A COMPARISON OF TWO RETROSPECTIVE MEASURES OF WEEKLY ALCOHOL CONSUMPTION: DIARY AND QUANTITY/FREQUENCY INDEX

Anthony P. Shakeshaft*, Jenny A. Bowman and Rob W. Sanson-Fisher1

Hunter Centre for Health Advancement and University of Newcastle, New South Wales and
1 Faculty of Medicine and Health Sciences, University of Newcastle, New South Wales, Australia

Received 22 December 1997; in revised form 22 December 1998; accepted 22 February 1999


    ABSTRACT
 TOP
 FOOTNOTES
 ABSTRACT
 INTRODUCTION
 SUBJECTS AND METHODS
 RESULTS AND DISCUSSION
 GENERAL DISCUSSION AND COMMENTS
 ACKNOWLEDGEMENTS
 REFERENCES
 
There is currently uncertainty regarding the relative performance of a retrospective diary (RD) and a quantity–frequency index (QFI) measure of weekly alcohol consumption. While some previous studies have found more consumption reported on an RD than a QFI, others have found the reverse. As yet, however, no study has compared computerized versions of these two measures. This cross-sectional study involved administration of a computerized survey in a community-based drug and alcohol treatment setting. Five hundred and eighty-six clients (420 males) attending counselling for a range of drug- and alcohol-related issues agreed to participate in the study. The major finding was that more alcohol consumption was reported on the RD, than on the QFI. Similarly, the RD detected a greater proportion of both heavy and high-risk drinkers than the QFI. It is argued that the RD may be preferable to the QFI as a measure of weekly alcohol consumption, for use in community-based treatment settings.


    INTRODUCTION
 TOP
 FOOTNOTES
 ABSTRACT
 INTRODUCTION
 SUBJECTS AND METHODS
 RESULTS AND DISCUSSION
 GENERAL DISCUSSION AND COMMENTS
 ACKNOWLEDGEMENTS
 REFERENCES
 
In an attempt to maximize the accuracy of self-reported alcohol consumption, measures incorporating a variety of time frames and various memory recall strategies have been developed (e.g. Sobell and Sobell, 1973Go; Midanik et al., 1989Go; Goransson and Hanson, 1994Go; Grant et al., 1995Go). Weekly alcohol consumption has been a commonly used method for assessing and reporting drinking behaviour. Its utility is indicated by the adoption of weekly consumption as the basis for government policy regarding recommended levels of alcohol consumption in many countries and its inclusion in many national surveys (e.g. National Health and Medical Research Council, 1992Go; Office for National Statistics, 1994Go; Single and Wortley, 1994Go).

Measures of individual weekly alcohol consumption include biochemical analysis of blood or breath samples and self-report. Biochemical measures are invasive, expensive and the empirical evidence for their reliability and validity is often unconvincing: they are either indicators of liver disease and not sensitive or specific to alcohol misuse, or they lack sensitivity in detecting hazardous consumption, possibly due to the short half-life of alcohol and the irregular consumption pattern of low-dependence drinkers (Allen et al., 1992Go; Whelan, 1992Go; Babor et al., 1994Go; Conigrave et al., 1995Go; Dawe and Mattick, 1997Go). Although self-reported alcohol consumption is similarly problematic (Pernanen, 1974Go; Popham and Schmidt, 1981Go; Watson et al., 1984Go; Brown et al., 1992Go), it is less invasive, relatively inexpensive, simpler to administer, acceptable to respondents and similar to collateral reports, in that there is no systematic trend for them to under- or over-estimate self-report (Miller et al., 1979Go; Midanik, 1982Go; Baranowksi, 1985; Strecher et al., 1989Go; Maisto and Connors, 1992Go).

A summary method, developed by Straus and Bacon (1953), asks about the average amount of alcohol consumed on a typical drinking day (quantity) and the average number of days on which alcohol is consumed (frequency). Several variations of this basic quantity–frequency index (QFI) have since been developed in an attempt to improve its accuracy (Cahalan, 1969Go; Bowman et al., 1975Go; Skinner and Sheu, 1982Go; Room, 1990Go; Midanik, 1994Go). With some exceptions (Wyllie et al., 1994Go; Grant et al., 1995Go), these modifications have generally resulted in an increase in the level of alcohol consumption reported on the QFI (Room, 1990Go; Sobell and Sobell, 1992Go). However, QFI measures have several disadvantages: respondents are required to generalize or abstract about their drinking, which tends to result in underestimation, an effect which is likely to be exacerbated by irregular consumption; the QFI provides little information about the overall pattern of drinking; accuracy in converting to mean consumption/week is compromised by the range of each category used, for example, seven to nine drinks, two to three times a week (Redman et al., 1987Go; Sobell and Sobell, 1992Go).

