EFFECTIVE AND COST-EFFECTIVE MEASURES TO REDUCE ALCOHOL MISUSE IN SCOTLAND: A LITERATURE REVIEW: page 12


	Home \| Topics \| About \| News \| Publications \| Consultations \| Search \| Links \| Contacts \| Help

Consultations > Health & Nutrition

< Previous | Contents | Next >

EFFECTIVE AND COST-EFFECTIVE MEASURES TO REDUCE ALCOHOL MISUSE IN SCOTLAND: A LITERATURE REVIEW

CHAPTER EIGHT BRIEF INTERVENTIONS

Summary

This chapter reviews the effectiveness and cost-effectiveness evidence relating to brief interventions used to treat problem drinking and alcohol abuse. The main findings are;

brief interventions are effective in reducing alcohol consumption for at least 12 months in patients who are not alcohol dependent;
pooled results from clinical trials show a 24% reduction in alcohol consumption;
those who received the intervention were twice as likely to change their behaviour compared with control groups;
brief interventions have fairly low costs and have been shown to be cost-effective in 3 economic studies;
using Scottish cost data the cost per life year lies in the range £1446 - £2628 assuming no cost savings; and
if resource savings are taken into account brief interventions may provide net benefits.
Savings from reduced future use of health care services need to be interpreted with care. It is more likely that resources will be released for alternative uses than that financial savings will be achieved.

Introduction

8.1 A brief intervention is a time-limited intervention focusing on changing patient behaviour with respect to alcohol consumption through motivational counselling. There is a more extensive literature in this area than for many other interventions. The studies reviewed are summarised briefly in tables 8.7 (effectiveness) and 8.8 (cost-effectiveness) at the end of the chapter.

Effectiveness of brief interventions

Types of brief interventions.

8.2 A brief intervention has been defined as having 5 essential steps (Fleming and Manwell 1999);

assessment of drinking behaviour and feed back;
negotiation and agreement of goal for reducing alcohol use;
familiarisation of patient with behaviour modification techniques;
reinforcement with self help materials;
follow up telephone support or further visits.

8.3 The counselling strategy used in brief interventions has also been summarised as FRAMES (Miller and Rollnick, 1991 cited in O’Connor and Schottenfeld, 1998);

Feedback	review problems experienced because of alcohol
Responsibility	patient is responsible for change
Advice	advise reduction or abstinence
Menu	provide options for changing behaviour
Empathy	use empathic approach
Self-efficacy	encourage optimism about changing behaviour

8.4 The precise content of the brief interventions evaluated in trials is variable, however. In this section, reviews have been included if they refer to brief interventions, minimal interventions or extended brief interventions. Brief interventions are mainly used to reduce alcohol consumption in people drinking above recommended levels but who are not dependent. Brief interventions may also have a role in improving compliance with other treatment regimens for alcohol dependent patients (Fleming and Manwell, 1999).

Effectiveness of brief interventions.

8.5 Most studies have found that brief interventions are effective in reducing alcohol consumption for at least 12 months in patients who are not alcohol dependent, and when compared with no intervention or usual care. A number of outcome measures are used including levels of alcohol consumption, change from heavy to moderate drinking and biochemical markers of alcohol consumption.

8.6 Formal meta-analysis of RCTs is difficult because of the variability in the content of the interventions, the population groups studied and the outcome measures used but this has been attempted in 3 reviews (Freemantle et al 1993; Poikolaninen, 1999; Wilk, Jensen and Havighurst, 1997). The former pooled results from 6 trials and estimated the effect of brief intervention as a 24% reduction in alcohol consumption (95% CI; 18%-31%). The second study analysed changes in alcohol consumption and concluded that there was no significant effect for brief interventions in either men or women. Extended brief interventions were effective for women; in men the effect size was similar but not significant. The last study analysed the proportion of patients reducing or moderating their drinking and concluded that those receiving motivational interventions were twice as likely as controls to change their behaviour (Odds Ratio 1.95; 95% CI; 1.66-2.30).

8.7 Five of the 7 studies included by Poikolaninen and all of the studies included by Freemantle et al were also included in the 8 studies pooled by Wilk et al. These meta-analyses were conducted to very high standards but have used different measures of effectiveness to arrive at different conclusions.

Table 8.1 Pooled results from 8 RCTs of brief interventions.

	Percentage moderating drinking		Odds Ratio	Number of studies
	Treatment group	Control group	(95% confidence interval	Number of studies
All trials	44	28	1.95 (1.66 — 2.30)	8
Quality trials	44	28	1.91 (1.61 — 2.27)	6

Females	50	27	2.42 (1.70 — 3.45)	3
Males	46	29	1.90 (1.57 — 2.31	5

1 session	46	30	1.83 (1.46 — 2.28)	5
>1 session	41	26	2.12 (1.66 — 2.70)	3

Outpatient	44	28	1.91 (1.61 — 2.27)	6
Inpatient	48	27	2.41 (1.40 — 4.15)	2

Source Wilk et al 1997

8.8 Some studies that have found minimal differences between intervention and control groups achieved reductions in alcohol consumption in both groups (Fleming and Manwell, 1999). This may be due to the research process drawing attention to drinking behaviour in the control group. One study that found increases in alcohol intake, though not statistically significant, is cited (Richmond et al, 1995 cited in Poikolainen, 1999). The same study showed a non-significant difference in the percentage moderating their alcohol use (Wilk, Jensen and Havighurst, 1997).

