Impact of the COVID-19 restrictions on the epidemiology of Cryptosporidium spp. in England and Wales, 2015-2021A time-series analysis

Background In England and Wales, cryptosporidiosis cases peak in spring and autumn, usually associated with zoonotic and environmental exposures (Cryptosporidium parvum, spring/autumn) and with overseas travel and water-based activities (Cryptosporidium hominis, autumn). Restrictions to control the COVID-19 pandemic prevented social mixing and access to swimming pools and restaurants for many months. Foreign travel from the UK also reduced by 74% in 2020. However, these restrictions potentially increased environmental exposures as people sought alternative countryside activities locally. To inform and strengthen surveillance programmes, we investigated the impact of COVID-19 restrictions on the epidemiology of C. hominis and C. parvum cases. Methods Cryptosporidium-positive stools, with case demographic data, are referred routinely for genotyping to the national Cryptosporidium Reference Unit (CRU). Cases were extracted from the CRU database (01 January 2015 to 31 December 2021). We defined two periods for pre- and post-COVID-19 restrictions implementation corresponding to the first UK-wide lockdown on 23 March 2020: pre-restrictions between week 1, 2015 and week 12, 2020, and post restrictions-implementation between week 13, 2020 and week 52, 2021. We conducted an interrupted time-series analysis, assessing differences in C. parvum and C. hominis incidence, trends and periodicity between these periods using negative binomial regression with linear-splines and interactions. Results There were 21,304 cases between 01 January 2015 and 31 December 2021 (C. parvum = 12,246; C. hominis = 9,058). Post restrictions-implementation incidence of C. hominis dropped by 97.5% (95%CI: 95.4%-98.6%; p<0.001). The decreasing incidence-trend observed pre-restrictions (IRR=0.9976; 95%CI: 0.9969-0.9982; p<0.001) was not observed post restrictions-implementation (IRR=1.0081; 95%CI: 0.9978-1.0186; p=0.128) due to lack of cases. No periodicity change was observed post restrictions-implementation. Where recorded, 22% of C. hominis cases had travelled abroad. There was also a strong social gradient, with those who lived in deprived areas experiencing a higher proportion of cases. This gradient did not exist post restrictions-implementation, but the effect was exacerbated for the most deprived: 27.2% of cases from the most deprived decile compared to 12.7% in the pre-restrictions period. For C. parvum, post restrictions-implementation incidence fell by 49.0% (95%CI: 38.4%-58.3%; p<0.001). There was no pre-restrictions incidence-trend (IRR=1.0003; 95%CI: 0.9997-1.0009; p=0.322) but a slight increasing incidence-trend existed post restrictions-implementation (IRR=1.0071; 95%CI: 1.0038-1.0104; p<0.001). A periodicity change was observed for C. parvum post restrictions-implementation, peaking one week earlier in spring and two weeks later in autumn. Where recorded, 8% of C. parvum cases had travelled abroad. The social gradient observed for C. parvum was inverse to that for C. hominis, and was stable pre-restrictions and post restrictions-implementation. Conclusion C. hominis cases were almost entirely arrested post restrictions-implementation, reinforcing that foreign travel is a major driver of seeding infections. Increased hand-hygiene, reduced social mixing, limited access to swimming pools and limited foreign travel affected incidence of most gastrointestinal (GI) pathogens, including Cryptosporidium, in the same period. C. parvum incidence fell sharply but recovered throughout the post restrictions-implementation period, back to pre-restrictions levels by the end of 2021; this is consistent with relaxation of restrictions, reduced compliance and increased countryside use. The effect on our results of changes in health-seeking behaviours, healthcare access and diagnostic laboratory practices post restrictions-implementation is uncertain, but it is likely that access to GPs and specimen referral rate to CRU decreased. Future exceedance reporting for C. hominis should exclude the post restrictions-implementation period but retain it for C. parvum (except the first six weeks post restrictions-implementation where the incidence fell sharply). Advice on infection prevention and control should be improved for people with GI symptoms, including returning travellers, to ensure hand hygiene and appropriate swimming pool avoidance.

1 Background 2 In England and Wales, cryptosporidiosis cases peak in spring and autumn, usually 3 associated with zoonotic and environmental exposures (Cryptosporidium parvum, 4 spring/autumn) and with overseas travel and water-based activities (Cryptosporidium 5 hominis, autumn). Restrictions to control the COVID-19 pandemic prevented social mixing 6 and access to swimming pools and restaurants for many months. Foreign travel from the UK 7 also reduced by 74% in 2020. However, these restrictions potentially increased 8 environmental exposures as people sought alternative countryside activities locally. To 9 inform and strengthen surveillance programmes, we investigated the impact of COVID-19 10 restrictions on the epidemiology of C. hominis and C. parvum cases.