An alternative measure of weekly consumption, the retrospective diary (RD), usually requires the nomination of the amount of alcohol consumed on each day during a specified week (Millwood and McKay, 1978Go). Although the RD asks respondents about their actual behaviour, rather than requiring an abstraction, it is still susceptible to memory errors, which may result in an under- or over-estimation of the quantity and/or frequency of drinking (Hilton, 1989Go). The RD is, however, particularly suitable for use in clinical settings, relative to the prospective diary and a timeline follow-back procedure: it takes greater opportunistic advantage of an existing contact with a clinical service by not relying on respondents to keep a future appointment; it is likely to be substantially quicker to complete than the timeline follow-back method, which requires between 10 and 30 min (Sobell and Sobell, 1992Go; Dawe and Mattick, 1997Go); and response differences do not appear to be clinically significant (Hilton, 1989Go; Samo et al., 1989Go; Lemmens et al., 1992Go; Sobell and Sobell, 1995Go).

While there is some empirical evidence for the reliability and stability of the QFI and the RD (Webb et al., 1991Go), the lack of a gold standard for alcohol consumption has engendered uncertainty regarding their validity (Conigrave et al., 1995Go; Grant et al., 1995Go). Although the majority of studies appear to conclude that the RD detects a higher level of self-reported alcohol consumption than the QFI (Redman et al., 1987Go; Webb et al., 1990Go; O'Hare, 1991Go; Sobell and Sobell, 1995Go), the evidence is not unequivocal: some studies found a higher level of consumption reported on the QFI than on the RD (Midanik et al., 1989Go, O'Callaghan and Callan, 1992Go; Single and Wortley, 1994Go; Wyllie et al., 1994Go).

Probably the most common methods of collecting data on alcohol consumption have been pen and paper questionnaires and face-to-face or telephone interviews (Babor and Del Boca, 1992Go; Wilson and McDonald, 1994Go). However, the use of computers has several potential advantages: improved speed and efficiency in data collection and processing; completion by unsupervised respondents after brief instruction, minimizing observer bias; greater interactive presentation of items, allowing questionnaires to be instantaneously and unobtrusively tailored, or branched, for each client; reduced incidence of missing data; easier and more standardized interpretation of results; a high level of user acceptability; greater likelihood that at-risk behaviours will be divulged; comparable, if not improved, reliability and validity to pen and paper questionnaires or interviews; and instant feedback tailored to the specific answers of each respondent (Gavin et al., 1992Go; Sobell et al., 1996Go; Bonevski et al., 1997Go; Kobak et al., 1997Go; Newell et al., 1997Go; Shakeshaft et al., 1998aGo).

The aim of this study was to assess the relative performance of computer-administered RD and QFI measures of weekly alcohol consumption.


    SUBJECTS AND METHODS
 TOP
 FOOTNOTES
 ABSTRACT
 INTRODUCTION
 SUBJECTS AND METHODS
 RESULTS AND DISCUSSION
 GENERAL DISCUSSION AND COMMENTS
 ACKNOWLEDGEMENTS
 REFERENCES
 
Sample
Clients attending a community-based drug and alcohol counselling service under the auspices of a regional Area Health Service in New South Wales, Australia, were asked for their consent to complete a computer survey at the time of their initial consultation. This service provides free counselling for substance mis-users. Clients whom clinicians judged to be under the influence of drugs or alcohol, or who were extremely distraught at the time of the first consultation, did not use the computer. Clients who were illiterate or unable to read English, were given the opportunity to complete the questionnaire with the assistance of a counsellor.

Measures
The RD and QFI measures, embedded within a larger questionnaire, were presented as follows.

One-week RD. Respondents were asked ‘Have you had any drinks containing alcohol in the last 7 days?' Clients who had not, received a score of zero on the RD. Clients who had, completed the RD, beginning with the previous day and working back through each day of the week (Millwood and McKay, 1978Go). Clients were only required to nominate the types of beverages consumed on each day, the types of beverage containers and the number of each of these consumed. The computer converted responses to standard drinks and summed them, giving the number of standard drinks consumed during the previous week. In line with the National Health and Medical Research Council of Australia (1992) guidelines, a standard drink is defined as the equivalent of 10 g of pure ethanol. All clients, irrespective of whether they indicated consuming alcohol in the last 7 days, were then asked if this is the amount of alcohol they would usually drink during 1 week (Redman et al., 1987Go).

QFI. The first two questions from the Alcohol Use Disorders Identification Test (AUDIT) questionnaire represent the QFI questions (Saunders et al., 1993Go). Possible responses to the frequency questions, ‘How often do you have a drink containing alcohol?' were: ‘never', ‘monthly or less', ‘two to four times a month', ‘two to three times a week' and ‘four or more times a week'. Possible responses to the quantity question, ‘How many "standard" drinks do you have on a typical day when you are drinking?' were: ‘one or two', ‘three or four', ‘five or six', ‘seven to nine' and ‘10 or more'. A graphic depicting a range of standard drinks was displayed with the second question. The maximum and minimum number of drinks that could be consumed was calculated for each combined quantity (the highest response category for the quantity question was allocated a score of 10) and frequency category (Redman et al., 1987Go; Webb et al., 1990Go). The mean of these two values represents the average number of standard drinks consumed/week. For example, three or four, two or three times a week is scored as nine standard drinks/week.