Population groups.

8.9 Brief interventions have been evaluated across a wide range of population groups, in terms of demographic characteristics; men, women, and the elderly. However, the most difficult cases, in terms of their drinking behaviour are often excluded or may self-select out of studies at the recruitment or assessment stage. Brief interventions have not been shown to be effective in pregnant women but this may be because of the high abstinence rates amongst the women at the time of the intervention (Chang et al, 1999 cited in NIAAA, 2000).

Settings.

8.10 The majority of studies have been conducted in primary care settings. Brief interventions have also been shown to be effective in inpatient settings, although the two studies cited are rated as being of relatively low quality and conducted more than 10 years ago (Annti-Poika et al, 1988; Chick et al, 1985 cited in Wilk et al 1997). One small study has also shown an effect in a hypertension clinic (Maheswaran et al, 1992 cited in Wilk et al 1997). Interventions in emergency room (A&E) settings have been effective in reducing alcohol consumption or securing referral into treatment (NIAAA, 2000).

Providers.

8.11 Doctors have delivered the brief interventions in the majority of studies. Nurses and health educators have also been effective in delivering brief interventions.

Intensity of intervention.

8.12 No significant difference in effectiveness was found between one session or more than one session (Wilk et al, 1997). Although the meta-analysis by Poikolainen, 1999, appears to support extended brief interventions, the effect size for brief interventions was similar but did not reach statistical significance.

Quality and relevance of the evidence.

8.13 Although the quality of the reviews is variable, overall they provide substantial support for the effectiveness of brief interventions in reducing alcohol consumption. The reviews by Fleming and Manwell, Freemantle et al, Poikolainen and Wilk et al are the most comprehensive. The only study cited suggesting that brief interventions increase alcohol consumption was not statistically significant. Potential for publication bias, that is the non-reporting of negative results, may still exist.

8.14 What is less clear is the generalisability of the results. The practitioners delivering interventions in research settings will be highly motivated and the effects may not be completely replicated in routine practice. Some groups are clearly excluded from the scope of the intervention but the process of consent for research studies means that the participants in the trial are self-selecting.

8.15 At least six of the reported trials were conducted in the UK and the effectiveness results should be relevant to Scotland.

The Cost-Effectiveness of Brief Interventions

8.16 Brief interventions for alcohol treatment have been shown to be cost-effective in published research. One of the major contributing factors to this high level of cost-effectiveness is that these interventions are relatively cheap in terms of inputs. The inputs to cost-effectiveness investigations of brief interventions include three major stages. Patients must be screened for alcohol problems, assessed for suitability for treatment and then costs of the actual treatment quantified and valued. The major input for the former two categories is the time of GPs plus any associated overheads and packages used. The costs of treatment will be highly dependent upon the intervention in question.

Resource inputs

8.17 Few studies exist which document the cost-effectiveness of alcohol treatments in sufficient detail as to be termed full economic evaluations. Fleming et al (2000) undertook a cost-benefit analysis of physician advice regarding problem drinking, in the USA. The authors evaluated a programme of brief advice in general practice for drinking over threshold limits versus no treatment. The no treatment group received a booklet on general health issues. The treatment group received the booklet and were scheduled to see their GP for the brief intervention, including a workbook of current health behaviour, a review of the prevalence of problem drinking, a list of the adverse effects of alcohol, a worksheet on drinking cues, a drinking agreement in the form of a prescription and drinking diary cards. The intervention was based on MRC protocols and consisted of two 15-minute appointments a month apart.

8.18 Patients received a follow up call from the clinic nurse two weeks after each GP meeting. The clinic resources and patient time and travel required to deliver the intervention are summarised below.

Table 8.2 Resource inputs to a brief intervention - example

Clinic Resources	Cost (1993 $)
Screening (8,962 patients)	30,736
Assessment (1,481 patients)	3,844
Primary intervention visit (392 patients)	10,266
Follow up intervention visit (392 patients)	10,266
Follow up phone calls	982
Training	8,839
Total clinic costs	64,933
Travel costs	3,646
Lost work time	11,631
Total economic cost of study	80,210

8.19 Practitioner time was estimated by using salary cost for the duration of the event and adding 25% for overheads. Patient time for the intervention included waiting and travel time. The total treatment cost of the clinic was estimated at $64 933 (£44 230). The cost per patient was $205 (£140) and total economic cost $80 210 (£54 640).

8.20 Wutzke et al (2001) investigated the question of whether brief interventions are cost-effective in reducing alcohol consumption. The direct costs of providing an alcohol treatment programme in primary care were examined, including costs of treatment included marketing the WHO's Drink-less package to GPs, training and support costs, and the cost of counselling at risk drinkers.

8.21 Three support strategies were used, a control group (no initial training or on-going support), a no support group (five minutes of initial training with no further contact or support), and a maximal support group (five minutes training plus alternate telephone and personal visits every two weeks). Marketing costs were taken from Gomel et al (1998) who evaluated three strategies for recruiting GPs to the WHO Drink-less package, and found the telemarketing option to be the most cost-effective. Cost of marketing was found to be AUS$2.16 per GP or AUS$5.35 per successful approach (not all agreed to use package).

