This exploratory study has highlighted several characteristics of Cryptosporidium cases, and of the environment in which they live, that might be correlated with spread of infection. The significant independent factor in multivariable analysis was an index case infected with C. hominis.
Despite a definite lean towards recruitment of families rather than other household compositions, the characteristics of our included participants were fairly typical of both North West England and Wales. Participants were mostly comprised of 25 to 44-year olds but we also had a decent proportion of young children represented (21% under 5 years), with male cases tending to be younger. This fits with what we know about the descriptive epidemiology of Cryptosporidium in England and Wales (Chalmers et al., 2009). The index case was a child under five years old in almost 30% of the recruited households and two-thirds of those were male. This is supported by other examinations of cases in the UK and beyond that demonstrate an increased prevalence in infants and young children [7, 8, 15, 16]. The households that took part originated from all socioeconomic areas, but there were slightly more households from the less deprived geographies. This is consistent with the profile of Cryptosporidium infection across England and Wales where the most deprived areas appear slightly underrepresented [17]. This might reflect difference in access to, or use of, services, or may be a reflection of differences in recruitment and participation [18, 19].
More than a quarter of index cases reported symptoms other than diarrhoea and vomiting, including nausea, abdominal pain, and headaches. Moreover, vomiting was not frequently reported at all, occurring in less than half of the index cases, which has been noted previously [20–22] Symptoms differed somewhat by age with nausea, headache, and stomach pain occurring more among older cases and vomiting in the younger cases, particularly males. This could be due to differences in the symptom profile of infecting species. Differences in symptom presentation have been identified before, in an outbreak of C. hominis, where headache and abdominal pain were more common in female cases [22]. However, as a high proportion of the cases were children, it is important to remember that we are relying on secondary reports of illness, usually via parents. Self-reported illness can be fraught with reliability issues, especially with non-clinically obvious symptoms like pain or nausea, which may be difficult for a young child to describe or articulate. Nonetheless, vomiting in children as a symptom of Cryptosporidium infection has been demonstrated before [8], and may be an important presentation to note; the presence of vomiting does not exclude it from a differential diagnosis in gastrointestinal investigations.
Of additional interest, this study revealed males reported a longer illness as did cases of C. hominis. A study of sporadic disease in the UK reported a mean duration of acute symptoms for patients with C. hominis that was two days longer than C. parvum cases [8]. One interesting finding in this study was that additional, and possibly secondary, cases in the home were not as long-lived as those in the index cases and might point to a decreased virulence in person-to-person spread [23]. This has been evidenced before in homes with transmission of gastrointestinal pathogens, where secondary cases’ average duration of illness was more than half that of primary cases [24].
Persistent and substantial burden of illness on the individual as well as on the home overall is well corroborated in other literature which has revealed duration of symptoms for Cryptosporidium infection far beyond IID of other aetiologies [25–28]. Crucially the longevity of illness might also amplify spread, by potentially increasing the length of time the oocysts are shed (Jokipii and Jokipii, 1986; Chappell et al., 1996; Chalmers; et al., 2016), although here there was no association between length of illness and burden of additional cases in the home. Nonetheless, complications this long lasting and potentially burdensome warrant further examination.
An analysis by species revealed that less than 20% of the C. parvum index cases reported additional illness in their household (19.6%), compared to 48% of the C. hominis indexes (p = 0.01). This result is in line with similar studies evidencing C. hominis as a species particularly associated with people, and probably the person-to-person transmission pathway. A case-control study in the Netherlands [31] found that C. hominis cases in particular were three times more likely than controls to have been exposed to a case in the home and were less likely to live in homes with lots of adults. Also, in those years where C. hominis was the predominant circulating species, other risk factors such as food items were reported as associated with decreased odds of illness. This work adds to the body of evidence that sources for C. hominis infection may be exclusively human and that person-person transmission is the most likely pathway (Hunter, Hughes, Woodhouse, Syed, et al., 2004; Hunter and Thompson, 2005).
Almost two additional cases occurred, on average, in households with an index case. Additionally, the analyses suggested that almost a third (31%) of people in the home could be expected to get ill from transmission of infection. This burden was even greater in households where the index case was infected with C. hominis or the index case was under five years old. Risk of infection in settings with young children has previously been demonstrated and is known to facilitate spread (Hannah and Riordan, 1988; Hunter, Hughes, Woodhouse, Syed, et al., 2004; Johansen et al., 2014). This work further buttresses that person-to-person is a specific transmission pathway but is first study to quantify the burden that this exacts on the home.