Procedure
Data were collected using MS-DOS-based colour, touchscreen personal computers and Iconauthor (AimTech Corporation, 1995Go) and dBase IV (Borland International, 1993Go) software packages. Questions were tailored automatically, in that respondents were only presented with screens relevant to them, based on their previous answers. Respondents were able to go back one screen to change an erroneous answer, but no further, in an attempt to minimize deliberate under-reporting. The time taken to complete each set of questions was automatically recorded by the computer. Satisfaction with the computer was assessed by a subsample of 134 clients who were asked: ‘overall, how would you rate using the computer?'

Analysis
A 7 x 7 contingency table (Table 1Go) compared alcohol consumption reported on the QFI and RD. For each of the seven categories and for both measures, the mean amount consumed was calculated, as was the overall mean for both measures. Medians were also calculated to indicate the skewness produced by the reporting of relatively high levels of consumption by very few drinkers. In addition, for both measures, the number of standard drinks consumed/week by each respondent was categorized according to National Health and Medical Research Council of Australia guidelines as safe [no more than 14 (females) or 28 (males)], hazardous [15 to 18 (females) or 29 to 42 (males)] or harmful [more than 28 (females) or 42 (males)]. Two 2 x 2 contingency tables indicated the level of agreement between the RD and QFI in detecting heavy and high-risk drinking: consumption which exceeds safe levels is categorized as heavy and harmful consumption equates to high risk. The statistical significance of differences was assessed using a continuity-corrected McNemar's {chi}2 statistic. All analyses were performed using SAS for Windows software, version 6.11 (SAS Institute Inc, 1995).Go


View this table:
[in this window]
[in a new window]
 
Table 1. The relationship between the amounts of alcohol reported to be consumed/week, in standard drinks, using a 1-week retrospective drinking diary (RD) and a quantity–frequency index (QFI)
 

    RESULTS AND DISCUSSION
 TOP
 FOOTNOTES
 ABSTRACT
 INTRODUCTION
 SUBJECTS AND METHODS
 RESULTS AND DISCUSSION
 GENERAL DISCUSSION AND COMMENTS
 ACKNOWLEDGEMENTS
 REFERENCES
 
Sample
Of the 658 clients who attended counselling during the data collection phase, 69 (10.5%) did not use the computer. Seven refused consent to participate; nine were under the influence of drugs or alcohol; 16 were extremely distraught; 18 did not use the computer for a variety of other reasons, such as immediate referral to detoxification unit; four were English illiterate and declined assistance from the counsellor; seven did not use the computer due to their time restraints, and eight because it was temporarily inoperative. A further three clients did not complete the full questionnaire and were excluded from the analyses. In all, a study sample of 586 clients (89.1%), was obtained.

Computer use and acceptability
Only 0.5% of all clients failed to complete the computer questionnaire. Of the subsample (n = 134), 26.1% rated the computer as excellent, 35.8% as very good, 27.6% as good, 9.0% as fair, and 1.5% as don't know. The mean time taken for all clients to complete the QFI was approximately 34 s (20% of the mean time taken to complete all 10 AUDIT questions), compared to 3 min, 38 s for the RD. Considering only those who had consumed alcohol in the last 7 days (n = 351; 60%), the mean time taken to complete the RD was 4 min, 48 s.

Weekly alcohol consumption
Compared to typical levels of drinking, 23.2% of clients indicated consuming more, 29.9% the same, and 46.9% less, than usual in the previous 7 days. Table 1Go identifies the total mean consumption reported on the RD (22.2 drinks/week) as higher than the QFI (16.3 drinks/week). This difference resulted from the larger number of clients who reported consumption in either of the highest two categories on the RD (n = 170), relative to the QFI (n = 131). Removing them resulted in a lower mean RD score (5.58 and 7.93 drinks/week respectively). Total median consumption was marginally higher on the RD (7.9 drinks/week) than the QFI (7.5 drinks/week). This lesser difference, compared to mean consumption, resulted from the larger number of clients who reported zero on the RD (n = 235), relative to the QFI (n = 78). Removing them increased the magnitude of this difference (26.7 and 14.0 drinks/week respectively). Within the seven categories, median differences between the RD and the QFI were statistically significant (Wilcoxon Signed Rank Test: P < 0.00).

The higher overall mean for the RD reflects two possibilities. First, clients' drinking in the 7 days prior to counselling (RD) may not be representative of their usual consumption (QFI). For example, 23% of clients indicated that their consumption during the previous 7 days was more than they would usually drink: the mean RD score for these clients (30.1 standard drinks) is almost double their mean QFI score (15.7 standard drinks). Curiously, the mean RD score for the 47% of clients who indicated drinking less in the previous week than usual (25.2 standard drinks), is much higher than their mean QFI score (16.1 standard drinks). As has been demonstrated previously (Strecher et al., 1989Go), this finding may suggest that clients are attempting to respond in a socially desirable manner.

