Alcohol and Alcoholism Vol. 38, No. 2, pp. 183-188, 2003
© 2003 Medical Council on Alcohol
IMPACT OF AN ALCOHOL MISUSE INTERVENTION FOR HEALTH CARE WORKERS — 2: EMPLOYEE ASSISTANCE PROGRAMME UTILIZATION, ON-THE-JOB INJURIES, JOB LOSS AND HEALTH SERVICES UTILIZATION
Behavioral Health Research Center of the Southwest, 612 Encino Place NE, Albuquerque, NM 87102, USA
Received 8 April 2002; first review notified 15 October 2002; accepted 4 November 2002
 |
ABSTRACT |
|---|
 TOPFOOTNOTES ABSTRACT INTRODUCTIONMETHODSRESULTSDISCUSSIONACKNOWLEDGEMENTSREFERENCES |
|---|
— Aims: We evaluated the effects of an enhanced substance misuse (SM) prevention/early intervention programme on referrals to an employee assistance programme, health care utilization rates, on-the-job injury rates and job termination rates among health care professionals employed in a managed care organization. Methods: The intervention was implemented at one site, with the remaining sites serving as the comparison group. Existing data from hospital databases were used to compare events occurring in the periods before and after initiation of the intervention. To account for baseline differences in age, gender and job class, logistic regression models produced adjusted means for events per employee month-at-risk. Results: We found that employee assistance referrals and non-SM-related in-patient hospitalizations increased significantly post-intervention, while rates of total out-patient SM-related visits decreased at both the intervention and comparison sites post-intervention. There was a small, statistically significant decrease in the monthly rate (OR = 0.92) of non-SM out-patient utilization at the intervention site, once the intervention was in place. No differences potentially attributable to the intervention were detected in job turnover or injury rates. Conclusions: We conclude that, while the intervention did not appear to affect health care utilization for SM-related problems, it was associated with increased referrals for employee assistance.
|
INTRODUCTION |
|---|
|
TOPFOOTNOTESABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONACKNOWLEDGEMENTSREFERENCES |
|---|
Alcohol and drug misuse by health care professionals is a problem that threatens the delivery and quality of health care. Although rates of substance misuse (SM) problems are difficult to estimate, the rates are at least as prevalent as those of other professional vocations (Bissell and Haberman, 1984
). One anonymous survey of registered nurses found that the prevalence of use of all substances was 32% (Trinkoff and Storr, 1998). The American Nurses Association (1984) estimated that 6–8% of nurses may have a drug or alcohol problem. Surveys of physicians and other data suggest that the prevalence of alcohol dependence among physicians may be as high as 12% (Moore et al., 1990). They may also be less likely to seek help for SM-related problems (Aach, 1992).
Health care workers labour sometimes under extremely stressful working conditions, work long hours, and have access to controlled substances that are commonly misused (Drug Addiction in Health Care Professionals, 2002). Clearly, SM prevention and early intervention programmes are needed for this population of employees. Yet, data are limited on how many such programmes exist, and their effectiveness in preventing SM, or in intervening with substance-misusing health care professionals (Hoffmann et al., 1997).
A large managed care organization (MCO) in the southwestern United States initiated an SM prevention programme, which was enhanced through the Workplace Managed Care Cooperative Agreement initiative funded by the Substance Abuse and Mental Health Services Administration’s Center for Substance Abuse Prevention. Named Project WISE (Workplace Initiative in Substance Education), this intervention included relatively low-cost elements, such as SM awareness training, information on how to reduce drinking, and brief counselling (Lapham et al., 2000). This consecutive paper is the second of two evaluating Project WISE. The first paper described the intervention and presented results of the analysis of health risk appraisal (HRA) data (Lapham et al., 2003). The present study evaluates the effects of Project WISE on employee assistance programme (EAP) referral rates for SM-related problems, health care utilization rates, on-the-job injury rates and job termination rates.
