Transboundary and Emerging Diseases
ORIGINAL ARTICLE
Performance of Diagnostic Tests for Bovine Tuberculosis in North American Furbearers and Implications for Surveillance D. J. O’Brien1, J. S. Fierke1, T. M. Cooley1, S. D. Fitzgerald2, M. K. Cosgrove1 and S. M. Schmitt1 1 2
Michigan Department of Natural Resources, Wildlife Disease Laboratory, Lansing, MI, USA Department of Pathobiology and Diagnostic Investigation, Michigan State University, East Lansing, MI, USA
Keywords: bovine tuberculosis; diagnosis; furbearing mammals; Mycobacterium bovis; spillover hosts; wildlife Correspondence: D. J. O’Brien. Wildlife Disease Laboratory, Michigan Department of Natural Resources, 4125 Beaumont Road, Room 250, Lansing, MI 48910-8106, USA. Tel.: +011 517 336 5035; Fax: +011 517 337 4920; E-mail: [email protected] Received for publication November 13, 2012 doi:10.1111/tbed.12093
Summary Risks of bovine tuberculosis (bTB) transmission from free-ranging wildlife to livestock remain a concern in the United States, in both known endemic areas and where spillover from recently-infected livestock herds occurs. Federal agriculture officials in the United States (US) have recommended surveillance of non-cervid furbearers to determine whether free-ranging wildlife in the vicinity of cattle herd breakdowns are bTB infected, yet the efficacy of common diagnostic tests in these species is largely unknown. We calculated the sensitivity, specificity, predictive values and positive likelihood ratios for bTB infection in carcasses of sixteen species of furbearers tested via: (i) the presence of gross lesions compatible with bTB; (ii) histopathology consistent with bTB; and (iii) the presence of acid-fast bacilli (AFB) on histopathology. The gold standard comparison test was mycobacterial culture of cranial visceral lymph nodes pooled for each animal. Forty-two animals distributed across six species cultured bTB positive from among 1522 furbearers tested over thirteen years. The sensitivity of all three tests was poor (10%, 22% and 24% for gross lesions, AFB and histopathology, respectively), while specificities (all 99%) and negative predictive values (all 97%) were high. Positive predictive values varied widely (31–75%). Likelihood ratios for culture positivity given a positive test result showed AFB on histopathology to be the most reliable test, and gross lesions the least, though confidence intervals were wide and overlapping. While non-cervid furbearers may prove useful in North American bTB surveillance, wildlife managers should be aware of factors that may abate their utility and complicate interpretation of surveillance.
Introduction A century ago, tuberculosis was the leading cause of human mortality in the United States (US), claiming nearly 150 000 lives in 1900 (Olmstead and Rhode, 2004). Although characterized at its outset in 1917 as ‘an impossible undertaking,’ the programme to eradicate bovine tuberculosis (bTB), caused by Mycobacterium bovis, from US cattle has been enormously successful, greatly reducing both the burden of human disease and the disease-related economic costs to livestock agriculture (Olmstead and Rhode, 2004; Palmer and Waters, 2011). However, global experience has demonstrated that eradication of bTB is
frequently complicated, and sometimes prevented entirely, by the establishment and persistence of M. bovis in freeranging wildlife reservoirs (Palmer, 2007), and bTB eradication efforts in North America are no exception (O’Brien et al., 2011; Shury and Bergeson, 2011). Recognizing the importance of wildlife reservoirs to nationwide bTB eradication plans, in 2011, the US Department of Agriculture’s Animal and Plant Health Inspection Service (USDA-APHIS) formulated and released guidelines for surveillance of bovine tuberculosis in wildlife (DeLiberto et al., 2011). The principal stated objective of those guidelines is ‘to determine whether bovine TB is present in wildlife’ in areas where bTB has been detected in a dairy,
© 2013 Blackwell Verlag GmbH • Transboundary and Emerging Diseases. 60 (Suppl. 1) (2013) 67–73
67
bTB Test Performance in Furbearers
D. J. O’Brien et al.
beef or captive cervid herd (DeLiberto et al., 2011, p. 4). These guidelines place considerable emphasis on surveillance of furbearing mammals of the Orders Didelphimorphia [e.g. Virginia opossums (Didelphis virginiana)] and Carnivora [e.g. raccoons (Procyon lotor), coyotes (Canis latrans)]. The basis for this emphasis is varied. Testing of ‘resident’ furbearers such as opossums and raccoons is ‘highly recommended’ to ‘clean and disinfect’ an infected livestock farm, despite the fact that ‘current research does not indicate that peridomestic mammals pose a risk of disease transmission to domestic livestock’ (DeLiberto et al., 2011, p. 6). The guidelines suggest testing of ‘transient’ furbearers such as coyotes which ‘may have greater potential to spread bovine TB through indirect contact (e.g. contamination of feed) or serve as sentinels for the disease,’ as well as ‘to assist in delineating the extent of bovine TB on the landscape’ (DeLiberto et al., 2011, p. 7). While the USDA-APHIS guidelines ‘were developed to provide guidance for State and Federal wildlife agencies that plan to conduct surveillance for bovine tuberculosis (TB) in wildlife’ (DeLiberto et al., 2011, p. 4), thus far, when such situations have arisen, the state agencies have typically been charged with that responsibility (see, e.g. Carstensen and DonCarlos, 2011), perhaps because of their statutory authority for managing non-migratory game species (Thorne et al., 2005). State wildlife management agencies frequently and perpetually lack adequate funding and personnel to carry out wildlife disease surveillance at an optimal level, particularly when unanticipated events such as bTB-infected livestock herds arise. Given the emphasis placed on furbearer surveillance by USDA-APHIS, it is conceivable that states might choose to rely on testing of furbearers as a primary basis for drawing conclusions concerning bTB surveillance and management. Yet, the efficacy of common screening diagnostic tests for bTB in these species is unknown. Consequently, our objective in this study was to establish the diagnostic sensitivity, specificity, predictive values and likelihood ratios for three common tests of infection with M. bovis in sixteen species of North American furbearers. Materials and Methods We drew on a large historical database of gross pathology, histopathology and mycobacterial culture results from Michigan Department of Natural Resources (MDNR) surveillance carried out in 1996–2008 in the northern half of Michigan’s Lower Peninsula, with sampling efforts concentrated in and around Deer Management Unit (DMU) 452, considered the core area of the bTB outbreak in whitetailed deer (WTD) and cattle (see O’Brien et al., 2002; Fig. 1). Details of furbearer sampling (Bruning-Fann et al., 1998, 2001; Schmitt et al., 2002; O’Brien et al., 2006), gross 68
and histopathology examinations (Bruning-Fann et al., 1998; Fitzgerald et al., 2000) and mycobacterial culture procedures (Schmitt et al., 1997; O’Brien et al., 2004) have all been described elsewhere. Not all tests were performed on all the cultured samples. We compared the diagnostic performance of three screening tests against the gold standard of mycobacterial culture: (i) the presence of gross lesions compatible with mycobacteriosis; (ii) histopathology consistent with M. bovis infection; and (iii) the presence of acid-fast bacteria (AFB) on histopathology as identified by the Ziehl–Neelson staining technique (Kent and Kubica, 1985). Diagnostic sensitivity, specificity, positive and negative predictive values (alternatively called the predictive value of a positive or negative test) were calculated as described by Thrusfield (1995), using freely available software (Lowry, 2012). Sensitivity is the proportion of infected (i.e. bTB culture positive) furbearers that are positive on a screening test. Specificity is the proportion of uninfected that test negative on screening. Positive and negative predictive values are the probabilities that a screening test positive is actually bTB infected, or a screening test negative actually is uninfected, respectively. We also calculated the positive likelihood ratio (Fagan, 1975), an index of how reliably a screening test detects bTB infection (Scherokman, 1997). The ratio of two conditional probabilities [the true-positive rate (sensitivity) divided by the false-positive rate (1 specificity); Lowry, 2012], it is essentially the odds that a positive test result would be expected from a genuinely bTB-infected furbearer. Results From 1996 through 2008, 1522 furbearers distributed across sixteen species (Schmitt et al., 2002; Table 2) were cultured for M. bovis. Of those, 42 (2.8%) cultured positive: 7/216 (3.2%) black bears (Ursus americanus); 3/56 (5.4%) bobcats (Lynx rufus); 19/380 (5.0%) coyotes; 2/379 (0.5%) opossums; 8/335 (2.4%) raccoons; and 3/29 (10%) red foxes (Vulpes vulpes). One other bobcat was bTB positive on polymerase chain reaction (PCR), but was culture negative (Bruning-Fann et al., 2001). All other species were culture negative [although a single grey fox (Urocyon cinereoargenteus) trapped by other researchers cultured positive in an unrelated study (Witmer et al., 2010), Table 1]. Performance of the three bTB screening tests compared to mycobacterial culture is summarized in Table 1. All were quite insensitive for detecting M. bovis infection, with histopathology the best of the three at only 24%. Specificities and negative predictive values were uniformly high, ranging between 99–100% and 97–98%, respectively. Positive predictive values varied widely (31–75%). Likelihood ratios for culture positivity given a positive screening test result
© 2013 Blackwell Verlag GmbH • Transboundary and Emerging Diseases. 60 (Suppl. 1) (2013) 67–73