Second, clients may report their alcohol consumption less accurately on one or other of these measures. The higher consumption reported on the RD, relative to the QFI, may indicate that the extent of this under-estimation is less for the former. The RD may be more accurate than the QFI, because it provides prompts to enhance recall (Pernanen, 1974Go; Little et al., 1977Go; Millwood and McKay, 1978Go). The computer version of the RD is also likely to enhance its accuracy: it is less daunting since respondents are only presented with one screen at a time, and it does not require respondents to calculate the overall number of standard drinks consumed during the 7 days. The enhanced accuracy of the computer version of the RD may be indicated by the finding that, of studies comparing the RD and the QFI (Webb et al., 1990Go; O'Hare, 1991Go; O'Callaghan and Callan, 1992Go; Single and Wortley, 1994Go; Wyllie et al., 1994Go), these results represent the greatest discrepancy between the two measures (5.9 standard drinks). Although two additional studies reported relatively large differences, both had substantial methodological problems: one study compared an RD based on a face-to-face interview with a self-administered QFI (Midanik et al., 1989Go) and the other compared an RD completed in the respondent's home with a QFI completed in a doctor's office (Redman et al., 1987Go). In both instances, the demand characteristics of completing an RD are likely to be greater than for a QFI, such that an RD may represent an under-estimation of actual consumption. Caution is required in making comparisons between studies involving different populations; however, given that demand characteristics are the same for the RD and the QFI in this study, it may be that a computerized RD results in a reduction in the rate of under-reporting, relative to the QFI.

Agreement between the RD and QFI
Residual plots indicated that the two measures were related in a linear fashion, with a slope of 1.05 (95% CI: 0.93–1.17). A Spearman rank correlation coefficient revealed a relationship which was statistically significantly different from 0 (r = 0.63; P < 0.00), although weaker than identified previously: r = 0.74 (Redman et al., 1987Go); r = 0.86 (Webb et al., 1990Go). Higher levels of mean consumption were reported on the RD for each of the seven categories: almost twice as much mean consumption was reported on the RD at low consumption levels (less than 14 standard drinks/week) and the difference diminished as consumption amounts increased. The only exception to this trend occurred for the 29–42 standard drinks level, where almost twice as much mean consumption was again reported on the RD, relative to the QFI.

The finding in previous studies (Redman et al., 1987Go; Webb et al., 1990Go) that greater prevalence is reported on the QFI than on the RD at high levels of consumption (more than 28 standard drinks/ week) was replicated for median, but not mean, scores. The different response options used for the QFI in this study may be one factor which contributes to the limited evidence for this reversal: QFI categories do not become proportionately larger with increased consumption levels to the same extent as previously (Webb et al., 1990Go), so that heavier drinkers are not required to make such gross choices about their drinking and are, therefore, less likely to be allocated to a heavier drinking category. In addition, Webb et al. (1990) argued that mean consumption estimates for the RD and the QFI are likely to become closer, as the proportion of heavy drinkers in the sample increases: given that this trend is apparent and that this clinical sample contains a greater proportion of heavy drinkers, it provides partial support for this explanation.

Heavy and high-risk alcohol consumption
Table 2Go shows that 150 (128 on the RD and 22 on the QFI only) males (35.7% of all males) and 55 (42 on the RD and 13 on the QFI only) females (33.1% of all females) were identified as heavy drinkers by at least one measure. The RD identified more heavy drinkers than the QFI: for males, n = 128 (30.5% of all males) and 94 (22.4% of all males), respectively; and for females, n = 42 (25.3% of all females) and 37 (22.3% of all females), respectively. The difference between the RD and the QFI was statistically significant for males ({chi}2 = 13.96; P < 0.01), but not for females ({chi}2 = 0.52; P > 0.05). For males and females combined, a greater proportion was detected by RD (128 males and 42 females: 29.0% of all respondents) than by QFI (94 males and 37 females: 22.4% of all respondents).


View this table:
[in this window]
[in a new window]
 
Table 2. Agreement between a 1-week retrospective drinking diary (RD) and a quantity–frequency index (QFI) in detecting heavy male and female drinkers
 
Table 3Go indicates that 114 (88 on the RD and 26 on the QFI only) males (27.1% of all males) and 39 (22 on the RD and 17 on the QFI only) females (23.5% of all females) were identified as high-risk drinkers by at least one measure. For males, the RD identified more high-risk drinkers than the QFI: n = 88 (21.0% of all males) and 63 (15.0% of all males), respectively. For females the RD identified fewer high-risk drinkers than the QFI: n = 22 (13.3% of all females) and 31 (18.7% of all females), respectively. The difference between the RD and the QFI was statistically significant for males ({chi}2 = 7.48; P < 0.01), but not for females ({chi}2 = 2.56; P > 0.05). For males and females combined, a greater proportion of high-risk drinkers was detected by the RD (88 males and 22 females: 18.8% of all respondents) than the QFI (63 males and 31 females: 16.0% of all respondents).


