< Previous | Contents | Next >

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

CHAPTER SEVEN SCREENING AND DETECTION

Introduction

7.1 This chapter concerns the screening instruments that are available for the detection of problem drinking, alcohol abuse and dependence and the laboratory tests that are available either to confirm such results or to monitor abstinence. The value of screening instruments lies in the detection of people for whom a brief intervention or more intensive treatment may be of benefit. Therefore, the important issue is the accuracy of the information that is provided. The studies reviewed are summarised briefly in tables 7.1 (effectiveness) and 7.2 (cost-effectiveness) at the end of the chapter.

Screening questionnaires

Types of intervention

7.2 A variety of questionnaires have been developed and the main instruments evaluated in the literature are summarised in chart 7.1. Most have been developed for screening purposes although two are used to assess the extent of alcohol problems (S-MAST and SADD). There is some overlap in the type of questions asked, such as those relating to the need to cut down on drinking or needing a drink to start the day (eye-opener). Other variations on these instruments may also be in use.

Chart 7.1 Summary of main instruments in use

Effectiveness

7.3 The accuracy of screening instruments is assessed in terms of both sensitivity and specificity. Sensitivity refers to the ability of the instrument to identify true positives (for example, 80% sensitivity implies that 8 out of 10 people will be identified). Specificity refers to the accuracy in excluding true negatives (for example, 80% specificity implies that for every 10 people without the condition of interest, 8 will be excluded). A good test is both sensitive and specific but there is usually a trade off between these performance measures.

7.4 An overall assessment of performance is given by ROC curves, which plot sensitivity against 1 minus specificity. The area under the ROC curve (AUROC) provides a measure of test performance with 1 equalling a perfect test and 0.5 indicating that the test provides no useful information. Only one review reports these figures and they were only available for two studies (Bradley et al 1998). The AUROC score for CAGE was 0.84, for AUDIT the range was 0.86 — 0.94 and for TWEAK the scores were 0.89 - 0.90.

7.5 The performance of the screening questionnaires is very variable and some are better at detecting certain levels of alcohol misuse than others. For general screening purposes, AUDIT is more effective in detecting at risk, hazardous or harmful drinking (sensitivity 51%-97%; specificity 78%-96%), whilst CAGE is superior for detecting alcohol abuse and dependency (sensitivity 43%-94%; specificity 70%-97%). These two tests perform better than other screening tests (Fiellin et al 2000). Screening performance also depends upon the choice of cut-off point. A lower cut-off point will tend to increase sensitivity and reduce specificity.

Population groups

7.6 One review has considered the use of questionnaires with women (Bradley et al 1998). AUDIT, CAGE and TWEAK were the optimal test for detecting alcohol dependence in women but sensitivity in female populations may be lower using traditional cut-off points. CAGE was found to be more sensitive in samples of black women than white women. TWEAK appears to be optimal for detecting heavy drinking or alcohol abuse and dependence in racially mixed female populations. T-ACE and TWEAK are more sensitive and specific than CAGE for screening pregnant women. The full Michigan Alcoholism Screening Test (MAST) has been found useful for psychiatric settings (Teitelbaum and Mullen 2000.)

Setting

7.7 Most screening takes place in outpatient and primary care settings. Comparisons of screening test performance across settings have not taken place.

Providers

7.8 Comparisons of different persons administering questionnaires were not reported in the reviews. Some cost-effectiveness information is provided below (para 7.16).

Quality and relevance of evidence

7.9 The assessment of the screening instruments is limited by the lack of an agreed diagnostic standard against which to confirm the screening results. A range of criterion standards has been used in the studies reviewed and differences in the performance of screening instruments may be due to the choice of criterion standard (Fiellin et al 2000).

Laboratory tests

Types of intervention

7.10 There is a range of routine and more specialised blood tests available to confirm evidence of problem drinking or alcohol abuse. The specific tests available for sustained alcohol abuse are;

Effectiveness

7.11 Laboratory tests are not useful in screening for alcohol related problems (Fiellin et all 2000; NIAAA 2000). They have a role in monitoring the treatment of alcohol disorders. One review of studies comparing CDT assay with GGT concluded that in studies up to June 1998 commercially available CDT assay tests were not significantly better than GGT as markers of excessive alcohol use (Scouller et al 2000).