D. J. O’Brien et al.
bTB Test Performance in Furbearers
showed the presence of AFB on histology to be the most reliable test, and the presence of gross lesions the least, although confidence intervals were wide and overlapping. Discussion Because it is an exemplary multihost pathogen (Francis, 1958; O’Reilly and Daborn, 1995), adequately responding to initial diagnosis of M. bovis in free-ranging wildlife necessitates determining not simply whether wildlife are collectively infected, but which species are, and are not, likely to be playing a significant role in the epidemiology of a particular outbreak (Corner, 2006). It was with that knowledge in mind that MDNR undertook large-scale bTB surveillance in multiple species beginning in 1995. The identification of furbearers as maintenance hosts of bTB in Ireland, Great Britain and New Zealand supported the plausibility, at least conceptually, of North American furbearers as potential maintenance hosts. However, that plausibility has yet to be validated. Examination of pathology and mycobacterial culture results from early in the course of surveillance led to the conclusion that furbearers in Michigan were likely to be spillover hosts only (BruningFann et al., 1998, 2001) and were likely playing no significant role in the maintenance of bTB nor its transmission to livestock (de Lisle et al., 2002; Schmitt et al., 2002). Since that time, data from a variety of experimental (Palmer et al., 2002; Johnson et al., 2008; Berentsen et al., 2010; Fenton et al., 2012) and field (Berentsen et al., 2011) studies in multiple species have thus far confirmed those early conclusions. Although unsupported by the accumulated data, some continue to believe that furbearers play a significant role in cattle herd infections (Atwood et al., 2009; Witmer et al., 2010). Coyotes were originally proposed as potential sentinel animals (de Lisle et al., 2002) because some bTB-positive coyotes were found in areas where bTB prevalence in WTD was very low (Schmitt et al., 2002; Fig. 2). Subsequent studies have produced mixed results however. A USDA group working in Michigan (Atwood et al., 2007; VerCauteren et al., 2008; Berentsen et al., 2011) concluded that coyotes
are good sentinels. However, researchers in Manitoba, Canada [another location like Michigan where bTB is present in both WTD and elk (Cervus elaphus)] found coyotes to be insensitive as sentinels there (Sangster et al., 2007). Even in Michigan, coyotes proved relatively insensitive as sentinels for bTB in WTD; only 1 of 17 (6%) M. bovis-positive coyotes came from a county where bTB had not already been detected via surveillance in deer (Berentsen et al., 2011). Thus in our view, considerable uncertainty remains to be resolved before furbearers can be confidently used in sentinel surveillance. In this study, the insensitivity of the three screening tests for detecting bTB infection was disappointing, but not particularly surprising. Both our previous reports (BruningFann et al., 2001; Schmitt et al., 2002; O’Brien et al., 2006) and those of others (Berentsen et al., 2011) have noted the paucity of gross and histopathological lesions of bTB in furbearers. Here, 13 of 1489 (0.9%), 24 of 1482 (1.6%) and 12 of 1483 (0.8%) samples were positive for gross lesions, histopathology and AFB, respectively, compared to 1 of 175 (0.6%) and 12 of 175 (6.9%) that had gross and histopathological lesions, respectively, in Berentsen et al. (2011). Even when gross or histologic lesions were present in our study, positive predictive values indicate only a 31–42% probability that those lesioned samples actually came from an animal with bTB. The presence of AFB on histopathology was a better predictor; samples with AFB had a 75% probability of originating from an M. bovis-infected animal. In addition, positive likelihood ratios indicate AFB to be the most reliable of the three tests, though differences between tests were not statistically significant. The uniformly high specificities and negative predictive values across all three tests are likely due to both the low sensitivities of the tests and the low prevalence of M. bovis infection. While other studies of Michigan furbearers have not specifically reported the performance of their diagnostic tests, histopathology and culture results from coyotes in Berentsen et al. (2011, Table 1) can be analysed by the same methods used here, yielding sensitivity of 47.1% (95% confidence limits: 23.9–71.5), specificity of 97.2% (92.6–99.1), and positive and negative predictive values of 66.7%
Table 1. Performance* of three diagnostic tests for Mycobacterium bovis versus culture in furbearers, Michigan, USA, 1996–2008 Predictive value Test
Sensitivity
Specificity
Positive
Negative
Positive likelihood ratio
Gross lesions Histopathology Acid-fast bacteria
9.5 (3.1–23.5) 24.4 (12.9–40.6) 22.0 (11.1–38.0)
99.4 (98.8–99.7) 99.0 (98.3–99.4) 99.8 (99.3–99.9)
30.8 (10.4–61.1) 41.7 (22.8–63.1) 75.0 (42.8–93.3)
97.4 (96.4–98.1) 97.9 (97.0–98.5) 97.8 (96.9–98.5)
15.3 (4.9–47.7) 25.1 (11.9–53.1) 105.5 (29.7–375.4)
*Point estimates expressed as percentages with 95% confidence limits calculated by the continuity-corrected efficient-score method (Newcombe, 1998).
© 2013 Blackwell Verlag GmbH • Transboundary and Emerging Diseases. 60 (Suppl. 1) (2013) 67–73
69
bTB Test Performance in Furbearers
D. J. O’Brien et al.
(35.4–88.7) and 94% (88.6–97.0), respectively. While those point estimates show somewhat higher sensitivity and positive predictive values for histopathology than our results, sensitivity is still poor, and confidence intervals overlap those of our study, suggesting little appreciable difference in test performance across studies. In our study, most furbearer samples were obtained skinned from licensed fur harvesters or as road casualties and had considerable exposure to soil saprophytes. Culturing M. bovis from among the many competing contaminant bacteria and fungi was sometimes challenging, a point noted by others (Berentsen et al., 2011). That said, only two of our 1522 (0.1%) culture samples returned no final results because of contamination. Repeated cultures with progressively stronger decontamination steps were sometimes necessary to eventually isolate M. bovis. Such decontamination could potentially lower the sensitivity of culture, and so the detection of bTB in furbearers at the population level. The desirability of obtaining fresh samples to minimize this problem has been discussed (VerCauteren et al., 2008), but it may be difficult to attain in state agency surveillance where animals harvested by the public are the most common source of samples. In addition, the potential disadvantages (e.g. limitations on inference from results) due to the smaller sample sizes of freshly-trapped animals need to be weighed against the theoretical advantage of increased sensitivity, which is thus far poorly quantified. When our analyses are restricted to histopathology versus culture in coyotes only, the resulting 95% confidence intervals for sensitivity [38.8% (18.3–63.9)], specificity [97.7% (95.4–98.9)], and positive and negative predictive values [46.7% (22.3–72.6) and 96.9% (94.3–98.3), respectively] overlap the confidence limits for freshly-trapped samples derived from Berentsen et al. (2011, Table 1) above. This suggests there is no significant difference between the diagnostic performances of histopathology on freshly-trapped animals versus publicly harvested carcasses. Indeed, the reliability of histopathology (as determined by positive likelihood ratios) was virtually identical [17.0 (6.9–41.6) in this study; 17.1 (5.7–50.7) in Berentsen et al., 2011], indicating that, at least for histopathology in coyotes, the use of fresh tissues does not result in more reliable diagnosis of bTB. Using furbearers in bTB surveillance presents other challenges as well. When state agencies design surveillance programmes for bTB, they frequently set out to determine whether the disease is present in a given proportion of the population with a set level of statistical confidence. Routine sample size tables (e.g. Beal, 1997) can be used to determine how many animals need to be tested in order to attain a particular power of bTB detection. It is relatively straightforward then to interpret the results of surveillance and report those interpretations to federal bTB regulators. However, the sampling parameters for testing furbearers in 70
order to make inference to the prevalence of bTB in a maintenance host are unknown. That, coupled with the fact that the furbearers themselves are extremely unlikely to be the source of livestock herd infections, makes it difficult to interpret what it means epidemiologically when M. bovis is detected in a furbearer. Some type of follow-up surveillance is likely to be necessary in order to determine what other wildlife species is maintaining bTB in that ecosystem. Interpretation is further obscured because the source of exposure for a bTB-positive furbearer is uncertain. As opportunistic scavengers, all of the Michigan furbearers that have cultured positive to date could have become infected by scavenging carcasses of inapparently-infected livestock as well as by scavenging free-ranging wildlife. In short, epidemiological interpretation of findings from bTB surveillance in furbearers is unlikely to be straightforward. Consequently, the stated objective of the USDA-APHIS guidelines ‘to determine whether bovine TB is present in wildlife’ may be of limited utility in and of itself. It also calls into question the purported advantage of using furbearers for surveillance. In many cases, easier, cheaper and more statistically interpretable alternatives to sampling furbearers are likely to exist. In Michigan, for example, where the bTB maintenance host, WTD, is known (and in contrast to what was anticipated by de Lisle et al., 2002), experience has shown it to be far more efficient and cost-effective to test deer than any other species. Surveillance samples from publicly harvested deer are more numerous and more widely distributed geographically than are samples from publicly harvested furbearers such as coyotes. Between 1997 and 2008, the mean annual harvest of coyotes in Michigan was 19454, (Frawley, 2001a, 2002, 2003, 2004a, 2006, 2007a,b, 2008, 2012) compared to the mean annual deer harvest of 491143 (Frawley, 1999, 2001b, 2004b, 2007c, 2009) over the same period. Harvest regulations stipulate which species of furbearers must be registered with MDNR and the pelts sealed before they may be sold. Coyotes, raccoons and opossums are not among them. Consequently, public harvest of these three species cannot be relied upon as a source of samples for bTB surveillance. Without the availability of harvested carcasses presented to MDNR, testing would have to rely on contributions from volunteers or some agency-furnished incentive for trappers to present carcasses for testing. Neither of these options is likely to reliably deliver large numbers of furbearers from the specific geographic areas in which bTB surveillance is required, and the payment of incentives may encourage submissions from outside of the area sought, resulting in bias. Alternatively, the agency would have to expend resources to trap the animals themselves from the desired surveillance locations. For these reasons, furbearers are unlikely to play any useful role in Michigan bTB surveillance for the foreseeable future. Such
© 2013 Blackwell Verlag GmbH • Transboundary and Emerging Diseases. 60 (Suppl. 1) (2013) 67–73
