METHODS

Identification and selection of published articles

An attempt was made to collect every article published in English in the scientific literature that contained any primary data on occupational risk factors for Lyme borreliosis. The following databases were searched: Medline, NIOSHTIC (NIOSHTIC was produced by the National Institute of Occupational Safety and Health (NIOSH) and was distributed as OSH-ROM; the July 1998 version, from Silverplatter Information, Norwood, MA, was used), Science Citation Index, and the EUCALB (European Union Concerted Action on Lyme Borreliosis) website. A general search of Medline in October 1999 conducted on “keyword = Lyme disease” found 4917 published articles. In an effort to select only those articles with occupational relevance, “keyword = Lyme disease” was then combined with multiple other keywords including: agriculture (n=3 articles), work(er) (n=67), employee (n=4), outdoor (n=28), occupation(al exposure) (n=9), park (n=58), exposure (n=187), military (n=18), vaccine (n=220), forest(er) (n=66) and farm(er) (n=11). A general search of NIOSHTIC for “Lyme” also yielded 25 articles for potential review. Additionally, all of the articles referenced on the EUCALB website as of January 2000 (n=260) were assessed for occupational relevance. Finally, the Science Citation Index was used to cross reference any article that had cited any of the following authors in its references: Arteaga F, Christiann F, Fahrer H, Gustafson R, Guy E, Kuiper H, Nuti M, Rath P, Santino I, Schwartz B, and Zhioua E. These authors were selected because they were most often cited in articles of occupational risk of Lyme disease.

Each database yielded multiple articles, many of which were cross referenced between and within databases. Through these techniques, 91 unique articles were preliminarily reviewed for possible relevance to the occupational risk of Lyme disease. Only articles which contained primary data on occupational populations were considered for further analysis. Data were considered primary if they were collected by the researchers and referenced to an occupational population or exposure. The 50 rejected papers made no connection between the study data and an occupational population or exposure. Ultimately, 41 unique articles with primary data on occupational Lyme borreliosis were selected for detailed analysis, representing a broad range of geographic locations and 36 unique authors (table 1).^7,9,17–55 One additional article by Zhioua et al,⁵⁶ was identified but was not reviewed separately because the data were derived from another included article (Fahrer et al).⁴¹ Control data for Christiann et al were obtained from a subsequent publication by Christiann et al.^21,57

Evaluation of use of articles

Initial criteria were established to categorise articles according to their use in assessing the occupational risk of Lyme disease (table 2). In developing these criteria, the primary motivation was the recent approval of a vaccine to prevent Lyme disease by the Food and Drug Administration and its impact on the occupational health care environment.¹ Current Occupational Safety and Health Administration requirements for vaccination of workers are limited to vaccination for hepatitis B virus in its bloodborne pathogens standard (29 CFR Part 1910.1030). This standard is designed to protect workers from a symptomatic, often serious, and sometimes fatal, illness. In evaluating the occupational risk of Lyme disease, similar attention was given to the morbidity associated with Lyme disease. Risk of symptomatic, clinically and laboratory confirmed infection was thus the primary outcome of interest.

Five major factors were evaluated in defining the use of the published articles for assessing the occupational risk of Lyme disease (table 2):

1. Study design - prospective studies had higher utility than did case-control or cross-sectional studies;

2. The occupational group had to be clearly defined;

3. A comparison group without occupational risk had to be included;

4. Definition of Lyme disease, with symptomatic, clinically confirmed infection of primary interest, by contrast with definition of cases on the basis of self reported symptoms ascertained by questionnaire or asymptomatic seropositivity only;

5. Laboratory confirmation had to be included, with western blot and the definition of the Centers for Disease Control considered to be better documentation than enzyme linked immunosorbent assay (ELISA) or indirect fluorescent antibody (IFA) only.⁵⁸ No previous studies of the occupational epidemiology of Lyme disease used more rigorous definitions of laboratory confirmation—such as culture or polymerase chain reaction (PCR) of skin tissue from erythema migrans lesions.

Measurement of occupational risk

After application of the criteria of usefulness to the 41 published articles with primary data on the occupational epidemiology of Lyme disease, two were defined as of high use, eight as moderate, 14 as low, and 17 had no use (table 3). For the 10 studies in the high or moderate categories, three epidemiological effect measures are reported and calculated, when possible, from the published data. These were, in order of least useful to most useful measure:

(1) Seroprevalence (odds) ratio—the prevalence of positive serological test results in the defined occupational group under study compared with the prevalence in a defined control group without occupational risk;

(2) Clinical prevalence (odds) ratio—the prevalence of symptomatic, clinically confirmed Lyme disease with laboratory confirmation in the defined occupational group compared with the prevalence in the control group;

(3) Clinical incidence ratio—the incidence of symptomatic, clinically confirmed Lyme disease with laboratory confirmation in the defined occupational group compared with the incidence in the control group. When calculated by the author of the published study, these data were reported directly. When sufficient data were included in the article to calculate a desired epidemiological effect measure, but not reported by the author, the effect measure was calculated using Intercooled Stata^TM, version 6 (Stata Corporation, College Station, TX).