View this table:
[in this window]
[in a new window]
 
Table 3. Agreement between a 1-week retrospective drinking diary (RD) and a quantity–frequency index (QFI) in detecting high-risk male and female drinkers
 
That the RD detects a greater proportion of heavy and high-risk drinkers than the QFI has been found previously among various populations (Redman et al., 1987Go; Webb et al., 1990Go; Werch, 1990Go; O'Callaghan and Callan, 1992Go). The only exception to this finding is for female high-risk drinkers, which may reflect their smaller numbers: a larger sample would produce a more accurate estimate of female heavy and high-risk drinking, which may have provided different results. Although separating results for men and women represents a different methodology to that used previously (Redman et al., 1987Go; Webb et al., 1990Go), utilizing cut-off points appropriate to gender, or a single cut-off point, appears to result in similar findings: for men and women combined, the RD detects a greater proportion of both heavy and high-risk drinkers, relative to the QFI. The smaller difference between the RD and the QFI for high-risk drinking, which is reversed for females, suggests that, while the RD appears to be preferable to the QFI for the detection of heavy drinking, it may not be substantially superior in detecting high-risk consumption (Webb et al., 1990Go).


    GENERAL DISCUSSION AND COMMENTS
 TOP
 FOOTNOTES
 ABSTRACT
 INTRODUCTION
 SUBJECTS AND METHODS
 RESULTS AND DISCUSSION
 GENERAL DISCUSSION AND COMMENTS
 ACKNOWLEDGEMENTS
 REFERENCES
 
Methodological considerations
This study appears to have minimal selection bias since 89% of clients attending counselling participated. As a result of all clients completing the RD and QFI in the same order, there is some possibility of unintended order effects. However, a previous study found only one significant interaction due to order of presentation (Webb et al., 1990Go) whereas another study, reporting more numerous order effects, had substantial methodological problems (Harford, 1994Go). It has also been found in another study that starting with high response options for the QFI yields different results, although the study was confounded by the QFI and RD being administered in different settings (Midanik et al., 1989Go).

The way in which this computerized version of the RD was presented to clients may be problematic. A major limitation of the RD is that it tends to require a disproportionate amount of time to complete. In order to minimize this problem, the RD version used in this study first checked whether clients drank at all in the week prior to attending counselling, rather than simply working backwards through the week. It may be that this method carries a relatively greater risk of clients forgetting whether they consumed alcohol in the previous week, since they have had no specific recall prompts. However, the possibility that clients presenting for an alcohol problem would forget whether they had consumed alcohol during the previous 7 days is probably minimal and any error due to the failure to prompt respondents from the beginning would, presumably, be in the direction of under- rather than over-estimation on the RD. As such, the differences between the RD and the QFI reported in this study are likely to be conservative: the usual RD measure may have resulted in higher levels of consumption being reported on the RD and an even greater discrepancy between the RD and the QFI.

In line with previous research (Bonevski et al., 1997Go; Newell et al., 1997Go; Shakeshaft et al., 1998aGo), clients appear to find the computer acceptable, as indicated by the negligible proportion of clients who failed to complete the questionnaire and the high acceptability reported by the majority of a subsample. However, computers do have some potential difficulties. First, they are susceptible to the same biases as any self-report measure, although evidence suggests they are an improvement on pen and paper methods (Bonevski et al., 1997Go; Kobak et al., 1997Go). Second, those with a physical disability or literacy problems may find them difficult to use. However, in this study, only 10.5% of clients did not use the computer. High levels of computer acceptability may reflect technological improvements, such as touchscreen interfaces and interactive software. Third, at approximately Australian $5000 each, they impose high initial costs. However, these costs dissipate over time: since 586 clients used the computer in this study over a 12-month period, a crude projection indicates that this equates to less than Australian $3/respondent over 3 years, although ongoing maintenance costs would add marginally to this figure. For data collection purposes, there are further cost savings from avoiding data entry. Fourth, there may be some resistance from clinicians. This issue is likely to require further investigation, given that the adoption of such innovations into routine clinical care is directly related to clinicians' levels of acceptability (Bonevski et al., 1997Go).

Implications
First, decisions about which measures to use may depend primarily on the purposes of the study (e.g. Room, 1990Go; Williams et al., 1994Go; Wyllie et al., 1994Go; Sobell and Sobell, 1995Go). For example, the primary advantage of the QFI is its brevity: it took respondents approximately 3 min less to complete than the RD. For those who had consumed alcohol in the previous 7 days, the difference is even greater, at over 4 min. Although the RD provides more information, there may well be situations in which this detail is of insufficient importance to justify the additional time burden. The various techniques used to augment the QFI (Werch, 1990Go; Russell et al., 1991Go; Williams et al., 1994Go; Wyllie et al., 1994Go) may begin to erode its brevity, and previous findings suggest that a relatively simple two-item QFI can result in comparable levels of self-reported consumption (Wyllie et al., 1994Go; Grant et al., 1995Go). However, given that less consumption overall was reported on this simple QFI, relative to the RD, a comparison between a more complex QFI and RD is indicated. Alternatively, the RD seems preferable for use in clinical settings: the majority of these treatment-seeking clients reported higher levels of consumption on the RD, which may enhance their motivation for change.

Second, the relatively high prevalence of heavy and high-risk drinking emphasizes the need to identify more closely patterns of at-risk drinking, that is, whether they are episodic or regular, as well as more precise estimates of the quantities consumed at these levels (Babor et al., 1994Go; Epstein et al., 1995Go; Rehm et al., 1996Go; Shakeshaft et al., 1998bGo). Such data would allow more accurate targeting of problem drinking behaviours, which may improve intervention effectiveness.