If most index cases are young children, and mums make up the burden of secondary cases, then it is plausible that the driver here is direct contact, in a caring capacity; undertaking activities which likely put the main carers in the home at high risk. In a study in the UK, Hunter et al. found that changing children’s nappies was a risk factor specific to C. hominis (Hunter, Hughes, Woodhouse, Syed, et al., 2004) and the Netherlands recently reported similar results (Nic Lochlainn et al., 2019), specifying that C. hominis cases were more likely than controls to have been exposed to a case in the home. Additionally, the authors reported corroborating indicators supportive of this exposure, including living in smaller homes, and living with children.
Siblings were also affected considerably by secondary infection, but where the adult was an index case, their children were less frequently the secondary case. It has been documented before that mums and siblings are most at risk of Cryptosporidium: in a follow-on study in Norway, with 12 and 13-year-old index cases, a 17% secondary transmission rate was mainly comprised of female caregivers and siblings [21]. This is mirrored among other gastrointestinal aetiologies [24], and indeed for E. coli O157 it has been suggested that separation of siblings might be a key intervention in reducing secondary cases [34].
Gender roles are known to influence both patterns of exposure to infectious agents and the treatment of infectious disease [35]. Caring for the sick carries an increased risk of exposure, especially for diseases that are spread directly from person-to-person and in most societies females are more likely to care for the sick than males [36]. The heterogeneity of contact within the home has been examined in respiratory diseases such as Influenza and Pertussis, and studies found that contacts between mother and children and between siblings are most prevalent [37]. Several cost and burden of illness studies have been undertaken in the Netherlands, which have considered the economic and societal impact of gastrointestinal infections. Overall, there is a considerable burden on productivity due to absence from work for the ill or the caregiver(s) [38], and one study estimated that in 15% of cases where a child was ill, a parent had to remain off work [39]. An additional analysis considering the longer-term manifestations of Cryptosporidium in particular reported similar burdens on productivity, with additional impact on disability adjusted life years (DALYs) due to recurring diarrhoea and long-term joint pain [40]. Further work confirming and examining this disparity for Cryptosporidium would be a welcome addition to work to describe the economic and societal burden of this disease.
This work did not reveal any considerable proportion of asymptomatic infection. Cryptosporidium was detected in 12/259 (4%) of household members’ samples of whom two were asymptomatic, giving a prevalence of asymptomatic infection of 2%. From the small number of relevant studies, carriage of Cryptosporidium appears to be low at between 0.1–1.3% [41, 42] although this has once been demonstrated as high as 9% following an outbreak of C. parvum in Norway [21]. Identification of true carriage is difficult as we tend to capture diarrhoeal cases and it is likely that all of the index cases here will have sought clinical assessment following symptoms. In addition, recrudescence of symptoms complicates the identification of differences between true asymptomatic infection and shedding of oocysts in an asymptomatic period. An asymptomatic prevalence of 2% would be in line with carriage expected for the UK [42] but the design of this work did not allow any examination of this contribution to spread.
The time between initial onset of illness in the home and sample retrieval from household contacts was variable, and often long. This does raise some uncertainty about the capability of the tests used to confirm infection. Given that we have already demonstrated differences in length of illness by species, it would not be implausible that asymptomatic, or indeed less protracted secondary infections, might lead to shorter shedding times [24]. If this were the case, the detection power of the tests may be reduced by the time samples are received at the laboratory. Also, a small sample size limits the power to truly detect asymptomatic infections. The true asymptomatic infection burden may well be under ascertained here and previous work reiterates that lack of detection by routine diagnostic methods does not necessarily equate to lack of infection (Chalmers et al., 2011; Chalmers et al., 2016). There are complex biological and social factors that affect surveillance data capture, of which illness severity has been shown to be important [44] and one person’s idea of ‘being ill’ might differ from another’s. Nonetheless, this result is not insignificant. If asymptomatic infections are indeed few, rather than this being something to be dismissed, actually this indicates that if infected you are more than likely to be ill, and we know that with this illness comes considerable symptomatic burden, and the risk of longer term sequelae [20, 45]. As such, this makes tackling preventable secondary transmission of infection a crucial issue of public health importance.
Cryptosporidium was detected in 12/259 (4%) of household members’ samples. This is much smaller than expected if the self-reported clinical illness does truly represent secondary infections. An explanation for this may be the lag time from illness to receiving household samples and the likelihood of detecting Cryptosporidium. Despite a range of laboratory testing methods, including PCR, the results demonstrated that confirmation was more likely in specimens taken during, or soon after, a case’s symptomatic period. The average time between the index cases' specimen date and the first household member specimen was 43 days. Oocysts might also be shed intermittently but the study design did not allow for repeat sampling.
Additionally, using a clinical indicator of BSS made no difference to microbiological confirmation: all of the diarrhoeic specimens were subsequently unable to be confirmed as infected with Cryptosporidium. Conversely, all those household member samples that were confirmed, had formed stools. This supports a prior recommendation to eliminate stool consistency as a testing inclusion criterion in local laboratories in England and Wales (McKerr & Chalmers, 2020 – personal communication).