D. J. O’Brien et al.
challenges to using furbearers in surveillance are not unique to Michigan. This is not to say, however, that furbearer surveillance will necessarily be of as limited value in other areas as it currently is in Michigan. Furbearers could prove useful for screening surveillance in presumed bTB-free areas where livestock herds have become infected for the first time, if many furbearer samples are easily and cheaply available, and samples from likely maintenance hosts such as deer or elk are not. New Zealand researchers have also demonstrated how spillover hosts such as furbearers can be important components in multispecies complexes that allow long-term persistence of bTB and its long-range geographic spread (Nugent, 2011). In such situations, surveillance of furbearers will be important. But in all cases, state wildlife management agencies that choose to base their bTB surveillance, and the decisions it informs, on furbearers, should keep in mind the performance limitations of common diagnostic tests as noted in our study. Moreover, wildlife managers should be aware of factors noted here that may abate the utility of bTB surveillance in furbearers and complicate its interpretation. Acknowledgements We thank D. E. Berry, Mycobacteriology Laboratory, Michigan Department of Community Health, and J. B. Payeur, National Veterinary Services Laboratory, U.S. Department of Agriculture Animal and Plant Health Inspection Service for culture of samples. This work was supported in part by the Federal Aid in Wildlife Restoration Act under Michigan Pittman-Robertson Project W-147-R. Conflicts of Interest The authors have no conflicts of interest to declare. References Atwood, T. C., K. C. Vercauteren, T. J. Deliberto, H. J. Smith, and J. S. Stevenson, 2007: Coyotes as sentinels for monitoring bovine tuberculosis prevalence in white-tailed deer. J. Wildl. Manage. 71, 1545–1554. Atwood, T. C., T. J. Deliberto, H. J. Smith, J. S. Stevenson, and K. C. Vercauteren, 2009: Spatial ecology of raccoons related to cattle and bovine tuberculosis in northeastern Michigan. J. Wildl. Manage. 73, 647–654. Beal, V. C., 1997: Efficacy of random samples of herds in Brucella or TB detection. Table 3. Number of free ranging cervidae tested or observed in order to detect infected cervids by random sampling. In: Regulatory Statistics, Volume 2B – Considerations About Animal Disease Program Evaluation and Surveillance Theory, 2nd edn. addendum to the preface, pp. 3. Veterinary Services, Animal and Plant Health Inspec-
bTB Test Performance in Furbearers
tion Service, U.S. Department of Agriculture, Hyattsville, Maryland. Berentsen, A. R., M. R. Dunbar, D. R. Marks, and S. Robbe-Austerman, 2010: Evaluation of potential shedding of Mycobacterium bovis in free-ranging raccoons. In: Timm, R. M., and K. A. Fagerstone (eds), Proceedings of the 24th Vertebrate Pest Conference, pp. 312–314. University of California-Davis, Sacramento. Berentsen, A. R., M. R. Dunbar, S. R. Johnson, S. RobbeAusterman, L. Martinez, and R. L. Jones, 2011: Active use of coyotes (Canis latrans) to detect bovine tuberculosis in northeastern Michigan. USA. Vet. Microbiol. 151, 126–132. Bruning-Fann, C. S., S. M. Schmitt, S. D. Fitzgerald, J. B. Payeur, D. L. Whipple, T. M. Cooley, T. Carlson, and P. Friedrich, 1998: Mycobacterium bovis in coyotes from Michigan. J. Wildl. Dis. 34, 632–636. Bruning-Fann, C. S., S. M. Schmitt, S. D. Fitzgerald, J. S. Fierke, P. D. Friedrich, J. B. Kaneene, K. A. Clarke, K. L. Butler, J. B. Payeur, D. Whipple, T. M. Cooley, J. M. Miller, and D. P. Muzo, 2001: Bovine tuberculosis in free-ranging carnivores from Michigan. J. Wild. Dis. 37, 58–64. Carstensen, M., and M. W. DonCarlos, 2011: Preventing the establishment of a wildlife disease reservoir: a case study of bovine tuberculosis in wild deer in Minnesota. USA. Vet. Med. Int. doi:10.4061/2011/413240. Available at http://www. hindawi.com/journals/vmi/mycobacterial.diseases.animals/ Corner, L. A. L., 2006: The role of wild animal populations in the epidemiology of tuberculosis in domestic animals: how to assess the risk. Vet. Microbiol. 112, 303–312. DeLiberto, T., T. Gidlewski, and K. Pedersen, 2011: Guidelines for Surveillance of Bovine Tuberculosis in Wildlife. Wildlife Services, Animal and Plant Health Inspection Service, U.S. Department of Agriculture. Available at http://www.aphis. usda.gov/animal_health/animal_diseases/tuberculosis/downloads/wildlife_tb_surv_manual.pdf (accessed November 1, 2012). Fagan, T. J., 1975: Nomogram for Bayes’s theorem. N. Engl. J. Med. 293, 257. Fenton, F. A., S. D. Fitzgerald, S. Bolin, J. B. Kaneene, J. Sikarskie, R. Greenwald, and K. P. Lyashchenko, 2012: Experimental aerosol inoculation and investigation of potential lateral transmission of Mycobacterium bovis in Virginia opossum (Didelphis virginiana). Vet. Med. Int. doi:10.1155/2012/ 842861. Available at http://www.hindawi.com/journals/vmi/ mycobacterial.diseases.animals/ Fitzgerald, S. D., J. B. Kaneene, K. L. Butler, K. R. Clarke, J. S. Fierke, S. M. Schmitt, C. S. Bruning-Fann, R. R. Mitchell, D. E. Berry, and J. B. Payeur, 2000: Comparison of postmortem techniques for the detection of Mycobacterium bovis in whitetailed deer (Odocoileus virginianus). J. Vet. Diagn. Invest. 12, 322–327. Francis, J., 1958: Tuberculosis in Animals and Man. A Study in Comparative Pathology. Cassell, London, UK. Frawley, B. J., 1999: Michigan Deer Harvest Survey Report, 1998 Seasons. Wildlife Division Report No. 3303. Michigan
© 2013 Blackwell Verlag GmbH • Transboundary and Emerging Diseases. 60 (Suppl. 1) (2013) 67–73
71
bTB Test Performance in Furbearers
D. J. O’Brien et al.
Department of Natural Resources, Lansing. Available at http://ww2.dnr.state.mi.us/publications/pdfs/HuntingWildlifeHabitat/Reports/WLD-library/3301-3399/3303.pdf (accessed November 1, 2012). Frawley, B. J., 2001a: 1997-2000 Michigan Furbearer Harvest Survey. Wildlife Division Report No. 3355. Michigan Department of Natural Resources, Lansing. Available at http://ww2. dnr.state.mi.us/publications/pdfs/HuntingWildlifeHabitat/ Reports/WLD-library/3301-3399/3355.pdf (accessed November 1, 2012). Frawley, B. J., 2001b: Michigan Deer Harvest Survey Report, 2000 Seasons. Wildlife Division Report No. 3344. Michigan Department of Natural Resources, Lansing. Available at http://ww2.dnr.state.mi.us/publications/pdfs/HuntingWildlifeHabitat/Reports/WLD-library/3301-3399/3344.pdf (accessed November 1, 2012). Frawley, B. J., 2002: 2001 Michigan Furbearer Harvest Survey. Wildlife Division Report No. 3379. Michigan Department of Natural Resources, Lansing. Available at http://ww2.dnr.state. mi.us/publications/pdfs/HuntingWildlifeHabitat/Reports/ WLD-library/3301-3399/3379.pdf (accessed November 1, 2012). Frawley, B. J., 2003: 2002 Michigan Furbearer Harvest Survey. Wildlife Division Report No. 3410. Michigan Department of Natural Resources, Lansing. Available at http://ww2.dnr.state. mi.us/publications/pdfs/HuntingWildlifeHabitat/Reports/ WLD-library/3400-3499/3410.pdf (accessed November 1, 2012). Frawley, B. J., 2004a: 2003 Michigan Furbearer Harvest Survey. Wildlife Division Report No. 3421. Michigan Department of Natural Resources, Lansing. Available at http://ww2.dnr.state. mi.us/publications/pdfs/HuntingWildlifeHabitat/Reports/ WLD-library/3400-3499/3421.pdf (accessed November 1, 2012). Frawley, B. J., 2004b: Michigan Deer Harvest Survey Report, 2003 Seasons. Wildlife Division Report No. 3418. Michigan Department of Natural Resources, Lansing. Available at http://ww2.dnr.state.mi.us/publications/pdfs/HuntingWildlifeHabitat/Reports/WLD-library/3400-3499/3418.pdf (accessed November 1, 2012). Frawley, B. J., 2006: 2004 Michigan Furbearer Harvest Survey. Wildlife Division Report No. 3459. Michigan Department of Natural Resources, Lansing. Available at http://ww2.dnr.state. mi.us/publications/pdfs/HuntingWildlifeHabitat/Reports/ WLD-library/3400-3499/3459%20.pdf (accessed November 1, 2012). Frawley, B. J., 2007a: 2005 Michigan Furbearer Harvest Survey. Wildlife Division Report No. 3472. Michigan Department of Natural Resources, Lansing. Available at http://ww2.dnr.state. mi.us/publications/pdfs/HuntingWildlifeHabitat/Reports/ WLD-library/3400-3499/3472.pdf (accessed November 1, 2012). Frawley, B. J., 2007b: 2006 Michigan Furbearer Harvest Survey. Wildlife Division Report No. 3480. Michigan Department of Natural Resources, Lansing. Available at http://ww2.dnr.state.
