TABLE 3.
Effectiveness of Workplace interventions for misuse of opioids and related drugs
| Outcomes | Studies | Study design | Results | Comments |
Quality |
|---|---|---|---|---|---|
| A. Intervention: employee education | |||||
| Illicit drug use | Brochu 1988 25 | Randomized‐controlled trial | Self‐reported marijuana or hashish use in the last 12 mo: Intervention 32%, Control 23% (variance = 0.05 and 0.02, respectively), t = 0.24; P > .01 | Education did not result in the reduction of illicit drug use. | Fair |
| Carpenter 2007 26 | Cross‐sectional | Self‐reported marijuana use in the last 30 d: aOR 0.791, SE 0.048, P < .01 | 21% lower odds of marijuana use. | Poor | |
| Cook 2000 28 | Randomized‐controlled trials |
Self‐reported illicit drug use: Pre‐Intervention:16 using illicit drugs Post‐test 1:5/16, McNemar test P = .02 Post‐test 2:2/9, McNemar test P = NS |
Data only presented for intervention group. Stress management education led to significant reduction in the use of illicit drugs in the short term (1 mo), but not long term (10 mo) | Poor | |
| Cook 2004 29 | Randomized‐controlled trial | Self‐reported illicit drug use in the past 30 d: Intervention 6%, Control 14% (χ 2 = 2.32, P = .128) | Education did not result in the reduction of illicit drug use | Fair | |
| Miller 2015 39 | Cross‐sectional | Self‐reported non‐medical prescription drug use in the last 30 d: aOR 0.98; 95% CI 0.85‐1.14, P = .834 | No association between education and drug misuse | Fair | |
| Patterson 2005 43 | Randomized‐controlled trial |
Likelihood to use over the counter drug to relax (Likert scale: 1‐5): Mean comparison, pre‐, and post‐intervention: Intervention 1:Pre 2.20, post 2.29; Intervention 2: Pre 2.30, post 2.15; Control: Pre 2.37, Post 2.26 ANOVA, F = 1.92, P > .05 |
Education did not result in the reduction of illicit drug use | Fair | |
| B. Intervention: drug testing | |||||
| Illicit drug use | Carpenter 2007 26 | Cross‐sectional |
Self‐reported marijuana use in the last 30 d: (AOR 0.697, SE 0.050) P < .01) |
31% lower odds of marijuana use | Poor |
| French 2004 31 | Cross‐sectional |
Any drug use: 1. Any drug testing: β = −0.31, SE 0.06, P < .01 2. Suspicion‐based: β = −0.35 SE 0.08, P < .01 3. Random: β = −0.38, SE 0.10 P < .01 |
Lower rate of illicit drug use among employees at worksites with any drug testing, random drug testing or suspicion‐based drug‐testing program | Fair | |
| Lee 2011 33 | Cross‐sectional |
Misuse of prescription pain relievers. Any drug testing: β = 0.2, SE 0.22 P = NS |
No association between drug testing and misuse of prescription pain relievers | Poor | |
| Messer 1996 38 | Retrospective cohort study |
Positive results on drug test: Non‐random drug test: Year 1 2.6%, Year 2:1.6%, Year 3 1.4%; 1.2% decline in year 3 compared to year 1. Random drug test: Year 1 2.3%, Year 2:2.1%, Year 3:1.5%, 0.8% decline in Year 3 compared to Year1 |
Introduction of random drug testing did not lead to a significant decline in positive drug tests compared to non‐random tests | Fair | |
| Miller 2015 39 | Cross‐sectional | Non‐medical prescription drug use in the last 30 d: aOR, 0.92, 95% CI 0.78‐1.07, P = .276 | No association between drug testing and drug misuse | Fair | |
| Work‐related Injuries | Feinauer 1993 30 | Retrospective cohort study |
OSHA reportable accidents over 5 y: Any Drug testing: β = −1.220, SE −0.068, t: −0.509, df: 43, P: NS) Post‐accident drug testing: β = −2.823, SE −0.225, t: −2.792, P < .01) Reasonable cause drug testing: β = −0.163, SE: −0.014, t: −0.115, P > .05 |
Post‐accident drug testing was effective in reducing workplace accidents Any drug test or reasonable cause drug testing did not reduce accident rates |
Fair |
| Lockwood 2000 35 | Interrupted time series (no control) |
