INTRODUCTION
Cryptosporidium is a protozoal parasite with an environmentally robust oocyst stage that is infectious when excreted. C. parvum, one of the major Cryptosporidium species in humans, can be associated with zoonotic infection. Calves can be a major reservoir as C. parvum-infected calves' manure contains high oocyst concentrations [Reference Preiser, Preiser and Madeo1, Reference Reif2]. In Sweden, C. bovis is the most common species in calves, but in herds with calf diarrhoea caused by Cryptosporidium, >90% of calves are infected with C. parvum [Reference Silverlås3–Reference Robertson, Caccio and Widmer7]. Other clinically important Cryptosporidium species include the anthroponotic species C. hominis, often involved in waterborne outbreaks, and the zoonotic species C. meleagridis, C. felis, and C. canis [Reference Bouzid8].
The infectious C. parvum and C. hominis dose for humans is low at ⩾10 oocysts, but might be higher for other Cryptosporidium species [Reference Smith and Nichols9]. The incubation period is usually 2–10 days (average 7 days), and the most common symptom is watery diarrhoea. Other symptoms can include stomach cramps, dehydration, nausea, vomiting, fever, or weight loss. The infection is self-limiting and can be asymptomatic in immunocompetent people, but more severe cases may require hospitalization [Reference Pohjola10].
Infection can be prevented by reducing people's exposure to contamination (e.g. contaminated food or drinking water), and promoted by reinforcing hand hygiene measures, particularly during and following contact with potentially infected symptomatic people, calves and other animals [Reference Heymann11].
While infection in cattle is not notifiable in Sweden, human Cryptosporidium infection has been notifiable since 1 July 2004 (Communicable Disease Act; Smittskyddslagen 2004: 168), with diagnosing clinicians and laboratories reporting all cases to the Public Health Agency of Sweden (until December 2013; the Swedish Institute for Communicable Disease Control (Smittskyddsinstitutet; SMI) and the County Medical Officer. In 2012, 238 cases were notified to SMI in Sweden. In the Uppsala region in 2008–2012, fewer than one case was notified each month on average (mean 0·8/month) [12].
Sweden has experienced several food- and waterborne outbreaks of Cryptosporidium in the years prior to this outbreak [Reference Mattsson13-Reference Gherasim16]. The outbreaks caused by contamination of the municipal drinking water with C. hominis in Östersund and Skellefteå in 2010 and 2011, were notable due to the large number of infected people (total 46000) [Reference Widerstrom17].
Veterinary studies and ambulatory clinic classes at the Swedish University of Agricultural Sciences
In 2013, 360 students at the Swedish University of Agricultural Sciences (Sveriges lantbruksuniversitet; SLU) in Uppsala, Sweden were veterinary students, of which 81 were in their fourth year, attending class between 21 January and 13 December 2013. Ten classes were taught on the SLU campus, while one class, coordinated by the animal hospital's ambulatory clinic, was taught off campus. The ambulatory clinic is an on-call centre with 10 staff. Its veterinarians triage phone calls at SLU each morning, then drive groups of students to farms in ambulance cars. This provides students with their first opportunity for intense, hands-on clinical education on farms, e.g. examining animals with diarrhoea. Veterinary students only have classes there in their fourth year spring and fifth year autumn semesters. By 29 March 2013, 40 fourth-year students had attended ambulatory clinic classes, visiting more than 50 different farms with animals including cattle, horses and sheep.
Initial alert and response
On 9 February 2013, a fourth-year veterinary student at SLU experienced the onset of diarrhoea and stomach cramps, sought healthcare, and tested positive for Cryptosporidium infection. On 5 March the student reported to the head of the ambulatory clinic that classmates were reporting similar symptoms. All had attended classes at the veterinary hospital's ambulatory clinic on the SLU campus.
SLU immediately initiated internal investigations to confirm the outbreak, and contacted SMI on 7 March. An investigation was initiated by an outbreak control team (OCT) comprising representatives from SMI, the National Veterinary Institute (Statens Veterinärmedicinska Anstalt; SVA) and SLU. Its aims were to describe the outbreak's size, identify potential infection source(s), and identify risk factors for infection to help prevent further cases and similar outbreaks in similar settings.
METHODS
Case-finding and hypothesis generation
During initial investigations and case-finding, a case was defined as a veterinary student or staff member at SLU with symptoms of diarrhoea (>2 loose stools within 24 h) with onset after 1 January 2013. On 5 March SLU initiated enquiries with SLU staff and fourth-year students to identify those with cryptosporidiosis-compatible symptoms. SLU contacted all fourth-year veterinary students on 6 March.
