Drug Abusers' Adherence to Medical Treatment Regimens

An Annotated Bibliography
April, 1998

Kathleen M. Carroll, Ph.D.
Yale University School of Medicine

Hyperlinks to sections within this text:

Discharge Against Medical Advice
Cystic Fibrosis
Hypertension
Diabetes
Chronic Pain
HIV, AIDS, and Tuberculosis
Pregnancy
General
Sample Articles Relevant to Compliance Where Drug Abuse Is Not Mentioned

Author's Note

Very few articles were available on the subject of drug abusers' adherence to medical treatment regimens. An extensive literature search included (a) searches back to 1980 on both Medline and Psychinfo databases for virtually all combinations of compliance, drug abuse, and many different forms of illness and medical treatments; (b) review of reference lists of all even slightly relevant articles and reviews; (c) review of references and index lists from the major texts on compliance and health care; and (d) consultation with colleagues knowledgeable in this area.

Many of the articles identified had several methodological limitations; many were retrospective chart reviews, and very few assessed drug abuse systematically using standardized instruments or diagnostic criteria. Areas in which drug abuse and a medical disorder or condition overlapped more closely (e.g., HIV, pregnancy) were by far the most plentiful and of the highest quality. The reasons for this apparently underinvestigated area may be many, but two key themes emerged from the review of this literature: First, compliance with medical treatment tends to be an understudied area (Aronson & Hardman, 1992; Bond & Hussar, 1991; Kruse, 1992). Second, drug abuse is frequently undetected in general medical practice (Kamerow, Pincus, & Macdonald, 1986), with only about 10% of those meeting criteria for a substance use or mental disorder detected by their primary care physicians.

In contrast, there are literally thousands of articles on drug abusers' compliance, and lack thereof, with their drug abuse or psychiatric treatment. This may be an area more ripe for compilation and analysis, even meta-analysis.

References

Aronson, J. K., & Hardman, M. (1992). Patient compliance. British Medical Journal, 305, 1009-1011.

Bond, W. S., & Hussar, D. A. (1991). Detection methods and strategies for improving medication compliance. American Journal of Hospital Pharmacy, 48, 1978-1988.

Kamerow, D. B., Pincus, H. A., & Macdonald, D. I. (1986). Alcohol abuse, other drug abuse, and mental disorders in medical practice: Prevalence, costs, recognition, and treatment. Journal of the American Medical Association, 255, 2054-2957.

Kruse, W. (1992). Patient compliance with drug treatment: New perspectives on an old problem. Clinical Investigator, 70, 163-166.

DISCHARGE AGAINST MEDICAL ADVICE

Baile, W. F., Brinker, J. A., Wachspress, J. D., & Engel, B. T. (1979). Signouts against medical advice from a coronary care unit. Journal of Behavioral Medicine, 2, 85-92.

This was a chart review study of 29 (1.2% of the 2,413 admissions during the 3-year study period) against medical advice (AMA) signouts from the coronary care unit of an urban general hospital. Drug abuse was not evaluated, but alcohol abuse was significantly associated with AMA discharge (14/29 AMA discharges versus 4/29 controls, p < .01).

Jankowski, C. B., & Drum, D. E. (1979). Diagnostic correlates of discharge against medical advice. Archives of General Psychiatry, 34, 153-155.

Noting that there have been few studies of patients discharged AMA from general (as opposed to psychiatric) hospitals, the authors conducted a chart review study of all patients discharged AMA from a university general hospital over a 1-year period. Seventy-three individuals (0.7% of the 10,600 total admissions) were discharged AMA, defined as signing a legal release form, leaving without notice. Six patients threatened to leave and signed a release but ultimately remained. Drug addiction was the diagnosis most strongly associated with AMA discharge and was present in 16 of the 73 patients (22%), compared with 0 of the controls. Alcoholism was noted in the charts of 42% of the AMA group versus 15% of the controls. The methods for diagnosing drug or alcohol dependence were not specified in the report.

Ochitill, H. N., Havassy, B., Byrd, R., & Peters, R. (1985). Leaving a cardiology service against medical advice. Journal of Chronic Disease, 1, 79-84.

