it may be because attempting to target a single risk factor alone is experienced as less empowering than holistic measures. It may be of value to evaluate the impact of incorporating alcohol interventions for deprived populations in a ‘community health’ package that targets a variety of clustered risks (e.g. economic, environmental, forensic, interpersonal, medical, nutritional, and psychological) simultaneously.
CONCLUSION
SES does not necessarily influence outcome of BI for hazardous and harmful drinking. However, participation effects cannot be ruled out. Fidelity with BI research methods is suggested, until the characteristics of those who decline to participate in BI research are better understood.
Acknowledgements — Sincere gratitude to Dr Peter Rice, Consultant Psychiatrist, Tayside Alcohol Problems Service, Sunnyside Royal Hospital, Montrose, and to two anonymous reviewers, for their discussion and comments on an initial version of this paper. All opinions voiced remain, however, the responsibility of the author alone.
Conflict of interest statement — I know of no conflicts of interest arising out of this paper. I am employed full time by NHS Tayside as Primary Care Facilitator with the Alcohol Liaison Service of Tayside Alcohol Problems Service, where a key task of my job is to promote evidence-based interventions for alcohol misuse in primary care. I am also a member of the management committee of the Nursing Council on Alcohol. I do not expect any financial gain or loss for myself or these bodies as a result of this paper.
REFERENCES
Aalto, M. and Sillanaukee, P. (
Aalto, M., Saksanen, R., Laine, P. et al. (
Aalto, M., Seppä, K., Mattila, P. et al. (
Adams, A., Ockene, J., Wheeler, E. V. et al. (
Anderson, P. and Scott, E. (
Ballesteros, J., Duffy, J. C., Querejeta, I. et al. (
Beich, A., Thorsen, T. and Rollnick, S. (
Bellis, M. A., Hughes, K., Tocque, K. et al. (
Bertholet, N., Daeppen, J.-B., Wietlisbach, V. et al. (
Bien, T. H., Miller, W. R. and Tonigan, J. S. (
Burge, S. K., Amodei, N., Elkin, B. et al. (
Burton, L. C., Paglia, M. J., German, P. S. et al. (
Chang, G., Behr, H., Goetz, M. A. et al. (
Chase, V., Neild, R., Sadler, C. W. et al. (
Copeland, L. A., Blow, F. C. and Barry, K. L. (
Córdoba, R., Delgado, M. T., Pico, V. et al. (
Curry, S. J., Ludman, E. J., Grothaus, L. C. et al. (
Edwards, A. G. K. and Rollnick, S. (
Erens, B. (
Fleming, M., Barry, K. L., Manwell, L. B. et al. (
Fleming, M., Manwell, L. B., Barry, K. L. et al. (
Fleming, M., Brown, R. and Brown, D. (
Fleming, M. F., Mundt, M. P., Manwell, L. B. et al. (
Fleming, M. F., Mundt, M. P., French, M. T. et al. (
Freeborn, D. K., Polen, R. P., Hollis, J. F. et al. (
Gordon, A. J., Conigliaro, J., Maisto, S. A. et al. (
Grossberg, P. M., Brown, D. D. and Fleming, M. F. (
Heather, N., Campion, P. D., Neville, R. G. et al. (
Israel, Y., Hollander, O., Sanchez-Craig, M. et al. (
Kristenson, H., Öhlin, H., Hultén-Nosslin, M.-B. et al. (
Kristenson, H., Österling, A., Nilsson, J.-Å. et al. (
Maisto, S. A., Conigliaro, J., McNeil, M. et al. (
Mäkelä, P. (
Manwell, L. B., Fleming, M. F., Mundt, M. P. et al. (
McIntosh, M. C., Leigh, G., Baldwin, N. J. et al. (
Miller, W. R. and Wilbourne, P. L. (
Moyer, A., Finney, J. W., Swearingen, C. E. et al. (
Mundt, M., French, M. T., Roebuck, M. C. et al. (
National Statistics (
Nilssen, O. (
Nilssen, O. (
Ockene, J., Adams, A., Hurley, T. G. et al. (
Poikolainen, K. (
Rehm, J., Room, R., Graham, K. et al. (
Reiff-Hekking, S., Ockene, J., Hurley, T. G. et al. (
Richmond, R., Heather, N., Wodak, A. et al. (
Rollnick, S., Hodgson, R. J. and Snail, S. (
Romelsjö, A., Andersson, L., Barrner, H. et al. (
Saitz, R., Horton, N. J., Sullivan, L. M. et al. (
Saunders, J. B., Reznik, R. B., Hanratty, S. J., Douglas, A. and Burns, F. H. (
Scott, E. and Anderson, P. (
Scott, H. K. (
Senft, R. A., Polen, R. P., Freeborn, D. K. et al. (
Seppä, K. (
Skutle, A. (
Tomson, Y., Romelsjö, A. and Åberg, H. (
Wallace, P., Cutler, S. and Haines, A. (
Wilk, A. I., Jensen, N. M. and Havighurst, T. C. (