72
mi.us/publications/pdfs/HuntingWildlifeHabitat/Reports/ WLD-library/3400-3499/3480.pdf (accessed November 1, 2012). Frawley, B. J., 2007c: Michigan Deer Harvest Survey Report, 2006 Seasons. Wildlife Division Report No. 3467. Michigan Department of Natural Resources, Lansing. Available at http://ww2.dnr.state.mi.us/publications/pdfs/HuntingWildlifeHabitat/Reports/WLD-library/3400-3499/3467.pdf (accessed November 1, 2012). Frawley, B. J., 2008: 2007 Michigan Furbearer Harvest Survey. Wildlife Division Report No. 3494. Michigan Department of Natural Resources, Lansing. Available at http://ww2.dnr.state. mi.us/publications/pdfs/HuntingWildlifeHabitat/Reports/ WLD-library/3400-3499/3494.pdf (accessed November 1, 2012). Frawley, B. J., 2009: Michigan Deer Harvest Survey Report, 2008 Seasons. Wildlife Division Report No. 3499. Michigan Department of Natural Resources, Lansing. Available at http://ww2.dnr.state.mi.us/publications/pdfs/HuntingWildlifeHabitat/Reports/WLD-library/3400-3499/3499.pdf (accessed November 1, 2012). Frawley, B. J., 2012: 2008 Michigan Furbearer Harvest Survey. Wildlife Division Report No. 3533. Michigan Department of Natural Resources, Lansing. Available at http://www.michigan.gov/documents/dnr/WLD_report_3533_2008_Michigan_furbearer_harvest_survey_388066_7.pdf (accessed November 1, 2012). Johnson, S. R., M. R. Dunbar, L. Martinez, R. L. Jones, R. Bowen, and P. Gordy, 2008: Experimental inoculation of coyotes with Mycobacterium bovis: susceptibility and shedding. In: Gower, M. R. (ed.), Proceedings of the One Hundred and Twelfth Annual Meeting of the United States Animal Health Association, pp. 94–97. United States Animal Health Association, St. Joseph, MO, and Richardson Printing, Kansas City, MO. Kent, P. T., and G. P. Kubica, 1985: Public Health Mycobacteriology: A Guide for the Level III Laboratory. U.S. Department of Health and Human Services, Atlanta, Georgia. de Lisle, G. W., R. G. Bengis, S. M. Schmitt, and D. J. O’Brien, 2002: Tuberculosis in free-ranging wildlife: detection, diagnosis and management. Rev. Sci. Tech. O.I.E. 21, 317–334. Lowry, R., 2012: VassarStats: Web site for Statistical Computation. Clinical Research Calculator 1. From an Observed Sample: Estimates of Population Prevalence, Sensitivity, Specificity, Predictive Values, and Likelihood Ratios. Available at http://vassarstats.net (accessed November 1, 2012). Newcombe, R. G., 1998: Two-sided confidence intervals for the single proportion: a comparison of seven methods. Stat. Med. 17, 857–872. Nugent, G., 2011: Maintenance, spillover and spillback transmission of bovine tuberculosis in multi-host wildlife complexes: a New Zealand case study. Vet. Microbiol. 151, 34–42. O’Brien, D. J., S. M. Schmitt, J. S. Fierke, S. A. Hogle, S. R. Winterstein, T. M. Cooley, W. E. Moritz, K. L. Diegel, S. D. Fitzgerald, D. E. Berry, and J. B. Kaneene, 2002: Epidemiology of
© 2013 Blackwell Verlag GmbH • Transboundary and Emerging Diseases. 60 (Suppl. 1) (2013) 67–73
D. J. O’Brien et al.
Mycobacterium bovis in free-ranging white-tailed deer, Michigan, USA, 1995–2000. Prev. Vet. Med. 54, 47–63. O’Brien, D. J., S. M. Schmitt, D. E. Berry, S. D. Fitzgerald, J. R. Vanneste, T. J. Lyon, D. Magsig, J. S. Fierke, T. M. Cooley, L. S. Zwick, and B. V. Thomsen, 2004: Estimating the true prevalence of Mycobacterium bovis in hunter-harvested white-tailed deer in Michigan. J. Wildl. Dis. 40, 42–52. O’Brien, D. J., S. M. Schmitt, S. D. Fitzgerald, D. E. Berry, and G. J. Hickling, 2006: Managing the wildlife reservoir of Mycobacterium bovis: the Michigan, USA, experience. Vet. Microbiol. 112, 313–323. O’Brien, D. J., S. M. Schmitt, S. D. Fitzgerald, and D. E. Berry, 2011: Management of bovine tuberculosis in Michigan wildlife: current status and near term prospects. Vet. Microbiol. 151, 179–187. Olmstead, A. L., and P. W. Rhode, 2004: An impossible undertaking: the eradication of bovine tuberculosis in the United States. J. Econ. Hist. 64, 734–772. O’Reilly, L. M., and C. J. Daborn, 1995: The epidemiology of Mycobacterium bovis infections in animals and man: a review. Tuber. Lung Dis. 76, 1–46. Palmer, M. V., 2007: Tuberculosis: a reemerging disease at the interface of domestic animals and wildlife. Curr. Top. Microbiol. Immunol. 315, 195–215. Palmer, M. V., and W. R. Waters, 2011: Bovine tuberculosis and the establishment of an eradication program in the United States: role of veterinarians. Vet. Med. Int. doi:10.4061/2011/ 816345. Available at http://www.hindawi.com/journals/vmi/ mycobacterial.diseases.animals/ Palmer, M. V., W. R. Waters, and D. L. Whipple, 2002: Susceptibility of raccoons (Procyon lotor) to infection with Mycobacterium bovis. J. Wildl. Dis. 38, 266–274. Sangster, C., D. Bergeson, C. Lutze-Wallace, V. Crichton, and G. Wobeser, 2007: Feasibility of using coyotes (Canis latrans) as
bTB Test Performance in Furbearers
sentinels for bovine mycobacteriosis (Mycobacterium bovis) infection in wild cervids in and around Riding Mountain National Park, Manitoba. Canada. J. Wildl. Dis. 43, 432–438. Scherokman, B., 1997: Selecting and interpreting diagnostic tests. Permanente J. 1, 4–7. Schmitt, S. M., S. D. Fitzgerald, T. M. Cooley, C. S. BruningFann, L. Sullivan, D. Berry, T. Carlson, R. B. Minnis, J. B. Payeur, and J. Sikarskie, 1997: Bovine tuberculosis in freeranging white-tailed deer from Michigan. J. Wildl. Dis. 33, 749–758. Schmitt, S. M., D. J. O’Brien, C. S. Bruning-Fann, and S. D. Fitzgerald, 2002: Bovine tuberculosis in Michigan wildlife and livestock. Ann. N. Y. Acad. Sci. 969, 262–268. Shury, T. K., and D. Bergeson, 2011: Lesion distribution and epidemiology of Mycobacterium bovis in elk and white-tailed deer in south-western Manitoba. Canada. Vet. Med. Int. doi:10. 4061/2011/591980. Available at http://www.hindawi.com/ journals/vmi/mycobacterial.diseases.animals/ Thorne, E. T., R. A. Humphries, D. J. O’Brien, and S. M. Schmitt, 2005: State wildlife management agency responsibility for managing diseases in free-ranging wildlife. In: Rahm, J. (ed.), Transactions of the 70th North American Wildlife and Natural Resources Conference. pp. 334–345.Wildlife Management Institute, Arlington, Virginia. Thrusfield, M., 1995: Veterinary Epidemiology, 2nd edn. Blackwell Science, Oxford, UK. VerCauteren, K. C., T. C. Atwood, T. J. DeLiberto, H. J. Smith, J. S. Stevenson, B. V. Thomsen, T. Gidlewski, and J. Payeur, 2008: Sentinel-based surveillance of coyotes to detect bovine tuberculosis. Michigan. Emerging Infect. Dis. 14, 1862–1869. Witmer, G., A. E. Fine, J. Gionfriddo, M. Pipes, K. Shively, K. Piccolo, and P. Burke, 2010: Epizootiologic survey of Mycobacterium bovis in wildlife and farm environments in northern Michigan. J. Wildl. Dis. 46, 368–378.