This study detected possible household transmission of C. ubiquitum, with an index case and a confirmed household infection. Unfortunately, the questionnaire element of the study was not returned for this home and so further examination of exposures was not possible. But this was an interesting find: it is an unusual species, sources of infections in humans are not entirely clear and transmission between people has never been demonstrated [46].
The variable most strongly associated with additional cases in the home was the infecting species of the index case. The Netherlands recently reported similar results [31], specifying that C. hominis cases were more likely than controls to have been exposed to a case in the home. Additionally, the authors reported corroborating indicators supportive of a person-to-person pathway, including living in smaller homes, and living with children. Although not independently associated with transmission in the logistic regression model, this study did highlight similar associated exposures, with homes with additional cases twice as likely to report the index case being a child less than five years old or attending a nursery.
This work continues to buttress the existing literature but highlights quite clearly that differences in species and transmission are quite likely. At risk homes can be identified as those where the index is less than five years old and/or is infected with C. hominis. Of particular risk are mums and caregivers, and siblings, and targeted hygiene advice should be specifically directed here.
The demonstration of a C. hominis-specific burden provides another argument for swift and complete characterisation of isolates, with results fed into local and national surveillance data. There is certainly a public health, and economic, argument for interventions to reduce not only primary infections with Cryptosporidium, but also subsequent spread. This might include work to provide more targeted advice for individual Cryptosporidium patients or during outbreaks, and these strategic and population-level approaches are critical given the lack of licensed treatment for this infection in the UK. This evidence does reinforce the importance of speciation and subtyping of isolates where at all possible, in order to better understand the clinical course of disease for the patient or population and administer appropriate interventions and advice.
A study of this kind is not without its limitations. Proving transmission is a difficult task, and studies oftentimes are unable to examine this to ascertain specific differences in cases or identify modifiable risk factors, and very specific study designs are required to examine this. The ubiquitous nature of Cryptosporidium and of its exposures make untangling these exposures and demonstrating causality difficult in a study of this set-up. Nevertheless, this study was intended as an exploratory piece, and the evidence presented suggests that Cryptosporidium does transmit readily in the home environment, and that person-to-person could be the transmission pathway.
Additionally, our limited understanding of the background prevalence of asymptomatic infection of Cryptosporidium, and its effectors, make it difficult to identify its importance in spread of disease in contained settings, and this study did not reveal a large amount of asymptomatic infection. The study design was not appropriate to demonstrate if an asymptomatic carrier was shedding oocysts or was infectious to others in the home. However, previous work on secondary transmission data has mainly stemmed initially from outbreaks, and data rarely include laboratory confirmation of secondary cases [21]. The epiCrypt study is unique in that it has allowed for an examination of secondary cases at both species level and with further typing. A larger scale study of sporadic infections would continue to build on our understanding of species-specific risks of spread and also could examine heterogeneity in subtype populations [47].
Difficulties arise distinguishing between primary and secondary infections as close contacts often have similar exposures [21] and the clinical course of Cryptosporidium infection can result in variable incubation, symptoms, and onset between individuals making verifying person-to-person transmission and differentiating true secondary cases from co-primary challenging. Further genotyping of some samples is ongoing which may support household level investigations of directionality and population mixing. The study design did not allow us to look for other enteric pathogens that cause diarrhoea, which could lead to both misattribution of index case illness to Cryptosporidium, and to overestimation of Cryptosporidium household transmission rates. We know that time constraints are a major contributor to issues in epidemiological observational studies, and in research are due in some part to ethical considerations [48, 49].
This work demonstrated that additional cases of Cryptosporidium occur in over a quarter of homes with a laboratory confirmed case. This is likely to affect up to a third of the household and cause considerable burden of illness. This is especially common where the index is a young child, with mums and other siblings are most at risk of secondary infection, and where homes have cases of C. hominis. This is important because current health care and public health systems are likely under ascertaining cases of sporadic illness, under examining person-to-person spread, and under-advising where specific clinical advice could be provided to high-risk households. Systematic changes that would provide improvement include species identification of all Cryptosporidium positive samples, fed routinely back into local health protection teams across England and Wales, and the consideration of specific clinical advice on prevention for high-risk households. This might include managing the patient’s expectations on the length of illness, and the possibility of relapse, and giving specific advice on preventing person-to-person spread [50].
Further work should expand on this research, which was only intended to be exploratory and low resolution. A better and closer examination of households and homes alongside a methodology to identify true secondary transmission more accurately should be the next step. This work should be designed in a way that allows correlations to be extrapolated more widely, and it is important that these are facilitated by all public health bodies across the UK.