This article was a retrospective, chart review study of 64 hospitalized patients (78% male, 63% white, mean age 60) who signed out from the Cardiology Service at San Francisco General Hospital Medical Center, where the common reasons for admission include symptoms or signs of ischemic heart disease, arrhythmias, and congestive heart failure. The 32 patients who signed out AMA were compared with 32 randomly selected control patients (admitted to the same service the same month). Determination of drug use required evidence of daily use of a substance, inability to reduce or stop drinking or drug use, or continued abuse despite a serious physical disorder (thus roughly parallel to DSM criteria). Alcohol abuse was higher in the AMA discharge groups with respect to controls (p < .08), and significantly so for those with physical signs of alcohol abuse (p < .005), but drug abuse did not distinguish the groups. The administration of anxiolytic medication was also significantly associated with AMA discharge. The lack of an association between drug abuse and AMA discharge may be associated with the relative age of the sample and hence the probable low rate of drug abuse.

CYSTIC FIBROSIS

Stern, R. C., Byard, P. J., Tomashefski, J. F., & Doershuk, C. F. (1987). Recreational use of psychoactive drugs by patients with cystic fibrosis. Journal of Pediatrics, 111, 293-299.

One hundred seventy-three adults with confirmed cystic fibrosis were surveyed about their current level of alcohol and drug use in an uncontrolled questionnaire study. No standardized instruments were used to assess drug use. Sixteen patients (9.2%) reported trying cocaine at least once, and 23% were regular marijuana users. Heavy marijuana use was associated with worsening of chronic bronchitis.

HYPERTENSION

Shea, S., Misra, D., Ehrlich, M. H., Field, L., & Francis, C. K. (1992). Predisposing factors for severe, uncontrolled hypertension in an inner-city minority population. New England Journal of Medicine, 327, 776-781.

Poorly controlled hypertension has been identified as a major health problem, particularly among minority populations, the poor, those with lower educational levels, and those with limited access to medical care. This is a report on a case-control study with 93 patients with severe uncontrolled hypertension who presented in emergency rooms and 114 control patients with hypertension. All patients were black or Hispanic. Data were collected from patient interviews, which included questions on the use of several categories of illicit drugs (marijuana, cocaine, "crack" cocaine, heroin, methadone, and other drugs) during the previous year. Multiple logistic regression models were used to identify risk factors for severe hypertension. As expected, uncontrolled hypertension was more common among subjects with no primary care physician and among those who did not comply with treatment for hypertension. Illicit drug use was reported in 20 of 93 case patients (21.5%) versus 8 of 1,146 controls (0.7%) (odds ratio: 3.6 [1.5-8.7], p < .002). However, illicit drug use was no longer significant after controlling for age, sex, race, ethnic group, educational level, and current smoking status (adjusted odds ratio, 1.3, [0.5-3.6], p = .60).

DIABETES

Glasgow, A. M., Tynan, D., Schwartz, R., Hicks, J. M., Turek, J., Driscol, C., O'Donnell, R. M., & Getson, P. R. (1991). Alcohol and drug use in teenagers with diabetes mellitus. Journal of Adolescent Health, 12, 11-14.

One hundred and one adolescents aged 12-20 (mean 14.9) (50% males; 63% white) with diagnosed diabetes mellitus with duration of at least 1 year were given an anonymous questionnaire about their drug use. Patients whose glycohemoglobin levels exceeded the upper limit of the reference range by 50% or more were oversampled to evaluate the relationship between diabetes control and substance use. Twenty (19.8%) reported trying at least one drug (most frequently marijuana), and 5% reported ongoing use (marijuana, inhalants, or stimulants/ tranquilizers). Of the 97 subjects who provided a urine specimen, one was positive for marijuana, another for marijuana and cocaine (the latter was in the poorly controlled group). Because of the small number of patients reporting drug use, all positive responses (except alcohol) were pooled. Reported alcohol use was not associated with higher glycohemoglobin levels, but the 20 patients who reported any drug use were significantly more likely to have higher glycohemoglobin levels than those who did not (9.3 versus 8.1, p = .02), but there was not a significant difference in the physician's estimate of control or compliance. Glycohemoglobin levels were also significantly higher among the 17 patients who reported having a parent with a drug or alcohol use problem.

