INTRODUCTION
The importance of leptospirosis as a re-emerging zoonotic disease in some countries has recently been emphasized [Reference Bharti1, Reference Adler and De la Peña2]. The disease is endemic worldwide but a higher incidence is observed in tropical and subtropical climates where the conditions of the temperature and humidity favour the survival of Leptospira in the environment. Moreover, most countries in tropical regions are developing countries, where there are usually higher opportunities for contact between potentially infected animals and humans [Reference Pappas3, Reference Levett4]. Leptospirosis is often associated with rodents but all mammal species, including livestock and companion animals, can be infected. Animals can shed the organism in their urine for an extended period. Humans are infected after direct or indirect contact with urine containing the bacteria. Leptospires enter the body through mucosal membranes or abrasions in the skin. Clinical signs of disease vary from mild, self-limiting influenza-like illness (ILI) lasting 3–5 days, to a severe condition that can involve renal and hepatic malfunction, and if not treated can be fatal [Reference Faine5]. Typical symptoms of the disease include myalgia, fever, nausea, vomiting, and headache [Reference Vickery6].
Leptospirosis is endemic in New Zealand livestock. Surveys have estimated seroprevalence in beef cattle, sheep and deer animals ranging from 48–65%, 33–48% and 26–43% for L. borgpetersenii serovar Hardjo type Hardjo-bovis (Hardjo) and 25–35%, 0–45% and 11–50% for L. interrogans serovar Pomona (Pomona), respectively [Reference Dreyfus7, Reference Subharat8]. Other production animals, including dairy cattle and pigs, are also at risk of leptospirosis but vaccination against serovars Hardjo and Pomona is thought to have largely reduced leptospirosis incidence and the risk of transmission to humans in New Zealand [Reference Marshall9, Reference Marshall and Manktelow10]. However, a recent pilot study in 44 conveniently selected dairy herds with a long history of vaccination against Leptospira (during the last 5 to >20 years), identified 13% urinary shedders by PCR and dark-field microscopy in 30% of herds [Reference Wilson11]. Although serovars were not determined in that study, it has been reported elsewhere that transient urinary shedding of Hardjo can occur in vaccinated cattle [Reference Zuerner12]. Serovars not included in cattle vaccines and reported from notified human cases in New Zealand include L. borgpetersenii serovar Ballum (Ballum) and L. borgpetersenii serovar Tarassovi (Tarassovi) [13].
Companion animals can also be exposed to Leptospira, although their role in transmission of Leptospira to humans is unknown. A cross-sectional study in New Zealand dogs found 15·0% of serum submissions seropositive to Hardjo, Pomona, Ballum, or L. interrogans serovar Copenhageni (Copenhageni). The latter serovar was the most prevalent (10·3%). Hardjo (3·5%) ranked second and was predominantly found in farm dogs [Reference Harland14]. Similar observations were made earlier in healthy dogs from rural or urban environments [Reference O'Keefe15].
In New Zealand the incidence of leptospirosis in humans is among the highest in industrialized countries being 2·6/100 000 people [Reference Pappas3]. The most common serovars in notified human cases during 2003–2013 were Hardjo, Pomona and Ballum [13]. The disease was identified as a common occupationally acquired infectious disease in the country and a higher incidence was described for males compared to females [Reference Thornley16]. Working in an abattoir and livestock farming were the most frequently reported occupations in Leptospira cases [13].
The risk of Leptospira exposure for workers of a sheep-only abattoir was quantified as 3–54 carcasses per day. Risk depended on work position and year [Reference Dorjee17, Reference Dorjee18]. The seroprevalence of Hardjo and Pomona in abattoir workers of another sheep abattoir was estimated as 9·5% [microscopic agglutination test (MAT) ⩾24, 2008] [Reference Benschop19]. During 2009/2010 a more comprehensive serosurvey of abattoir workers estimated that 10·9% of the workers had evidence of exposure to Hardjo or Pomona (MAT ⩾48): 5·4% in beef, 12·9% in sheep and 17·5% in deer abattoir workers [Reference Dreyfus20].