Third, the apparent instability in the agreement between the RD and the QFI as consumption levels increase indicates that these measures may not detect specific amounts of alcohol consumed at relatively high levels reliably. This suggests that outcome studies should report the proportion of drinkers moving between categories as a more reliable, although less specific, estimate of treatment effectiveness.

Fourth, there appears to be greater potential for improving the reliability and validity of the RD, relative to the QFI. In particular, greater use of computer technology may improve the validity of the RD.


    ACKNOWLEDGEMENTS
 TOP
 FOOTNOTES
 ABSTRACT
 INTRODUCTION
 SUBJECTS AND METHODS
 RESULTS AND DISCUSSION
 GENERAL DISCUSSION AND COMMENTS
 ACKNOWLEDGEMENTS
 REFERENCES
 
This study was primarily supported by a grant from the National Health and Medical Research Council of Australia. The authors would like to acknowledge the valuable assistance provided by the community counsellors in the collection of data and the statistical advice provided by the Hunter Centre for Health Advancement's statistics group, particularly Ms Sally Burrows.


    FOOTNOTES
 TOP
 FOOTNOTES
 ABSTRACT
 INTRODUCTION
 SUBJECTS AND METHODS
 RESULTS AND DISCUSSION
 GENERAL DISCUSSION AND COMMENTS
 ACKNOWLEDGEMENTS
 REFERENCES
 
* Author to whom correspondence should be addressed at: Hunter Centre for Health Advancement, Locked Bag No. 10, Wallsend, NSW, 2287, Australia. Back


    REFERENCES
 TOP
 FOOTNOTES
 ABSTRACT
 INTRODUCTION
 SUBJECTS AND METHODS
 RESULTS AND DISCUSSION
 GENERAL DISCUSSION AND COMMENTS
 ACKNOWLEDGEMENTS
 REFERENCES
 
AimTech Corporation (1995) IconAuthor: User Manual for Windows and OS/2 Version 7.0. Nashua.

Allen, J. P., Litten, R. Z. and Anton, R. (1992) Measures of alcohol consumption in perspective. In Measuring Alcohol Consumption: Psychosocial and Biochemical Methods, Litten, R. Z. and Allen, J. P. eds, pp. 205–226. Humana Press, New Jersey.

Babor, T. F. and Del Boca, F. K. (1992) Just the facts: enhancing measurement of alcohol consumption using self-report methods. In Measuring Alcohol Consumption: Psychosocial and Biochemical Methods, Litten, R. Z. and Allen, J. P. eds, pp. 3–19. Humana Press, New Jersey.

Babor, T. F., Longabaugh, R., Zweben, A., Fuller, R. K., Stout, R. L., Anton, R. F. and Randall, C. L. (1994) Issues in the definition and measurement of drinking outcomes in alcoholism treatment research. Journal of Studies on Alcohol 55 (Suppl.), 101–111.

Baranowski, T. (1985) Methodologic issues in self-report of health behaviour. Journal of School Health 55, 179–182.[ISI][Medline]

Bonevski, B., Sanson-Fisher, R. W., Campbell, E. M. and Ireland, M. C. (1997) Do general practice patients find computer health risk surveys acceptable? A comparison with pen-and-paper method. Health Promotion Journal of Australia 7, 100–106.

Borland International (1993) Using dBase IV Version 2.0. Borland International, Scotts Valley.

Bowman, R. S., Stein, L. I. and Newton, J. R. (1975) Measurement and interpretation of drinking behaviour. Journal of Studies on Alcohol 36, 1154–1172.[ISI][Medline]

Brown, J., Kranzler, H. R. and Del Boca, F. K. (1992) Self-reports by alcohol and drug abuse inpatients: factors affecting reliability and validity. British Journal of Addiction 87, 1013–1024.[ISI][Medline]

Cahalan, D. (1969) A multivariate analysis of the correlates of drinking-related problems in a community study. Social Problems 17, 234–247.[ISI]

Conigrave, K. M., Saunders, J. B. and Whitfield, J. B. (1995) Diagnostic tests for alcohol consumption. Alcohol and Alcoholism 30, 13–26.[Abstract]

Dawe, S. and Mattick, R. P. (1997) Review of Diagnostic Screening Instruments for Alcohol and Other Drug use and Other Psychiatric Disorders. Australian Government Publishing Service, Canberra.

Epstein, E. E., Kahler, C. W., McCrady, B. S., Lewis, K. D. and Lewis, S. (1995) An empirical classification of drinking patterns among alcoholics: binge, episodic, sporadic, and steady. Addictive Behaviors 20, 23–41.[ISI][Medline]

Gavin, D. R., Skinner, H. A. and George, M. S. (1992) Computerized approaches to alcohol assessment. In Measuring Alcohol Consumption: Psychosocial and Biochemical Methods, Litten, R. Z. and Allen, J. P. eds, pp. 21–39. Humana Press, New Jersey.