© 2013 Blackwell Verlag GmbH • Transboundary and Emerging Diseases. 60 (Suppl. 1) (2013) 67–73
73
ORIGINAL ARTICLE
Performance of Diagnostic Tests for Bovine Tuberculosis in North American Furbearers and Implications for Surveillance D. J. O’Brien1, J. S. Fierke1, T. M. Cooley1, S. D. Fitzgerald2, M. K. Cosgrove1 and S. M. Schmitt1 1 2
Michigan Department of Natural Resources, Wildlife Disease Laboratory, Lansing, MI, USA Department of Pathobiology and Diagnostic Investigation, Michigan State University, East Lansing, MI, USA
Keywords: bovine tuberculosis; diagnosis; furbearing mammals; Mycobacterium bovis; spillover hosts; wildlife Correspondence: D. J. O’Brien. Wildlife Disease Laboratory, Michigan Department of Natural Resources, 4125 Beaumont Road, Room 250, Lansing, MI 48910-8106, USA. Tel.: +011 517 336 5035; Fax: +011 517 337 4920; E-mail: [email protected] Received for publication November 13, 2012 doi:10.1111/tbed.12093
Summary Risks of bovine tuberculosis (bTB) transmission from free-ranging wildlife to livestock remain a concern in the United States, in both known endemic areas and where spillover from recently-infected livestock herds occurs. Federal agriculture officials in the United States (US) have recommended surveillance of non-cervid furbearers to determine whether free-ranging wildlife in the vicinity of cattle herd breakdowns are bTB infected, yet the efficacy of common diagnostic tests in these species is largely unknown. We calculated the sensitivity, specificity, predictive values and positive likelihood ratios for bTB infection in carcasses of sixteen species of furbearers tested via: (i) the presence of gross lesions compatible with bTB; (ii) histopathology consistent with bTB; and (iii) the presence of acid-fast bacilli (AFB) on histopathology. The gold standard comparison test was mycobacterial culture of cranial visceral lymph nodes pooled for each animal. Forty-two animals distributed across six species cultured bTB positive from among 1522 furbearers tested over thirteen years. The sensitivity of all three tests was poor (10%, 22% and 24% for gross lesions, AFB and histopathology, respectively), while specificities (all 99%) and negative predictive values (all 97%) were high. Positive predictive values varied widely (31–75%). Likelihood ratios for culture positivity given a positive test result showed AFB on histopathology to be the most reliable test, and gross lesions the least, though confidence intervals were wide and overlapping. While non-cervid furbearers may prove useful in North American bTB surveillance, wildlife managers should be aware of factors that may abate their utility and complicate interpretation of surveillance.
Introduction A century ago, tuberculosis was the leading cause of human mortality in the United States (US), claiming nearly 150 000 lives in 1900 (Olmstead and Rhode, 2004). Although characterized at its outset in 1917 as ‘an impossible undertaking,’ the programme to eradicate bovine tuberculosis (bTB), caused by Mycobacterium bovis, from US cattle has been enormously successful, greatly reducing both the burden of human disease and the disease-related economic costs to livestock agriculture (Olmstead and Rhode, 2004; Palmer and Waters, 2011). However, global experience has demonstrated that eradication of bTB is
frequently complicated, and sometimes prevented entirely, by the establishment and persistence of M. bovis in freeranging wildlife reservoirs (Palmer, 2007), and bTB eradication efforts in North America are no exception (O’Brien et al., 2011; Shury and Bergeson, 2011). Recognizing the importance of wildlife reservoirs to nationwide bTB eradication plans, in 2011, the US Department of Agriculture’s Animal and Plant Health Inspection Service (USDA-APHIS) formulated and released guidelines for surveillance of bovine tuberculosis in wildlife (DeLiberto et al., 2011). The principal stated objective of those guidelines is ‘to determine whether bovine TB is present in wildlife’ in areas where bTB has been detected in a dairy,
© 2013 Blackwell Verlag GmbH • Transboundary and Emerging Diseases. 60 (Suppl. 1) (2013) 67–73
67
bTB Test Performance in Furbearers
D. J. O’Brien et al.
beef or captive cervid herd (DeLiberto et al., 2011, p. 4). These guidelines place considerable emphasis on surveillance of furbearing mammals of the Orders Didelphimorphia [e.g. Virginia opossums (Didelphis virginiana)] and Carnivora [e.g. raccoons (Procyon lotor), coyotes (Canis latrans)]. The basis for this emphasis is varied. Testing of ‘resident’ furbearers such as opossums and raccoons is ‘highly recommended’ to ‘clean and disinfect’ an infected livestock farm, despite the fact that ‘current research does not indicate that peridomestic mammals pose a risk of disease transmission to domestic livestock’ (DeLiberto et al., 2011, p. 6). The guidelines suggest testing of ‘transient’ furbearers such as coyotes which ‘may have greater potential to spread bovine TB through indirect contact (e.g. contamination of feed) or serve as sentinels for the disease,’ as well as ‘to assist in delineating the extent of bovine TB on the landscape’ (DeLiberto et al., 2011, p. 7). While the USDA-APHIS guidelines ‘were developed to provide guidance for State and Federal wildlife agencies that plan to conduct surveillance for bovine tuberculosis (TB) in wildlife’ (DeLiberto et al., 2011, p. 4), thus far, when such situations have arisen, the state agencies have typically been charged with that responsibility (see, e.g. Carstensen and DonCarlos, 2011), perhaps because of their statutory authority for managing non-migratory game species (Thorne et al., 2005). State wildlife management agencies frequently and perpetually lack adequate funding and personnel to carry out wildlife disease surveillance at an optimal level, particularly when unanticipated events such as bTB-infected livestock herds arise. Given the emphasis placed on furbearer surveillance by USDA-APHIS, it is conceivable that states might choose to rely on testing of furbearers as a primary basis for drawing conclusions concerning bTB surveillance and management. Yet, the efficacy of common screening diagnostic tests for bTB in these species is unknown. Consequently, our objective in this study was to establish the diagnostic sensitivity, specificity, predictive values and likelihood ratios for three common tests of infection with M. bovis in sixteen species of North American furbearers. Materials and Methods We drew on a large historical database of gross pathology, histopathology and mycobacterial culture results from Michigan Department of Natural Resources (MDNR) surveillance carried out in 1996–2008 in the northern half of Michigan’s Lower Peninsula, with sampling efforts concentrated in and around Deer Management Unit (DMU) 452, considered the core area of the bTB outbreak in whitetailed deer (WTD) and cattle (see O’Brien et al., 2002; Fig. 1). Details of furbearer sampling (Bruning-Fann et al., 1998, 2001; Schmitt et al., 2002; O’Brien et al., 2006), gross 68
and histopathology examinations (Bruning-Fann et al., 1998; Fitzgerald et al., 2000) and mycobacterial culture procedures (Schmitt et al., 1997; O’Brien et al., 2004) have all been described elsewhere. Not all tests were performed on all the cultured samples. We compared the diagnostic performance of three screening tests against the gold standard of mycobacterial culture: (i) the presence of gross lesions compatible with mycobacteriosis; (ii) histopathology consistent with M. bovis infection; and (iii) the presence of acid-fast bacteria (AFB) on histopathology as identified by the Ziehl–Neelson staining technique (Kent and Kubica, 1985). Diagnostic sensitivity, specificity, positive and negative predictive values (alternatively called the predictive value of a positive or negative test) were calculated as described by Thrusfield (1995), using freely available software (Lowry, 2012). Sensitivity is the proportion of infected (i.e. bTB culture positive) furbearers that are positive on a screening test. Specificity is the proportion of uninfected that test negative on screening. Positive and negative predictive values are the probabilities that a screening test positive is actually bTB infected, or a screening test negative actually is uninfected, respectively. We also calculated the positive likelihood ratio (Fagan, 1975), an index of how reliably a screening test detects bTB infection (Scherokman, 1997). The ratio of two conditional probabilities [the true-positive rate (sensitivity) divided by the false-positive rate (1 specificity); Lowry, 2012], it is essentially the odds that a positive test result would be expected from a genuinely bTB-infected furbearer. Results From 1996 through 2008, 1522 furbearers distributed across sixteen species (Schmitt et al., 2002; Table 2) were cultured for M. bovis. Of those, 42 (2.8%) cultured positive: 7/216 (3.2%) black bears (Ursus americanus); 3/56 (5.4%) bobcats (Lynx rufus); 19/380 (5.0%) coyotes; 2/379 (0.5%) opossums; 8/335 (2.4%) raccoons; and 3/29 (10%) red foxes (Vulpes vulpes). One other bobcat was bTB positive on polymerase chain reaction (PCR), but was culture negative (Bruning-Fann et al., 2001). All other species were culture negative [although a single grey fox (Urocyon cinereoargenteus) trapped by other researchers cultured positive in an unrelated study (Witmer et al., 2010), Table 1]. Performance of the three bTB screening tests compared to mycobacterial culture is summarized in Table 1. All were quite insensitive for detecting M. bovis infection, with histopathology the best of the three at only 24%. Specificities and negative predictive values were uniformly high, ranging between 99–100% and 97–98%, respectively. Positive predictive values varied widely (31–75%). Likelihood ratios for culture positivity given a positive screening test result
© 2013 Blackwell Verlag GmbH • Transboundary and Emerging Diseases. 60 (Suppl. 1) (2013) 67–73
