Relation of the Pre-employment Drug Testing Result to Employment Status: A One-year Follow-up

DAVID CHARLES PARISH, MD, MPH"

AH employees h i red over a s ix -month p e r i o d a t a large hos- pital underwen t pre -employment u r i n a r y toxicology screening. Resalts o f the screening were kep t con f iden t ia l Af ter a y e a r o f employmeng the p e r s o n n e l f o l d e r s o f all employees s tudied were reviewed. Twenty-two o f 180 em- p loyees (12%) had tested pos i t i ve f o r drug use. Employees in c ler ica l /a ide pos i t i ons were s igni f icantly more likely to test pos i t i ve than were employees in p ro f e s s iona l p o s i t i o n s (1796 vs. 696). Drug-posi t ive employees were also more l ikely to be y o u n g a n d male. Compar ison o f Job per fo rm- a n t e variables, j o b retention, superv i sor evaluations, a n d reasons f o r t erminat ion showed no difference between drug-posi t ive a n d drug-negat ive employees. Eleven drug- negative employees were f i r ed dur ing the study; n o drug- pos i t i ve employee was f i red . There was a s trongly signifi- can t di f ference between clerical a n d p r o f e s s iona l employees on each o f these variables. This s tudy d id no t

f i n d a relat ion between drug use a n d j o b per formance . The widespread use o f drug screening p r i o r to employment makes f u r t h e r studies o f this issue importartL Key words: pre -employment drug testing; u r i n a r y toxicology; screen- ing. J GEN INTERN MED 1989; 4:44--47.

HOSPITALS AND OTHER EMPLOYERS have uti l ized urinary toxicology screens of potential employees for many years. The legality and validity of such testing have been questioned. ~ Federal employees have chal lenged a move to have all employees undergo screening; the court decisions in this case have found urinary drug screening in the absence of probable cause to be an invasion of privacy. The social value of screening pro- grams has been the subject of reviews and editorials 2, 3 that raise questions regarding the accuracy of screen- ing, the failure to include legal but abusable drugs such as alcohol, and the ethics of making employment deci- sions based on non-work-related behavior.

The research li terature on drug screening focuses on technical aspects of screening. No studies could be found addressing the basic quest ion of whe ther drug usage correlates with impairment of job performance. To assess the relationship between positive urinary tox- icology and job performance, all persons hired over a six-month per iod at a hospital were screened. This re- port compares the job performance characteristics of employees who tested positive on the urinary toxicol- ogy screens with those of employees who tested negative.

*Assistant Professor. Department of Internal Medicine. Mercer Uni- versity School of Medicine. 857 Orange Terrace, Macon. Georgia 31201.

Address correspondence and reprint requests to Dr. Parish.

44

STUDY POPULATION AND METHODS

Employees hired at a large teaching hospital in a moderate-sized urban area (SMSA < 1 ,000,000) during a six-month per iod underwent standard urinary toxi- cology screening during their pre-employment physi- cal. Prospective employees were notified by the per- sonnel department during the interview process that screening would be a part of their physical examination and that the results would be used as part of a s tudy of the usefulness of urinary drug testing. All employees hired during this six-month per iod were tested. No at- tempt was made to see whether the screening program caused potential employees to drop out of the inter- v iew process. No employee refused screening.

The program was begun in response to a request from administration. Following review of the available li terature and extensive discussion, an agreement was made to study the yield and usefulness of screening. The screening program was discont inued after the six- month study period.

One hundred ninety-two employees underwent urinary toxicology screening. Samples were obtained in the Employee Health Clinic during the pre-employ- ment physical examination. No at tempt was made to direct ly observe specimen collection. The restroom used for col lect ion is less than 10 feet from the exami- nation room and the nurse 's station; no at tempt to sub- mit a cold specimen or a di luted specimen was de- tected. Specimens were coded using a random number system and sent to the laboratory for analysis. Results were repor ted by the laboratory using only the random numbers assigned.