Goransson, M. and Hanson, B. S. (1994) How much can data on days with heavy drinking decrease the underestimation of true alcohol consumption? Journal of Studies on Alcohol 55, 695–700.[ISI][Medline]

Grant, K. A., Tonigan, J. S. and Miller, W. R. (1995) Comparison of three alcohol consumption measures: a concurrent validity study. Journal of Studies on Alcohol 56, 168–172.[ISI][Medline]

Harford, T. C. (1994) The effects of order of questions on reported alcohol consumption. Addiction 89, 421–424.[ISI][Medline]

Hilton, M. E. (1989) A comparison of a prospective diary and two summary recall techniques for recording alcohol consumption. British Journal of Addiction 84, 1085–1092.[ISI][Medline]

Kobak, K. A., Taylor, L. vH., Dottl, S. L., Greist, J. H., Jefferson, J. W., Burroughts, D., Mantle, J. M., Katzelnick, D. J., Norton, R., Henk, H. J. and Serlin, R. C. (1997) A computer-administered telephone interview to identify mental disorders. Journal of the American Medical Association 278, 905–910.[Abstract]

Lemmens, P., Tan, E. S. and Knibbe, R. A. (1992) Measuring quantity and frequency of drinking in a general population survey: a comparison of five indices. Journal of Studies of Alcohol 53, 476–486.[ISI][Medline]

Little, R. E., Schultz, F. A. and Mandell, W. (1977) Describing alcohol consumption: a comparison of three methods and a new approach. Journal of Studies on Alcohol 38, 554–562.[ISI][Medline]

Maisto, S. A. and Connors, G. J. (1992) Using subject and collateral reports to measure alcohol consumption. In Measuring Alcohol Consumption: Psychosocial and Biochemical Methods, Litten, R. Z. and Allen, J. P. eds, pp. 73–96. Humana Press, New Jersey.

Midanik, L. (1982) The validity of self-reported alcohol consumption and alcohol problems: a literature review. British Journal of Addiction 77, 357–382.[ISI][Medline]

Midanik, L. T. (1994) Comparing usual quantity/ frequency and graduated frequency scales to assess yearly alcohol consumption: results from the 1990 United States National Alcohol Survey. Addiction 89, 407–412.[ISI][Medline]

Midanik, L. T., Klatsky, A. L. and Armstrong, M. A. (1989) A comparison of 7-day recall with two summary measures of alcohol use. Drug and Alcohol Dependence 24, 127–134.[ISI][Medline]

Miller, W. R., Crawford, V. L. and Taylor, C. A. (1979) Significant others as corroborative sources for problem drinkers. Addictive Behaviors 4, 67–70.[ISI][Medline]

Millwood, J. E. and McKay, A. M. (1978) Measurement of alcohol consumption in the Australian population. Community Health Studies 2, 123–132.[ISI]

National Health and Medical Research Council (1992) Is There a Safe Level of Daily Consumption of Alcohol for Men and Women? Guideline, 2nd edn, Australian Government Publishing Service, Canberra.

Newell, S., Girgis, A. and Sanson-Fisher, R. W. (1997) Are touchscreen computer surveys acceptable to medical oncology patients? Journal of Psychosocial Oncology 15, 37–46.[ISI]

O'Callaghan, F. V. and Callan, V. J. (1992) Young adult drinking behaviour: a comparison of diary and quantity-frequency measures. British Journal of Addiction 87, 723–732.[ISI][Medline]

O'Hare, T. (1991) Measuring alcohol consumption: a comparison of the retrospective diary and the quantity– frequency methods in a college drinking survey. Journal of Studies on Alcohol 52, 500–502.[ISI][Medline]

Office for National Statistics (1994) Living in Britain: Results from the 1994 General Household Survey. HMSO, London.

Pernanen, K. (1974) Validity of survey data on alcohol use. In Research Advances in Alcohol and Drug Problems, Vol. 1, Gibbens, R. J., Israel, Y., Kalant, H., Popham, R. E., Schmidt, W. and Smart, R. G. eds, pp. 335–374. Wiley, New York.

Popham, R. E. and Schmidt, W. (1981) Words and deeds: the validity of self-report data on alcohol consumption. Journal of Studies on Alcohol 42, 355–358.[ISI][Medline]

Redman, S., Sanson-Fisher, R. W., Wilkinson, C., Fahey, P. P. and Gibberd, R. W. (1987) Agreement between two measures of alcohol consumption. Journal of Studies on Alcohol 48, 104–108.[ISI][Medline]

Rehm, J., Ashley, M. J., Room, R., Single, E., Bondy, S., Ferrence, R. and Giesbrecht, N. (1996) On the emerging paradigm of drinking patterns and their social and health consequences. Addiction 91, 1615–1621.[ISI][Medline]

Room, R. (1990) Measuring alcohol consumption in the United States: methods and rationales. In Research Advances in Alcohol and Drug Problems, Vol. 10, Kozlowski, L. T., Annis, H. M., Cappel, H. D., Glaser, F. B., Goodstadt, M. S., Israel, Y., Kalant, H., Sellers, E. M. and Vingilis, E. R. eds, pp. 39–80. Plenum Press, New York.