Veterinarians are not commonly listed in notified cases. During 2003–2013, only two cases (one veterinary technician in 2012 and one veterinarian in 2013) were notified [13]. However, veterinarians are probably at high risk of leptospirosis due to potential exposure to urine of infected animals. A previous study conducted in New Zealand found 1/86 veterinarians seropositive for Hardjo (MAT ⩾100) [Reference Robinson and Metcalfe21]. Another serosurvey found no antibody titres for Leptospira in 302 veterinary students during 2010/2011 [Reference Fang22] but no recent estimate of the extent of exposure to Leptospira in veterinarians was available. This study aimed to estimate the seroprevalence of five Leptospira serovars and to evaluate risk factors for exposure in New Zealand veterinarians.
METHODS
Study design
A cross-sectional study was designed to estimate seroprevalence of Leptospira and assess risk factors associated with Leptospira serostatus. Veterinarians attending the Deer Special Interest group conference of the New Zealand Veterinary Association in Queenstown (May 2012) or the New Zealand Veterinary Association annual conference in Hamilton (June 2012) were eligible for inclusion. A station was set up in the exhibition hall of each conference where certified phlebotomists sampled blood from voluntary participants. Samples were kept refrigerated during the day at 4 °C. Tubes with blood were sent overnight inside bio-bottles™ (Bio-Bottle New Zealand Ltd) to the Hopkirk Institute, Massey University for serum extraction. Sera were kept frozen until testing. A modified MAT as described by Faine et al. [Reference Faine5] was used to detect antibodies against Hardjo, Pomona, Ballum, Copenhageni and Tarassovi. In summary, eight twofold dilutions of serum were made using sterile 0·9% saline solution as a diluent. An equal volume of live Leptospira antigen was then added to each dilution to make a final serial dilution that ranged from 1:24 to 1:3072. The last dilution able to agglutinate more than 50% of leptospires was taken as the final titre for each serovar.
Recording of risk factors
A questionnaire was designed to capture information about exposure to potential risk factors during the previous 18 months. This time-frame of exposure was selected based on the duration of titres previously estimated as 10 months for Pomona and 29 months for Hardjo [Reference Dreyfus23].
The questionnaire was available for participants at the conferences either online (Survey Gizmo®, www.surveygizmo.co.uk/) or paper. It covered personal information, animal exposure in the work place, animal and environmental exposure outside the work place, previous leptospirosis illness, ILI in the last 18 months, and general opinions about leptospirosis.
ILI was defined as an episode of illness accompanied by some or all of the following signs/symptoms: fever, headache, myalgia, photophobia, sweating, and severe general debility.
Statistical analysis
Antibody titres (24–3072) were described graphically using numbers (Fig. 1). Seroprevalence and 95% confidence intervals were estimated using the titre cut-off of ⩾48 for a positive sample.
Fig. 1. Distribution of the number of samples at each microscopic agglutination test titre for Hardjo, Pomona, Ballum, Copenhageni and Tarassovi.
Gender and age were described using percentages and summary statistics [median, minimum value (min), first quartile (Q1), third quartile (Q3) and maximum value (max)], respectively. Differences in the time spent working with different animal species between females and males were assessed using Wilcoxon's rank-sum test. The association between gender and home slaughter of pigs or cattle was explored using logistic regression.
The percentage of time spent in different activities within the veterinary practice was described and kernel density plots were used to graphically compare percentages in seropositive and seronegative veterinarians.
The number of previous leptospirosis episodes and the most common signs and symptoms associated with these episodes were recorded. The association between Leptospira serostatus and past episodes of leptospirosis was explored using prevalence ratios.
The number of ILIs in the last 18 months was recorded. ILI was categorized according to the time veterinarians spent off work due to this illness: no illness, off work <3 days, off work ⩾3 days. The relationship between Leptospira serostatus and ILI (yes/no) or time off work categories was explored. The population attributable fraction (PAF) was estimated. PAF is a measure of effect in the population. In this case it estimates the percentage of ILI in the population of veterinarians during the last 18 months that may be attributed to Leptospira exposure.
For inclusion in the multivariable analyses, the percentage of time spent in contact with different animal species was categorized as: none (0%), low (>0–25%), mid-low (>25–50%), mid-high (>50–75%), and high (>75%). For example, a veterinarian could be exposed to dairy cattle for 60% (mid-high) of his/her working time, to pets for 10% (low) and office work for the remaining 30% (mid-low) of work-related time.