SLU conducted unstructured telephone interviews with potential cases to identify common exposures, and reviewed ambulatory clinic records to record their farm visits. On 11 and 12 March SMI made a site visit to SLU to identify plausible risk factors for infection. On 12 March the OCT sent a document to all fourth-year students describing the outbreak, reinforcing standard control measures (including hand hygiene and hygiene measures for cattle farm visits), and directing symptomatic students to contact healthcare and SLU. On 12 March, SLU also searched for cases in other student years by re-contacting all SLU staff who taught clinical students.
Additional cases were sought in the electronic national surveillance system (SmiNet) by searching for cases reported between 1 December 2012 and 30 April 2013 in Uppsala County.
Calf faecal samples
Faecal samples were collected by SLU's ambulatory clinic staff from five calves aged <6 weeks from cattle herds at two farms suggested by hypothesis-generating interviews in April 2013. If present, calves with diarrhoea were sampled preferentially.
Microbiological investigations
Symptomatic students' faecal specimens were sent to SMI for detection of Cryptosporidium oocysts by modified Ziehl–Neelsen staining. Additional stool parasites were investigated by light microscopy after formol/ethyl acetate concentration. The presence of Giardia cysts was analysed using a direct immunofluorescent test (Agua-Glo, Waterborne Inc., USA). DNA was extracted directly from stool specimens using the QIAamp DNA mini kit (Qiagen, Germany) according to the manufacturer's recommendations. Disruption of oocysts was performed before extraction using a BulletBlender (Techtum, Sweden). Species identification was performed by amplification of the small subunit rRNA (SSU rRNA) gene followed by restriction fragment length polymorphism analysis [Reference Xiao18, Reference Xiao19]. Subtyping was achieved by sequencing an amplified fragment of the 60-kDa glycoprotein (GP60) gene [Reference Alves20].
SVA performed microscopic analysis, SSU rRNA and GP60 typing of calf faecal samples. Molecular analysis of SSU rRNA was performed using Sanger sequencing [Reference Xiao19]. Subtyping of calf samples were performed using the same laboratory protocol as on the human samples.
Epidemiological cohort study
The OCT initiated a cohort study to identify risk factors for symptomatic Cryptosporidium infection among fourth-year SLU veterinary students in January–April 2013, and to inform recommendations to prevent further cases.
For this epidemiological investigation, a probable case was defined as a veterinary student at SLU with diarrhoea (>2 loose stools within 24 h) between 21 January (the start of the semester) and 5 April 2013; a confirmed case was a probable case with laboratory-confirmed Cryptosporidium infection; cases were excluded if their stool specimens were negative for Cryptosporidium.
An online questionnaire was generated using Survey Generator™ (http://www.alstra.se/en/survey-tool), and sent electronically to the cohort on 15 March. The questionnaire collected data on cryptosporidiosis symptoms experienced since 25 December 2012 (i.e. >2 incubation periods prior to the semester, to permit a reliable description of the outbreak's start), SLU-centred factors (e.g. contact with the ambulatory clinic, farms and symptomatic animals; lunch in cars), and non-SLU-centred factors (e.g. close contact with cases outside SLU, non-SLU animal contact). A list of farms visited by cases was provided by SLU (i.e. farms A–P); respondents were asked which farm they visited. Students who reported symptoms after questionnaire completion were sent a new questionnaire, and their original response was discarded.
Descriptive analyses examined cases' clinical, geospatial and temporal characteristics, and common exposures.
Cohort data analyses estimated associations between being a case and potential risk factors using χ 2 and Fisher's exact tests if there were fewer than five exposed or unexposed cases or non-cases, or fewer than 20 respondents to a question. Ordered categorical variables with few cases per category were recoded as binary variables following univariate analysis and/or logistic regression. Test for trend for ‘weeks spent in the ambulatory clinic’ was analysed assuming a categorical variable using a non-parametric test, and by univariate logistic regression.
Multivariable analyses by logistic regression of variables which had evidence of an association by univariate analysis are not presented, as no statistically significant associations were identified due to insufficient statistical power. All analyses were performed in Stata/SE v. 12.0 (StataCorp LP, USA) and Microsoft Excel 2010 (Microsoft Corp., USA).
RESULTS
Hypothesis generation
Hypothesis-generating interviews with eight cases identified that all had attended class at the ambulatory clinic in 2013. During that class, they visited 16 different farms, 4–6 each. All had visited farms A and/or B. Clinic staff recalled that these two farms had calves with diarrhoea in 2013. Clinic staff did not know of any clinic co-workers, or farm workers who had experienced cryptosporidiosis-compatible symptoms in the previous 6 months.