RESULTS

Descriptive summary of published literature

The published literature on the occupational epidemiology of Lyme disease included 41 articles with a mean (SD) occupational group sample size of 367 (378) and a mean (SD) comparison group sample size of 196 (333) (table 4). Of these studies, two (4.9%) were conducted (or initiated) before 1985; 23 (56.1%) were conducted from 1986 to 1990; 10 (24.3%) were conducted from 1991 to 1995; and six (14.6%) were conducted from 1996 to 1999. A total of 14 (34.2%) used a prospective study design (including one study with a retrospective cohort design, by Bowen et al (1984)⁴⁸); 20 (48.8%) were cross sectional in design; four (9.8%) were case-control studies; and three (7.3%) were case series. Sixteen (39%) studies presented data only for occupationally exposed subjects without reference to a control group, and 25 studies (61%) presented data for both occupational and comparison groups.

The definition of Lyme disease varied among the studies. Twenty three (56.1%) studies solely relied on serological assessment. Sixteen (39.0%) studies used both serological assessment and clinical evaluation by a physician and two (4.9%) studies used only clinical evaluation by a physician without any laboratory confirmation. Of the 39 studies with serological data, 15 (38.5%) confirmed positive results by ELISA or IFA with western blot.

Evaluation of occupational risk in studies of high and moderate use

Of the 10 studies with high and moderate use, seven reported the prevalence of seropositivity among the occupational group under study, which ranged from 1.0% to 44.8%. When compared with the seroprevalence among controls, six of the seven studies documented a significantly increased relative odds of seropositivity ranging from 3.7 to 9.9 (table 5), whereas the association found in one study (Zhioua et al²²) did not reach statistical significance. However, only two of the seven studies (Kuiper et al^31,32) matched the comparison subjects to the occupationally exposed subjects for age and residence. The other studies did not reach an important goal of the comparison group, which would be the ability to separate occupational risk from residential and recreational risk, because it seemed that comparison subjects also had a lower risk of residential and recreational tick exposure. Of these seven studies, only two studies (Kuiper et al^31,32) performed western blots as confirmatory evidence in subjects with positive ELISA or IFA test results according to current standards recommended by the Centers for Disease Control.⁵⁸ It should be noted that all seven studies were performed in Europe where confirmatory western blots are not standardised and four of the seven studies predated the 1995 Centers for Disease Control criteria for Lyme disease serological testing.

Of the 10 studies with high and moderate use, six presented data on the prevalence of clinically confirmed Lyme disease in the occupational group under study, ranging from 0.4% to 6.0%. Many of these studies reported lifetime prevalence—that is, diagnosis of disease compatible with Lyme disease at any time in the past. However, only two (Chmielewska-Badora³⁶ and Gustafson et al³⁹) reported prevalence of symptomatic Lyme disease among the occupational group under study and a comparison group. Chmielewska-Badora³⁹ documented one case of Lyme disease among 261 farmers and foresters compared with no cases in 50 blood donor controls (odds ratio (OR) not calculable). Gustafson et al³⁹ documented a lifetime history of 17 definite and five probable cases of Lyme disease among 362 orienteers compared with two cases of Lyme disease among 150 controls (OR=4.8, 95% confidence interval (95% CI) 1.1 to 20.0). The conclusion of definite disease was based on a past diagnosis of erythema migrans made by a physician.

The highest quality epidemiological measure of effect in this review is that of incidence of symptomatic, clinically confirmed disease. Of the 10 studies with high and moderate use, five reported annual incidence of symptomatic, clinically confirmed Lyme disease in the occupational groups under study. Vos et al³⁵ reported both the clinical incidence and seroconversion rates. Overall, relatively few new cases of occupationally acquired Lyme disease were documented in these incidence studies. Both Kuiper et al³¹ and Vos et al³⁵ found no new cases of symptomatic Lyme disease during their follow up period. Vos et al³⁵ found a risk of seroconversion among the occupational group that was lower than that of the comparison group, although the relative risk did not reach significance. Fahrer et al^40,41 documented 15 new cases of Lyme disease, equivalent to an annual incidence of 0.8% over the mean follow up interval of 6.5 years. This annual incidence was no different than that found among the non-exposed comparison group. Only Bowen et al⁴⁸ reported an increased 2 year cumulative incidence of clinical Lyme disease in an outdoor occupational group (3.8%), compared with an indoor occupational group (0.8%), with a relative risk (95% CI) of 4.9 (1.9 to 12.6). These cases were collected retrospectively from among cases reported to the state of New Jersey in 1981 and 1982; the cases that worked at the Naval Weapons Station in Monmouth County were then categorised by indoor and outdoor work.

