Cancer morbidity among Danish female pharmacy technicians

HANSEN J, OLSEN Cancer morbidity among Danish female pharmacy technicians. Scand J Work Environ Health 1994;20:22-6. OBJECTIVES - Pharmacy technicians maintain a substantial production and packing of pharmaceuticals and other chemicals, many of which are carcinogens. This study re ports on cancer incidence among Danish female pharmacy assistants and dispensers. M ETHODS - Al together, 8499 compulsory members were identified in the archives of the Association of Danish Pharmacy Technicians and followed through the files of the Danish Cancer Registry (1970-1990); observed figures were compared with those expected on the basis of national cancer incidence rates. R ESULTS - The overall standardized incidence ratio [SIR) for cancer was 1.0 [N = 219, 95% confi dence interval (95% CI) 0.8-1.1). A 1.5-fold (N = 34; 95% CI 1.1-2.1) elevated risk of nonmelanoma skin cancer was found, especially for long-term pharmacy assistants (N = 15, SIR 2.8, 95% CI 1.6 4.6). An increased risk for non-Hodgkin's lymphoma appeared among long-term pharmacy dispensers (N = 5, SIR 3.7, 95% CI 1.2- 8.9). In the entire group, the risk of tobacco-related tumors was sig nificantly reduced (N = 8, SIR 0.4, 95% CI 0.2-0.9), together with the probably socioeconomic-as sociated cervical cancer risk (N = 18, SIR 0.6, 95% CI 0.4-0.9). CONCLUSIONS - Sunlight is usually the dominant cause of nonmelanoma and melanoma skin cancer, but occupational factors may have contributed in this study in view of the uncommon localization observed for many of these cancers and the unelevated melanoma risk. In addition to the increased risk of non-Hodgkin's lymphomas, which may have been associated with exposure to organic solvents, the results do not indicate any other notable cancer risks during the follow-up.

A substa ntia l number of kn own and suspec ted carcinogens are pharmaceutical s and related che micals (1) . Although the main source of human exposure to pharmaceuticals is their use in medical therapy, people involved in manufacturing and packing such agents are also exposed during long periods (2). Th e approximately 320 Danish pharmaci es ha ve a 100year tradition of compounding pharmaceutical s, and they account for ab out one-quarter of all the dru gs used in Denmark (3). Extemporaneou s preparations and a large number of reagents and chemica l mixture s are also prepared for technical purposes at pharma cie s. This work is normally maintained by pharmacy ass istants and dispensers. The aim of this study wa s to evalua te the can cer risk amo ng a histori cal cohort of pharmacy technicians.

Subjects and methods
Pharmacy technicians undergo three years of formal tra ining as apprentices at a pharmacy before the y are licensed as a pharmacy assi stant or dispenser. Th e

22
training of ph armacy dispen sers beg an around the tim e of the Second W orld War. Th e first ph armacy ass istants were licen sed in 1962. Pharmacy assistants follo w cours es in chemistry, pharmacology , and human biology , whereas the dispensers are especially qual ified in bookkeeping and business-related subjects. Both gro ups receive training in the low-scale compounding of pharmaceuticals. Afte r training, particularly the dispensers are occupied in the packing and bottling of tablets, mixtures , and ointments, whil e the as sistants normally compound pharmaceutical s and rel ated agen ts. Since 1970, all trained technici an s working in Danish pharmacies have be en compulsory member s of the Ass oci ation of Danish Pharmacy Technici an s. Onl y about I % of its member s are men , who have not been incl uded in thi s study . About 90% of the members work in pharmacies , and the remainder are employed mainl y in pharmaceutical and related industries.
Th e study popul ati on was intended to include all female pharmacy technicians licensed as of I January 1970 or later. Individual files for active and retired members were obtained from one of two register s, kept by the Union. The first, which was comput eri zed , covered all members act ive in 1985 and all subseque nt new members (N = 6415). The second, a manual regi ster based on paper cards, co mprised information on persons who had either died or were withdrawn before data we re transferred to computer files in 1985 (N =2645). Each record included the name and either a unique lO-digit personal identity number (including six digits for date of birth) or solely date of birth, along with date of license , type of final education (ie, pharmacy assistant or dispenser), date of membership, and eventual withdrawal from the Union . Present job activity was categorized as working at a pharmacy, in the pharmaceutical industry , or in nonpharmaceutical job s. An additional 39 persons were found in both the paper and computer records. The reasons for exclusion of a total of 555 individual s are shown in table I . The major reasons were training unfinished by the end of 1990 (the end of foll ow-up) (N = 317) and male gender (N = I 17). The identities of all except 360 women (4.0%) were verified in the Cent ral Population Register on the basis of the ir identit y number. Each of the 360 women whose identific ation number was invalid or missing was searched for manually in the files of the Central Population Register on the basis of date of birth, and the record was checked visually takin g into account gender and present and historic names according to a previousl y described procedure (4) . Only 45 women (0.5%) remained unidentified and were consequently excluded from the study. The 8499 identified women were searched for in the Register for date of death or emigration .
Subsequently, each woman was searched for in the files of the Danish Cancer Regi stry during the period 1 January 1970 to the end of 1990. The cancer register contains all new cases of malignant disease diagnosed in Denmark since 1943 (5). A modified version of the International Classification of Diseases, seventh revision (ICD-7) was used for classifying the diagnosis (5); from 1978 on, cancer s were also classified according to ICD-O (6), which also specifies tumor morphology. Only one skin cancer of each morphological type was recorded per person; thus, if one person was diagnosed as having two or more skin tumor s of identical histologi cal types, the person was classified as having one primar y skin tumor, multiple sites.
Since all of the women in the cohort had had three years of training in a pharmacy before obtaining their license , the period of risk was consid ered to begin three year s before that date or on I January 1970, whichever occurred later. The end of follow-up was taken as date of death, date of emigration, or 31 January 1990, for those known to be living in Denmark at the closing date. The expected number of site-specific cancers in the cohort was calculated by multiplying the personyears at risk during the follow-up period in each fiveyear age and calendar-year group by the corresp onding site-specific incidence rates in the general Danish female population. The standardized incidence ratio (SIR) was taken as the ratio between the total number of observed and the total number of expected cases (7). Their corresponding exact 95% confiden ce intervals (95% CI) were calculated (8).  The distributions of histological types and anatomic sites of the nonmelanoma skin cancers observed in the cohort were compared with those of a random sample of 680 nonmel anoma skin cancer cases in the general female population with a corresponding distribution of year of birth and diagnosis from the files of the Danish Cancer Registry.