CHRONIC PAIN

Berndt, S., Maier, C., & Schutz, H. W. (1993). Polymedication and medication compliance in patients with chronic non-malignant pain. Pain, 52, 331-339.

One hundred nine patients with chronic, nonmalignant pain (68% female) were asked about their current medication intake during the 24 hours prior to a routine examination and were told a urine test would be done to verify their statements about current drug intake. Polymedication (defined as daily consumption of three or more medications and consistent with noncompliance with physicians' pain prescription) was seen in 41 patients (38%). Only 68% of self-reports of current medication were found to be congruent with urine toxicology screens. Twenty-one percent had concealed drug consumption from their physician. Of the drugs concealed, 222 (54%) were psychotropic, most frequently benzodiazepenes, hypnotics, and antidepressants. The consumption of analgesics was also frequently denied falsely.

A history of drug abuse (not defined) was strongly associated with polymedication in this sample, with polymedication seen in 16 of 24 patients (63%) with a history of drug abuse, compared with 26% (31/85) of patients without such a history (p < .009), and was the most strongly associated with noncompliance of the factors studied (including age, duration of pain, intensity of pain, and type of pain). A history of drug abuse was also strongly associated with false statements concerning their medication use, as false statements were seen in 54% (13/24) of individuals with a drug abuse history versus 31% (26/85) of those without a drug abuse history. Poor compliance was associated with poorer long-term outcome.

Sullivan, M. D., Turner, J. A., & Romano, J. (1991). Chronic pain in primary care: Identification and management of psychosocial factors. Journal of Family Practice, 32, 193-199.

This is a general review of psychosocial factors associated with chronic pain, which is the most common chief complaint presented to primary care physicians. Focusing on psychiatric disorders that retard recovery from illness or injury, the authors describe data associated with depression, anxiety, and dementia. Substance abuse is noted as follows: "Substance abuse may arise in conjunction with the pain problem (e.g. opiates, benzodiazepenes) or precede it (e.g., alcohol, marijuana, cocaine). In either form, substance abuse impedes rehabilitation and needs to be addressed aggressively. Abuse can manifest itself as signs of medication dependence, including escalating doses, maneuvering in the health care system to get drugs, and withdrawal symptoms." No references are given in this section.

HIV, AIDS, TB

Cowan, F. M., Jones, G., Bingham, J., Flegg, P. J., MacCallum, L. R., Whitelaw, J., Hargreave, D., Gray, J. A., Welsby, P. D., & Brettle, R. P. (1989). Use of zidovudine for drug misusers infected with human immunodeficiency virus. Journal of Infection, 18 (Suppl. 1), 59-66.

This study prospectively examined the safety of zidovudine among injecting drug users with HIV in an HIV clinic in Scotland. Of 26 patients who were determined to have contracted HIV through intravenous drug use, 58% missed at least one appointment. Eleven (42%) had no missed clinic visits, 12 (46%) missed one to three clinic visits, and 3 (12%) missed four to five clinic visits. Of the 26, the most reliable attenders were those on a methadone maintenance program (p = .05). The authors noted that none of the patients who contracted HIV through homosexual contact missed any clinic visits (n = 13).

Elk, R., Grabowski, J., Rhodes, H., Spiga, R., Schmitz, J., & Jennings, W. (1993). Compliance with tuberculosis treatment in methadone-maintained patients: Behavioral interventions. Journal of Substance Abuse Treatment, 10, 371-382.

With the recognition that noncompliance with treatment for tuberculosis is complicated and, therefore, generally poor, two studies were conducted to evaluate behavioral strategies for enhancing compliance among methadone-maintained patients with tuberculosis (TB) as well as reducing their drug use. In the first study, nine opiate users who tested positive for tuberculosis received their methadone contingent on prophylactic isoniazid (INH) ingestion. Compliance with INH was 100% in eight patients (94% for the ninth); however, subjects' cocaine use remained high, and seven patients were discharged before the end of the study due to drug-positive urines (patients received methadone decreases for drug-positive urines). A second study was conducted to address the high dropout levels in the first study. Two methadone patients who were positive for TB received methadone contingent on INH ingestion, but also were positively reinforced for decreases in cocaine use. Again, compliance with INH was high (above 97%) and had over 40% cocaine-free urine samples compared with 0% at baseline. This study is consistent with the broader behavioral literature and points to the promise of positive contingencies for promoting compliance with medical treatments in this population.