|
METHODS |
|---|
|
TOPFOOTNOTESABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONACKNOWLEDGEMENTSREFERENCES |
|---|
The impact of Project WISE was assessed by comparing outcomes before and after the implementation of the intervention for the intervention site (the main hospital and its support services personnel) and a comparison site, consisting of the satellite facilities. Information for the analysis was obtained from four data sources. The Human Resources database provided information on employee worksite, age, gender, job classification and hire/termination dates, and covered the entire study period from 1 July 1996 to 1 July 2000. The MCO Employee Health database provided injury types and date of injury, and covered the period from 1 January 1997 to 1 July 2000. The EAP database housed EAP referrals by date and reason for referral, and covered the period of time from 1 January 1997 to 1 July 1999. The EAP provider recorded each EAP referral according to the source (self, supervisor) and reason for the referral (attention deficit disorder, anxiety, depression, family/children, grief, marital, SM evaluation/counselling, work relations). Included in the analysis were all self- or management-initiated referrals for an SM-related problem.
The MCO Patient Database, derived from the hospital billing database, provided dates and types of health care services rendered for each employee, and covered the full study period, from 1 July 1996 to 1 July 2000. The post-intervention period was from 1 September 1998 to 1 July 2000. The unit of analysis for this investigation was person-months of employment, the ‘at risk’ period for an EAP referral, injury, job termination or medical service. Medical services were classified as non-SM in-patient visits, non-SM days in which one or more out-patient services were provided, and total days of in-patient or out-patient visits for SM-related reasons. SM diagnoses ICD-9 codes included misuse of alcohol and other drugs, excluding tobacco, and were selected and grouped based on criteria provided by the Workplace Managed Care Cross-site Evaluation Team and Workplace Managed Care Steering Committee (Galvin, 2000). ICD-9 codes indicating SM-related medical conditions for all in-patient and out-patient services were as follows: 291, 292, 303–305, 357.5, 357.6, 425.5, 535.3, 571.0–571.3, 571.5, 648.3, 655.4, 655.5, 760.7 (excluding 760.74 and 760.79), 779.5, 790.3, 962.0, 965.0, 967–970, 977.0, 977.3, 980, V70.4 and V79.1. Non-SM-related utilization is defined as all visits where the primary, secondary or tertiary diagnoses did not contain any of the above ICD-9 codes. Injuries included in the analysis were limited to accidental injuries (i.e. excluding communicable diseases). All voluntary and involuntary job terminations, excluding those from a reduction in work force, were included in this analysis.
The mean number of events per person-month was calculated separately for the intervention and comparison sites before 1 September 1998, and after 1 September 1998. First, univariate comparisons were made using the Kruskal–Wallis test among medians for each of the outcome measures. These tests were based on summary measures over all person-months and were not adjusted for differences in person-months employed. Then, average monthly events were calculated separately by the following predictors: group (intervention, comparison), time (pre-, post-), the interaction between group and time, and demographic factors including age (
30, 31–40, 
41 years), gender and job class (professional/technical, executive administrative, administrative support, other). Finally, the effect of the intervention was evaluated statistically by fitting a logistic regression model for each outcome. This model compared intervention site employees, whose period of employment included the time after 1 September 1998, with all other employees, while adjusting for employment site and effects of time.
Each employee had multiple person-months, during which an event, such as an EAP referral or job termination, could occur. Data used throughout this analysis were longitudinal, and individual employee data were inter-correlated. Therefore, SAS/PROC GENMOD was used for analysis, as it allowed the use of generalized estimating equations (GEEs) for fitting repeated measures logistic regression models to each outcome measure in this analysis (Liang and Zeger, 1986). These estimation procedures provided unbiased parameter estimates and standard errors for repeated measures data. A similar analytical framework was recently used by Parthasarathy et al. (2001) to study changes in health care utilization with alcohol and drug treatment.