8.22 Screening and counselling 'at risk' drinkers followed strict protocols and took 5 minutes of GP time, and the costs were taken from the Medicare Fee Schedule. Table 8.3 shows the estimated average costs of delivering the intervention (1996 Australian dollars). Total costs for implementing the intervention nationally would be AUS$4.3 million for the control strategy, AUS$7.5 million for the no support and AUS$12.8 million for the maximal support option. The cost per patient counselled in UK currency were £7.56 for the control group, £6.80 for the no support group and £7.60 for the maximal support group.

Table 8.3 Intervention costs for Drink-less

	Control	No Support	Maximal support
Recruitment	5.35	5.35	5.35
Training	35.56	44.66	138.68
Counselling	171.70	351.90	544.00
Total cost per GP	212.61	401.91	688.03
Number of people counselled per GP	10	21	32
Cost per patient counselled	21.26	19.14	21.50

8.23 In a study of alcohol treatment in Sweden, Lindholm (1998) used results from controlled trials showing short-term effectiveness combined with epidemiological studies of alcohol and mortality links. In the model, if people reduce their drinking, life expectancy is expected to increase. The differences in mortality between moderate and heavy drinkers are calculated in terms of life years gained by switching drinkers from the 'high' to 'moderate' drinking level. Different relative risks and proportions changing from 'high' to 'moderate' are presented to assess the impact on cost-effectiveness, based on cohorts of 'high' and 'moderate' drinkers. The intervention is a hypothetical programme of primary care advice to reduce individual's drinking from a 'high' to a 'moderate' level. GP and nurse advice are modelled.

8.24 The costs per patient of the intervention are CAGE screening of 120ECU, GP visit of 130 ECU, visit to district nurse of 40 ECU and GT tests costing 1 ECU. Yearly health care costs per individual for moderate and heavy drinkers are presented, ranging from 700ECU (£432) (moderate 40-44 year olds) to 2 800ECU (£1 730) (Heavy, 65-69 year olds). Costs are discounted at 5%.

Effectiveness of brief interventions: Alcohol consumption and health outcomes

8.25 The population in Fleming's study were patients aged 18-65 years attending routine general practice appointments and reporting drinking above threshold limits (defined as men consuming >14 drinks and women consuming >11 drinks per week), and comprised 482 men and 292 women. Health outcomes were based on self-reported alcohol consumption, at six and twelve months, completing a timeline follow back. The average number of drinks in the past seven days declined by 39.5% at six months and 40.0% at 12 months. Binge drinking reduced by 49.1% (6 months) and 45.7% (12 months), which was significantly greater than the reduction in the control group.

8.26 Wutzke's estimates of post treatment consumption came from the WHO Phase 2 trial and outcomes were also presented as life years saved. Estimates from the WHO trial found that baseline alcohol consumption fell by 28% in the intervention group after treatment. Potential deaths from alcohol related illnesses were then applied to the data to estimate the number of life years saved following implementation of the programme. The estimated life years saved were 674, 1285 and 1972 in the control, no support and maximal support strategies.

Wider consequences

8.27 Wutzke's work was based exclusively on costs and benefits to the health care provider and the individual patient. Fleming provided detailed treatment costs, and unit costs were provided for clinic costs for screening, assessment, primary visit, a follow up visit for patients in the intervention group and clinic and training costs, using 1993 dollars. Fleming also estimated the wider economic costs from a societal perspective, including patient and health care costs and consequences, and cost savings to the legal system following treatment. Patient costs included travel and lost work time. However, health benefits to patients were measured in terms of drinking outcomes but not included in the aggregated results. The exposition therefore resembles the cost-offset literature rather than a cost-effectiveness study. The estimated magnitudes of these wider economic costs are provided below.

Cost-effectiveness

8.28 In the Drink-less study, the average cost per life year saved was estimated to be AUS$645, AUS$581 and AUS$653 for the control, no support and maximal support strategies respectively, compared to 'do nothing' baseline. Compared to the control, each year of life costs $1,223 (£435) comparing minimal support with control and AUS$1,873 (£666) comparing maximal intervention with control. The results are based on the assumption that effects last 10 years. A discount rate of 3% was applied to all costs and benefits occurring in the future. The Drink-less package appears to offer a cost-effective means of reducing alcohol consumption and saving life years. Costs are very low compared to many programmes currently in operation in Australia.

8.29 The wider economic costs and benefits in Fleming et al's analysis were not combined with health outcomes, hence are not strictly cost-effectiveness results. Savings in emergency room visits and hospital use were $195,448 (£133,134) compared to the control. However, whilst the cost of hospitalisations were significantly different from zero, the difference in the cost of emergency departments was not statistically significant. In terms of crime costs, a difference in the costs of legal events (substance abuse, theft, forgery, assault) was $18,963 but not statistically significant. A difference of $209,108 for motor vehicle events was also not significant. Total savings of $228,071 also failed to satisfy significance testing (p=.14). The method of calculation was unclear.

8.30 Summing the total economic costs and benefits, the average benefit per subject was $1151 and the benefit-cost ratio 5.6:1 (or $56,263 in total benefit for every $10,000 invested in such programmes). However, it should be noted that such programmes may be subject to diminishing returns in a particular area and such benefit cannot be bought by continually investing sums of $10,000 into treatment as the greatest gains are likely to be exhausted first. However, as with many US studies, this claims to be a cost-benefit study but is actually another cost-offset study, due to the omission of any valuation of patient health benefits which are implicitly valued as zero. Some of the crime categories have few cases, making costs hard to extrapolate. Cost data are, on the whole, well presented although methodologies are excluded in the case of criminal justice costs. Fleming found the total economic benefit of the brief intervention to be $423,519 (£288,490) (95% CI $35,947-$884,848), the results being significantly positive although the 95% confidence interval is particularly large.