Cost-effectiveness

7.12 The economic literature with respect to screening for alcohol problems is very limited. The sequence of events through screening, referral for treatment and eventually changes in patient outcomes means economic analysis in this field is seldom undertaken. The key economic study regarding screening centres on the marketing of the Drink-less package in New Zealand. There is also limited UK evidence.

Drink-less

7.13 McCormick et al (1999) investigated cost-effective methods of encouraging general practitioners to take up screening and early intervention for problem use of alcohol in a marketing trial. The authors analysed the costs and consequences of marketing techniques designed to encourage GPs to receive an early intervention and screening package for the problem use of alcohol. An RCT of New Zealand GPs, comparing mail, telemarketing and personal marketing was employed for the 'Drinkless' early intervention and screening package developed with the WHO collaborative study for early interventions for 'at risk' alcohol consumption. GPs in New Zealand were identified from a database. From a total of 369 GPs, 186 were assigned to mail marketing, 87 to telemarketing and 96 to personal marketing. Of the 369, 40 were either ineligible or un-contactable.

7.14 The costs of the marketing exercise included promotional material, postage, telephone charges, travel costs (time and transport), receptionist time and waiting time. Direct mail costs per doctor were estimated at $5.11, telemarketing $2.92 and personal marketing $16.54. Outcomes were measured by the number of GPs willing to receive the package. Telemarketing appeared to be the most cost-effective means of persuading GPs to receive and employ the Drink-less package. Sixty-four of the 87 GPs contacted agreed to receive the package. The cost per doctor receiving was $3.97. Direct mail cost $10 per doctor agreeing and personal marketing cost $20.36.

Quality and relevance of evidence

7.15 McCormick's study shows a cost-effective means of disseminating information but information about how effective the Drink-less intervention is in practice is required. Little is known about whether the GPs actually used the packages. A follow up would be needed as receiving a package does not amount to using it, and thus health improvements could be negligible. More information on the treatment costs is provided in chapter 8.

Cost-effectiveness of alternative providers.

7.16 Tolley and Rowland (1991) investigated the cost-effectiveness of screening patients for alcohol problems at York District Hospital, UK. Over a 21-month period, doctors, nurses and a specialist worker screened orthopaedic and medical admissions to the Hospital. The costs of the intervention were calculated by using the mid-points of the salary ranges for the relevant grades and computing a cost for each screening based on the hourly rate of pay. The costs of screening were £0.10 for a nurse, £0.11 for a doctor and £0.15 for a specialist worker. As a percentage of admissions (positive screenings in parentheses), nurses screened 48% (7.8%), doctors 27% (9.8%) and the specialist worker 21% (12.5%). In terms of the lowest cost per positive screening, doctors cost £1.17, compared to £1.20 for the specialist worker and £1.29 for nurses.

7.17 The authors concluded that the specialist worker had a higher effectiveness rate defined by the percentage of positive screenings made, whilst nurses managed to screen a higher percentage of all admissions than doctors or the specialist. The authors conducted a marginal cost analysis based on the additional positive cases identified per 1000 screenings of different combinations of health care professionals delivering the interventions. By employing a specialist worker compared to a nurse, the specialist worker would identify an additional 88 cases per 1000 admissions at a marginal cost per positive case identified of £0.56. However, it should be noted that other considerations are important. Hospital managers may decide that the cost of employing a full time specialist is too great compared to the additional cases detected. Furthermore it may mean that a specialist cannot devote any time not screening to other duties. Nurses may be able to use any spare time more productively. The study provides some evidence as to the relative cost-effectiveness of screening for alcohol problems, but evidence is far from conclusive and likely to be affected by staffing levels and other resource issues within the health care provider.

Quality and relevance of evidence

7.18 The crucial shortcomings of both studies are the outcomes employed. The initial screening process identifies positive and negative screens, of which positives are referred for further treatment, yielding a further chain of events before successful health outcomes can be quantified. However, evaluations of screening programmes to date have only considered the first link, and the process through to patient outcomes is not covered by the current literature.

Further research

7.19 The available screening instruments appear to work well but evaluating their performance is hampered by the lack of an agreed diagnostic gold standard. Research to resolve this issue would be useful. Continuing research on the performance and role of laboratory tests will be needed as they are developed. There is a lack of research relating to comparisons of screening settings and providers.

Table 7.1 Summary of Screening and Detection Effectiveness Studies Reviewed

< Previous | Contents | Next >