12
Cryptosporidium-positive stools, with case demographic data, are referred routinely for 13 genotyping to the national Cryptosporidium Reference Unit (CRU). Cases were extracted 14 from the CRU database (01 January 2015 to 31 December 2021). We defined two periods 15 for pre-and post-COVID-19 restrictions implementation corresponding to the first UK-wide 16 lockdown on 23 March 2020: "pre-restrictions" between week 1, 2015 and week 12,2020,17 and "post restrictions-implementation" between week 13, 2020 and week 52, 2021. We 18 conducted an interrupted time-series analysis, assessing differences in C. parvum and C.

19
. CC-BY 4.0 International license It is made available under a perpetuity.
is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint The copyright holder for this this version posted September 26, 2022. ;https://doi.org/10.1101https://doi.org/10. /2022 NOTE: This preprint reports new research that has not been certified by peer review and should not be used to guide clinical practice. hominis incidence, trends and periodicity between these periods using negative binomial 20 regression with linear-splines and interactions.

30
This gradient did not exist post restrictions-implementation, but the effect was exacerbated 31 for the most deprived: 27.2% of cases from the most deprived decile compared to 12 is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint The copyright holder for this this version posted September 26, 2022. ;https://doi.org/10.1101https://doi.org/10. /2022

Data summary 74
Cryptosporidium is a notifiable agent in the UK which diagnostic laboratories must 75 report to local health protection teams. Submission of Cryptosporidium-positive stools 76 to the CRU is voluntary, but allows characterisation of the species. We used these 77 data, where the specimen originated from English and Welsh diagnostic laboratories, 78 to describe the epidemiology of Cryptosporidium spp. between 2015 and 2021. 79