Urinary toxicology was performed under the su- pervision of a clinical chemist employed full-time by the laboratory. Protocols for handling specimens found positive on an initial screen included initial confirma- tion within the laboratory by a separate methodology and submission of all confirmed positive tests to a refer- ence laboratory for independent corroboration. Speci- mens were tested for the fol lowing substances: barbitu- rates, opiates, benzodiazepenes , p ropoxyphene , meperidine, THC (marijuana), amphetamines, co- caine, phencycl idine, and phenothiazines. Table 1 shows the methods used for screening and confirma- tion.4.5 The use of internal confirmation prior to sub- mission of samples to the reference laboratory resulted in no unconfirmed positives. False-negative results were not sought.

JOURNAL OF GENERAL INTERNAL MEDICINE, Volume 4 (Jan/Feb), 1989

TABLE 1

Analytic Methods Used for Drug Screening and Confirmation*

4S

Drug(s) Screening Internal Reference Tested for Method Confirmation Laboratory

THC (marijuana) EMIT®-dau TLC GC/MS Opiates EMIT®-dau TLC GC or GC/MS Meperidine TLC TLC* GC/MS Propoxypene TLC TLC, metabollte GC/MS Benzodiazepenes EMIT®-dau TLC GC or GC/MS Cocaine and metabolltet EMIT®-dau TLC GC or GC/MS Barbiturates TLC GC GC or GC/MS Amphetamines TLC TLC GC or GC/MS Phenothiazines TLC TLC, metabolite GC or GC/MS PCP EMIT®-dau TLC GC/MS

*Using internal confirmation prior to sending specimens to the reference laboratory resulted in no uncomfirmed positives. TLC = thin-layer chromatography; EMIT = enzyme-multiplied immunoassay technique; GC = gas chromatography; MS = mass spectrophotometry.

tCocaine screen done by TLC, metabollte by EMIT. tAlternate method for TLC utilized (HPTLC).

Characteristics of job performance were assessed by review of personnel files of study employees. The personnel department reviewed the files and extracted information on job evaluation, disciplinary actions, promotions, commendations, termination of employ- ment, and absenteeism. The reviewer had no access to information concerning the individual employee's drug screening status. Summary information about each employee was forwarded to the author for compilation of results.

Each job was characterized as clerical/aide or pro- fessional based on the degree of training required for the job and amount of independent decision making required to perform the job. The categorization was made without knowledge of any individual's drug test- ing status.

Statistical calculations were performed using the SAS package (1985) run on a PRIME 50 computer. 6 Comparisons were made using the chi-square statistic. Several subgroups included cell sizes too small to allow valid calculations to be performed.

RESULTS

One hundred ninety-two sequentially hired em- ployees underwent urinary toxicology screening. Twelve employees were excluded from the study. Six had no results reported by the laboratory, four were excluded because of two instances of assignment of the same random number to two employees, and two never reported to work and therefore had no employee folder available to review.

The fi~al study population was 180 employees. Twenty-two employees tested positive for drug usage, representing 12% of the study group. Drugs identified were THC (marijuana) (14), propoxyphene (3), barbi- turates (3), phenothiazine (1), and multiple drugs (phenothiazine, barbiturates, codeine, and propoxy- phene) (1).

Demographic characteristics are shown in Table 2.

The employees were representative of the racial makeup of the community at large. The high percent- age of women in the study group (73%) reflects the high percentage of total hospital employees in female- dominated job categories (e.g., nursing, ward clerk, nurse's aide). The new employees tended to be young.

Employees testing positive were more likely to be young and male; this tendency is supported by popula- tion surveys of drug usage. 7 Test-positive employees were significantly more likely to be in clerical/aide positions (chi square = 4.33, p = 0.0037). This find- ing is also in keeping with population surveys. 8

Table 3 shows comparisons of the employees on the various job-related criteria assessed during the study. Comparison of job retention at one year revealed a higher turnover rate among drug-positive employees

TABLE 2

Demographic Characteristics and Job Categories by Drug-testing Status in 180 Employees

Drug-testing Status

Positive Negative No. (%) No. (%) Total

Age (years) <30 16 (14) 96 (86) 112 (62) >30 6 (9) 62 (91) 68 (38)

Race White 16 (13) 111 (87) 127 (71 ) Black 6 (11) 47 (89) 53 (29)

Sex Male 9 (19) 39 (81) 48 (27) Female 13 (10) 119 (90) 132 (73)

Job category* Professional 5 (6) 73 (94) 78 (43) Clerical/aide 17 (17) 85 (83) 102 (57)

TOTAL 22 (12) 158 (88) 180 (100)

*p = 0.04 comparing professional and clerical workers.