OSHA reportable accidents: Pre‐employment drug test vs. pre‐employment + Random drug test. Pre‐intervention slope = 0.21 Post‐intervention slope = −0.04 Change in slope = t test = −2.70, P < .01 |
Introduction of random drug testing led to a reduction in OSHA reportable accidents | Poor | |
| Marques 2014 37 | Retrospective cohort study |
Workplace accidents: Untested employees: 47.0% Random drug test: 19.4% Adjusted P < .001 |
Employees randomly selected for drug testing were less likely to have workplace accidents following the test, compared to untested employees | Fair | |
| Messer 1996 38 | Retrospective cohort study |
Mean accidents rates/1 000 000 miles: Random drug test: 1.5%, Non‐random drug test: 1.9%, P = NS Passenger injury rates/100 000 miles: Random drug test: 3.9%, non‐random drug test: 5.2%, t (62) = 1.85, P = .045 |
A change from non‐random to random drug test led to a decline in passenger injuries, but not overall accidents | Fair | |
| Ozminkowski 2003 42 | Interrupted time series (No control) |
Regression odds of a workplace accident: aOR: −0.5856; P = .0532 |
Random drug testing led to lower accident rates, but the change was not statistically significant | Fair | |
| Schofield 2013 45 | Retrospective cohort study |
All workplace injuries: No program versus pre‐employment/post‐accident: RR = 0.85, CI = 0.72‐1.0, P = NS No program versus pre‐employment/post‐accident/random/suspicion: RR = 0.97 95% CI = 0.86‐1.10), P = NS |
Drug testing was not associated with a significant reduction in workplace injuries | Fair | |
| Waehrer 2016 50 | Cross‐sectional |
No work lost injuries: IRR 0.859, SE 0.062, P < .01 Injuries resulting in job loss: IRR 0.92, SE 0.054, P = NS |
Drug testing was associated with a reduction in injuries that did not result in loss of work, but not injuries that resulted in work loss | Fair | |
| Healthcare Cost | Morantz 2008 41 | Controlled interrupted time series |
Total worker compensation claims: aOR = −0.123, SE 029, P < .01 |
Introduction of drug testing led to a significant decline in total worker compensation claims | Fair |
| Ozminkowski 2003 42 | Interrupted time series (No control) |
Any substance abuse or related expenditure: aOR = −1.0356, P = .3504 |
Random drug testing did not lead to a reduction in substance abuse or related expenditure | Fair | |
| Productivity | Shepard 1998 46 | Cross‐sectional |
Productivity: Log sales/employee Any drug testing:regression coefficient: −0.192, SE 0.077, P < .01 Pre‐employment drug test: regression coefficient: −0.16, SE 0.082, P < .05 Random drug test: regression coefficient: −0.285, SE 127, P < .02 |
Any form of drug testing was associated with a 19% reduction in productivity. Pre‐employment and random drug testing was associated with a 16% and 29% reduction in productivity, respectively | Poor |
| C. Employee Assistant Programs | |||||
| Illicit drug use | Carpenter 2007 26 | Cross‐sectional | Self‐reported marijuana use in the last 30 d: aOR 1.01, SE 0.064, P > .05 | No association between EAP and illicit drug use | Poor |
| Miller 2015 39 | Cross‐sectional | Self‐reported non‐medical prescription drug use: aOR 0.85, 95% CI 0.72‐1.00, P = .047 | EAP was associated with 15% lower non‐medical prescription drug use | Fair | |
| Work‐related Accident | Castro 2000 27 | Historically controlled trial |
Number of Accidents: Mean number of accidents‐ Pre‐EAP: 2.22, SD 1.9 Post‐EAP: 1.0 (SD 1.32) Mean difference; −1.21 (SD 2.49), t‐value = −2.79; P = .009 |
Introduction of EAP led to a significant reduction in the number of workplace accidents | Poor |
| Waehrer 2016 50 | Cross‐sectional |
Injuries with no loss of work: IRR 0.867, SE 0.063, P < .01 Injuries with work loss: IRR 0.923, SE 0.056, P = NS |
EAP was associated with a reduction in injuries that resulted in no loss of work, but not injuries that resulted in work loss | Fair | |
| Healthcare Cost | Castro 2000 27 | Historically controlled trial |
