Background

Cryptosporidiosis is a clinical disease, typically affecting the intestinal tract of humans and animals who have ingested the protozoan parasite Cryptosporidium in its oocyst (infective) stage [1]. Transmission of Cryptosporidium occurs predominantly via the faecal-oral route, or through consumption of contaminated food or water and therefore the prevalence of human Cryptosporidium infections is higher in low-resource settings [2]. However, Cryptosporidium infections are not infrequent in industrialized countries [3], with large outbreaks being reported in Sweden [4], the USA [5, 6] and the UK [7] following contamination of public water supplies.

While asymptomatic carriage is possible [8, 9], human cryptosporidiosis typically presents as an acute, gastroenteritis-like illness characterized by profuse, watery diarrhoea, frequently accompanied by abdominal pain/cramps, vomiting and weight loss, as well as more non-specific symptoms such as fatigue, low-grade fever, nausea and muscle weakness [10]. In immunocompetent hosts, cryptosporidiosis is generally self-limiting; however, disease severity can be influenced by host factors, such as age, immune status and nutritional status, as well as pathogen factors e.g. Cryptosporidium species and subtype [11].

Alongside ongoing interest in the acute symptomology of human cryptosporidiosis, there is also growing evidence to suggest that, rather like some bacterial causes of gastroenteritis and giardiasis [12,13,14], Cryptosporidium infection may have longer-term health consequences. Seven studies [15,16,17,18,19,20,21], with follow-up periods ranging from 2 months to 3 years, have investigated numerous potential post-Cryptosporidium infection sequelae including diarrhoea, abdominal pain, vomiting, loss of appetite, irritable bowel syndrome (IBS) [21], joint pain and fatigue, while case reports document incidences of reactive arthritis [22,23,24], Reiter’s syndrome [25], acute pancreatitis [26, 27] and haemolytic uremic syndrome [28], in the context of Cryptosporidium infection. There is also some emerging evidence, recently reviewed, of a possible association between cryptosporidiosis and cancer [29].

Due to resource limitations, public health professionals currently face the challenge of identifying and prioritising specific infectious diseases whose quantified burden of disease estimates justify the allocation of interventions and funding for research [30]. The Global Enteric Multicentre Study [31] identified Cryptosporidium as the second most common cause of moderate-to-severe diarrhoea (MSD; defined as diarrhoeal disease with presence of the suggestive features of sunken eyes, wrinkled skin, hospitalization, receipt of intravenous hydration, or dysentery) in children less than 2 years-old within sub-Saharan Africa and south Asia, while in 2016, the European Network for Foodborne Parasites (Euro-FBP) ranked Cryptosporidium spp. as the second highest priority foodborne parasite in northern and western Europe, and the eighth highest priority in eastern and south-western Europe [30]. However, actual burden of disease estimates for Cryptosporidium still vary widely [11] and it remains difficult to quantify the true burden of cryptosporidiosis, as current estimates only account for the morbidity and mortality associated with the acute illness, while the potential contributions of long-term manifestations are not included [32, 33]. A recent study from the Netherlands [2] found that long-term manifestations contributed nearly 10% of the total Disability-Adjusted Life Years (DALYs) and costs when included in burden of disease models for Cryptosporidium, suggesting a higher public heath burden and cost than previously estimated.

Accurate estimations of the burden of disease associated with Cryptosporidium will inform decisions regarding the allocation of diagnostic, surveillance and interventional measures to prevent and control Cryptosporidium infections. Due to the potential morbidity and mortality associated with long-term sequelae of human cryptosporidiosis, an accurate estimation of the proportion of cases that develop such sequelae is needed to quantify true burden of disease estimates for Cryptosporidium.

The objectives of this review were: (i) estimate the proportion of people that self-report health sequelae post-Cryptosporidium infection; (ii) estimate the risk of specific sequelae following Cryptosporidium infection; and (iii) explore potential risk factors associated with developing sequelae following Cryptosporidium infection in industrialised countries.

Methods

Search strategy

We searched for studies in PubMed, ProQuest and Web of Science databases, with no limitations on publication year or language. The reference lists from relevant papers identified during our electronic searches were also reviewed for additional relevant papers which may warrant inclusion in our review. Search terms were initially developed and piloted using PubMed and, to ensure consistency, the same search terms were used when searching ProQuest and Web of Science databases. Databases were searched using the following keywords: Cryptosporid*, Complications, Sequel*, Post-infecti*, Long term and Chronic. The full electronic search strategies are documented in Additional file 1: Table S1. The review was registered with PROSPERO, registration number CRD42019141311.

Selection of studies

All citations identified using the final search strategies were exported to Mendeley® reference managing software for organisation and removal of duplicates. The titles and abstracts of the remaining articles were screened for relevance by one reviewer (BC), after which, the remaining articles were independently screened by two reviewers (BC and APD) to ensure consistent application of the pre-determined inclusion/exclusion criteria (Additional file 1: Table S1).

Studies from any country were included in qualitative synthesis, but only those in industrialised countries were included in quantitative analysis. An industrialised country was defined using Organisation for Economic Co-operation & Development (OECD) membership.

Final inclusion of studies was decided by consensus, with any conflicts being reviewed by a third reviewer (RMC). The full text was retrieved and reviewed for articles where the title and abstract had been deemed relevant by reviewers.