The hypothesis that infection of these students followed contact with the ambulatory clinic and/or farm A and/or farm B in 2013 was found to be plausible during the March site visit. Therefore the OCT recommended that calf faecal specimens be taken from farms A and B, and initiated an analytical study to test the hypothesis.
Environmental investigations
There were no cattle with diarrhoea at farm A when faecal specimens were collected, while there were calves with diarrhoea at farm B during collection.
Microbiological investigations
Cryptosporidium infection was confirmed by microscopy and polymerase chain reaction in 6/7 faecal specimens submitted for analysis. No other stool parasites were found by extended microscopy; anti-Giardia monoclonal antibody was negative for the presence of Giardia cysts in all eight specimens. Molecular analysis identified C. parvum subtype IIaA16G1R1b (100% identical to EU647728) and C. parvum subtype IIdA24G1 (100% identical to HQ005751) in five and two specimens, respectively.
C. parvum subtype IIaA16G1R1b was identified in calf faecal specimens that were taken in April 2013 from farm B, but no Cryptosporidium oocysts could be found in specimens from farm A.
Epidemiological investigations
The cohort of 81 fourth-year students returned 69 questionnaire responses (response rate 84%); 91% of the respondents were female, representing the demographics of fourth-year students (~80% female). Four responses were excluded from analysis (one met the exclusion criteria; three did not indicate symptom onset dates, essential data for classification of cohort members as cases or non-cases).
The cases definition was met by the questionnaire responses of 13 students, therefore the attack rate (AR) in the cohort was 20% (13/65 students). Six of these were confirmed cases, and seven were probable cases. Their mean age was 26 years with all except one being female.
The cases' symptom onset dates were between 24 January and 18 March 2013, i.e. none prior to the 2013 spring semester (Fig. 1). Ten cases reported watery diarrhoea for at least 1–2 days (71%). Two cases sought medical care, one of whom was admitted to hospital for a day. Ten (71%) cases took at least a day off class due to this illness (mean 4 days, maximum 10 days, sum of all missed days = 52 days).
The ambulatory clinic had been attended by all 13 cases identified in the cohort study, of which 12 attended clinic classes; the thirteenth case attended the clinic for just 1 day. Cases had used all five clinic cars; all but one had eaten lunch or snacks in the cars. Cases reported seeing animals with diarrhoea at farms A, B, C and D only (Table 1). The only animals reported with diarrhoea were ‘cows’ or ‘calves’. Eleven of 13 cases visited at least one of these farms (85%, Table 2).
AR, Attack rate; RR, relative risk; CI, confidence interval.
* P value calculated using χ 2 test, all other P values calculated using Fisher's exact test.
Cohort study data analysis identified 13 risk factors with strong evidence of an association with being a case (P⩽0·05), and six others with some evidence (P⩽0·125) (Table 2).
The strongest association was ambulatory clinic attendance [relative risk (RR) 10·9, 95% confidence interval (CI) 1·5–79·3]. The AR for students who spent a day or more in the ambulatory clinic was 35% (13/37), and higher in those attending 4 weeks of class (7/19 students, AR 37%) than 1 week (5/15 students, AR 33%) (P = 0·02).
The next strongest associations were eating in cars (RR 9·1, 95% CI 1·3–65·8), and drinking tap water on farms (RR 4·4, 95% CI 2·0–9·7). All cases reported washing their hands fewer than twice per visit to a cattle farm, whereas 13/37 non-cases (AR 37%) reported hand-washing at least twice per visit, providing strong evidence for it being protective (P = 0·03).
There was strong evidence in the cohort data of a risk associated with visits to farm B (P = 0·009), and also farms G and I (P < 0·05), but no statistically significant risk associated with visits to farms A, C or D, nor from visiting ‘at least one of farms A, B, C or D’, i.e. the four farms on which students recalled seeing animals (calves) with diarrhoea (0·2 < P < 1·0, Table 2).
The univariate cohort analysis was restricted to the subset reporting ambulatory clinic attendance. All variables were tested, and two factors retained evidence of association with being a case (P < 0·1): drinking tap water on a farm (RR 3·3, 95% CI 2·0–5·5, P = 0·04), and attending farm B (RR 1·2, 95% CI 1·2–5·5, P = 0·073).