Unconditional associations between each of the putative risk factors and Leptospira serostatus were evaluated using logistic regression. Variables with a P value ⩽0·35, tested by the likelihood ratio test (LRT), were selected for inclusion in the multivariable logistic model. Backward elimination of variables from a full multivariable model was based on the LRT with significance threshold of P > 0·05. Akaike's Information Criterion was used to select the final multivariable model [Reference Akaike24]. Goodness of fit of the multivariable model was assessed using Pearson and deviance χ 2 tests and the Le Cessie–van Houwelingen test [Reference le Cessie and van Houwelingen25]. Observations with large Pearson residuals and delta-beta values were removed temporarily to investigate their influence on the coefficients and standard errors of the final multivariable model.
Analyses were performed using R software version 3.03 [26].
Ethical standards
The authors assert that all procedures contributing to this work comply with the ethical standards of the relevant national and institutional committees on human experimentation and with the Helsinki Declaration of 1975, as revised in 2008. The study was approved by the Massey University Human Ethics Committee, application number 12/09.
RESULTS
Data from 277 veterinarians, working in New Zealand that were sampled and had completed the questionnaire, were used for analysis.
Titre distribution
Antibody titres for Leptospira (MAT ⩾ 48) were detected in 14 veterinarians. The most frequent serovar was Pomona (n = 7) followed by Hardjo (n = 6). Antibodies detected for Pomona had a maximum titre of 96, while for Copenhageni and Ballum the maximum titre was 48. Comparatively higher antibody titres of up to 1536 were observed for Hardjo. Figure 1 shows the titre distribution for Hardjo, Pomona, Ballum, Copenhageni and Tarassovi serovars.
Seroprevalence
Seroprevalence of Leptospira (⩾48 MAT titre), all serovars included, was 5·1% [95% confidence interval (CI) 2·8−8·3]. One participant was seropositive for both Hardjo and Pomona. No veterinarian was seropositive for Tarassovi. Table 1 summarizes seroprevalence for each serovar individually and for any of the five serovars.
Table 1. Number of participants positive and negative and seroprevalence for individual serovars at a titre of ⩾48
CI, Confidence interval.
* One sample positive for Hardjo and Pomona.
Age and gender
The median age of study participants was 42 years (min = 22, Q1 = 33, Q3 = 53, max = 73 years).
Thirty-nine per cent of veterinarians were females (n = 109). Male participants were in general older (median 47, min = 23, Q1 = 37, Q3 = 57, max = 73 years) than female participants (median = 35, min = 22, Q1 = 29, Q3 = 45, max = 59 years). There was no significant difference (P = 0·40) between the seroprevalence in men (10/168 positives) and women (4/109 positives). Females spent significantly (P < 0·001) more time on average working with dogs and cats (23·5%, 95% CI 17·5–29·6) than males (11·4%, 95% CI 8·2–14·6). On the other hand, males spent significantly (P < 0·001) more time working with deer (6·7%, 95% CI 4·1−9·3) than females (2·7%, 0·5−5·0%). Time spent with dairy cattle, beef cattle, and sheep was not significantly different between males and females. Home slaughter of cattle (P = 0·54) or pigs (P = 0·41) were not associated significantly with gender.
Occupational exposure to animals
Most veterinarians (n = 155) worked at least a quarter of their time with dairy cattle. By contrast, a small proportion of veterinarians worked more than 25% of their time with beef (n = 18), sheep (n = 15), or deer (n = 12). Few (n = 24) worked 100% of their time with a single species, while 22 had no animal contact at work.
Figure 2 shows the smoothed distribution of the proportion of time that seropositive and seronegative veterinarians spent working with different animal species. These plots suggest that seropositive veterinarians were not more involved than seronegative veterinarians in any of the extreme ends of exposure to any species, but some differences were apparent for middle portions of exposure to some species.
Fig. 2. Smoothed proportion of seropositive and seronegative participants (density) by percentage of time spent in contact with different animal species at work.