D. J. O’Brien et al.
bTB Test Performance in Furbearers
showed the presence of AFB on histology to be the most reliable test, and the presence of gross lesions the least, although confidence intervals were wide and overlapping. Discussion Because it is an exemplary multihost pathogen (Francis, 1958; O’Reilly and Daborn, 1995), adequately responding to initial diagnosis of M. bovis in free-ranging wildlife necessitates determining not simply whether wildlife are collectively infected, but which species are, and are not, likely to be playing a significant role in the epidemiology of a particular outbreak (Corner, 2006). It was with that knowledge in mind that MDNR undertook large-scale bTB surveillance in multiple species beginning in 1995. The identification of furbearers as maintenance hosts of bTB in Ireland, Great Britain and New Zealand supported the plausibility, at least conceptually, of North American furbearers as potential maintenance hosts. However, that plausibility has yet to be validated. Examination of pathology and mycobacterial culture results from early in the course of surveillance led to the conclusion that furbearers in Michigan were likely to be spillover hosts only (BruningFann et al., 1998, 2001) and were likely playing no significant role in the maintenance of bTB nor its transmission to livestock (de Lisle et al., 2002; Schmitt et al., 2002). Since that time, data from a variety of experimental (Palmer et al., 2002; Johnson et al., 2008; Berentsen et al., 2010; Fenton et al., 2012) and field (Berentsen et al., 2011) studies in multiple species have thus far confirmed those early conclusions. Although unsupported by the accumulated data, some continue to believe that furbearers play a significant role in cattle herd infections (Atwood et al., 2009; Witmer et al., 2010). Coyotes were originally proposed as potential sentinel animals (de Lisle et al., 2002) because some bTB-positive coyotes were found in areas where bTB prevalence in WTD was very low (Schmitt et al., 2002; Fig. 2). Subsequent studies have produced mixed results however. A USDA group working in Michigan (Atwood et al., 2007; VerCauteren et al., 2008; Berentsen et al., 2011) concluded that coyotes
are good sentinels. However, researchers in Manitoba, Canada [another location like Michigan where bTB is present in both WTD and elk (Cervus elaphus)] found coyotes to be insensitive as sentinels there (Sangster et al., 2007). Even in Michigan, coyotes proved relatively insensitive as sentinels for bTB in WTD; only 1 of 17 (6%) M. bovis-positive coyotes came from a county where bTB had not already been detected via surveillance in deer (Berentsen et al., 2011). Thus in our view, considerable uncertainty remains to be resolved before furbearers can be confidently used in sentinel surveillance. In this study, the insensitivity of the three screening tests for detecting bTB infection was disappointing, but not particularly surprising. Both our previous reports (BruningFann et al., 2001; Schmitt et al., 2002; O’Brien et al., 2006) and those of others (Berentsen et al., 2011) have noted the paucity of gross and histopathological lesions of bTB in furbearers. Here, 13 of 1489 (0.9%), 24 of 1482 (1.6%) and 12 of 1483 (0.8%) samples were positive for gross lesions, histopathology and AFB, respectively, compared to 1 of 175 (0.6%) and 12 of 175 (6.9%) that had gross and histopathological lesions, respectively, in Berentsen et al. (2011). Even when gross or histologic lesions were present in our study, positive predictive values indicate only a 31–42% probability that those lesioned samples actually came from an animal with bTB. The presence of AFB on histopathology was a better predictor; samples with AFB had a 75% probability of originating from an M. bovis-infected animal. In addition, positive likelihood ratios indicate AFB to be the most reliable of the three tests, though differences between tests were not statistically significant. The uniformly high specificities and negative predictive values across all three tests are likely due to both the low sensitivities of the tests and the low prevalence of M. bovis infection. While other studies of Michigan furbearers have not specifically reported the performance of their diagnostic tests, histopathology and culture results from coyotes in Berentsen et al. (2011, Table 1) can be analysed by the same methods used here, yielding sensitivity of 47.1% (95% confidence limits: 23.9–71.5), specificity of 97.2% (92.6–99.1), and positive and negative predictive values of 66.7%
Table 1. Performance* of three diagnostic tests for Mycobacterium bovis versus culture in furbearers, Michigan, USA, 1996–2008 Predictive value Test
Sensitivity
Specificity
Positive
Negative
Positive likelihood ratio
Gross lesions Histopathology Acid-fast bacteria
9.5 (3.1–23.5) 24.4 (12.9–40.6) 22.0 (11.1–38.0)
99.4 (98.8–99.7) 99.0 (98.3–99.4) 99.8 (99.3–99.9)
30.8 (10.4–61.1) 41.7 (22.8–63.1) 75.0 (42.8–93.3)
97.4 (96.4–98.1) 97.9 (97.0–98.5) 97.8 (96.9–98.5)
15.3 (4.9–47.7) 25.1 (11.9–53.1) 105.5 (29.7–375.4)
*Point estimates expressed as percentages with 95% confidence limits calculated by the continuity-corrected efficient-score method (Newcombe, 1998).
© 2013 Blackwell Verlag GmbH • Transboundary and Emerging Diseases. 60 (Suppl. 1) (2013) 67–73
69
bTB Test Performance in Furbearers
D. J. O’Brien et al.
(35.4–88.7) and 94% (88.6–97.0), respectively. While those point estimates show somewhat higher sensitivity and positive predictive values for histopathology than our results, sensitivity is still poor, and confidence intervals overlap those of our study, suggesting little appreciable difference in test performance across studies. In our study, most furbearer samples were obtained skinned from licensed fur harvesters or as road casualties and had considerable exposure to soil saprophytes. Culturing M. bovis from among the many competing contaminant bacteria and fungi was sometimes challenging, a point noted by others (Berentsen et al., 2011). That said, only two of our 1522 (0.1%) culture samples returned no final results because of contamination. Repeated cultures with progressively stronger decontamination steps were sometimes necessary to eventually isolate M. bovis. Such decontamination could potentially lower the sensitivity of culture, and so the detection of bTB in furbearers at the population level. The desirability of obtaining fresh samples to minimize this problem has been discussed (VerCauteren et al., 2008), but it may be difficult to attain in state agency surveillance where animals harvested by the public are the most common source of samples. In addition, the potential disadvantages (e.g. limitations on inference from results) due to the smaller sample sizes of freshly-trapped animals need to be weighed against the theoretical advantage of increased sensitivity, which is thus far poorly quantified. When our analyses are restricted to histopathology versus culture in coyotes only, the resulting 95% confidence intervals for sensitivity [38.8% (18.3–63.9)], specificity [97.7% (95.4–98.9)], and positive and negative predictive values [46.7% (22.3–72.6) and 96.9% (94.3–98.3), respectively] overlap the confidence limits for freshly-trapped samples derived from Berentsen et al. (2011, Table 1) above. This suggests there is no significant difference between the diagnostic performances of histopathology on freshly-trapped animals versus publicly harvested carcasses. Indeed, the reliability of histopathology (as determined by positive likelihood ratios) was virtually identical [17.0 (6.9–41.6) in this study; 17.1 (5.7–50.7) in Berentsen et al., 2011], indicating that, at least for histopathology in coyotes, the use of fresh tissues does not result in more reliable diagnosis of bTB. Using furbearers in bTB surveillance presents other challenges as well. When state agencies design surveillance programmes for bTB, they frequently set out to determine whether the disease is present in a given proportion of the population with a set level of statistical confidence. Routine sample size tables (e.g. Beal, 1997) can be used to determine how many animals need to be tested in order to attain a particular power of bTB detection. It is relatively straightforward then to interpret the results of surveillance and report those interpretations to federal bTB regulators. However, the sampling parameters for testing furbearers in 70
order to make inference to the prevalence of bTB in a maintenance host are unknown. That, coupled with the fact that the furbearers themselves are extremely unlikely to be the source of livestock herd infections, makes it difficult to interpret what it means epidemiologically when M. bovis is detected in a furbearer. Some type of follow-up surveillance is likely to be necessary in order to determine what other wildlife species is maintaining bTB in that ecosystem. Interpretation is further obscured because the source of exposure for a bTB-positive furbearer is uncertain. As opportunistic scavengers, all of the Michigan furbearers that have cultured positive to date could have become infected by scavenging carcasses of inapparently-infected livestock as well as by scavenging free-ranging wildlife. In short, epidemiological interpretation of findings from bTB surveillance in furbearers is unlikely to be straightforward. Consequently, the stated objective of the USDA-APHIS guidelines ‘to determine whether bovine TB is present in wildlife’ may be of limited utility in and of itself. It also calls into question the purported advantage of using furbearers for surveillance. In many cases, easier, cheaper and more statistically interpretable alternatives to sampling furbearers are likely to exist. In Michigan, for example, where the bTB maintenance host, WTD, is known (and in contrast to what was anticipated by de Lisle et al., 2002), experience has shown it to be far more efficient and cost-effective to test deer than any other species. Surveillance samples from publicly harvested deer are more numerous and more widely distributed geographically than are samples from publicly harvested furbearers such as coyotes. Between 1997 and 2008, the mean annual harvest of coyotes in Michigan was 19454, (Frawley, 2001a, 2002, 2003, 2004a, 2006, 2007a,b, 2008, 2012) compared to the mean annual deer harvest of 491143 (Frawley, 1999, 2001b, 2004b, 2007c, 2009) over the same period. Harvest regulations stipulate which species of furbearers must be registered with MDNR and the pelts sealed before they may be sold. Coyotes, raccoons and opossums are not among them. Consequently, public harvest of these three species cannot be relied upon as a source of samples for bTB surveillance. Without the availability of harvested carcasses presented to MDNR, testing would have to rely on contributions from volunteers or some agency-furnished incentive for trappers to present carcasses for testing. Neither of these options is likely to reliably deliver large numbers of furbearers from the specific geographic areas in which bTB surveillance is required, and the payment of incentives may encourage submissions from outside of the area sought, resulting in bias. Alternatively, the agency would have to expend resources to trap the animals themselves from the desired surveillance locations. For these reasons, furbearers are unlikely to play any useful role in Michigan bTB surveillance for the foreseeable future. Such
© 2013 Blackwell Verlag GmbH • Transboundary and Emerging Diseases. 60 (Suppl. 1) (2013) 67–73