4 6 Parish, eRE-EMPLOYMENT DRUG TESTING

TABLE 3

Job Retention and Job Performance Characteristics of 180 Employees

Subgroups by Job Category and Testing Status Testing Status

Professional Clerical/Aide Totals Job Category

Totals

Drug Drug Drug Drug Drug- Drug- Profes- Clerical/ Positive Negative Positive Negative positive negative sional Aide (n = 5) (n = 73) (n = 17) (n = 85) (n = 22) (n = 158) (n = 78) (n = 102) Total

Job retention Employed >1 year 2 (40) 57 (78)* 10 (59) 45 (53) 12 (55) 102 (65) 59 (76) 55 (54)* 114 (63)

Resigned With notice 2 (40) 12 ( 16)* 3 (18) 16 ( 19)* 5 (23) 28 ( 18)* 14 (18) 19 ( 19)t 33 (18) Without notice 1 (20) 3 (4) 4 (24) 14 (16) 5 (23) 17 (11 ) 4 (5) 18 (18) 22 (12)

Fired 0 ( - - ) 1 (1) 0 ( - - ) 10(12) 0 ( - - ) 11 (7) 1 (1) 10(10) 11 (6)

Total employed <1 year 3 (60) 16 (22) 7 (41) 40 (47) 10 (45) 56 (35) 19 (24) 47 (46) 66 (37)

Supervisor evaluation None 4 (80) 28 (38)* 10 (59) 55 (65) 14 (64) 83 (53) 32 (41) 65 (64)* 97 (54) Poor 0 ( - - ) I ( I) 2(12) 3 (4) 2 (9) 4 (3) I ( I) 5 (5) 6 (3) Good 0 ( - - ) 17 (23) 4 (24) 19 (22) 4 (18) 36 (23) 17 (22) 23 (23) 40 (22) Excellent I (20) 27(37) I (6) 8 (9) 2 (9) 35(22) 28(36) 9 (9) 37(21)

Other entries in personnel file

Commendations 0 ( - - ) 10(14)* 2(12) 10(12) 2 (9) 20(13) 10(13) 12(12)* 22(12) Warnings 0 ( - - ) 5 (7) 9 (53) 34 (40) 9 (41) 39 (25) 5 (6) 43 (42) 48 (27) No entry 5 (100) 58 (79) 6 (35) 41 (48) 11 (50) 99 (63) 63 (81) 47 (46) 110 (61)

*X 2 not calculated due to small cell size. tp = 0.04 comparing professional and clerical workers. $p < 0.001 comparing professional and clerical workers.

(45%) than among drug-negative employees (35%). This difference was not significant. There was a strongly significant difference in the turnover rate of clerical (46%) versus professional employees (26%) (chi square = 8.98, p = 0.003). Clerical employees who tested positive for drugs had a lower turnover rate (41%) than did clerical employees who tested negative (47%).

There was a strongly significant difference in the qualities of evaluation received by clerical and profes- sional employees (chi square= 21.74, p < 0 . 0 0 1 ) . Professional employees were much more likely to re- ceive favorable evaluations (36%) than were clerical employees (9%). A nonsignificant trend toward better evaluation among drug-negative employees as a total group resulted from the higher proportion of drug-pos- itive employees in the clerical/aide category. Compari- son of only clerical/aide employees showed no differ- ence in the qualities of evaluation. The number of professional employees testing positive for drug use was too small to allow valid comparisons.

A comparison of other markers of job performance --commendations, warnings, promotions, and p roba t ion- - showed a highly significant difference in likelihoods of receiving a warning between clerical/ aide employees (42%) and professional employees (6%) (chi square = 29.92, p < 0 . 0 0 1 ) . There was a nonsignificant increase in the likelihood of receiving a warning among drug-positive (41%) compared with

drug-negative (25%) employees. A comparison of drug-positive and drug-negative clerical/aide em- ployees showed no difference. The warnings given were of several types. Many were for minor infractions related to arriving at work a few minutes early or late. A few were for more serious behavior such as rudeness, poor job performance, and sleeping on the job. None of the warnings given drug-positive employees were for serious infractions. Clerical employees were signifi- cantly more likely to be fired or to leave without notice than were professional employees (chi square = 6.38, p = 0.043). Eleven drug-negative employees were fired during the study; no drug-positive employee was fired. Reasons given for leaving were similar among drug-positive and drug-negative employees. However, most groups were too small for valid statistical compar- isons to be made.