Four confirmed cases visited Farm B on 1 day in the week commencing 25 February 2013, during their only week of ambulatory clinic classes. Students only used one car that day. They recalled entering a building with calves with diarrhoea and high-risk activities for zoonotic infections, such as de-horning and neutering calves. Their symptom onsets were all during the subsequent week. C. parvum IIaA16G1R1b was identified from these four students' faecal samples, and from environmental calf faecal specimens at farm B (Figs 1 and 2).
DISCUSSION
This investigation benefited greatly from the rapid availability of microbiological results from farms and students during the collaborative, intersectoral investigation. Microbiological subtype analysis is often not available during outbreak investigations; the university's initial investigations were swift and there was a high response rate from the veterinary students to an electronic questionnaire. Together these microbiological and epidemiological data provided evidence of a link between C. parvum infections that affected cattle, and students having contact with the animals' diarrhoea.
The transmission of C. parvum infection in those students who attended farms with calves with diarrhoea is likely to have been associated with direct animal contact. Those without this direct contact were most likely infected via contaminated objects in the working environment, e.g. on the farms themselves, in the veterinary hospital's ambulatory clinic, and/or while eating within the ambulatory clinic's cars.
Zoonotic outbreaks of C. parvum infection in veterinary students are well documented, and commonly associated with calf contact concomitant with lapses in hygiene measures [Reference Preiser, Preiser and Madeo1, Reference Reif2, Reference Pohjola10, Reference Gait21-Reference Kiang23]. Investigation of one of these outbreaks was also initiated by a concerned veterinary student together with their university [Reference Grinberg22]. In 2003, a C. parvum outbreak in 68 students in Minnesota related to infected calves had two equally sized epidemic waves, in two sequential school semesters. The recurrence was attributed to inadequate implementation of recommended control measures for barrier protection and hand hygiene [Reference Kiang23]. Subsequent to our investigation into this current outbreak, Cryptosporidium infection was diagnosed in a fourth-year veterinary student in early May 2013, and another student in July 2013, i.e. after the initial control measures were recommended and SMI's 19 April 2013 outbreak report. This may indicate that the implemented control measures were not sufficient, transmission outside of SLU, or a missed risk factor.
Most of the identified risk factors were plausible. Students who attend ambulatory clinic classes will attend cattle farms, potentially high-risk locations for zoonotic C. parvum infection if the calves have diarrhoea [Reference Haagsma24]. Clinic attendance necessarily includes intense hands-on clinical work on farm animals including calves. Indeed cases reported procedures likely to have resulted in risky contact with infectious material, and so there was potential exposure to high oocyst burdens. Eating in clinic cars was a near ubiquitous experience for clinic attendees, also presenting a potential site for oocyst ingestion. Students were usually out all day, typically eating lunch in the cars. Uniform protective clothing is provided to students by SLU, washed following each day's farm visits with detergent at 40 °C and hung to dry. On the next day it is stored in boxes between trips. Perhaps this protocol provides a possible site for distribution of oocysts; washing this clothing at 40 °C may be insufficient to inactivate oocysts [Reference Heymann11, Reference Semenza25].
One notable finding was the epidemiological evidence for consumption of farms' tap water being a risk factor. Indeed the animal facilities at farm A have their own water supply, and farms B, C and D also have their own supplies. As no environmental samples were taken from water supplies on any farm visited by cases there are no direct data to exclude these supplies as a source for infections. However, only three cases reported consuming farm tap water, and there were no notifications in SmiNet relating to farm workers in the Uppsala region. In hindsight, we should have taken environmental samples and would have included questions in the cohort study's questionnaire on the tap's location, and how tap water was consumed (e.g. using a container, using hands or drunk from directly). However, the published literature and the data we collected all suggest that veterinary students' direct and/or indirect contact with infected calves' diarrhoea is the more plausible hypothesis in this outbreak.
Other identified risk factors also require more interpretation. Hand sanitizer gels do not kill oocysts, yet there was some evidence that non-cases were more likely to use it than cases. Perhaps this is a reflection of these students’ inclination towards more rigorous hygiene practices generally. Moreover, there was some statistical evidence that those attending farms G, I, J, O and P were more likely to be cases, but these exposures explain three or fewer cases each (Table 2). In terms of impact, few students sought healthcare due to this outbreak, but 52 days of study were missed by the 81 fourth-year students.