Fischl, M. A., Richman, D. D., Hansen, N., Collier, A. C., Carey, J. T., Para, M. F., Hardy, W. D., Dolin, R., Powderly, W. G., Allan, J. D., Wong, B., Merigan, T. C., McAuliffe, V. J., Hyslop, N. E., Rhame, F. S., Balfour, H. H., Spector, S. A., Volberding, P., Pettinelli, C., Anderson, J., & The AIDS Clinical Trials Group. (1990). The safety and efficacy of zidovudine (AZT) in the treatment of subjects with mildly symptomatic human immunodeficiency virus Type I (HIV) infection: A double-blind, placebo-controlled trial. Annals of Internal Medicine, 112, 727-737.

This study was a multicenter, double-blind, randomized controlled trial of AZT in 711 (5% women, 91% white, mean age 35) subjects with mildly symptomatic HIV infection. Compared with placebo, AZT was found to delay progression of HIV disease and produced little toxicity. Noncompliance was defined as the failure to keep consecutive follow-up appointments or to take study medication as directed. One hundred sixty-four subjects left the study (93 in the placebo and 71 in the AZT group due to self-withdrawal or noncompliance). Compliance analyses by group (e.g., drug abusers) were not provided in this paper.

Marwich, C. (1992). Do worldwide outbreaks mean tuberculosis again becomes "Captain of all these men of death"? Journal of the American Medical Association, 267, 1174-1175.

Remarking on the spread of multi-drug-resistant tuberculosis in the United States, the author notes, "Because many HIV-infected patients are intravenous drug abusers or are homeless or both, they are a very difficult group to treat. In addition, there are anecdotal reports to suggest that the combination of HIV and tuberculosis infections is particularly virulent...." No data or references are provided.

Samet, J. H., Libman, H., Steger, K. A., Dhawan, R. K., Chen, J., Shevitz, A. H., Dewees-Dunk R., Levenson, S., Kufe, D., & Craven, D. E. (1992). Compliance with zidovudine therapy in patients infected with human immunodeficiency virus, Type 1: A cross-sectional study in a municipal hospital clinic. American Journal of Medicine, 92, 495-502.

Noting that no previous studies have examined factors associated with compliance with AZT, this cross-sectional study surveyed 83 patients (80% male, 46% white, mean age 36) infected with HIV. Sixty-three percent of the sample reported IV drug use as their risk factor for HIV (28% male homosexuality, 8% immigration from endemic area, and 34% heterosexual contact). The compliance survey, administered by a study nurse, included questions on frequency and types of illicit drugs used and participation in drug treatment programs, but did not assess drug use using standardized instruments. Compliance was defined as the percentage of prescribed zidovudine doses reportedly taken in the previous week (values greater than 80% were defined as compliant). Compliance histories were validated by serum and urine AZT levels. Overall, compliance was high, with 67% of the 83 patients reporting greater than 80% compliance (33% were fully compliant). Missing doses were most frequently attributed to forgetfulness (75% of patients who missed at least one dose). The mean compliance percentage for patients with a history of IV drug use was 83%, compared with 92% for those without such a history (p = .02). Although 15 of 53 IV drug users reported use of IV drugs within the past 3 months, recent drug use was not related to reported compliance. The odds ratio for a history of IV drug use on poorer compliance was 2.7 (p < .06).

Wall, T. L., Sorenson, J. L., Batki, S. L., Delucchi, K. L., London, J. A., & Chesney, M. A. (1995). Adherence to zidovudine (AZT) among HIV-infected methadone patients: A pilot study of supervised therapy and dispensing compared to usual care. Drug and Alcohol Dependence, 37, 261-269.

