Identifying Predictors for Relapse at a Rural Medical Center

Dean Stueland, M.D.
Medical Director
Program Manager
Marshfield Alcohol and Drug Treatment Program
Marshfield Clinic
St. Joseph's Hospital
Marshfield, Wisconsin

It is difficult in treating chronic medical illnesses to know how to measure clinical improvement. For example, in treating neoplastic diseases, one measure of success is 5 years survival. Another potential outcome measurement is time to relapse. In other chronic illnesses, such as diabetes mellitus, survival or the time from disease onset to the development of associated conditions is monitored. The treatment of addiction requires a similar orientation. Experienced clinicians as well as health planners are well aware of the propensity for persons with addiction to relapse. A reasonable measurement of success, therefore, is to consider the number of persons who remain free of the disease or effects of the disease over a monitored time period (Institute of Medicine 1990).

It is possible to do surveys of patients or clients or of persons acquainted with the patient in order to establish whether relapse has occurred (Desmond et al. 1995). Relapse, however, should often be apparent to clinicians involved in the patient's medical care. At the least, medically apparent relapse measures an outcome of relapse.

In analogous chronic diseases, preexisting medical, social, or environmental variables often help predict outcome. Often these variables are easily obtained and interact with treatment in determining outcome. In evaluating the science of addiction treatment, it is important to know how social, psychological, and physical variables predict relapse (Institute of Medicine 1990).

Addiction in rural America may be different from urban areas. The usage in rural American may differ from urban areas not only in observed rates of use but also by agents used in addiction. Moreover, the influence of predictors for relapse may be different in rural areas (Wertz et al. 1995).

Purpose

The purpose of this study is to look at predictors for relapse in a defined rural population. This study is meant to be exploratory. It looks at several potential predictors of relapse. These predictors are compared to the incidence of observed relapse. Such information is useful in evaluating the effectiveness of treatment. In addition, this information can be useful to clinicians in individualizing treatment to a patient's need.

Methods

The inpatient unit of the Marshfield Alcohol and Drug Treatment Program is part of the medical complex that includes the Marshfield Clinic and the adjacent St. Joseph's Hospital. The Marshfield Medical Complex is a rural referral center that potentially serves patients from central and northern Wisconsin. Persons who live near the Marshfield Center would need to travel a considerable distance in order to seek treatment elsewhere. Previous studies have demonstrated the utility of using this type of defined population in medical studies (Nordstrom et al. 1994). Followup is further facilitated by the use of a combined medical record.

A form was developed for data collection and presented to the Institutional Review Board (IRB) of the Marshfield Medical Research Foundation. With IRB approval and appropriate methods for maintaining anonymity, the treatment and medical records of all patients who were admitted to the inpatient unit during the years 1990 through 1993 were reviewed during 1995. The data abstracted included social variables, psychological and personality disorder diagnoses, medical treatment, and the most recent RAATE (Mee-Lee 1988) estimate of status. For patients who had more than a single treatment episode during that time, only information collected during the first episode was used for prediction. In addition, the clinic medical record and any hospital readmissions were reviewed for evidence of relapse in order to establish if and when relapse occurred. From that information, a variable was defined as "relapse within 1 year."

Information was maintained in a Statistical Program for the Social Sciences (SPSS) file, and analysis was done by simple frequencies and measures of differences in percentages. Although all admissions to the inpatient unit had data abstraction, this study is based only on patients who initially lived in the five county area around Marshfield.

Results

During the 4 years of observation, there were 562 admissions to the inpatient unit from the five county area. Of this group, 78 (13.9 percent) had evidence of relapse within 1 year. Overall, there were 392 (69.8 percent) males, but gender was not associated with the probability of relapse within 1 year.

The relationship of marital status to the probability of relapse is shown in table 1. For those who are married, 20 (9.0 percent) had relapsed. For single patients, 34 (18.9 percent) had relapsed within 1 year.

The probability of relapse is related to the level of education (table 2). Those who had a high school education had a probability of relapse (13.8 percent) that was nearly identical to those who had less than a high school education (14.0 percent). However, for those who had greater than a high school education, the probability of relapse was 6.0 percent.

Occupation was also related with the 1-year relapse rate (table 3). The highest relapse rate was for those who were students. The lowest relapse rate was for those who do odd jobs, followed by those who work in manufacturing and management.

Potential referral sources were compared with the probability of relapse (table 4). Only the referral sources of an AODA (alcohol and other drugs of abuse) counselor and a judge or lawyer recommendation were found to be related to the 1-year relapse rate. Seventy-four of the 77 patients who were referred by an AODA counselor (96.1 percent) had not relapsed within a year.

The sources of care at the time of referral were compared with the 1-year relapse rate (table 5). The only care provider related to the probability of relapse was probation officer. The presence of other providers was not related to relapse rate.

Nubmers in parentheses are percentages.

Note: Each patient was classified by whether or not that referral source was involved at the time of admission

The 1-year relapse rate is compared with social and psychological parameters (table 6). The strongest relationship was shown for patients who had had a prior admission for AOD treatment. In addition, the presence of a personality disorder, a drug conviction or the diagnosis of hyperactivity were related to the probability of relapse. Of the 58 patients who had prior admissions for AOD treatment, 19 (32.8 percent) had evidence of relapse within the year. Of the 26 patients with hyperactivity, 8 (30.8 percent) had also relapsed within a year.

The potential medical diagnosis or findings on admission were compared with the probability of relapse (table 7). The presence of intoxication at the time of admission was related to the probability of relapse, but the other four diagnoses were not. Of the 77 patients who were intoxicated at the time of admission, 17 (22.1 percent) had relapsed within a year.