History of previous leptospirosis episodes
Leptospirosis was previously diagnosed by general practitioners and laboratory test (serology) 11 times in 10 participants. Nine were able to recall symptoms of disease, which were: headache (n = 9), sweating (n = 9), fever (n = 8), myalgia (n = 7), sore eyes (n = 6), photophobia (n = 6), severe debility (n = 5) and meningitis (n = 1). Eight recalled being off work a median time of 7 days (min = 2, Q1 = 3, Q3 = 21, max = 105 days) and one recalled being hospitalized because of the disease. Eight of these episodes of leptospirosis occurred from 43 years to 16 years prior to participation in this study, while one occurred 3 months before sampling. Having clinical leptospirosis in the past was significantly associated (P < 0·001) with seropositivity [prevalence ratio (PR) = 10·7%, 95% CI 4·0–28·3]. This strong association was also observed when excluding the recent episode of leptospirosis (PR 8·9, 95% CI 2·9–26·9). Previous leptospirosis was excluded from the multivariable model since it could mask the effect of other risk factors associated with serostatus.
Leptospira exposure and illness
A total of 118/275 veterinarians experienced ILI in the previous 18 months. Of Leptospira seropositives, 8/14 had ILI in the last 18 months compared to 110/261 in seronegative veterinarians. ILI cases in seropositive veterinarians was therefore 1·4 times greater than in seronegative veterinarians (P = 0·28). However, seropositive veterinarians were more likely to spend ⩾3 days off work due to ILI than seronegative veterinarians (Table 2) (P = 0·09). Assuming a causal relationship between Leptospira exposure and ILI, the percentage of ILI in the last 18 months associated with ⩾3 days off work that was attributable to Leptospira exposure in the population of veterinarians, was 8·3% (PAF).
Table 2. Association between Leptospira serostatus and time off work due to ILI in the last 18 months
ILI, Influenza-like illness; CI, confidence interval; PR, prevalence ratio; PAF, population attributable fraction.
* PR and PAF were estimated using values for ILI and not ill in the last 18 months.
† PR and PAF were estimated using values for ⩾3 days ill and <3 days or not ill in the last 18 months.
‡ Four veterinarians, seronegative to Leptospira, had ILI but did not state number of days off work.
The ILI variable was excluded from the multivariable model as it was likely to be a consequence of leptospirosis rather than a cause, hence not a confounder for other risk factors.
Unadjusted associations
Unadjusted odds ratios (ORs) (Table 3) suggested that home slaughtering of cattle, sheep or deer in the previous 18 months significantly increased the odds of seropositivity compared to veterinarians who were not involved in such activities. Categories of the time spent with beef cattle, sheep, deer or other animals; hunting wild pigs, wild goats, rabbits, or birds; owning cattle, sheep, horses, dogs or cats; home slaughter of deer; age and gender were largely not associated (LRT P > 0·35) with Leptospira serostatus (results not shown in Table 3).
Table 3. Unadjusted associations between potential risk factors and Leptospira serostatus (P ⩽ 0·35)
LRT, Likelihood ratio test; OR, Odds ratio; CI, confidence interval; n.a., not identifiable as no seropositives were observed.
Multivariable analysis
In the final multivariable model, home slaughter of cattle or pigs was positively associated with Leptospira seropositivity. For example, the odds of being seropositive for veterinarians involved in home slaughter of cattle in the previous 18 months were 4·6 times the odds of being seropositive for those not involved in this activity, after controlling for the effects of home slaughter of pigs and occupational exposure to dogs and cats. Moreover, veterinarians spending from 50% to 75% of their time working with dogs or cats had significantly higher odds of being seropositive than those not working with dogs or cats (Table 4).
Table 4. Results of a multivariable logistic regression model between Leptospira serostatus and potential risk factors
LRT, Likelihood ratio test; OR, Odds ratio; CI, confidence interval; n.a., not identifiable as no seropositives were observed.