Workers compensation claims in dollars: Pre‐EAP: 6041.17 (SD: 8705.50) Post‐EAP: 2523.59 (SD: 17 339.19), mean diff: −3517.59 (SD: 3525.04) P = .326 |
Introduction of EAP did not lead to a reduction in total worker compensation claims | Poor |
| Sweeney 1995 49 | Controlled interrupted time series |
Mental health/chemical dependency claim/costs: EAP user‐non‐user claims: n = 45 pairs, mean difference = −0.05, P = .7217 EAP user‐non‐user, cost (dollars), mean difference: n = 45 pairs, x = −26.55, P = .515 |
EAP did not result in a significant change in mental health/chemical dependency claims or costs | Fair | |
| Absenteeism | Castro 2000 27 | Historically controlled trial |
Sick leaves hours: pre‐EAP: 177.84, Post‐EAP: 64.62, diff: 113.22, SD: 417.757, P = .164 |
Introduction of EAP did not lead to a significant reduction in absenteeism due to sick leaves | Poor |
| D. Written workplace drug‐free policy | |||||
| Illicit drug use | Carpenter 2007 26 | Cross‐sectional | Self‐reported marijuana use in the last 30 d: aOR 0.697, SE 0.050, P < .01) | Written policy associated with 31% lower self‐reported marijuana use | Poor |
| Lee 2011 33 | Cross‐sectional |
Misuse of prescription pain relievers. Any drug testing: β = 0.2 (0.22) P = NS |
No association between workplace policy and misuse of prescription pain relievers | Poor | |
| Miller 2015 39 | Cross‐sectional | Self‐reported non‐medical prescription drug use: (AOR 0.85, 95% CI 0.73‐1.00, P = .045) | Written policy associated with 15% lower non‐medical prescription drug use | Fair | |
| Pidd 2016 44 | Cross‐sectional |
Use of illicit drugs in the last 12 mo. AOR, 1.0, 95% CI 0.81‐1.24, P = .98 |
No association between workplace policy and use of illicit drugs | Poor | |
| Work‐related injuries | Waehrer 2016 50 | Cross‐sectional |
No work lost injuries: IRR 1.066, SE 0.075, p = NS Injuries with work loss: IRR 1.043, SE 0.043, P = NS |
A written drug‐free workplace policy was not associated with a reduction in workplace injuries | Fair |
| E. Restructuring employee health benefits | |||||
| Healthcare cost | Sturm 2000 48 | Retrospective cohort study |
Cost of substance abuse care: Fully managed Behavioral Health organization versus cost‐sharing with workplace: Cost of out‐patient care: regression coefficient = 0.428, P < .01 Cost of in‐patient care: regression coefficient = −0.101, P = NS |
The total cost of out‐patient, but not in‐patients care was lower in organizations that fully contracted out management of substance abuse treatment to Managed Behavioral Health Organizations | Fair |
| Healthcare utilization | Lo Sasso 2004 34 | Retrospective cohort study |
Out‐patient visit utilization: Regression coefficient: −0.069, SE 0.031 P < .05 Inpatient treatment days: Regression coefficient: −0.016, SE 0.012, P < .0 |
Increase in co‐payment level was associated with a statistically significant decrease in the number of outpatient and in‐patient treatment visits | Fair |
| Sturm 2000 48 | Retrospective cohort study |
Access to substance abuse care: Fully managed Behavioral Health organization vs cost sharing with workplace: Access to care: OR = 1.13, P = NS |
No difference in access to care for employees in organizations that fully contracted out management of substance abuse treatment to Managed Behavioral Health Organizations compared to those who did not | Fair | |
| F. Combined interventions | |||||
| Illicit drug use | Gómez‐Recasens 2018 32 | Non‐randomized single‐arm study (EE + Drug testing) |
Illicit drug use, saliva drug test (Drager drug test) Baseline: 75/1103 (6.8%) Year 1:65/990 (6.6%); baseline vs Year 1, P = .332 Year 2:47/700(6.7%); baseline vs Year 2, P = .143 Year 3:43/625 (6.9%) baseline vs Year 3, P = .108 Year 1 vs Year 2: P = .039 Year 2 vs Year 3:P = .754, |
There was a significant decline in illicit drug use in year 2 compared to year 1, but not at any other time interval | Fair |