Results
The characteristics of the cohort are given in table 2. It can be seen that about 90% of the women obtained their license in the 1960s or later and that 73% of the entire group was followed-up for more than 15 years. The cancer morbidit y experience of more than 143 000 female person-years (95 135 for pharmacy assistants and 48 669 for dispensers) was available. Of the 8499 women included , 214 developed cancer, of whom five had two tumors , yielding a total of 2 19 malignant neopla sms, with 86 among pharmacy assistants and 133 among dispensers, in the 21-year period of follow -up . The overall cancer risk was uniform (SIR 1.0). Table 3 shows the site-specific risks of the entire cohort. The risk for nonmelanoma skin cancer was sign ificantly elevated. Exclu sion of this gener ally nonfatal cancer gave a relati ve risk of 0.9 (95% CI 0.8-1.0) for all other sites combined. In the subgroups of pharmacy assistants and dispensers, the risk for nonmelanoma skin cancer was 2.0 (N = 16, 95% CI 1.2-3.1) and 1.2 (N =18,95% CI 0.7-2.0), respectively. No significant change in the risk for these skin cancers was seen according to 20 years' duration of union membership, or for the equivalent time since first employment at a pharmacy (table 4). For women being sole-Scand J Work Environ Health 1994, vo l 20, no I   cinomas (N =3) was obse rved for pharmacy techn icians than for the random sample of cases in the general population (table 5), and 27% of the nonmelan orna skin cancers among the union members were multiple, compared with only 16% among the population sample. For the specified sites, a significant lower proportion of nonmelanoma skin cancers appeared for the head locality than in the population sample. Nonsignificantl y ele vated risks were found for Hodgkin' s disease among the pharmacy assistants (SIR 1.7, N = 4) and for non-Hodgkin' s lymphoma among the pharmacy dispensers (SIR 2.4, N = 5).
When a duratio n of at least 15 years of work as a pharmacy dispenser was taken into account, the SIR for non-Hodgkin' s lymphoma was 3.7 (N =5,95% CI 1. 2-8.9). A significant deficit was observ ed for cervical cancer in the combi ned group (SIR 0.6, N = 5), on the basis of equally low risks in the two groups of members. The risk for cancers (of the buccal cavity , phary nx, larynx, lung and bladder) related to tobacco smoki ng was also significantly decrease d (N = 8, SIR 0.4 , 95% CI 0.2-0.9).