8.31 In Lindholm's study, cost-effectiveness was estimated based on assumptions regarding the proportion of drinkers changing from 'heavy' to 'moderate'. Based on 20% changing and a relative risk of mortality of heavy to moderate drinkers of 2, the 25 visit intervention cost 200ECU (£123.50) per life year saved (LYS). With a relative risk of mortality of 1.25, with 2% changing from heavy to moderate drinking status, the cost was 144,000ECU (£88,900) / LYS. If nurses give advice, as opposed to GPs, savings are greater than costs if effectiveness exceeds 10%. With the lowest effectiveness and relative risk figures, cost was 108,000ECU (£66,700) / LYS for the 25-visit and 20,000ECU (£12,350) / LYS for the 5-visit intervention programme.

8.32 Lindholm concludes that brief interventions delivered by GPs are cost-effective if the lasting treatment effect is about 1%. A 25-visit intervention is considered to be cost-effective if effectiveness is 2% and relative risk of mortality is greater than 1.5. Lower wages amongst nurses causes the cost-effectiveness of a nurse led intervention to be greater. However, the effectiveness of a nurse-led intervention is unknown. Although the message is clear from Lindholm's results, the author points to potential bias in trials as participants are often different from non-participants. Furthermore, the results are sensitive to changing the parameters involved, as changes in the assumptions behind the model result in large changes in the cost-effectiveness ratios.

Quality and relevance of evidence

8.33 The studies outlined above illustrate brief interventions to be relatively cost-effective, due to fairly high levels of effectiveness and low costs. Fleming's study is effectively a cost-offset study, as individual health benefits are subsumed. The results are unlikely to have much relevance to Scotland. UK data do not tend to show the large benefit:cost ratios found in the USA. It also must be remembered that the utilisation of health care in the USA shows very different patterns to the UK, and the health care system is based on an insurance system that operates differently to the UK. Cost differences for health care interventions are also likely between the two countries.

8.34 A crucial difference between the studies is the way in which health outcomes are dealt with. Outcomes are expressed in alcohol consumption in Fleming's work, whilst Wutzke and Lindholm use life years saved. The advantage of the latter is that comparison with other health care interventions is facilitated. Wutzke's study offers potential for modelling a similar strategy in Scotland. Local costs could be estimated to replicate the study and assess the potential cost-effectiveness for this package in Scotland.

8.35 Finally, it should be emphasised that generalising the results of economic evaluations should be undertaken with extreme caution. Evaluations carried out in insurance-based health care systems and overseas where the prevalence and characteristics of alcohol related problems are different to Scotland should not be considered as directly applicable to Scotland. Incentives to use treatment and other health care services, and the impact of treatment on other costs, such as crime, should not be taken and applied directly to the Scottish scenario.

Modelling of results for Scotland

8.36 The published literature on brief interventions can be used to estimate the cost-effectiveness of these interventions in the UK, by applying local costs to the main study findings. However, evidence is scarce as few papers include resource use and unit costs in their analysis. Fleming et al (2000) conducted a cost-benefit analysis of brief GP advice with regard to problem drinking. They included health care costs and wider social costs using data from Project TrEAT, a randomised controlled trial. Based on a control group (n=382) and a treatment group (n=392), alcohol use, accident and emergency (A and E), hospital days and legal events were recorded for the two populations.

Costs

8.37 In this simulation, local (UK) costs are applied to the resource use data to estimate the cost-effectiveness ratios that may be expected if the treatment programme was replicated in the UK. Obviously these are only estimates and the results, especially treatment effectiveness, will be determined by particular characteristics of the population in the parent study. Table 8.5 shows the costs of the brief intervention programme, based on UK costs. GP costs (including overheads) of £1.92 per minute are used, whilst practice nurse time is estimated at £0.30 per minute (Netten and Curtis, 2000). It should be noted that all GP time is costed using a 'higher bound' estimate including overhead costs for GP practice.

Table 8.5 Costs of the brief intervention programme, using UK costs 1999-2000

Unit costs:
GP consultation cost (inc overheads)	£1.92 / minute
Practice nurse (£18/hour)	£0.30 / minute
Training costs:
1 hour GP plus 30 min 'booster'	£172.80
1 hour practice nurse plus 1 hour consultation	£36.00
Screening costs:
5 mins with practice nurse	£1.50
Assessment:
5 mins with practice nurse	£1.50
Cost of brief intervention
15 mins with GP (£1.92 x 15 mins)	£20.80
Follow up =
15 mins with GP = (£1.92 x 15 mins)	£20.80
Two follow up phone calls (10 mins practice nurse)	£3.00

8.38 The intervention requires that patients are screened and assessed before being given a brief intervention. Fleming's study showed that 8,962 patients were screened and assessments were conducted for 1,481 patients to provide primary interventions for 392 patients. Therefore 22.8 screenings and 3.78 assessments are required for every patient receiving a primary care intervention. The total intervention costs for one individual receiving a brief intervention are derived as follows:

costs for a total of 22.8 patients screened	£34.20
costs for a total of 3.78 patients assessed	£5.67
intervention of 15 minutes	£20.80
follow up also of 15 minutes	£20.80
two telephone follow ups
5 minutes of practice nurse time each	£3.00
training costs per patient	£2.27
total cost	£86.74

Training costs will vary according to assumptions about workload. The figure used here assumes 92 interventions per year. No other implementation costs have been included.