The mean of each outcome measure [monthly EAP referral rates, monthly injury rates, monthly risk of job termination, SM-related in-patient utilization rates, SM-related out-patient utilization rates, non-SM in-patient days, non-SM days of out-patient visits, and emergency department (ED) utilization rates] was modelled as a function of the previously-listed predictors. Log odds ratios for each predictor were adjusted for all other terms in the model. Adjusted odds ratios estimated by the analysis were interpreted as approximate relative risks, averaged over the study population.
|
RESULTS |
|---|
|
TOPFOOTNOTESABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONACKNOWLEDGEMENTSREFERENCES |
|---|
Demographic factors related to outcome measures
The per-employee EAP referral rate was highest for those in the 31–40 year age group (Table 1
). Men had slightly more SM-related EAP referrals, and executive/administrative employees had a rate of EAP referral that was more than twice all other job categories. There were no significant differences among the age, gender and job class groups for in-patient SM or out-patient SM utilization. The average number of SM in-patient days was very small overall.
|
Total means of non-SM in-patient visits were greatest for employees under 30 years, and over 40 years, of age. Women had more than double the mean utilization rate of men, while employees in the executive/administrative and administrative support jobs had slightly elevated rates of non-SM in-patient utilization. Mean out-patient visits per month increased with age. Women had a much larger average number of monthly out-patient visits, compared with men. Executive/administrative and administrative support employees shared the highest rate of non-SM out-patient visits. Utilization rates of the ED were different among the age and gender groups of employees, with the youngest age group and women having the highest rates. Women had nearly twice the injury rate of men, and employees in the ‘other’ job class had five times the injury rate of those in executive/administrative positions. The monthly hazard of job termination decreased with age, but appeared to be roughly the same for men and women. Administrative support personnel suffered the highest rates of job termination, whereas professional/technical and executive/administrative staff had the lowest hazard of job termination.
Univariate comparisons of outcomes pre- and post-intervention
Univariate comparisons revealed that average SM-related EAP referral rates per employee decreased markedly in the comparison site, but showed a slight increase in rates at the intervention site (Table 2). Individual counts of EAP referrals for all reasons decreased from 246 total referrals pre-intervention to 163 post-intervention for the intervention site and from 179 total referrals pre-intervention to 90 post-intervention for the comparison site. The percentage of referrals for SM-related reasons increased from 8% of all referrals to 14.7% for the intervention site, but decreased from 10 to 1% for the comparison site in the post-intervention period. In-patient visits for SM-related conditions did not differ between the sites after the intervention was implemented. Out-patient SM utilization decreased at both sites (P < 0.001 for the overall differences) following implementation of the intervention. Average non-SM out-patient visit rates decreased over time at both sites, but the decrease was larger at the intervention site (P < 0.001 for the overall difference among medians). Average ED visit rates decreased over time at both sites. Average monthly injury rates increased at both sites, as did the monthly job termination incidents (P < 0.001 and P < 0.05, respectively).
|
Multivariate analysis
EAP referral rate. SM-related EAP referrals increased at the intervention site after 1 September 1998 (Table 3
), compared with the same site pre-intervention and the comparison site. The estimated odds of referral among these employees were over 16 times the monthly rate of all other employees. Note the very large confidence interval for the estimate. This indicates a wide range of effect sizes for which data were consistent, and thus a considerable degree of uncertainty as to the true benefit in EAP referral rates attributable to the intervention. Employees under 41 years were 2.4–3.4 times as likely to be referred as were employees over 40 years (P < 0.05 and P < 0.01, respectively). Males were more than 2.5 times as likely to be referred to the EAP in any given month as female employees. No significant job class effects were detected in this analysis.
|
Health care utilization rates (SM in-patient, SM out-patient, non-SM in-patient, non-SM out-patient, ED visits). There were no significant predictors of SM in-patient utilization. SM out-patient utilization rates dropped after 1 September 1998 (Table 3
). Employees at either site were roughly half as likely to use SM out-patient services per month employed after 1 September 1998, but this effect was not specific to either employment site. Thus, this drop reflects secular trends in SM out-patient utilization that are in effect throughout the workforce. Administrative support personnel had roughly 1.5 times the odds of an SM utilization as professional/technical staff at either employment site.