The identification of two subtypes may indicate mixed infection on one farm, or two infections on two or more farms. C. parvum subtype IIaA16G1R1b (EU647728) is fairly common in Sweden; it was identified in 16% of subtyped C. parvum cases in the Stockholm area between 2006 and 2008, including a possible foodborne outbreak [Reference Insulander14], and in several sporadic cryptosporidiosis cases in 2012–2013 (SMI, unpublished data). It was identified as the most prevalent subtype in a study conducted in Swedish herds with calf diarrhoeal problems between 2010–2012, identified in 33% (26/79) of the herds and 34% of calves with determined C. parvum subtype [Reference Silverlås and Blanco-Penedo4, Reference Silverlås5]. Therefore the strong epidemiological evidence for transmission of this subtype at farm B does not exclude the possibility that other herds were a source. Use of sequence-based methods with higher discriminatory power, such as a multi-locus sequence typing scheme for several micro and mini satellites, may have been able to provide stronger evidence of a microbiological link.
C. parvum subtype IIdA24G1 (HQ005751) has been rarely identified in Sweden and elsewhere. It was identified as the aetiological agent in a foodborne outbreak in two Swedish cities in 2010 [Reference Gherasim16]. Otherwise it has only been found twice in Sweden before this outbreak, in sporadic cases in 2011 and 2013 (J. Beser, unpublished data). Subtype IIdA24G1c, which is highly homologous to IIdA24G1, has been identified in four calves from two herds in Sweden [Reference Silverlås and Blanco-Penedo4].
There was epidemiological evidence for subtype IIdA24G1 at farm A, but it was not identified at either farm tested in this study. Farm A had been sampled for presence of Cryptosporidium for research purposes twice in 2012, i.e. prior to this outbreak; only C. bovis was identified. This is not unusual: >95% herds in Sweden have calves positive for C. bovis, with 30–40% calves <10 weeks infected, while C. parvum is rarely identified from non-diarrhoeal calves. Therefore, with three repeated samplings negative for C. parvum, it is unlikely that this Cryptosporidium species was present at this farm.
Factors inherent to the study's design and execution will have affected the obtained results. The cohort was relatively small, providing too few people per category to generate strong statistical evidence from multivariable analyses to control any confounding and detect possible effect modification. The questionnaire elucidated catalogued each case's symptoms; five probable cases experienced watery diarrhoea for more than a day, no probable case reported blood in stool. Even though these are consistent with cryptosporidiosis, no stool sample was available to confirm the categorization. Case-finding was strong in the fourth-year group, but far weaker in other year groups, and so cases may have remained unidentified by the OCT. The definition of a probable case was non-specific to cryptosporidiosis and diarrhoeal symptoms may have been due to other causes. Plausibly, the infections with the common subtype may have occurred on another farm, from a contaminated working environment or person-to-person. Indeed, students can move more or less freely throughout teaching areas, including seminar rooms opposite the ambulatory clinic. The data collected suggest that person-to-person transmission was unlikely (other than indirect transmission in clinic cars or at the ambulatory clinic) as there were no symptomatic persons in the cases’ households or in student peers who did not attend the ambulatory clinic. This investigation focused on identifying the source of infectious Cryptosporidium oocysts rather than mapping their distributions. For this reason specimens were taken from calves and not from ambulatory clinic cars (even though soiling was identified in these cars during the site visit), and no environmental testing was performed on protective clothing or the clinic environment.
A current study at SMI invites all laboratories in Sweden to submit stool samples from human Cryptosporidium cases for species differentiation and subtyping. Cryptosporidium species and subtype identification in calf diarrhoeal samples are currently being performed at SLU. Such projects should permit discussion of the place of the IIaA16G1R1b and IIdA24G1 subtypes within the Swedish Cryptosporidium landscape within the next few years.
The identified risk factors enable us to propose evidence-based recommendations for universities to prevent C. parvum infections in veterinary students. We recommend that veterinary schools (a) continue to reinforce information to students on risk factors for Cryptosporidium infection, especially the risks from calves with diarrhoea; and on hygiene recommendations, particularly related to eating and drinking during clinical studies; (b) consider recommending that students do not eat in transport cars; and (c) ensure that current routines for washing protective clothes and vehicles are sufficient to kill Cryptosporidium oocysts.
As C. parvum can transmit person-to-person and result in absenteeism in students, healthcare professionals, particularly at primary care, should be sensitized to test for Cryptosporidium infection in veterinary students presenting with gastrointestinal symptoms. This would permit early identification of C. parvum outbreaks and targeted interventions to end transmission.
ACKNOWLEDGEMENTS
The authors thank SLU Animal Hospital Ambulatory Clinic staff and students for their time and assistance; EPIET supervisors Lara Payne Hallström and Gianfranco Spiteri and EPIET coordinator Marion Muehlen for advice on study design, analysis and comments on the manuscript.
DECLARATION OF INTEREST
None.