Consequences

Resource savings

8.39 In terms of health care utilisation, the table below shows the number of events per patient in the previous 12 months at follow up. An average patient in the intervention group experienced 0.07 fewer A and E visits and 0.53 fewer hospitalised days compared to the control group. Based on a cost of £44 per A and E attendance and £199 per hospital day (Information and Statistics Division 2000), the cost savings per patient are £3.08 (A and E) plus £105.47 (hospital days).

Table 8.6 Health care utilisation by intervention and control patients

	Intervention	Control
A and E visits	0.27 per patient	0.34 per patient
Days hospitalised	0.32 per patient	0.85 per patient

(Source Fleming et al 2000)

8.40 Fleming's estimates of legal consequences include too few events to provide a meaningful comparison. However, if we look at motor vehicle events, there are 78 events in the treatment group and 95 in the control at follow up. This converts to 0.20 per patient (Treatment) and 0.25 (control). The difference of 0.05 can be multiplied by the cost per vehicle crime of £890 (Home Office, 2000) to give a saving of £44.50 per individual.

Health outcomes

8.41 Fleming et al found the average reduction in alcohol consumption was 45.7 % at the 12 month follow up. This is the same order of magnitude as the treatment groups in studies included in the effectiveness reviews.

8.42 Based on Wutzke et al (2001) examining brief interventions in New Zealand, using a basic 'no support' strategy, a total of 6,743 life years were estimated to be saved for a population of 204,587 counselled. This is 0.033 life years per patient.

Costs and consequences

8.43 If we assume the cost per patient of delivering a brief intervention is £86.74 and that life years saved are 0.033 then the cost of saving one life year is £2,628. (A lower estimated cost of £1,446 per life year saved can be generated by modelling the intervention on the basis used in Wutzke.) In terms of cost offsets, based on Fleming's (2000) estimates, and using Scottish costs, the estimated health care savings from A and E attendances and hospitalised days avoided is £108.55. Therefore a cost saving of £21.81 per patient can be estimated. In addition, if vehicle crimes are included the saving increases sharply to £66.31. However, some care should be exercised in interpreting these results. The saving represents the value of resources that are released but it is unlikely that financial savings of the same magnitude could be realised.

Limitations of the simulation.

8.44 The simulation provides some outline figures against which actual policies could be monitored. The figures for costs include an allowance for additional GP or nurse time for screening and delivering the intervention. In practice, it is unclear how this additional resource would be allocated but it is not assumed that these interventions are undertaken by substituting other primary care activity. The costs include some training element but do not include any direct implementation costs. A direct programme of implementation may be required although marketing activities have in isolation a limited impact on take-up of interventions. In this simulation, the take-up does not alter the individual cost-effectiveness figures as the costs (excluding training) vary directly with the number of brief interventions undertaken. The direct costs of the intervention are low and therefore the results are highly sensitive to significant variations in assumptions. Indeed if health care cost savings arise from such interventions, as indicated in the Fleming et al study, there is potential for the NHS to save resources over time by implementing such interventions. However, there is no UK research on the size of such potential savings in a Scottish setting.

Further research

8.45 Brief interventions have been relatively well researched but evidence is required of their effectiveness when used routinely outwith a study setting. Specific interventions may be required for the successful implementation of a brief intervention programme and the costs and effectiveness of implementation strategies should be evaluated. Any further studies on brief interventions, per se, should focus on their longer-term effectiveness (most studies have 6-12 months follow up) and the added effectiveness when combined with medication. There is also potential to use monitoring information from pilot schemes to improve the potential economic simulations by substituting some actual costing data from these pilot schemes. The improved models could then be used to assess future schemes for their performance both in terms of costs and outcomes with reference to this research.