Discussion
The pharmacy technicians had a 53% significantly higher risk for nonmelanoma skin cancer than the general population. The risk appeared to be conce ntrated in the subgroup of pharmacy assistants (SIR 2.0, N = 18), especially among those with stable employment at a pharmacy (SIR 2.8, N = 15). Exposure to sunlight is the major risk fact or for both melanoma and nonmelanoma skin cancer (9), but the topographica l and morphologica l patterns in our gro up indicate that other factors may also be involved. First, the risk for malignant melanoma was not elevated, indicating that exposure to sunlight had not been much more freque nt for the pharmacy technicians than for the general population. Seco nd, the nonmelano ma skin cancers among the pharmacy technicians were not more frequently distributed for sites habitually expose d to sunlig ht (ie, face, scalp , and neck). Although we observed a higher relative frequency of squamous-cell carc inomas, which are more strongly associa ted with sun expos ure than basal-ceil carcin omas ( 10), it was nonsignificant and based on only three cases. Furthermore, nonmelanoma skin cancer is one of the major occupational cance rs, and the International Agency for Research on Cancer has recently reported that 17 pharmaceuticals may cause skin ca ncer in humans (1). Among these, espec ially arsen ic salts and coa l tars were widely used in Danish pharmacies up until the 1980s (3). An elevated risk of nonmelanoma skin cancer has, to our knowledge, not previously been described among employ ees handling pharmaceutica ls. Most occupational cancer studies are, however, based on mortality data, and, since this disease in general is Table 4. Risk o f nonmelanoma skin cancer (1970 -1990), according to du ration and time si nce start of membership in t he Assoc iation of Danish Pharmacy Technicians. (0 = observed , SIR = standard ized in ci dence ratio , 95% CI = 95% confiden ce in terva l)  nonfatal, it might have been ove rlooked . Finally, the topograp hica l distribution of nonmelanoma skin cancers in our study was similar to that among Danish patients with psoriasis trea ted with various carcinogenic pharmaceuticals (I I), which may indicate common causal facto rs (eg, coa l tars and arsenic compounds) . The five cases of non-Hodgkin 's lymphoma were all found in the subgro up of pharmacy dispensers, who had a sign ifica nt, almo st fourfo ld increa se in their stan dardize d incidence ratio when a latency of at least 20 years of work was co nsidered. This risk is in acco rdance with the results of studie s of workers exposed to organ ic solvents, health professionals, and patients exposed to pharmace utica ls (12)(13)(14)(15)(16). The fact that none of the cases was found among pharmacy assis tants might be due to fact that the latency of around 20 years had not yet been achieved for abou t two-thirds of this group .
Follow-up was completed for 99.5 % of the cohort mem bers; therefore the selection bias was reduced to a minimum. It has been estimated that the Dan-ish Cancer Register covers about 98% of all malignant neop lasms (17). However, some genera l underreporting of nonmelanoma skin cancer may exist , owing to the normally nonlethal charac ter of the disease ( 18), but the interpretation of the relative distribution of topography and morphology would not be affected.
National rates were used as refere nce in this study because the numbers of employees in Danis h pharmacies are similar to local population densities and therefore bias cau sed by geog rap hic variations in risks may have been diminished . Social differences may exi st between pharmacy technicians and the general popu latio n, but skin cancer is, in genera l, unaffected by social class (19). However, a slig ht overreporting of skin cancer may have occurred among the pharm acy technicians, since health professionals may pay more attenti on to skin diseases. Risk was cons idered to begin three years before licensing. Since person-years before the coho rt-defining even t generally tends to be rather free of risk. some minor diluti on of the risk may have occ urred .
Scand J Work Environ Health 1994, vo120, no I More appropriate health care in thi s occupa tional gro up than in the general population may also have reduced ca nce r ris ks. For instance a lower smo ki ng freque ncy in thi s populat ion has bee n repo rted previously (20) and has been further corroborated by the finding of a significantly decreased risk for smoking-related tumors; a correspondingly low risk for lung cancer has previously been reported for similar groups of health care per sonnel (2 1). Th e signific antly redu ced cervical can cer ris k observed in this study is al so fo und among women with similar jobs (22) and ma y likewi se reflect a mor e healthy life -style.
The fa ct that pharmacy di sp ensers and ass istants ha ve different tasks and are co nsequently expos ed to different le vel s of vari ou s carcinogens ma y explain their diffe rent le vel s of ris k. Althou gh work as a pharmacy technician implies possible exposures to carcinogen ic chemicals, no information abo ut specific jobs or exposures was available. However , some pharmacy technicians mainly have counter and administrative tas ks and are ther efore not usu ally exposed to ca rci nog e ns . Th is frac tion ma y ha ve diluted the risk of cancer for the truly exp osed . Finally, mo st of the cohort members, es peci ally the pharmacy as sist ants, were relativel y young at the end of the folJo w-up. Sin ce most can cers occur in olde r people, an y tum or s already ini tiated in the ph arm ac y technici ans will not be observed unti l some time in the future. Further follow-up and individual exposure assessments are needed to clarify the cancer risks of this group.