DISCUSSION

Drug use is common in the United States. Given the high frequency of use of drugs in the population, it is certain that hospitals and other major employers will have significant numbers of employees who use drugs.

Drug use on the job is generally agreed to be detri- mental to performance. Tests of motor and intellectual function while intoxicated support this perception. Drug addiction is frequently associated with impair- ment of job performance, usually related to family and economic stresses caused by addiction.

JOURNAL OF GENERAL INTERNAL MEDICINE, Volume 4 (Jan/Feb), 1989 47

The relationship of drug use off the job to overall job performance is a subject of great heat and little light. There are few population surveys of drug use among workers; no studies comparing the job perform- ances of non-job-related drug users with those of com- plete abstainers were found. While drug abusers con- stitute a subset of the population of drug users, the relationship between use and abuse is complicated and poorly understood. Pre-employment drug screening programs presume that drug users are less desirable employees than non-drug users. The National Institute on Drug Abuse estimated in 1986 that 40% of Fortune 500 companies use some form of urinary drug screen- ing of employees. 9 However, only three institutions have reported the yields of their screening programs. One study was done by the Employee Health Service at Columbia Presbyterian Hospital in New York in 1981 lo; the others were done in private industry. H All yielded such low percentages of positive screens that the programs were discontinued. No known study prior to this report has hired all individuals tested to compare job performances. We found no significant difference between individuals who tested positive and those who tested negative using broad categories of job perform- ance. The categories were drawn broadly to ensure con- fidentiality; review of routinely generated personnel data prevented any special handling of the group being studied.

Further studies of the relation of drug use to job performance are needed. The size of the drug-positive group in this study was clearly not large enough to detect subtle differences in job performance or reten-

tion. Other institutions should attempt to replicate this study and expand the range of job characteristics stud- ied. If further studies show impairment in performance with certain drugs or in certain jobs, techniques of identifying relevant drug use could be supported. If no relation between drug use and job performance can be found, we should transfer the money now being spent on drug screening to other programs. The data pres- ently available do not support the thesis that positive drug screening predicts poor job performance.

REFERENCES 1. Stille A. Drug testing. Nat Law J 1986;8:29:1,22-4. 2. Rosenstock L, Cullen MR. Routine urine testing for evidence of drug

abuse in workers: the scientific, ethical, and legal reasons notto do it. J Gen Intern Med 1987;2:135-7.

3. Lundberg GD. Mandatory unindicated urine drug screening: still chemical McCarthyism. JAMA 1986;256:3003-5.

4. Sutherimer CA, Yarborough R, Hepler BR, Sunshine I. Detection and confirmation of urinary cannabinoids. J Analyt Toxicol 1985; 9:156-60.

5. Jones DW, Adams D, Martel PA, Rousseau RJ. Drug population in one thousand geographically distributed urine specimens. J Analyt Toxi- col 1985;9:125-9.

6. SAS user's guide. Version 5, edition period. SAS Institute, Inc., Carey, NC, 1985.

7. Raveis VH, Kandel DB. Changes in drug behavior from the middle to late twenties. Initiation, persistence and cessation of use. Am J Public Health 1987;77:607-11.

8. Miller JD, ed. National survey on drug abuse: main findings 1982. Rockville, MD, National Institute on Drug Abuse, 1983.

9. Hoyt DW, Finnigan RE, Nee T, Shults TF, Butler TJ. Drug testing in the workplace--are methods legally defensible. JAMA 1987; 258:504-9.

10. Lewy R. Pre-employrnent qualitative urine toxicology screening. J Occup Med 1983;25:8:579-80.

11. HilkerRRH.AsmaFE, DaghestaniAN.RossRLAdrugabuserehabili- ration program. J Occup Med 1975; 17:351-4.