DISCUSSION
In this sample of veterinarians, 14/277 were seropositive (5·1%, 95% CI 2·8−8·3). A previous study investigated the seroprevalence of veterinarians in New Zealand for Hardjo, Pomona, Ballum, Copenhageni, Tarassovi, plus serovars Canicola, Autumnalis, Andaman and Medanensis not tested in our study, observing 1/86 (1·2%) veterinarians seropositive for Hardjo [Reference Robinson and Metcalfe21]. The MAT titre cut-off used in that investigation was one dilution higher (⩾100) than the one used in the present study. At an equivalent MAT titre of 96, 7/277 veterinarians were seropositive (2·5%, 95% CI 1·0−5·1); which is similar to that of the previous study.
The selection of the MAT titre cut-off depends on the purpose for testing. For a clinical case of leptospirosis based on MAT, the World Health Organization (WHO) considered a MAT titre of ⩾400 in a single or paired serum sample, or a fourfold increase in MAT titre in acute and convalescent phase in subjects experiencing clinical signs compatible with leptospirosis [27]. There is not a defined cut-off for antibody MAT titres when assessing Leptospira exposure in apparently healthy individuals working in high-risk occupations, hence the cut point must be considered when comparing seroprevalence studies.
For example, exposure to Hardjo and Pomona in apparently healthy abattoir workers was estimated at 10·9% (95% CI 8·5–13·9) using the MAT titre of ⩾48 [Reference Dreyfus20], which is 2·5 times higher than the exposure observed to these two serovars in veterinarians (4·3%, 95% CI 2·3–7·5). Exposure to Hardjo, Pomona and Ballum has also been assessed in apparently healthy veterinary students where none of the 302 participants was positive for these serovars at a titre of ⩾48 [Reference Fang22]. This finding suggests that exposure to Leptospira most likely occurs during the years of active work.
Leptospirosis is likely under-reported in New Zealand, especially for mild infections. The disease in people is characterized by non-specific symptoms [Reference Vickery6] that may be confused with flu. We observed a marginally significant association between Leptospira exposure and ILI (⩾3 days off work) in the previous 18 months. Although, the cross-sectional design used does not allow assessment of when exposure and outcome occurred; we believe that some of the ILIs recorded are a consequence of Leptospira exposure. This is consistent with recent findings in abattoir workers where new Hardjo or Pomona infections increased the risk of ILI by 1·9 (95% CI 1·3–2·7) in a year and the proportion of ILI cases attributed to Hardjo or Pomona exposure (PAF) was 10·0% [Reference Dreyfus23]. If we assume a causal relationship between Leptospira exposure and being off work for ⩾3 days with flu-like signs of illness in the last 18 months, it can be said that 8·3% of these cases could be prevented if Leptospira exposure of veterinarians was controlled (Table 2).
A significant risk factor for Leptospira exposure was home slaughtering of cattle (n = 32) or pigs (n = 11). The average predicted seroprevalence increased from 3·2% (95% CI 1·0−16·9) in veterinarians not performing these activities to 16·4% (95% CI 4·4−46·0) in veterinarians performing any of these activities. This result contrasts with findings from the serosurvey in veterinary students of New Zealand, where no student was seropositive and yet 20·9% stated that they performed home slaughter in the 18 months prior to sampling [Reference Fang22]. A plausible explanation for the difference observed between students and veterinarians is the presumably shorter length of time that students were exposed to home slaughter due to their young age. A recent study of abattoir workers did not find a significant association between home slaughter and Leptospira seropositivity [Reference Dreyfus20] but no information was available about the species slaughtered at home or the frequency of this practice. Possibly, the additional risk of Leptospira exposure from home slaughter in abattoir workers was relatively small compared to veterinarians.
Home slaughter of sheep was not a significant risk factor in the multivariable model, most likely due to its significant association with slaughtering cattle (OR 2·5, P = 0·02) and pigs (OR 9·5, P = 0·001). Hence, those involved in home slaughter appear to process more than one species. It is therefore impossible to make inferences about which home-slaughtered species was the likely source of infection for veterinarians.
A positive correlation between Leptospira seroprevalence and the time spent with a specific species was expected, especially since earlier studies demonstrated Leptospira exposure in livestock [Reference Dreyfus7, Reference Subharat8] and that between 21% and 41% of seropositive animals potentially shed Leptospira in their urine [Reference Dorjee17, Reference Fang28]. Instead, seroprevalence tended to be higher when veterinarians worked with multiple species (Fig. 2). If any species, it was the contact with pets and dairy cattle that tended to be positively associated with seropositivity. However, the data are too sparse to draw clear conclusions about the comparative risk of infection from any particular animal species that veterinarians are handling during their clinical work.