Table 8.7 Summary of Brief Intervention Effectiveness Studies Reviewed

Author(s) and Date	Search Strategy	Inclusion / Exclusion Criteria	Quality Assessed	Number of Studies Reviewed	Target Group	Main Findings
Anderson P. Effectiveness of general practice interventions for patients with harmful alcohol consumption. British Journal of General Practice. 1993; 43: 386-9.	Not reported.	Not reported.	Limited assessment of quality.	6 studies included. Also included was a WHO multi-centre study.	Men and women who were heavy drinkers.	The results suggest that very brief advice leads to reductions in alcohol consumption of 25-35% and reductions in the proportions of excessive drinkers of around 45%. The studies provide some understanding of the effective components of brief interventions. First, the target of brief intervention should be a reduction in the consumption of alcohol. Secondly, age, socio-economic status and marital status do not appear to predict outcome. Thirdly, initial level of consumption predicted outcome, with heavier drinkers in the WHO study reducing their alcohol consumption by a significantly greater amount at follow-up than lighter drinkers, although this finding was not repeated in the Oxford study. Fourthly, in the WHO study, among those with a long-term alcohol problem, brief counselling worked best, while among those with a recent problem, simple advice worked best. This suggests that the effect of minimal intervention is enhanced when the patient has experienced a recent problem caused by alcohol. Finally, there is greater evidence of a treatment effect among men than women. Further work in this area is needed, and it may be that sex-specific intervention strategies should be evaluated.
Ashenden R et al. A systematic review of the effectiveness of promoting lifestyle change in general practice. Family Practice. 1997; 14(2): 160-76.	Yes English language only.	Included trials of lifestyle advice in GP setting with random allocation between experiment and comparison group	Yes	37 reviewed 6 trials included	Men and women drinking above safe levels	2 out of 5 studies found significant change in alcohol consumption. Effect size was greater for men than women. 3 out of 6 studies found significant change from heavy to moderate drinking. Proportion of women shifting was greater than or equal to proportion of men. 2 out of 5 studies found significantly lower GGT levels. Trials included 4 from UK, and one each from Sweden and US. Comparison groups could be no intervention, usual care or different intensity of advice.
Babor, TF et al. Alcohol-related problems in the primary health care setting: A review of early intervention strategies. British Journal of Addiction. 1986; 81: 23-46.	Not reported.	Not reported.	Assessed for some studies.		Poorly reported in majority of included studies.	The development of effective, inexpensive, early interventions is still in its early stages. Low intensity, brief interventions have much to recommend as the first approach to the problem drinker in the primary care setting. Given the tentative nature of many of the conclusions drawn from this review, systematic research on early intervention should be given high priority by both national and international health agencies.
Dinh-Zarr T, DiGuiseppi C, Heitman E, Roberts I. Preventing injuries through interventions for problem drinking: A systematic review of randomised controlled trials. Alcohol & Alcoholism 1999; 34(4): 609-21.	Yes	Studies included had subjects randomly assigned to experimental and control groups; and interventions designed to reduce or eliminate alcohol consumption, or to prevent injuries or their antecedents; and if outcome measures included injuries or their antecedents.	Yes	19 RCTs	Alcohol dependence, alcohol abuse, or other problem drinking .	This systematic review considers the effect of a range of interventions on injuries and deaths. The results suggest that treatment for problem drinking may reduce injuries and their antecedents, but there were a number of methodological weaknesses and effect sizes were often imprecise. Because injuries account for much of the morbidity and mortality from problem drinking, further studies are warranted to confirm these effects. The review did not provide strong support for reduced alcohol consumption as the mechanism for injury reduction.
Drummond DC. Alcohol interventions: Do the best things come in small packages? Addiction. 1997; 92(4): 375-9.	Not reported.	Not reported.	Review mainly concerns quality of the evidence base	Not reported	Not reported.	The results of RCTs of brief interventions are not generalisable to groups that are typically excluded from studies either explicitly or by self-selection, such as cases of severe alcohol dependence, concurrent mental illness or serious social problems. Few studies have compared specialist and generalist interventions. Many subjects do not improve and differences between control and intervention groups are not always maintained. Little is known of the effective ingredients and the most effective methods of delivery. Trials tend to be conducted in settings with motivated practitioners. Reviews of brief interventions have been overly selective, and meta-analysis in this area is problematic. Further research is needed into the question of which drinkers are most likely to benefit from which type of intervention.
Fleming MF, Manwell LB. Brief intervention in primary care settings: A primary treatment method for at-risk, problem, and dependent drinkers. Alcohol Research and Health. 1999; 23(2):128-37	Not reported	Not reported	Not reported	13 trials included	At risk, problem and dependent drinkers. Various population groups	Review included 4 studies each from UK and US, 2 from Sweden, one each from Canada and Norway and one cross-national study. Brief intervention is effective with at risk and problem drinkers. Effect is similar for men and women and across age groups (10 — 30% of patients changing drinking behaviour). Two studies measured hospitalisation and found significant reduction in length for intervention groups. Brief interventions may also improve compliance with treatment for dependent patients and may facilitate referral into specialist treatment for those who fail to respond to the brief intervention.
Freemantle et al Brief interventions and alcohol use 1993 Quality in Health Care 2: 267-273	Yes	Details in Effective Health Care Bulletin 1993.	Details in Effective Health Care Bulletin 1993	7 RCTs of brief interventions versus assessment only controls; 13 trials of brief interventions versus specialist treatments	Heavy drinkers, Problem drinkers	Meta-analysis of 6 of the 7 brief intervention studies produced a pooled result of 24% reduction in alcohol consumption (95% CI 18%-31%). The other study was conducted with hypertensive patients. The studies comparing brief interventions to specialist treatment were too varied to be pooled in a formal meta-analysis. In general, these studies showed no evidence of extra benefit from more specialist interventions. However, when brief interventions are unsuccessful, more specialist treatment may be effective.
Fuller RK, Hiller-Strmhofel S. Alcoholism treatment in the United States: an overview. Alcohol Research & Health. 1999; 23(2): 69-	Not reported.	Not reported.	Not reported	2 studies on brief interventions	Not reported.	Brief interventions were considered as part of a wider review of treatments. Both studies are covered by another review (Fleming and Manwell)
Kahan M et al. Effectiveness of physician-based interventions with problem drinkers: A review. Canadian Medical Association Journal. 1995; 152(6): 851-9.	Yes.	Included RCTs examining the effectiveness of interventions by physicians in reducing alcohol consumption among problem drinkers attending health care facilities. Trials involving subjects attending alcohol treatment clinics and trials delivered solely by non-physician were excluded.	Yes.	11 studies involving 4048 subjects.	Problem drinkers. 3 studies presented separate results for men and women, 1 study involved women only.	The trials support the use of short-term interventions by physicians for patients with drinking problems, although, the results for women were inconsistent and the trials did not provide convincing evidence of reductions in alcohol-related morbidity. Even though short-term interventions may yield only modest reductions in alcohol consumption, their public health impact is potentially enormous. Further research is needed to determine which patients are best suited for short-term interventions, the optimal intensity of treatment and which components of short-term interventions are most effective. Research is also needed to establish which strategies are effective in inducing physicians to use short-term interventions. Given the evidence for the effectiveness of short-term interventions and the minimal amount of effort and time required, physicians are advised to implement these strategies in their practice