Impact statement 80
Cryptosporidium infections in industrialised countries can cause serious disease and 81 lead to complicated and lasting sequelae, especially in the immunocompromised. 82 Even in the general population, as well as long term gastrointestinal upset, joint pain, 83 headache and eye pain have also been identified more frequently following 84 cryptosporidiosis (1). There is an established association between cryptosporidiosis 85 and colorectal cancer, although no conclusive evidence regarding causality in either 86 direction (2-5). There has never been such a dramatic reduction in international travel 87 in the modern era than during the COVID-19 pandemic, which is a key driver of C. 88 hominis infections. Conversely, pressure on outdoor amenities has rarely been higher, 89 which posed an increase in the likelihood of infection and cross-contamination for C. 90 is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint The copyright holder for this this version posted September 26, 2022. ;https://doi.org/10.1101https://doi.org/10. /2022 Introduction 96 Cryptosporidiosis is a zoonotic disease caused primarily by the protozoan parasites 97 Cryptosporidium hominis and Cryptosporidium parvum. It is most common in children 98 aged between one and five years (6-8). People with weak immune systems, especially 99 severe T-cell deficiencies, are usually more seriously affected (9,10). The most 100 common symptom is mild to severe watery diarrhoea, often accompanied by 101 abdominal cramps, nausea/vomiting, low-grade fever, weight loss and dehydration (7). 102 Symptoms begin three to 12 days (average five to seven days) after infection. In 103 healthy people, symptoms usually last about one to two weeks but can persist for up 104 to a month. The symptoms may be cyclical, where patients seem to get better for a 105 few days, then feel worse again before the illness ends. In the immunocompromised, 106 illness can be severe and protracted and sometimes fatal. There is an association with 107 previous infection and developing colorectal cancer, although no causative proof (2-108 5). Long-term sequelae such as diarrhoea, abdominal pain, nausea, fatigue and 109 headache are common (1) and infection can cause cognitive deficit and failure to thrive 110 in malnourished young children in moderate-to-low income countries (11). 111 Diagnosis is performed by microscopy (acid-fast or fluorescent staining) or 112 immunoassay to detect oocyst antigens or PCR to detect DNA. Genotyping by PCR 113 is used as a reference test to differentiate species; C. parvum and C. hominis may be 114 further sub-typed by sequencing part of the gp60 gene (12,13)  is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint The copyright holder for this this version posted September 26, 2022. ; https://doi.org/10.1101/2022.09.26.22280357 doi: medRxiv preprint and consumption of contaminated food items (7,16,17). The parasites are resistant to 125 chlorine but large enough to be captured by appropriate water filtration systems (18). 126 The majority of outbreaks in England and Wales are linked to animal contact at 127 open/petting farms (exclusively C. parvum) and swimming pools (vast majority C. 128 hominis) (13). 129 UK cryptosporidiosis cases display a seasonal trend: a late spring peak for C. parvum 130 (often associated with greater countryside activities, opening of farm-based leisure 131 activities and the lambing season) and an early autumn peak for C. hominis (often 132 associated with overseas travel and summer activities such as swimming) (19,20). 133 COVID-19 greatly limited foreign travel for most UK residents in 2020; official figures 134 show a 74% reduction in visits abroad for any reason (21), whereas the previous 135 decade had seen a steady growth in foreign travel. There was a more than 500% 136 increase in the number of people seeking holidays or leisure activities within the UK in 137 2020 (22) and also an increase in the number of people using outdoor spaces in the 138 UK in 2020-21, including walking, cycling or "wild" swimming (23-25). Furthermore, 139 people might have undertaken these activities in areas new to them and where they 140 were unaware of locally understood health risks. People were also more likely to wash 141 their hands and less likely to use swimming pools and restaurants because of 27 is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint The copyright holder for this this version posted September 26, 2022. ; https://doi.org/10. 1101 were imported into Stata V14 and cleaned to remove quality control specimen data 159 and duplicate reports. We retained only cases with C. hominis and C. parvum 160 infections who were resident in England and Wales or, if the case's address was not 161 known, where the specimen had been sent to the CRU from a laboratory in England 162 and Wales. We merged the CRU data using case's resident postcode with Office for 163 National Statistics (ONS) data for deprivation (30)  is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint The copyright holder for this this version posted September 26, 2022. ; https://doi.org/10.1101/2022.09.26.22280357 doi: medRxiv preprint weekly case-total variation was between zero to 180 cases pre-restrictions and zero 194 to four cases post restrictions-implementation. For C. parvum, the range was zero to 195 162 pre-restrictions and six to 63 post restrictions-implementation. A periodicity of 52 196 weeks was observed for C. hominis and of 26 and 52 weeks for C. parvum. 197 Age data was complete but sex classification was unknown for 0.4% of C. hominis 198 (n=35) and 0.2% (n=29) of C. parvum cases. Age distribution was similar pre-and 199 post-restrictions for C. hominis (SI figure 1) and C. parvum (SI figure 2), with children 200 aged 0-9 years experiencing a higher proportion of cases. 201 Cases of C. hominis were more likely to be female pre-restrictions but male post 202 restrictions-implementation(SI figures 3). For C. parvum, cases were more likely to be 203 female pre-restrictions and post restrictions-implementation (SI figure 4). 204 Postcodes were not recorded for approximately 10% of cases (C. hominis n=858; C. 205 parvum n=1,139), meaning that deprivation and rural/urban ranking could not be 206 assigned. There was a social gradient observed for C. hominis cases pre-restrictions, 207 where the proportion of cases increased with deprivation. This social gradient didn't 208 remain post restrictions-implementation but there was a large increase in the 209 proportion of cases from the most deprived decile (SI figure 5), where 27.2% of C. 210 hominis cases were from the most deprived decile compared to 12.7% pre-restrictions. 211 For C. parvum cases, the social gradient was the inverse of C. hominis but remained 212 stable pre-and post restrictions-implementation, with at 7.6% and 6.7% respectively 213 in the most deprived decile (SI figure 6). 214 Cases were more likely to live in an urban area for both species pre-restrictions and 215 post restrictions-implementation. The protective nature of rural residence was more 216 pronounced for C. hominis cases (SI figure 7) than for C. parvum cases (SI figure 8). 217 is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint The copyright holder for this this version posted September 26, 2022. ;https://doi.org/10.1101https://doi.org/10. /2022 three. For C. parvum cases, the most common travel destinations pre-restrictions were 228 Portugal (n=69), Spain (n=65), Turkey (n=64), France (n=63), Pakistan (n=51) and 229 India (n=35). These cases were again distributed fairly evenly across all deprivation 230 deciles with the exception of those who had been to Pakistan, with 63% (n=32) of 231 cases ranked in deciles one to three. is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