An important and consistent finding among the studies was that most subjects found to be seropositive, in either prevalence or incidence studies, had no current or past symptoms compatible with Lyme disease. For example, among the studies with high and moderate use, Kuiper et al,^31–32 Chmielewska-Badora,³⁶ Hauser et al,²³ and Zhioua et al²² reported that 82%, 83%, 99%, 100%, and 100% of seropositive subjects were asymptomatic, respectively. Similarly, Vos et al³⁵ and Fahrer et al⁴⁰ reported that 93% and 98% of people with incident seroconversion had no symptoms consistent with Lyme disease.

DISCUSSION

The goal of this exercise was to find if outdoor workers who live, pursue leisure activities, and work in areas endemic for Lyme disease have an increased risk of the disease compared with people who only live and pursue leisure activities in those same areas. Although Lyme disease would seem to be an obvious risk of outdoor work, the scientific literature has not clearly documented the magnitude of the risk of symptomatic, clinically confirmed disease. Many studies have documented the occurrence of Lyme disease in outdoor workers, but few have attempted to document the risk of confirmed, symptomatic disease comparing occupationally exposed people and controls using the current standards of diagnosis of the disease. Most published studies with primary occupational data were cross sectional in design and documented an increased risk of seropositivity among workers compared with controls. However, the use of these studies in guiding decisions about the need to vaccinate outdoor workers in the United States is limited by: (a) lack of clinical evaluation of subjects; (b) reliance on serological evaluation with ELISA or IFA only, without confirmation by western blot⁵⁸; (c) inadequate assessment of occupational exposure to ticks—for example, lack of measurement of the number of outdoor hours worked or years of service, lack of assessment of protective behaviour or clothing, and failure to adjust for non-occupational exposures; and (d) a predominance of data from European studies, where Borrelia burgdorferi strain differences may alter the clinical presentation of Lyme disease.

In this evaluation of the scientific literature on the occupational risk of Lyme disease, three epidemiological effect measures were evaluated, in order from lowest to highest inferential value: (a) odds ratios for seropositivity from cross sectional studies; (b) odds ratios for symptomatic, clinically confirmed Lyme disease from cross sectional studies; and (c) incidence ratios (relative risks) for symptomatic, clinically confirmed Lyme disease, from longitudinal studies.

Despite the apparent increased risk of seropositivity in occupationally exposed people compared with controls, it should be noted that the choice of an appropriate comparison group can influence the magnitude of this risk. Of the seven studies in table 5 that had an increased risk of seropositivity in the occupational groups under study, five used a comparison group with a decreased residential risk of Lyme disease. Comparing seroprevalence among occupationally exposed people, who generally live in areas endemic for Lyme disease, with controls, who live in areas where Lyme disease is lowest or absent, would bias estimations of occupational risk in the direction of increased risk. Thus, the increased risks of seropositivity for outdoor work presented in table 5 may over estimate the occupational risk.

One consistent deficiency is that less than half of the 41 published studies required clinical confirmation by a physician. What is perhaps most surprising about the published studies is that three studies published in the 1990s all document no increased incidence of symptomatic, clinically confirmed Lyme disease in outdoor workers (Kuiper et al³¹; Vos et al³⁵; and Fahrer et al⁴⁰). Although one of these studies,³¹ did not report data from a comparison group, no cases of Lyme disease were found in the occupational group under study. Of the 41 studies, only Bowen et al⁴⁸ documented an increased incidence of symptomatic Lyme disease among a very specific group of outdoor workers and controls. However, all cases were diagnosed during 1981 and 1982, and thus may not represent either current standards for diagnosis of Lyme disease or current levels of risk after 20 years of experience with the disease among outdoor working populations. Thus, although we think that the findings of Bowen et al⁴⁸ are internally valid, they may not be generalisable, especially to current practice.

Among cross sectional studies that evaluated symptomatic Lyme disease, only one (Gustafson et al³⁹) reported the second most useful effect measure (OR for symptomatic disease); this study reported an increased relative odds for symptomatic Lyme disease over a lifetime, but the methods may raise concerns about recall bias. Many studies documented an increased risk of seropositivity, but most seropositive subjects were asymptomatic, and seropositivity was defined with ELISA or IFA only, which is not the current standard for serological testing. An important point is that, to our knowledge, there are no studies that suggest that asymptomatic seropositive subjects are at risk of developing symptomatic disease or late sequelae of infection. On the contrary, at least one study has reported that there is no increased risk for the development of symptomatic disease in such subjects over an average follow up interval of 6.5 years (Fahrer et al⁴⁰). Serological studies also document significant seroreversion rates; asymptomatic seropositive subjects are seronegative, and still asymptomatic, on repeat testing (Fahrer et al⁴⁰; Schwartz et al⁵⁴). For example, Schwartz et al⁵⁴ reported annual seroreversion rates of 43%, 23%, and 53% from 1988–89, 1989–90, and 1990–91, respectively.