There was a large decrease in the non-SM in-patient utilization rates after 1 September 1998 (Table 3). Employees had monthly odds of in-patient utilization that were nearly half those before 1 September 1998. Furthermore, the intervention site had lower overall rates of in-patient utilization than the comparison site (OR = 0.74, P < 0.05). However, after 1 September 1998, employees at the intervention site had increased rates by a factor of 1.5 (P < 0.05). This mitigated the overall drop in utilization seen after 1 September 1998, for both workforces. Employees in the 30 age group had a significantly higher rate of in-patient utilization (OR = 1.4, P < 0.001), while men had a significantly lower rate of utilization (OR = 0.4, P < 0.001).
There was a slight historical trend toward lower non-SM out-patient utilization rates after 1 September 1998 (OR = 0.92, P < 0.01) (Table 4). This trend was apparent at both sites. The interaction effect of period at risk by site of employment was highly significant (P < 0.01) and suggests a greater decrease in the monthly rate (OR = 0.92) of non-SM out-patient utilization at the intervention site, once the intervention was in place. Table 4 shows a significant age effect on non-SM out-patient utilization monthly rates. Employees under the age of 41 years had significantly lower rates of out-patient utilization than employees in the oldest age group. Men also had nearly half the monthly odds of out-patient utilization as women (P < 0.001). All job classes in this analysis had approximately 1.3 times the odds of monthly out-patient utilization as the professional/technical employees.
|
Employees at the intervention site had a significantly higher average rate of ED utilization than employees at the comparison site (OR = 1.4, P < 0.001) (Table 4
). However, the time period main effect and site by time period interaction effects were not significant at the 0.05 test level. Employees in the 30 years age group had slightly higher odds of ED utilization (OR = 1.13, P < 0.05) than employees in the oldest age groups, and men had somewhat lower odds of ED utilization than women (OR = 0.76, P < 0.01). Employees in the administrative support and their job classes had significantly greater odds of ED utilization than professional/technical employees.
Injury rates. The increase in the average per-capita injury rate was marked after 1 September 1998 (Table 4). The odds of injury for those employed after that date were more than 1.7 times those employed prior to the date (P < 0.01). No significant site or site by time period effects were apparent here, however. Men had significantly lower on-the-job injury rates than women (OR = 0.69, P < 0.01), and employees in the executive/administrative job classes had much lower odds of injury than those in the professional/technical job class (OR = 0.29, P < 0.001).
Job loss. Intervention site employees had a significantly lower risk of losing their jobs than comparison site employees (OR = 0.82, P < 0.001) (Table 4). However, there was no site by time effect. The hazard of job termination increased with age, and employees in the administrative support job class had significantly greater risk of losing their jobs than employees in the professional/technical fields (OR = 1.13, P < 0.001).
|
DISCUSSION |
|---|
|
TOPFOOTNOTESABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONACKNOWLEDGEMENTSREFERENCES |
|---|
It was hypothesized that, if the intervention were effective, the number of EAP referrals would increase, as supervisors became trained to recognize the signs of alcohol- and other drug-related problems and make appropriate referrals, and as employees were educated about the adverse health effects of hazardous alcohol use. Univariate analysis revealed that average EAP referral rates per employee decreased markedly in the comparison site, but showed a slightly increased rate at the intervention site. After controlling for age, gender and job classification effects, the odds of referral for the site by time period interaction showed an odds of over 16, which is highly significant statistically and indicated a significant increase in SM-related EAP referrals at the intervention site in the post-intervention period. This difference may be attributable to the intervention of Project WISE. It is also possible that the intervention may have been due to chance or to factors unrelated to the implementation of Project WISE. The average monthly rates of SM-related EAP referrals did not rise dramatically post-intervention. However, the total percentage of SM-related referrals rose from 8 to 14.7%. Yet, the statistical significance in the logistic regression analysis resulted largely from the drop in referrals at the comparison site, for which there is no clear explanation. Nevertheless, whatever was operating to reduce SM-related EAP referrals at the comparison sites apparently did not reduce SM-related referrals at the intervention site. Of interest, this drop was demonstrated before a system-wide change in the EAP, from an in-house department to an externally contracted company. EAP services were provided by one full-time staff member whose office was located in one of the comparison group sites. It seems unlikely that the differential drop in referrals at the comparison site could be explained by an impending contract change.