| Pidd 2016 44 | Cross‐sectional (Written workplace drug‐free policy ± drug testing) |
Self‐reported use of illicit drugs in the last 12 mo: aOR, 0.99, 95% CI 0.72‐1.36, P = .95 |
No association between workplace policy ± drug testing and use of illicit drugs | Poor | |
| Pidd 2016 44 | Cross‐sectional (written workplace drug‐free policy + EE or EAP) |
Self‐reported use of illicit drugs in the last 12 mo: aOR, 0.90, 95% CI 0.69‐1.18, P = .46 |
No association between Written workplace policy + EE or (EAP and the use of illicit drugs | Poor | |
| Pidd 2016 44 | Cross‐sectional (EE + drug testing + Written workplace drug‐free policy ± EAP) |
Self‐reported use of illicit drugs in the last 12 mo. aOR, 0.72, 95% CI 0.53‐0.98, P = .04 |
A comprehensive policy was associated with 28% lowers odds of illicit drug use | Poor | |
| Work‐related injuries | Spicer 2005 47 | Controlled Interrupted time‐series analysis (EE + EAP) |
Workplace injuries rates: aRR, 0.9984; 95% CI, 0.9975‐0.9994 |
The combined intervention led to modest (1%) but significant reduction in workplace injuries | Poor |
| Miller 2007 40 | Controlled interrupted time series (EE + EAP + Drug testing) |
Injuries: Injuries avoided: 824‐849, P = .035‐.040 |
The combined intervention led to significant reduction in workplace injuries | Fair | |
| Wickizer 2004 51 | Retrospective cohort study (EE + Drug testing + EAP + Supervisor training + Written workplace drug‐free policy) |
Injury rates per 100 person‐years (Intervention‐comparison companies): Pre‐intervention = 12.13, 95% CI 11.59‐12.67) During Intervention = 8.80, 95% CI 8.36‐9.23, P < .05), Post‐Intervention = 7.36 95% CI 6.44‐8.29, P < .05 |
Organizations that adopted the combined policy experienced a greater decline in workplace injuries (3.3/100 person years) | Fair | |
| Lockwood 1998 36 | Interrupted time‐series analysis |
Workplace accidents: Slope Pre‐intervention = −0.01 Post‐intervention = −0.01 Change in slope: t(99) = 0.03, P = .976 |
The combined program did not lead to significant reduction in workplace accidents | Fair | |
| Healthcare Cost | Lockwood 1998 36 | Interrupted time‐series analysis (EE + Drug testing + EAP + Supervisor training + Written workplace drug‐free policy) |
Health insurance claims: Slope Pre‐intervention = 3.04 Post‐intervention = 1.57 Change in slope: t(50) = −0.55, P = .59 |
The introduction of the combined intervention did not lead to a reduction in health insurance claims | Fair |
|
Miller 2007 40 (EE + EAP + drug testing) |
Controlled interrupted time series |
Injury costs avoided in 1999 (millions of $): 32.7‐33.3, P < .01 |
The combined intervention led to a reduction in the cost of workplace injuries | Fair | |
| Absenteeism | Lockwood 1998 36 |
Interrupted time‐series analysis ( EE + Drug testing + EAP + Supervisor training + Written workplace drug‐free policy) |
Absenteeism: Slope Pre‐intervention = 1.05 Post‐intervention = −0.94 Change in slope: t(61) = −1.79, P = .08 |
The combined program did not lead to a significant reduction in absenteeism | Fair |
| Productivity | Lockwood 1998 36 | interrupted time‐series analysis (EE + Drug testing + EAP + Supervisor training + Written workplace drug‐free policy) |
Productivity: Slope Pre‐intervention = 3.67 Post‐intervention = −3.04 Change in slope: t(102) = −1.06, P = .29 |
The combined program did not lead to a significant change in productivity | Fair |
Abbreviations: ANOVA, analysis of variance; aOR, adjusted odds ratio; aRR, adjusted relative risk; CI, confidence Interval; df, degrees of freedom; EAP, employee assistance program; EE, Employee education; IRR, incidence rate ratio; NS, not statistically significant; OSHA, Occupational Safety and Health Administration of the United; RR, relative risk; SD, standard deviation; SE, standard error.