This is a random assignment pilot study evaluating an intervention intended to enhance AZT compliance among HIV-infected methadone-maintained patients, where adherence was previously estimated at 50%. The 8-week intervention was based on the Centers for Disease Control and Prevention's (CDC's) successful tuberculosis prevention project and involved on-site dispensing and monitoring of AZT adherence by a registered nurse, as well as monitoring of adverse side effects and monthly feedback about participants' CD4 cell count and MCV. Twenty-seven patients (52% male, 32% Caucasian, mean age 41) were randomly assigned to treatment, and two dropped out before completing the study (one was jailed and one was discharged due to treatment program rules). A three-item AZT adherence scale was developed for this project and found to have high internal consistency (alpha = 0.82). Subjects were selected for self-reported nonadherence to AZT. The authors noted that correlations between measures of compliance have been found to have low to moderate correlations. Thus, four different AZT adherence indicators were used (self-report, MCV, Medication Event Monitoring System [MEMS], and pill counts), with correlations in this study ranging from .01 (pill count/self-report) to .64 (MCV/MEMS). Compared with treatment as usual (which included a physical examination, an explanation of AZT including benefits and side effects, the opportunity to refill prescriptions, and regular clinical care and monitoring), results favored the experimental condition, but significant differences were seen only on some measures (MCV and MEMS) and some follow-up points (with significant differences at 1-month follow-up). MEMS data indicated better adherence on weekdays (50% adherence) than weekends when patients did not receive interventions (30% adherence). The authors stress the importance of multiple measures of adherence and discuss the strengths and weaknesses of the indexes used in this study.

PREGNANCY

Elk, R., Schmitz, J., Spiga, R., Rhoades, H., Andres, R., & Grabowski, J. (1995). Behavioral treatment of cocaine dependent pregnant women and TB-exposed patients. Addictive Behaviors, 20, 533-542.

This paper reported on two multiple baseline studies of a behavioral treatment; both used behavioral techniques (incentives) to increase adherence to medical treatments. The first study evaluated five TB-positive patients without active disease who required chemoprophylactic treatment with isoniazid (INH). All met current DSM-III-R criteria for cocaine and opiate dependence and had more than 50% cocaine-positive urine samples during the baseline period. Methadone was dispensed contingent on INH ingestion, and thus compliance was extremely high (97.6%). The second study included seven pregnant women (four met DSM-III-R criteria for cocaine dependence; three met criteria for opiate and cocaine dependence). In both studies, patients received monetary reinforcers for cocaine-free urine samples or decreases in cocaine metabolites detected in urines, with schedules of urine specimen collection and payments varying across the two studies. Retention in both studies was high; only one patient dropped out of the TB study and all of the pregnant women were retained through delivery. In the pregnancy study, reinforcement for compliance with scheduled prenatal visits was offered as a weekly bonus if all required specimens for the week were cocaine free. Overall compliance with prenatal care throughout the study was high, with a mean rate of 72.5%. For both groups, compared to baseline, there was a significant decrease in BE (the cocaine metabolite benzoylecgonine) during the contingency phase of the study, but the effect was somewhat stronger for the TB group.

Fraser, A. C., & Cavanagh, S. (1991). Pregnancy and drug addiction: Long-term consequences. Journal of the Royal Society of Medicine, 84, 530-532.

This was a retrospective study of 86 pregnant, drug-dependent women who delivered between 1968 and 1974. A log-linear analysis suggested that poorer prenatal attendance was associated with lower likelihood of abstinence after delivery, as 38% (25/66) of those who continued their drug abuse had good prenatal attendance compared with 70% (14/20) of those who stopped using drugs after delivery.

Melnikow, J., Alemagno, S. A., Rottman, C., & Zyzanski, S. J. (1991). Characteristics of inner-city women giving birth with little or no prenatal care: A case-control study. Journal of Family Practice, 32, 283-288.

Risk of low birth weight and perinatal mortality is roughly double among women who receive no prenatal care. In a retrospective case review study from a center treating primarily low-income, inner-city women, 120 women who received inadequate prenatal care (fewer than three visits) were compared with 120 women who received more adequate care (the sample was 42% white, 45% black, 7% Hispanic; mean age 23). Because this was a retrospective chart review, alcohol and drug use data were not systematically collected; these data were recorded as categorical only (yes/no). Drug use was found in 23% of the inadequate care groups versus 6% of the control group (p < .01); the odds ratio from logistic regression for drug use predicting inadequate prenatal care was 2.4 (0.9-6.5 CI). Regression analyses identified higher parity, tobacco use, and drug abuse as the three factors most strongly associated with inadequate prenatal care in this population.