It was also hypothesized that the intervention may lead to increased health care utilization for SM treatment. Analysis of HRA data (see the preceding paper by Lapham et al., 2003) revealed that, among binge drinkers, the proportion of those who reported a desire to reduce drinking in the post-intervention period increased significantly. This effect was limited to those at the intervention site. Despite this, neither in-patient nor out-patient levels of service increased following the intervention. On the contrary, the out-patient utilization of SM treatment services dropped at both sites, following the intervention. If the prevention/intervention programmes were effective in persuading binge drinkers to reduce the number of days in which they drank heavily, it might have been that treatment services were not needed for these individuals. However, HRA data suggest that binge drinking rates were not affected by the Project WISE intervention. Studies have demonstrated that a large percentage of heavy drinkers reduce their alcohol intake on their own, without outside interventions (Walters, 2001). A review of brief intervention programmes in health care settings, consisting of elements similar to those of Project WISE (providing information, brief advice, self-help manuals) can reduce drinking, especially among those who are less serious problem drinkers (Babor et al., 1986; Bien et al., 1993). Other explanations for the lower utilization of SM treatment services include system-wide changes in service delivery, low penetration of the intervention and/or reluctance of employees to identify themselves as having an SM problem. The latter is supported by focus group data showing that employees were suspicious of using in-house services in the event they developed SM problems and fearful that this information may not remain confidential (S. Lapham and N. O. Lewis, in preparation).
Many health care utilization events may be the result of illness and conditions not influenced by SM or risky drinking. ED utilizations may be reduced if substance misusers received intervention services, but did not change significantly over time after the intervention was introduced to the intervention site. A small increase in the rate of out-patient visits for SM was observed at the intervention site, although this effect was not significant at the 0.05 test level. A large positive trend toward increasing in-patient utilization rates and a small negative trend toward decreasing non-SM out-patient utilization were observed at the intervention site after the introduction of the intervention (P < 0.05). We could find no likely explanation for these unanticipated findings.
Project WISE had no demonstrated effect on job losses or injury rates, as might be anticipated with effective prevention programmes. However, Project WISE was short-lived, with a post-intervention evaluation period of only 22 months. The analysis was also complicated by a number of factors. First, data censoring was an unavoidable problem in this evaluation. The complete service uses and charges for some workers were not fully observed because of staff turnover. To address this problem, we calculated person-months as an adjusting variable in fitted models. Secondly, not all data sources were available for the entire study period because data were either not collected or lost. Staff turnover, a lack of a computerized record-keeping system prior to project initiation, and misplaced hard-copy files were noted as the reasons for missing data. Therefore, certain planned analyses (e.g. biochemical indicators of SM, absenteeism) were not possible. Missing from the retrospective data pool were worker’s compensation claims, absenteeism rates and job satisfaction survey data. EAP data were only partially available from January 1997 forward, at which time a computerized tracking system was initiated. Data collection for the EAP was truncated at 1 year post-intervention due to outsourcing of services.