The substances found on toxicology screen (table 8) or as determined by dependency (table 9) were compared with the probability of relapse within 1 year. The presence of alcohol within the urine was noted to be related to the risk of relapse within 1 year.

The diagnosis of amphetamine dependence predicted sobriety. Of the 152 patients who had alcohol in the urine on admission, 33 (21.7 percent) had relapsed within 1 year (see table 8). Of the 56 patients who were dependent on amphetamines, 2 (3.6 percent) had relapsed within 1 year (see table 9).

The five categories of RAATE were compared to relapse (table 10). The third component, medical conditions (acuity of biomedical problems), was related to relapse. The relationship to medical condition was ordinal so that those who had the better medical condition were less likely to relapse.

Note: Data are missing for 96 patients

Note: The total represents the number with that value. The N and percentage apply to the total who had that value at the time of the most recent RAATE. An increasing value indicates increasing problems.

Discussion

It is clear that certain social and psychological variables were related to the risk of relapse. Having a more stable marital stiuation or increased education reduced the probability of relapse. Similarly, certain occupations had less probability of relapse. It would appear that work adds an important structure that prevents relapse since those who had no work or students were most likely to relapse. The small number who had more than a single part-time job (odd jobs) actually had the best prognosis.

The referral source was also predictive. Those who were referred by an AODA counselor tended to have a better prognosis. The involvement of legal personnel (such as a judge or lawyer) tended to predict a better prognosis. Both of these variables may indicate a commitment to sobriety, from whatever source, that can be measured at the time of admission. In a similar fashion, the presence of someone under the care of a probation officer suggests preexisting structure.

This study also supports the concept that patients who have had prior admissions and relapses are more likely to relapse again. This is not significantly different from other chronic diseases. This study also suggests that certain concurrent medical and psychological diagnoses, such as hyperactivity or personality disorder, increase the probability of relapse. Although the number is small, drug conviction seems to indicate a better prognosis, but that is consistent with the fact that patients under the care of a probation officer or referred for legal reasons tended to show less relapse within a year.

Specific medical diagnoses, except for intoxication, did not predict relapse. On the other hand, those who were medically stable were less likely to relapse than those who still had ongoing medical problems. This phenomenon may be partly explained by the fact that the search for evidence of relapse used the medical record. Along with the fact that intoxication on admission portrays a poorer prognosis, alcohol in the urine was the only agent that also portrayed a poor prognosis.

In general, the type of dependency did not predict relapse. The only possible exception is the presence of amphetamine dependence. This, however, was the least common diagnosis. Furthermore, persons with amphetamine dependence may be less likely to seek medical care when they relapse.

Certain limitations should be noted. Central Wisconsin is a largely rural area and would seem to be consistent with many rural agricultural communities. However, the social, medical, and psychological parameters are likely to be somewhat specific to an area. Several variables were examined. It is possible, especially in a retrospective study, that spurious relationship may be found. Although there was checking of the data abstraction and data entry, it is important to remember that the data were obtained from clinical practice. When the information on diagnosis and RAATE was obtained, the participants were not aware that the data would be used for research purposes. Although this may introduce some imprecision in the collecting of the predictor variables, this study underscores the fact that routinely collected clinical data are predictive.

The presence in the medical record of evidence of relapse may understate the overall relapse. However, the presence of medical relapse is a functional marker of clinical importance. Although this very well may understate the overall relapse rate, it is not clear that this should necessarily be true for any specific predictor variable.

Implications

This study suggests that psychosocial predictors in a rural area are similar to those that have been found in other areas. It also suggests that information in the medical record can be reasonably combined with AOD information to facilitate care and overall planning.

From a research viewpoint, it is likely that some of the predictor variables are related. Further studies in a rural addiction center are indicated including, possibly, the use of multi-variable analysis. This preliminary information does suggest that patients can be stratified by risk of relapse.

Acknowledgments

Several individuals with the program made a significant contribution to this study: Sheila Weix, R.N., CARN, and Lanny Parker, CADC III, assisted with the conceptualization and design of the study and development of the abstraction instrument. Richard Berg, M.S., assisted in statistical analysis.

References

Desmond, D.P.; Maddux, J.F.; Johnson, T.H.; and Confer, B.A. Obtaining follow-up interviews for treatment evaluation. Journal of Substance Abuse Treatment 12(2):95-102, 1995.

Institute of Medicine. Broadening the Base of Treatment for Alcohol Problems. Washington, DC: National Academy Press, 1990.

Mee-Lee, D. An instrument for treatment progress and matching The Recovery Attitude and Treatment Evaluator (RAATE). Journal of Substance Abuse Treatment 5:183-186, 1988.

Moos, R.H.; Brennan, P.L.; and Mertens, J.R. Diagnostic subgroups and predictors of one-year re-admission among late-middle-aged and older substance abuse patients. Journal of Studies in Alcohol 55(2):173-183, 1994.

Nordstrom, D.L.; Remington, P.L.; and Layde, P.M. The utility of HMO data for the surveillance of chronic diseases. American Journal of Public Health 84(6):995-997, 1994.

Statistical Package for the Social Sciences (SPSS/PC+), Version 4.0.1.

Wertz, J.S.; Cleaveland, B.L.; and Stephen, R.S. Problems in the application of the Addiction Severity Index (ASI) and rural substance abuse services. Journal of Substance Abuse 7(2):175-188, 1995.

Back to Top