In New Zealand dogs, Copenhageni was the most commonly observed serovar, followed by Hardjo which was commonly present in dogs from rural locations or farm working dogs [Reference Harland14, Reference O'Keefe15]. However, only one veterinarian (who did not work with dogs or cats) was seropositive for Copenhageni. This contrasts with six veterinarians seropositive for Hardjo; three of whom had no or only minor exposure to dogs or cats (0−3%), and the other three had moderate exposures (20−65%). Furthermore, none of the 19 veterinarians who worked >75% of the time with dogs or cats were seropositive.
A strong and highly significant positive association between previous leptospirosis episodes and seropositivity to Leptospira observed in this survey was also reported from abattoir workers [Reference Dreyfus20]. Antibodies are usually detectable from 5 to 10 days after infection but they may persist for months or even years after infection at low antibody titres [Reference Cumberland29]. Re-testing of 69 meat inspectors with previous positive titres, irrespective of their clinical history, showed that MAT titres of 384−192, 96, and 48−24 persisted for at least 30, 43 and 52 months, respectively, in some individuals [Reference Blackmore, Schollum and Moriarty30]. Nevertheless, the persistence of titres is difficult to assess because often individuals at high risk of infection are studied and it is unknown whether persistent titres are the product of a slow decay, or continued re-exposure. The latter hypothesis could be questionable since seroconversion in seronegative meat inspectors was estimated in 4·4% a year [Reference Blackmore, Schollum and Moriarty30]. Previous episodes of clinical Leptospirosis in veterinarians of this study occurred between 16 and 43 years before sampling; it is therefore unlikely that the strong association between previous leptospirosis and seropositivity was due to a slow decay of antibodies, thus they may rather reflect continuous re-exposure to Leptospira resulting in mild or subclinical infections.
Gender was not significantly associated with Leptospira exposure (P = 0·9) in this group of veterinarians. National data show males have a higher incidence of leptospirosis than females which may be due to the predominance of males in high-risk occupations [Reference Thornley16]. For example, male abattoir workers were more likely to be seropositive and seroconverted more often to Leptospira than female workers [Reference Dreyfus20, Reference Dreyfus23]. In our data, females represented 39·4% of participants and they spent on average more time with dogs or cats and less time with deer than males. Other occupational exposures and the proportion of veterinarians involved in home slaughter of cattle or pigs were not different between females and males.
The 277 participant veterinarians represented 11% of the veterinarians registered in New Zealand [31] (n = 2521) in the year of the study. The median age of veterinarians in New Zealand was 43 years, similar to our study (42 years) but a slightly higher proportion of females (62%) and males (31%) under 40 years was observed in comparison to population demographics of veterinarians (59% and 23%, respectively). Potential selection bias may have been introduced by design as only veterinarians attending the conferences had the opportunity to participate in the study. The effect this might have had on seroprevalence and ORs is difficult to predict and subject to speculation. Nonetheless, we can see no reason to believe that selection bias affected the overall conclusions presented in this study.
CONCLUSION
Leptospira seroprevalence in veterinarians is about half as high as that of abattoir workers. High-risk activities for Leptospira exposure in veterinarians included home slaughter of cattle or pigs. No clear associations were found with the animal species that veterinarians were in contact with during professional activities. The borderline association between Leptospira exposure and ILI resulting in ⩾3 days off work suggested that 8·3% of these ILI cases were attributable to Leptospira infection during the 18 months prior to sampling. However, caution should be exercised when interpreting this result since the cross-sectional design does not warrant a causal relationship.
ACKNOWLEDGEMENTS
We express our gratitude to all the veterinarians that voluntarily participated in the study and made it possible, to Neville Hack for all the technical assistance and support provided, and to MSD Animal Health, Zoetis and Vibac for computers and online connection for participants to access the questionnaire.
This study was funded by MSD Animal Health, Zoetis and Vibac, New Zealand Government through an International Doctoral Research Scholarship and by Massey University.
DECLARATION OF INTEREST
None.