Russell, M., Welte, J. W. and Barnes, G. M. (1991) Quantity–frequency measures of alcohol consumption: beverage-specific versus global questions. British Journal of Addiction 86, 409–417.[ISI][Medline]

Samo, J. A., Tucker, J. A. and Vuchinich, R. E. (1989) Agreement between self-monitoring, recall, and collateral observation measures of alcohol consumption in older adults. Behavioral Assessment 11, 391–409.[ISI]

SAS Institute Inc. (1995) SAS/STAT Users Guide, Version 6.11. SAS Institute Inc., Cary, NC.

Saunders, J. B., Aasland, O. G., Babor, T. F., De La Fuente, J. R. and Grant, M. (1993) Development of the alcohol use disorders identification test (AUDIT): WHO collaborative project on early detection of persons with harmful alcohol consumption-II. Addiction 88, 791–804.[ISI][Medline]

Shakeshaft, A. P., Bowman, J. A. and Sanson-Fisher, R. W. (1998a) Computers in community-based drug and alcohol clinical settings: are they acceptable to respondents? Drug and Alcohol Dependence 50, 177–180.[ISI][Medline]

Shakeshaft, A. P., Bowman, J. A. and Sanson-Fisher, R. W. (1998b) Comparison of three methods to assess binge consumption: one week retrospective drinking diary, AUDIT and quantity/frequency. Substance Abuse 19, 191–203.[Medline]

Single, E. and Wortley, S. (1994) A comparison of alternative measures of alcohol consumption in the Canadian National Survey of alcohol and drug use. Addiction 89, 395–399.[ISI][Medline]

Skinner, H. A. and Sheu, W. J. (1982) Reliability of alcohol use indices: the Lifetime Drinking History and the MAST. Journal of Studies on Alcohol 43, 1157–1170.[ISI][Medline]

Sobell, L. C. and Sobell, M. B. (1992) Timeline follow-back: a technique for assessing self-reported alcohol consumption. In Measuring Alcohol Consumption: Psychosocial and Biochemical Methods, Litten, R. Z. and Allen, J. P. eds, pp. 41–72. Humana Press, New Jersey.

Sobell, L. C. and Sobell, M. B. (1995) Alcohol consumption measures. In Assessing Alcohol Problems. A Guide for Clinicians and Researchers, Allen, J. P. and Columbus, M. eds, pp. 55–73. National Institute on Alcohol Abuse and Alcoholism, Bethesda, MD.

Sobell, L. C., Brown, J., Leo, G. I. and Sobell, M. B. (1996) The reliability of the alcohol timeline follow-back when administered by telephone and by computer. Drug and Alcohol Dependence 42, 49–54.[ISI][Medline]

Sobell, M. B. and Sobell, L. C. (1973) Individualized behavior therapy for alcoholics. Behavior Therapy 4, 49–72.[ISI]

Straus, R. and Bacon, S. D. (1953) Drinking in College. Yale University Press, New Haven, CT.

Strecher, V. J., Becker, M. H., Clark, N. M. and Prasada-Rao, P. (1989) Using patients' descriptions of alcohol consumption, diet, medication compliance and cigarette smoking: the validity of self-reports in research and practice. Journal of General Internal Medicine 4, 160–166.[ISI][Medline]

Watson, C. G., Tilleskjor, C., Hoodecheck-Schow, E. A., Pucel, J. and Jacobs, J. (1984) Do alcoholics give valid self-reports? Journal of Studies on Alcohol 45, 344–384.[ISI][Medline]

Webb, G. R., Redman, S., Sanson-Fisher, R. W. and Gibberd, R. W. (1990) Comparison of a quantity– frequency method and a diary method of measuring alcohol consumption. Journal of Studies on Alcohol 51, 271–277.[ISI][Medline]

Webb, G. R., Redman, S., Gibberd, R. W. and Sanson-Fisher, R. W. (1991) The reliability and stability of a quantity–frequency method and a diary method of measuring alcohol consumption. Drug and Alcohol Dependence 27, 223–231.[ISI][Medline]

Werch, C. E. (1990) Two procedures to reduce response bias in reports of alcohol consumption. Journal of Studies on Alcohol 51, 327–330.[ISI][Medline]

Whelan, G. (1992) Biological markers of alcoholism. Australian and New Zealand Journal of Medicine 22, 209–213.[ISI][Medline]

Williams, G. D., Proudfit, A. H., Quinn, E. A. and Campbell, K. E. (1994) Variations in quantity– frequency measures of alcohol consumption from a general population survey. Addiction 89, 413–420.[ISI][Medline]

Wilson, A. and McDonald, P. (1994) Comparison of patient questionnaire, medical record, and audio tape in assessment of health promotion in general practice consultation. British Medical Journal 309, 1483–1485.[Abstract/Free Full Text]

Wyllie, A., Zhang, J. and Casswell, S. (1994) Comparison of six alcohol consumption measures from survey data. Addiction 89, 425–430.[ISI][Medline]