Assessment of occupational risk was further limited by the paucity of studies that attempted to define occupational risk factors more carefully (in terms of hours outdoors in specific tick infested habitats, specific high risk tasks, or controlling for use of preventive behaviour) or that controlled for non-occupational risk factors in the assessment of occupational risk. Hours of recreation outdoors, deer sightings near the home, pet ownership, rural residence, and personal preventive behaviour have all been shown to be risk or protective factors for Lyme disease or seropositivity for antibodies to Borrelia burgdorferi.^9,10,49,54

Despite the limitations of the scientific literature, it would seem obvious that people who work outdoors in tick infested areas should be at increased risk of Lyme disease; however, an increased risk of symptomatic, clinically confirmed Lyme disease has not been documented in outdoor workers. Many of the studies were not specifically designed to evaluate the occupational risk of symptomatic, clinically confirmed Lyme disease, and the current serological testing guidelines were developed after most of these studies were published. However, the three best published epidemiological studies do not suggest that an occupational risk exists (Vos et al³⁵; Kuiper et al³¹; Fahrer et al⁴⁰). It may be that workers in tick infested habitat become knowledgeable about the disease and use personal preventive behaviour to minimise their risk.^9,55

Several studies suggest that such personal protective behaviour as tick checks, tucking trousers into socks, or use of permethrin or DEET may decrease the risk of seropositivity or tick bites. Antitick saliva antibody (ATSA) and antirecombinant tick calreticulin antibody are two biomarkers of tick exposure that have been used in epidemiological studies of exposure to ticks.^8,59,60 For example, in a study of military personnel on manoeuvres in tick infested areas of Arkansas, people who tucked their trousers into their socks were significantly less likely to be ATSA seropositive than subjects who did not tuck in their trousers (OR 2.8 (95% CI 1.1 to 7.1, Schwartz 1996).⁸ Similarly, outdoor workers in New Jersey who did not use insect repellants were more likely to be ATSA seropositive (OR (95% CI) = 2.0 (1.0-4.0), Schwartz and Goldstein).⁹

Environmental application of insecticides has been shown to decrease the abundance of ticks, but this is more likely to be useful in the prevention of residentially acquired Lyme disease than in the prevention of occupational disease.¹¹ It is also known that duration of tick feeding is an important determinant of the risk of infection, so tick checks and early tick removal are likely to be effective in the prevention of the disease.⁶¹ It should be noted, however, that other studies have not shown that personal preventive behaviour is effective in disease prevention.⁵⁵ All of these strategies have been recommended by the Centers for Disease Control as effective measures to decrease the risk of acquiring Lyme disease.¹

There are no other personal strategies to prevent Lyme disease that have been proved to prevent the disease as effectively as the Lyme disease vaccine. The Lyme vaccine efficacy study was a randomised, placebo controlled trial in 10 936 subjects aged 15–70 years.⁶² Lyme disease was carefully documented with culture, polymerase chain reaction, or western blot seroconversion among subjects with symptoms compatible with Lyme disease. Among study subjects under the age of 65, after three doses the vaccine was 90% effective in preventing laboratory confirmed, symptomatic Lyme disease. The Centers for Disease Control, The Medical Letter, and other authors have all commented on use of the vaccine.^1,13,14

Another important factor to consider is the natural history of Lyme disease compared with other diseases preventable by vaccine. Lyme disease is non-fatal, relatively easy to diagnose, and relatively easy to treat with oral antibiotics. By contrast, hepatitis B virus, for example, can be fatal, can have a chronic carrier state, and is not effectively treated. Although vaccination against Lyme disease may help to decrease morbidity, it would seem less imperative than for an organism such as hepatitis B virus.

When deciding whether or not to vaccinate a workforce against Lyme disease, the occupational physician should consider several factors—the strength of the scientific evidence presented here, the availability of resources, the history of risk of Lyme disease in the workforce under consideration, and the concerns of workers. The scientific literature shows an increased risk of occupational exposure to Borrelia burgdorferi, as assessed by the seroprevalence studies, but fails to document an increased risk of development of symptomatic Lyme disease. The lack of documented clinical risk makes it difficult to rely exclusively on the current scientific literature. Factors that may favour vaccination would include documented increased risk of Lyme disease in the workforce under consideration for vaccination; a high level of concern about the disease from workers; high level of outdoor activity in tick infested areas; poor compliance with personal protective clothing or behaviour; and availability of financial resources for occupational health programmes. Further studies that explicitly assess the risk of developing occupationally acquired Lyme disease are necessary to help guide the physician on whether or not to vaccinate.