In summary, health care workers should have SM prevention services available, as this employee group works in high stress environments with easy access to mood-altering chemicals. The implementation of Project WISE, an SM prevention/early intervention programme designed specifically for this population, was associated with a differential increase in SM-related EAP referrals at the intervention site, but not with differential increases in the provision of treatment services for SM-related problems, job loss or injury rates. This singular, but encouraging finding merits further study.
|
ACKNOWLEDGEMENTS |
|---|
|
TOPFOOTNOTESABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONACKNOWLEDGEMENTSREFERENCES |
|---|
We thank Karen Kranich, Chris Glidden, Dr Maggie Gunter and Dr Ben Klein for their work in implementing Project WISE, all the employees who participated in the programme, Dr Ted Miller for reviewing the manuscript, and Joyce Welt for manuscript preparation. This study was funded by a grant from the Center for Substance Abuse Prevention of the Substance Abuse and Mental Health Services Administration (grant no. 5 U1K SP08152).
|
FOOTNOTES |
|---|
|
TOPFOOTNOTESABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONACKNOWLEDGEMENTSREFERENCES |
|---|
* Author to whom correspondence should be addressed.
|
REFERENCES |
|---|
|
TOPFOOTNOTESABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONACKNOWLEDGEMENTSREFERENCES |
|---|
Aach, R. D. (1992) Alcohol and other substance abuse and impairment among physicians in residency training. Annals of Internal Medicine 116, 245–254.
American Nurses Association (1984) Addictions and Psychological Dysfunctions in Nursing: The Profession’s Response to the Problem. American Nurses Publishing, Kansas City, MO.
Babor, T., Ritson, E. B. and Hodgson, R. J. (1986) Alcohol-related problems in the primary health care setting: a review of early intervention strategies. British Journal of Addiction 81, 23–46.[CrossRef][Web of Science][Medline]
Bien, T. H., Miller, W. R. and Tonigan, J. S. (1993) Brief interventions for alcohol problems: a review. Addiction 88, 315–336.[CrossRef][Web of Science][Medline]
Bissell, L. and Haberman, P. (1984) Alcoholism in the Professions. Oxford University Press, New York.
Drug Addiction in Health Care Professionals (2002) Office of Diversion Control, Liaison and Policy Section. http://www.deadiversion.usdoj.gov/pubs/brochures/drug_hc.htm
Galvin, D. (2000) Workplace managed care: collaboration for substance abuse prevention. Journal of Behavioral Health Services and Research 27, 125–130.[CrossRef]
Hoffmann, J. P., Larison, C. and Sanderson, A. (1997) An analysis of worker drug use and workplace policies and programs. Workplace Managed Resources Updates, Rockville, MD: Substance Abuse and Mental Health Services Administration, Office of Applied Statistics, A-2 Analytic Series.
Lapham, S., Chang, G. and Gregory, C. (2000) Substance abuse intervention for health care workers: a preliminary report. Journal of Behavioral Health Services and Research 27, 131–143.
Lapham, S., Gregory, C. and McMillan, G. (2003) Impact of an alcohol misuse intervention for health care workers — 1: frequency of binge drinking and desire to reduce alcohol use. Alcohol and Alcoholism 38, 176–182.
Liang, K. and Zeger, S. L. (1986) Longitudinal data analysis using generalized linear models. Biometrika 73, 13–22.
Moore, R. D., Mead, L. and Pearson, T. A. (1990) Youthful precursors of alcohol abuse in physicians. American Journal of Medicine 88, 332–336.[CrossRef][Web of Science][Medline]
Parthasarathy, S., Weisner, C., Hu, T. and Moore, C. (2001) Association of outpatient alcohol and drug treatment with health care utilization and cost: revisiting the offset hypothesis. Journal of Studies on Alcohol 62, 89–97.[Web of Science][Medline]
Trinkoff, A. M. and Storr, C. L. (1998) Substance use among nurses: differences between specialties. American Journal of Public Health 88, 581–585.
Walters, G. D. (2001) Spontaneous remission from alcohol, tobacco, and other drug abuse: seeking quantitative answers to qualitative questions. American Journal of Drug and Alcohol Abuse 26, 443–460.[CrossRef]
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||