D. J. O’Brien et al.
challenges to using furbearers in surveillance are not unique to Michigan. This is not to say, however, that furbearer surveillance will necessarily be of as limited value in other areas as it currently is in Michigan. Furbearers could prove useful for screening surveillance in presumed bTB-free areas where livestock herds have become infected for the first time, if many furbearer samples are easily and cheaply available, and samples from likely maintenance hosts such as deer or elk are not. New Zealand researchers have also demonstrated how spillover hosts such as furbearers can be important components in multispecies complexes that allow long-term persistence of bTB and its long-range geographic spread (Nugent, 2011). In such situations, surveillance of furbearers will be important. But in all cases, state wildlife management agencies that choose to base their bTB surveillance, and the decisions it informs, on furbearers, should keep in mind the performance limitations of common diagnostic tests as noted in our study. Moreover, wildlife managers should be aware of factors noted here that may abate the utility of bTB surveillance in furbearers and complicate its interpretation. Acknowledgements We thank D. E. Berry, Mycobacteriology Laboratory, Michigan Department of Community Health, and J. B. Payeur, National Veterinary Services Laboratory, U.S. Department of Agriculture Animal and Plant Health Inspection Service for culture of samples. This work was supported in part by the Federal Aid in Wildlife Restoration Act under Michigan Pittman-Robertson Project W-147-R. Conflicts of Interest The authors have no conflicts of interest to declare. References Atwood, T. C., K. C. Vercauteren, T. J. Deliberto, H. J. Smith, and J. S. Stevenson, 2007: Coyotes as sentinels for monitoring bovine tuberculosis prevalence in white-tailed deer. J. Wildl. Manage. 71, 1545–1554. Atwood, T. C., T. J. Deliberto, H. J. Smith, J. S. Stevenson, and K. C. Vercauteren, 2009: Spatial ecology of raccoons related to cattle and bovine tuberculosis in northeastern Michigan. J. Wildl. Manage. 73, 647–654. Beal, V. C., 1997: Efficacy of random samples of herds in Brucella or TB detection. Table 3. Number of free ranging cervidae tested or observed in order to detect infected cervids by random sampling. In: Regulatory Statistics, Volume 2B – Considerations About Animal Disease Program Evaluation and Surveillance Theory, 2nd edn. addendum to the preface, pp. 3. Veterinary Services, Animal and Plant Health Inspec-
bTB Test Performance in Furbearers
tion Service, U.S. Department of Agriculture, Hyattsville, Maryland. Berentsen, A. R., M. R. Dunbar, D. R. Marks, and S. Robbe-Austerman, 2010: Evaluation of potential shedding of Mycobacterium bovis in free-ranging raccoons. In: Timm, R. M., and K. A. Fagerstone (eds), Proceedings of the 24th Vertebrate Pest Conference, pp. 312–314. University of California-Davis, Sacramento. Berentsen, A. R., M. R. Dunbar, S. R. Johnson, S. RobbeAusterman, L. Martinez, and R. L. Jones, 2011: Active use of coyotes (Canis latrans) to detect bovine tuberculosis in northeastern Michigan. USA. Vet. Microbiol. 151, 126–132. Bruning-Fann, C. S., S. M. Schmitt, S. D. Fitzgerald, J. B. Payeur, D. L. Whipple, T. M. Cooley, T. Carlson, and P. Friedrich, 1998: Mycobacterium bovis in coyotes from Michigan. J. Wildl. Dis. 34, 632–636. Bruning-Fann, C. S., S. M. Schmitt, S. D. Fitzgerald, J. S. Fierke, P. D. Friedrich, J. B. Kaneene, K. A. Clarke, K. L. Butler, J. B. Payeur, D. Whipple, T. M. Cooley, J. M. Miller, and D. P. Muzo, 2001: Bovine tuberculosis in free-ranging carnivores from Michigan. J. Wild. Dis. 37, 58–64. Carstensen, M., and M. W. DonCarlos, 2011: Preventing the establishment of a wildlife disease reservoir: a case study of bovine tuberculosis in wild deer in Minnesota. USA. Vet. Med. Int. doi:10.4061/2011/413240. Available at http://www. hindawi.com/journals/vmi/mycobacterial.diseases.animals/ Corner, L. A. L., 2006: The role of wild animal populations in the epidemiology of tuberculosis in domestic animals: how to assess the risk. Vet. Microbiol. 112, 303–312. DeLiberto, T., T. Gidlewski, and K. Pedersen, 2011: Guidelines for Surveillance of Bovine Tuberculosis in Wildlife. Wildlife Services, Animal and Plant Health Inspection Service, U.S. Department of Agriculture. Available at http://www.aphis. usda.gov/animal_health/animal_diseases/tuberculosis/downloads/wildlife_tb_surv_manual.pdf (accessed November 1, 2012). Fagan, T. J., 1975: Nomogram for Bayes’s theorem. N. Engl. J. Med. 293, 257. Fenton, F. A., S. D. Fitzgerald, S. Bolin, J. B. Kaneene, J. Sikarskie, R. Greenwald, and K. P. Lyashchenko, 2012: Experimental aerosol inoculation and investigation of potential lateral transmission of Mycobacterium bovis in Virginia opossum (Didelphis virginiana). Vet. Med. Int. doi:10.1155/2012/ 842861. Available at http://www.hindawi.com/journals/vmi/ mycobacterial.diseases.animals/ Fitzgerald, S. D., J. B. Kaneene, K. L. Butler, K. R. Clarke, J. S. Fierke, S. M. Schmitt, C. S. Bruning-Fann, R. R. Mitchell, D. E. Berry, and J. B. Payeur, 2000: Comparison of postmortem techniques for the detection of Mycobacterium bovis in whitetailed deer (Odocoileus virginianus). J. Vet. Diagn. Invest. 12, 322–327. Francis, J., 1958: Tuberculosis in Animals and Man. A Study in Comparative Pathology. Cassell, London, UK. Frawley, B. J., 1999: Michigan Deer Harvest Survey Report, 1998 Seasons. Wildlife Division Report No. 3303. Michigan
© 2013 Blackwell Verlag GmbH • Transboundary and Emerging Diseases. 60 (Suppl. 1) (2013) 67–73
71
bTB Test Performance in Furbearers
D. J. O’Brien et al.
Department of Natural Resources, Lansing. Available at http://ww2.dnr.state.mi.us/publications/pdfs/HuntingWildlifeHabitat/Reports/WLD-library/3301-3399/3303.pdf (accessed November 1, 2012). Frawley, B. J., 2001a: 1997-2000 Michigan Furbearer Harvest Survey. Wildlife Division Report No. 3355. Michigan Department of Natural Resources, Lansing. Available at http://ww2. dnr.state.mi.us/publications/pdfs/HuntingWildlifeHabitat/ Reports/WLD-library/3301-3399/3355.pdf (accessed November 1, 2012). Frawley, B. J., 2001b: Michigan Deer Harvest Survey Report, 2000 Seasons. Wildlife Division Report No. 3344. Michigan Department of Natural Resources, Lansing. Available at http://ww2.dnr.state.mi.us/publications/pdfs/HuntingWildlifeHabitat/Reports/WLD-library/3301-3399/3344.pdf (accessed November 1, 2012). Frawley, B. J., 2002: 2001 Michigan Furbearer Harvest Survey. Wildlife Division Report No. 3379. Michigan Department of Natural Resources, Lansing. Available at http://ww2.dnr.state. mi.us/publications/pdfs/HuntingWildlifeHabitat/Reports/ WLD-library/3301-3399/3379.pdf (accessed November 1, 2012). Frawley, B. J., 2003: 2002 Michigan Furbearer Harvest Survey. Wildlife Division Report No. 3410. Michigan Department of Natural Resources, Lansing. Available at http://ww2.dnr.state. mi.us/publications/pdfs/HuntingWildlifeHabitat/Reports/ WLD-library/3400-3499/3410.pdf (accessed November 1, 2012). Frawley, B. J., 2004a: 2003 Michigan Furbearer Harvest Survey. Wildlife Division Report No. 3421. Michigan Department of Natural Resources, Lansing. Available at http://ww2.dnr.state. mi.us/publications/pdfs/HuntingWildlifeHabitat/Reports/ WLD-library/3400-3499/3421.pdf (accessed November 1, 2012). Frawley, B. J., 2004b: Michigan Deer Harvest Survey Report, 2003 Seasons. Wildlife Division Report No. 3418. Michigan Department of Natural Resources, Lansing. Available at http://ww2.dnr.state.mi.us/publications/pdfs/HuntingWildlifeHabitat/Reports/WLD-library/3400-3499/3418.pdf (accessed November 1, 2012). Frawley, B. J., 2006: 2004 Michigan Furbearer Harvest Survey. Wildlife Division Report No. 3459. Michigan Department of Natural Resources, Lansing. Available at http://ww2.dnr.state. mi.us/publications/pdfs/HuntingWildlifeHabitat/Reports/ WLD-library/3400-3499/3459%20.pdf (accessed November 1, 2012). Frawley, B. J., 2007a: 2005 Michigan Furbearer Harvest Survey. Wildlife Division Report No. 3472. Michigan Department of Natural Resources, Lansing. Available at http://ww2.dnr.state. mi.us/publications/pdfs/HuntingWildlifeHabitat/Reports/ WLD-library/3400-3499/3472.pdf (accessed November 1, 2012). Frawley, B. J., 2007b: 2006 Michigan Furbearer Harvest Survey. Wildlife Division Report No. 3480. Michigan Department of Natural Resources, Lansing. Available at http://ww2.dnr.state.