261
The time-series data for C. parvum had a periodicity of 26 and 52 weeks, with biannual 262 peaks visible in spring and in autumn. An interaction was found between our COVID 263 variable and both Fourier waves so these were retained in our final model. In the 264 forecast model of the previous five years' time-series data beyond week 13, 2020 for 265 C. parvum, we observed a larger number of cases than predicted had COVID-19 not 266 occurred, but not in the same order of magnitude as predicted for C. hominis ( figure  267 2). Incidence of C. parvum had recovered to pre-restrictions levels by the end of 2021. 268 . CC-BY 4.0 International license It is made available under a perpetuity.
is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint The copyright holder for this this version posted September 26, 2022. ; https://doi.org/10.1101/2022.09.26.22280357 doi: medRxiv preprint is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint The copyright holder for this this version posted September 26, 2022. ;https://doi.org/10.1101https://doi.org/10. /2022   is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint The copyright holder for this this version posted September 26, 2022. ;https://doi.org/10.1101https://doi.org/10. /2022 Discussion 299 COVID-19 restrictions had a significant effect on the number of cases of 300 gastrointestinal illness (GI) in England and Wales including Cryptosporidium(28), and 301 in this study we have identified that they impacted both C. hominis and C. parvum. 302 Whilst C. parvum cases were reduced by around half, C. hominis cases were almost 303 entirely arrested. Whilst our model for C. hominis detected a change in periodicity 304 (table 1 It is not possible to quantify exactly the extent to which COVID-19 restrictions reduced 314 social mixing or any commensurate effect on person-to-person transmission, which is 315 a limitation. C. hominis cases had been declining in the five years leading to COVID, 316 despite foreign travel increasing. It is possible that other interventions were influencing 317 these data, such as improved compliance and awareness in swimming pool filtration 318 and avoidance with GI symptoms. We do know that public venues were closed for 319 much of our post restrictions-implementation period, including known sources of 320 is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint The copyright holder for this this version posted September 26, 2022. ;https://doi.org/10.1101https://doi.org/10. /2022 Another limitation in our analysis is the effect that reduced healthcare provision and 333 health-seeking behaviours had on specimen submission post restrictions-334 implementation, particularly at the start of the COVID-19 pandemic. There is 335 international evidence showing that people sought help less often, and those who were 336 only mildly ill did not seek help at all (37). There was genuine fear about being infected 337 with SARS-CoV-2 in health care settings early on when lack of an effective vaccination 338 or definitive therapy were at the forefront of many people's minds (38,39). This might 339 have attenuated people's opinion of diarrhoea and vomiting symptoms, or decisions 340 to seek medical care. Likewise, there might have been disruptions in access to, and 341 application of, the stool specimen and diagnostic process, resulting in fewer cases 342 being detected. 343 The fact that people living in the most deprived areas experienced the highest 344 proportion of C. hominis cases both pre-restrictions and post restrictions-345 implementation highlights potential opportunities for public health interventions. Of 346 those in the most deprived decile where travel information was available post 347 restrictions-implementation, 83% had surnames suggestive of a non-white ethnic 348 background (n=5). This, coupled with the fact that in 36% of all cases where there had 349 been foreign travel involved Pakistan, provides a compelling opportunity to target 350 interventions to outgoing and returning travellers. However, the limited data available 351 for travel history and ethnicity means we cannot draw absolute conclusions, despite 352 the length of our study period. Ethnicity is often poorly reported for many diseases and 353 this reduces the ability to understand the true impact of cryptosporidiosis in specific 354 communities. Promotion of accurate ascertainment of ethnicity and travel history 355 should be sought at the point of specimen collection to improve future surveillance. 356 This would require a change to the sample submission form sent to the laboratory and 357 training for clinicians taking history. The fact that cases of both species were more 358 likely to live in an urban residence should be interpreted with caution. Our data, being 359 laboratory-based, did not include details of recent trips to the countryside, occupation 360 or contact with animals and its relevance is diminished further where foreign travel is 361 indicated. 362 Our results also demonstrate the need to improve rates of specimen submission to the 363 CRU where there is currently under-representation, especially from areas that serve 364 large clusters of ethnic minority communities. A submission bias exists in that of the 365 approximately 4,500 human Cryptosporidium cases a year in England and Wales (40), 366 only around half are submitted to the CRU for genotyping on average. This proportion 367 . CC-BY 4.0 International license It is made available under a perpetuity.
is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint The copyright holder for this this version posted September 26, 2022. ;https://doi.org/10.1101https://doi.org/10. /2022