GENERAL

Gottlieb, N. H., Mullen, P. D., & McAlister, A. L. (1987). Patient's substance abuse and the primary care physician: Patterns of practice. Addictive Behaviors, 12, 23-32.

In this article, 442 of 500 primary care physicians in Texas responded to a survey about their self-efficacy and outcome expectations regarding patients' smoking, alcohol problems, over-the-counter (OTC) drug problems, and illicit drug use. Physicians were most likely to collect patient history of smoking (72.7% reported routinely taking a history), followed by alcohol problem histories (50%), and least likely to collect patient histories of drug abuse (19.7%). Of the type of substances evaluated, the physicians were most likely to be certain that their patients would follow through on recommendations to reduce use of OTC drugs, followed by smoking, alcohol, and illicit drugs.

SAMPLE ARTICLES RELEVANT TO COMPLIANCE WHERE DRUG ABUSE IS NOT MENTIONED

Baker, D. W., Stevens, C. D., & Brook, R. H. (1991). Patients who leave a public hospital emergency department without being seen by a physician: Causes and consequences. Journal of the American Medical Association, 266, 1085- 1090.

One hundred eighty-six patients presenting at a public hospital ER were compared with a random sample (n = 211) of patients who waited. There were no significant differences between patients who left and those who were seen in terms of chief complaint, acuity ratings, or health status. Drug use was not assessed or reported.

Bickler, C. B. (1985). Defaulted appointments in general practice. Journal of the Royal College of General Practitioners, 35, 19-22.

Missed appointments in a general private practice over a 2-month period were studied. The rate of missed appointments was 11.7%. Missed appointments were associated with particular doctors (the rates varied from 7.2% to 14.6% across eight physicians), day of the week (Mondays were lowest at 9.7%; Friday was highest at 14.9%), and length of time between scheduling and appointment date (5.2% defaults for same day, 19.1% for 14-24 day delay). The authors noted that "additional information was taken in special circumstances, for example, for patients who were temporary residents, drug addicts," but no data on the drug abusers or their rate of missing appointments were provided.

Bond, W. S., & Hussar, D. A. (1991). Detection methods and strategies for improving medication compliance. American Journal of Hospital Pharmacy, 48, 1978-1988.

This is an excellent review of the reliability and feasibility of compliance detection methods, as well as practical strategies for improving patient compliance with drug therapy. There is no mention of substance abuse as a risk factor.

Kruse, W. (1992). Patient compliance with drug treatment: New perspectives on an old problem. Clinical Investigator, 70, 163-166.

This review of patient compliance notes that it is an understudied area. The author mentions that, until recently, limited methodology of compliance measurement hampered major progress in research. No reference is made to drug abuse as a risk group.

Macharia, W. M., De Leon, G., Rowe, B. H., Stephenson, B. J., & Haynes, R. B. (1992). An overview of interventions to improve compliance with appointment keeping for medical services. Journal of the American Medical Association, 267, 1813-1817.

This study was a meta-analysis of the 23 randomized trials that provided quantitative data on the effectiveness of interventions to improve attendance at appointments for supervised administration of care. Across studies, the average rate of compliance was 58%. Interventions determined to improve compliance included mailed reminders and telephone prompts, orientation statements, contracting with patients, and physician prompts. None of the studies included involved drug abusers; two were studies of alcoholics.

Rudd, P., Byyny, R. L., Zachary, V., LoVerde, M. E., Titus, C., Mitchell, W. D., & Marshall, G. (1989). The natural history of medication compliance in a drug trial: Limitations of pill counts. Clinical Pharmacology and Therapeutics, 46, 169-176.

Investigators performed pill counts among 121 ambulatory hypertensive subjects for 1 year. Patients were selected for high compliance. Weekly pill counts indicated marked intrasubject and intersubject variability for any one medication, between medications, and over time, which was marked by averaging compliance over time. Compliance lapses appeared to be random. No specific mention was made of drug abuse.