Ketola E et al Effectiveness of individual lifestyle interventions in reducing cardiovascular disease and risk factors 2000 Ann Med 32 : 239-251	Yes	Included RCTs of lifestyle interventions with at least 60 subjects, followed up for at least 1 year.	Yes	42 studies of which 3 were alcohol interventions and 4 were multifactorial with alcohol consumption as an endpoint	Heavy drinkers, Problem drinkers.	Two of the alcohol specific interventions reported decreases in heavy drinking (brief interventions included in other reviews). These studies reported no CVD endpoints. Only one multifactorial study reported a statistically significant reduction in alcohol consumption.
Mattick RP, Jarvis T. A summary of recommendations for the management of alcohol problems: the quality assurance in the treatment of drug dependence project. Drug and Alcohol Review. 1994; 13: 145-55.	Not reported.	Not reported.	Not reported.	Not reported	Not reported	In situations where there is no time available to intervene with excessive drinkers, for example A&E, it is recommended that leaflets be made available that set out the currently accepted "responsible" limits for alcohol consumption, and list strategies for cutting back, as well as appropriate contact points for referral. Where a few minutes are available, brief one-to-one, face-to-face interventions are recommended. At a minimum there should be screening and identification of excessive alcohol consumption, clear and firm advice to cut down consumption, a description of the safe/responsible level of consumption and a follow-up visit. For drinkers with more severe problems, brief intervention over a few sessions (1-5), is recommended. Where assessment or previous failures in treatment suggest the need, longer out-patient intervention extended over multiple sessions will be required. In-patient/residential and day patient interventions for the most seriously affected and dependent individuals are recommended.
McCrady B S Alcohol use disorders and the Division 12 Task Force of the American Psychological Association, Psychology of Addictive Behaviours 2000 14(3) : 267-276	Not reported	Published studies reporting positive findings	Yes	62 studies of all treatment types of which 13 were brief interventions	Various	This paper reviews alcohol treatment studies against the guidelines of the Task Force. No treatment for alcohol abuse or dependence had been rated as efficacious by the Task Force. This study finds that brief interventions and relapse prevention do meet the criteria but that insufficient studies had been published when the Task Force list was published.
Modesto-Lowe V, Boornazian A. Screening and brief intervention in the management of early problem drinkers: Integration into health care settings. Dis Manage Health Outcomes. 2000; 8(3): 129-37.	Not reported.	Not reported.	Not reported.	4 meta-analysis are included. The number of trials included in 3 of the 4 meta-analyses are reported to be 32, 11, and 12 which involved 6000, 4048, and 3948 subjects respectively.	Men and Women who are early problem drinkers or heavy drinkers.	Brief intervention reduces drinking among early problem drinkers. Although most early problem drinkers do not go on to become alcohol dependent, they are a legitimate source of concern due to their significant numbers and the costs involved in treating their health and social problems. Alcohol dependent drinkers are likely to need specialised treatment (e.g. detoxification, rehabilitation and Alcoholic Anonymous meetings) and are unlikely to benefit from brief interventions. However, if the individual’s drinking is found to be consistent with hazardous or harmful drinking or with alcohol abuse, brief intervention may be an effective tool. The WHO cross cultural trial on brief interventions found that heavy drinkers not reporting any recent adverse consequences benefited from a more extended intervention whereas those who had experienced a recent negative consequence benefited from the intervention of shorter duration. Barriers to implementing brief interventions identified include: the physician’s failure to screen patients, lack of skills and time, organisational issues, lack of professional reward and lack of diagnostic aids for alcohol disorders. Moralistic attitudes, stereotypes about the nature of alcoholism and poor communication between physicians and non-physician providers have also impeded progress in patient identification and treatment. Despite these research findings, screening and brief interventions have yet to exert a significant influence in clinical practice.
Mullen PD et al. A meta-analysis of trials evaluating patient education and counselling for three groups of preventative health behaviours. Patient Education & Counselling. 1997; 32(3): 157-73.	Yes	Published and unpublished studies measuring the effect of education or counselling interventions.	Not reported	74 studies only 4 of which dealt with alcohol	Various	The 4 alcohol studies included were from 1987 and 1988 and have been included in other larger reviews reported in this section.
National Institute on Alcohol Abuse and Alcoholism, 10^th Special Report to the US Congress on Alcohol and Health June 2000.						Although not a systematic review, this is an authoritative overview of research, particularly in the US. The section on brief interventions cites results from Bien et al, 1993 and Wilk et al 1997. Other specific findings include the effectiveness of interventions in emergency room settings and in specific populations (young and pregnant women).
O’Connor PG, Schottenfeld RS. Patients with alcohol problems. New England Journal of Medicine. 1998; 338(9): 592-602.	Not reported	Not reported	Not reported	Cites 3 trials and 2 other reviews	Heavy or excessive drinkers	Brief interventions were considered as part of a wider review of treatments. All 3 trials are covered by another review (Fleming and Manwell) and the 2 reviews are included here (Bien, Miller and Tonigan, 1993; Wilk, Jensen and Havighurst, 1997.)
Poikolaninen K. Effectiveness of brief interventions to reduce alcohol intake in primary health care populations: A meta-analysis. Preventative Medicine.1999; 28(5): 503-9.	Not reported	Included studies on general population or GP population with random allocation, no intervention in control group, alcohol intake or GGT as outcome measure and follow up 6-12 months.	Not reported	7 trials included	Men and women. Problem drinkers and heavy drinkers	Study effects were pooled using fixed effects model. Change in alcohol consumption was not significant for men or women for very brief interventions. Extended brief interventions produced significant reduction for women. Average effects could not be reliably estimated for other pooled data.
Walitzer KS and Connors GJ Treating problem drinking 1999 Alcohol Research and Health 23(2): 138-43	Not reported	Not reported	Not reported	Not reported	Problem drinkers with low level of dependence	This paper reviews moderation treatment for problem drinkers. Drinking reduction appears to be an effective goal and may have positive benefits among women. Evidence is cited in support of bibliotherapy (self-help material), telephone therapy and motivational approaches.
Watson HE. Minimal interventions for problem drinkers: A review of the literature. Journal of Advanced Nursing.1999; 30(2):513-9.	Not reported	Not reported	Not reported	13 studies included	Problem drinkers and heavy drinkers. Various population characteristics	The review covers studies with the general population, media-recruited studies, studies in general practice and acute care settings. Most of the studies reviewed are included in other reviews and no formal meta analysis is carried out. The review does provide some additional information about the methods of the primary studies and the content of the interventions.
Wilk AI et al. Meta-analysis of randomised control trials addressing brief interventions in heavy alcohol drinkers. Journal of General Internal Medicine. 1997; 12(5): 274-83.	Yes English language only	Included randomised trials with no intervention in control groups, sample size at least 30 and motivational intervention with self-help orientation	Yes	12 trials met the inclusion criteria	Heavy or problem drinkers	Eight of the studies contained enough data to be included in the formal meta-analysis. There is considerable overlap with the studies included by Freemantle et al and Poikolaininen. A pooled odds ratio showed that heavy drinkers receiving a brief intervention were almost twice as likely to decrease and moderate their drinking compared with those who received no intervention (OR 1.95; 95% CI 1.66-2.3). Odds ratios were higher for more than one session compared with one session ( 2.12 and 1.83 respectively), for women compared with men (2.42 and 1.9) and for interventions in inpatient settings compared with outpatient settings (2.41 and 1.91) but none of these differences were statistically significant.