72
mi.us/publications/pdfs/HuntingWildlifeHabitat/Reports/ WLD-library/3400-3499/3480.pdf (accessed November 1, 2012). Frawley, B. J., 2007c: Michigan Deer Harvest Survey Report, 2006 Seasons. Wildlife Division Report No. 3467. Michigan Department of Natural Resources, Lansing. Available at http://ww2.dnr.state.mi.us/publications/pdfs/HuntingWildlifeHabitat/Reports/WLD-library/3400-3499/3467.pdf (accessed November 1, 2012). Frawley, B. J., 2008: 2007 Michigan Furbearer Harvest Survey. Wildlife Division Report No. 3494. Michigan Department of Natural Resources, Lansing. Available at http://ww2.dnr.state. mi.us/publications/pdfs/HuntingWildlifeHabitat/Reports/ WLD-library/3400-3499/3494.pdf (accessed November 1, 2012). Frawley, B. J., 2009: Michigan Deer Harvest Survey Report, 2008 Seasons. Wildlife Division Report No. 3499. Michigan Department of Natural Resources, Lansing. Available at http://ww2.dnr.state.mi.us/publications/pdfs/HuntingWildlifeHabitat/Reports/WLD-library/3400-3499/3499.pdf (accessed November 1, 2012). Frawley, B. J., 2012: 2008 Michigan Furbearer Harvest Survey. Wildlife Division Report No. 3533. Michigan Department of Natural Resources, Lansing. Available at http://www.michigan.gov/documents/dnr/WLD_report_3533_2008_Michigan_furbearer_harvest_survey_388066_7.pdf (accessed November 1, 2012). Johnson, S. R., M. R. Dunbar, L. Martinez, R. L. Jones, R. Bowen, and P. Gordy, 2008: Experimental inoculation of coyotes with Mycobacterium bovis: susceptibility and shedding. In: Gower, M. R. (ed.), Proceedings of the One Hundred and Twelfth Annual Meeting of the United States Animal Health Association, pp. 94–97. United States Animal Health Association, St. Joseph, MO, and Richardson Printing, Kansas City, MO. Kent, P. T., and G. P. Kubica, 1985: Public Health Mycobacteriology: A Guide for the Level III Laboratory. U.S. Department of Health and Human Services, Atlanta, Georgia. de Lisle, G. W., R. G. Bengis, S. M. Schmitt, and D. J. O’Brien, 2002: Tuberculosis in free-ranging wildlife: detection, diagnosis and management. Rev. Sci. Tech. O.I.E. 21, 317–334. Lowry, R., 2012: VassarStats: Web site for Statistical Computation. Clinical Research Calculator 1. From an Observed Sample: Estimates of Population Prevalence, Sensitivity, Specificity, Predictive Values, and Likelihood Ratios. Available at http://vassarstats.net (accessed November 1, 2012). Newcombe, R. G., 1998: Two-sided confidence intervals for the single proportion: a comparison of seven methods. Stat. Med. 17, 857–872. Nugent, G., 2011: Maintenance, spillover and spillback transmission of bovine tuberculosis in multi-host wildlife complexes: a New Zealand case study. Vet. Microbiol. 151, 34–42. O’Brien, D. J., S. M. Schmitt, J. S. Fierke, S. A. Hogle, S. R. Winterstein, T. M. Cooley, W. E. Moritz, K. L. Diegel, S. D. Fitzgerald, D. E. Berry, and J. B. Kaneene, 2002: Epidemiology of
© 2013 Blackwell Verlag GmbH • Transboundary and Emerging Diseases. 60 (Suppl. 1) (2013) 67–73
D. J. O’Brien et al.
Mycobacterium bovis in free-ranging white-tailed deer, Michigan, USA, 1995–2000. Prev. Vet. Med. 54, 47–63. O’Brien, D. J., S. M. Schmitt, D. E. Berry, S. D. Fitzgerald, J. R. Vanneste, T. J. Lyon, D. Magsig, J. S. Fierke, T. M. Cooley, L. S. Zwick, and B. V. Thomsen, 2004: Estimating the true prevalence of Mycobacterium bovis in hunter-harvested white-tailed deer in Michigan. J. Wildl. Dis. 40, 42–52. O’Brien, D. J., S. M. Schmitt, S. D. Fitzgerald, D. E. Berry, and G. J. Hickling, 2006: Managing the wildlife reservoir of Mycobacterium bovis: the Michigan, USA, experience. Vet. Microbiol. 112, 313–323. O’Brien, D. J., S. M. Schmitt, S. D. Fitzgerald, and D. E. Berry, 2011: Management of bovine tuberculosis in Michigan wildlife: current status and near term prospects. Vet. Microbiol. 151, 179–187. Olmstead, A. L., and P. W. Rhode, 2004: An impossible undertaking: the eradication of bovine tuberculosis in the United States. J. Econ. Hist. 64, 734–772. O’Reilly, L. M., and C. J. Daborn, 1995: The epidemiology of Mycobacterium bovis infections in animals and man: a review. Tuber. Lung Dis. 76, 1–46. Palmer, M. V., 2007: Tuberculosis: a reemerging disease at the interface of domestic animals and wildlife. Curr. Top. Microbiol. Immunol. 315, 195–215. Palmer, M. V., and W. R. Waters, 2011: Bovine tuberculosis and the establishment of an eradication program in the United States: role of veterinarians. Vet. Med. Int. doi:10.4061/2011/ 816345. Available at http://www.hindawi.com/journals/vmi/ mycobacterial.diseases.animals/ Palmer, M. V., W. R. Waters, and D. L. Whipple, 2002: Susceptibility of raccoons (Procyon lotor) to infection with Mycobacterium bovis. J. Wildl. Dis. 38, 266–274. Sangster, C., D. Bergeson, C. Lutze-Wallace, V. Crichton, and G. Wobeser, 2007: Feasibility of using coyotes (Canis latrans) as
bTB Test Performance in Furbearers
sentinels for bovine mycobacteriosis (Mycobacterium bovis) infection in wild cervids in and around Riding Mountain National Park, Manitoba. Canada. J. Wildl. Dis. 43, 432–438. Scherokman, B., 1997: Selecting and interpreting diagnostic tests. Permanente J. 1, 4–7. Schmitt, S. M., S. D. Fitzgerald, T. M. Cooley, C. S. BruningFann, L. Sullivan, D. Berry, T. Carlson, R. B. Minnis, J. B. Payeur, and J. Sikarskie, 1997: Bovine tuberculosis in freeranging white-tailed deer from Michigan. J. Wildl. Dis. 33, 749–758. Schmitt, S. M., D. J. O’Brien, C. S. Bruning-Fann, and S. D. Fitzgerald, 2002: Bovine tuberculosis in Michigan wildlife and livestock. Ann. N. Y. Acad. Sci. 969, 262–268. Shury, T. K., and D. Bergeson, 2011: Lesion distribution and epidemiology of Mycobacterium bovis in elk and white-tailed deer in south-western Manitoba. Canada. Vet. Med. Int. doi:10. 4061/2011/591980. Available at http://www.hindawi.com/ journals/vmi/mycobacterial.diseases.animals/ Thorne, E. T., R. A. Humphries, D. J. O’Brien, and S. M. Schmitt, 2005: State wildlife management agency responsibility for managing diseases in free-ranging wildlife. In: Rahm, J. (ed.), Transactions of the 70th North American Wildlife and Natural Resources Conference. pp. 334–345.Wildlife Management Institute, Arlington, Virginia. Thrusfield, M., 1995: Veterinary Epidemiology, 2nd edn. Blackwell Science, Oxford, UK. VerCauteren, K. C., T. C. Atwood, T. J. DeLiberto, H. J. Smith, J. S. Stevenson, B. V. Thomsen, T. Gidlewski, and J. Payeur, 2008: Sentinel-based surveillance of coyotes to detect bovine tuberculosis. Michigan. Emerging Infect. Dis. 14, 1862–1869. Witmer, G., A. E. Fine, J. Gionfriddo, M. Pipes, K. Shively, K. Piccolo, and P. Burke, 2010: Epizootiologic survey of Mycobacterium bovis in wildlife and farm environments in northern Michigan. J. Wildl. Dis. 46, 368–378.
© 2013 Blackwell Verlag GmbH • Transboundary and Emerging Diseases. 60 (Suppl. 1) (2013) 67–73
73
