A Simple Scale of Gorski's Warning Signs for Relapse(*).(Statistical Data Included)

ABSTRACT. Objective: Though it has enjoyed widespread popularity, Gorski's post-acute withdrawal syndrome (PAWS) model of relapse has been subjected to little scientific scrutiny. A scale to operationalize Gorski's 37 warning signs was developed and tested in a larger prospective study of predictors of relapse. Of central interest were: (1) whether the warning signs hypothesized by Gorski are interrelated in a meaningful single factor and (2) whether the hypothesized syndrome would accurately predict subsequent relapses. Method: A sample of 122 individuals (84 men) entering treatment for alcohol problems was followed at 2-month intervals for 1 year. The Assessment of Warning-signs of Relapse (AWARE) scale was administered at each assessment point, and the occurrence of both slips (any drinking) and relapses (heavy drinking) was monitored during each subsequent 2-month interval. Principal factor analysis was used to study the factor structure of the warning signs. Results: Of the 37 warning signs, 28 clustered as a robust single factor with excellent internal consistency (Cronbach's alpha: 0.92-0.93). A conservative evaluation of test-retest stability across 2-month intervals estimated reliability at r = 0.80. After covarying for prior drinking status, clients' AWARE scores significantly predicted subsequent slips and relapses. Relapse rates for clients with highest AWARE scores, as projected by regression equations, were 33 to 46 percentage points higher than those for clients with lowest AWARE scores, after taking into account prior drinking status. Conclusions: This scale of Gorski's warning signs appears to be a reliable and valid predictor of alcohol relapses. (J. Stud. Alcohol 61: 759-765, 2000)

TERENCE GORSKI'S relapse prevention model has achieved popularity among treatment providers in the United States. Central to the model is the concept of a protracted alcohol withdrawal syndrome (PAWS) and a 37-step progression to relapse. The steps are presented as warning signs of proximity to a relapse (Gorski and Miller, 1982).

Despite its popularity and widespread dissemination over the past 2 decades, very little research has been devoted to the validity of this conceptual model, and not a single controlled trial has been published to document the efficacy of treatment based on this approach. In addition, key conceptual elements of the model, including the PAWS and hypothesized 37-step progression, remain unconfirmed by peer-reviewed scientific research. Instruments to measure key components of the model have been unavailable or undocumented.

As a small contribution toward testing Gorski's model, we designed a scale to operationalize the 37 steps or warning signs of relapse: the Assessment of Warning-signs for Relapse (AWARE). Each item has a 7-point Likert-type scale for responses ranging from "never" (1) through "sometimes," "often" and "always" (7). We developed the scale for use in a multisite prospective study of relapse precipitants (Lowman et al., 1996) and found it to be a reasonably good predictor of relapse (Miller et al., 1996). This brief report describes the scale, its factor structure and predictive validity, and our recommendations for its use in future research.

Method

Based on Gorski's narrative descriptions of the PAWS, we drafted single items to parallel each of the 37 hypothesized steps toward relapse. The AWARE was administered only at the Albuquerque site, at baseline and at 2, 4, 6, 8, 10 and 12-month follow-ups for 122 subjects (84 men) treated for alcohol dependence at the University of New Mexico Center on Alcoholism, Substance Abuse and Addictions (CASAA). Follow-up completion rates ranged from 97% at 2 months to 86% at 12 months.

Sample

Clients entered the study at an average age of 33.5 years. A majority (51%) were of non-Hispanic white origin, with 35% Hispanic, 7% Native American, 3% black and the remainder self-identifying with other racial-ethnic groups. Most were single (46%) or divorced (26%), with less than one third married (16%), separated (9%) or widowed (3%). Only 12% had full-time and 24% part-time employment at the time of intake. Severity of alcohol dependence and negative life consequences of drinking were high, with scores near the median for treatment-seeking populations (Miller et al., 1996).

Definitions of slip and relapse

At each follow-up point, a slip was defined as any use of alcohol following at least 4 days of abstinence. A relapse in the same period was defined as any heavy drinking day following a period of at least 4 days of abstinence. "Heavy drinking" was defined by a criterion amount of alcohol, adjusted for gender and body weight, that would result in a blood alcohol level over 200 mg% during a few hours of drinking (see Miller et al., 1996, p. S157). For a 140-pound woman or a 120-pound man, for example, the criterion number of standard drinks was five (2.5 oz absolute ethanol). Drinking outcomes were also measured via a variety of continuous variables (e.g., percent days abstinent, total alcohol consumption). Although we had expected superior predictability of continuous measures, in fact we were equally able to predict dichotomous outcomes (slip and relapse) across the 12-month follow-up period (Miller et al., 1996). This brief report focuses only on the AWARE scale and its relationship to these dichotomous outcomes. Additional details of the full sample and methodology of the parent study have been reported elsewhere (Lowman et al., 1996).

Statistical analyses

To determine whether the 37 elements of the PAWS formed a meaningful syndrome, we completed both factor analyses (FA) and principal component analyses (PCA) for the AWARE at baseline and at 12-month follow-up. We also…