Table 8.5 Summary of Brief Intervention Cost-Effectiveness Studies Reviewed

Study	Type of Analysis	Population	Intervention	Outcomes	Results
Fleming M, Manwell LB. Brief intervention in primary care settings: A primary treatment method for at-risk, problem, and dependent drinkers. Alcohol Research and Health. 2000; 23(2):128-37.	Cost-Benefit Analysis	Patients aged 18-65 reporting drinking above threshold limits attending routine general practice appointments (men consuming >14 drinks and women consuming >11 drinks)	Brief advice in general practice for drinking over threshold limits	Health care cost savings, crime savings	Total economic benefit of the brief intervention was $423,519 (95% CI $35947-$884848). Savings in emergency room visits and hospital use = $195448, and savings in crime costs = $228071. Average benefit per subject = $1151. Benefit cost ratio = 5.6:1
Lindholm L. Alcohol advice in primary care: Is it a wise use of resources? Health Policy. 1998; 45:47-56	Economic evaluation	Hypothetical cohorts of 'high' and 'moderate' drinkers in Sweden	A hypothetical intervention of primary care advice to reduce individual's drinking from a 'high' to a 'moderate' level. GP and nurse advice are modelled.	Life Years saved. Health care cost savings - however, their role in the model is not fully documented.	Cost-effectiveness is estimated based on the assumptions regarding the proportion of drinkers changing from 'heavy' to 'moderate'. Based on 20% changing and a relative risk of mortality of heavy to moderate drinkers of 2, the 25 visit intervention costs 200ECU / LYS. With RR=1.25, 2% changing the cost is 144000ECU / LYS. If nurses give advice as opposed to GPs, savings are greater than costs if effectiveness exceeds 10%. With the lowest effectiveness and RR figures, cost is 108000ECU/LYS for 25-visit and 20,000ECU/LYS for the 5 intervention programme.
Wutzke SE et al. Cost effectiveness of brief interventions for reducing alcohol consumption. Social Science and Medicine. 2001; 52:863-70	Evaluation / Simulation	Based on a simulation of the total Australian population	Brief intervention including marketing the WHO Drink-less package, advertised to GPs through tele-marketing.	Life Years Saved. Estimates from the WHO trial found that baseline alcohol consumption fell by 28% in the intervention group after treatment. Potential deaths from alcohol related illnesses were then applied to the data to estimate the number of life years saved. Estimate of 17 per male and 11 per female	Estimated life years saved = 674, 1285 and 1972 in the control, no support and maximal support strategies. Average cost per life year saved = AUS$645, AUS$581 and AUS$653 for the three strategies, compared to 'do nothing'. Compared to the control, each year of life costs $1223 comparing minimal support with control and AUS$1873 comparing maximal intervention with control.

< Previous | Contents | Next >


	Home \| Topics \| About \| News \| Publications \| Consultations \| Search \| Links \| Contacts \| Help

Crown Copyright | Privacy policy | Content Disclaimer | General enquiries