Data extraction

Data were extracted from eligible studies and collated in a Microsoft Word document. We recorded post-Cryptosporidium infection health sequelae data as reported in the individual papers (e.g. prevalence, cumulative incidence, etc.). Relative risks or odds ratios were recorded where data were available. We also extracted the following study characteristics from each paper if available: name of authors, year of publication, study location/setting, study design, year(s) of study, study duration and duration of follow-up, number of included study participants, participation rate, study population demographics (including age and gender distributions), Cryptosporidium species data, the diagnostic method to ascertain Cryptosporidium infection and the types of sequelae reported. Additionally, where available, data on the incidence/prevalence of post-infectious IBS following Cryptosporidium infection and the IBS diagnostic criterion applied were collected.

Quality assessment

The methodological quality of the included studies was assessed using the Newcastle-Ottawa Scale (NOS) for nonrandomised studies [34]. NOS was used to score studies using three domains: (i) the selection of the study groups; (ii) the comparability of the groups; and (iii) the determination of either the exposure or outcome of interest in case-control or cohort studies, respectively. Scores ranged between 5–8 (Additional file 1: Table S2).

Statistical analysis

The proportion of Cryptosporidium cases that developed specific sequelae was calculated by dividing the number of individuals developing a sequela by the total number of Cryptosporidium cases. Where data were available from two or more appropriate studies, we used a random-effects meta-analysis model to obtain pooled estimates of prevalence for the outcomes of interest (i.e. sequelae) across the eligible studies. For this analysis, a study could be included more than once if sequelae data were reported longitudinally at different time periods. Data analyses were performed using Meta XI [35].

Assessment of heterogeneity and reporting biases

Forest plots and the I2 statistic were used to assess heterogeneity between the studies. Values of 0–40%, 30–60%, 50–90% and 75–100% were interpreted as; might not be important, may represent moderate heterogeneity, may represent substantial heterogeneity and considerable heterogeneity, respectively [36]. Funnel plots were used to assess for publication bias and small-study effects.

Stratified analysis was performed for the following subgroups; time (less than 6 months post-infection and more than 6 months post-infection) and Cryptosporidium sp. (e.g. C. parvum vs C. hominis).

Results

Data synthesis

The number of papers identified, included and excluded is presented according to the requirements of the PRISMA statement [37] in Fig. 1. Fifteen studies were identified for qualitative synthesis and eight of these were identified as being set in industrialised countries and of sufficient quality for additional quantitative synthesis.

Fig. 1
figure 1

PRISMA flowchart

Full size image

The qualitative synthesis is shown in Table 1. Quantitative synthesis results are shown in Tables 2, 3 and 4 and Figs. 2 and 3.

Table 1 Fifteen studies included in the qualitative synthesis
Full size table
Table 2 Pooled estimates for the prevalence of post-Cryptosporidium sequelae using a random effects model
Full size table
Table 3 Pooled prevalence of post-Cryptosporidium sequelae estimated by a random effects model, according to clinical manifestation by time period post-infection
Full size table
Table 4 Pooled risk ratio of individual post-Cryptosporidium sequelae
Full size table
Fig. 2
figure 2

Reported sequelae up to 36 months post-Cryptosporidium infection

Full size image
Fig. 3
figure 3

Reported sequelae up to 36 months post-Cryptosporidium infection by species (%)

Full size image

Qualitative synthesis

Electronic searching returned 1251 PubMed, 2161 ProQuest and 3227 Web of Science abstracts. After removal of duplicates, screening and assessment, 15 articles were suitable for inclusion in the qualitative synthesis and the data extracted from these studies are summarized in Table 1.

The 15 shortlisted studies included 3670 Cryptosporidium cases. The studies comprised 8 cohort studies and 7 case-control studies. Seven studies were conducted in children, with the remaining 8 studies including both adults and children. The length of duration of follow-up ranged from 2 months to 9 years. Studies were conducted in South America (3 studies), Africa (1 study), South Asia (2 studies) and Europe (9 studies) and were all based in a community setting. The selected studies were published between 1992 and 2019. The studies investigated a range of potential sequelae; diarrhoea (3 studies), developmental delay (2 studies), stunting of growth (4 studies) and multiple gastrointestinal and non-gastrointestinal symptoms (8 studies).

Quantitative synthesis

Adequate information to estimate post-Cryptosporidium infection sequelae was available in 8 of the 15 studies [4, 15,16,17,18,19,20,21]. The pooled estimates for each of the sequelae are shown in Table 2. Data for each individual sequela are available in Additional file 2.

The eight studies were conducted in Europe between 2004 and 2019; four in Sweden, three in the UK and one in the Netherlands. The sequelae investigated were mostly gastrointestinal, with some non-gastrointestinal symptoms such as joint pain and eye pain and most recruited cases were adults. This was in contrast to studies in non-industrialised countries which focused on growth, nutrition and cognitive detriment in children.

The most frequently investigated sequelae are listed in Table 2 and included diarrhoea, abdominal pain, vomiting, fatigue, joint pain, eye pain and headache.

The most common reported long-term sequelae were diarrhoea (25%), abdominal pain (25%), nausea (24%), fatigue (24%) and headache (21%). The distribution of gastrointestinal manifestations and non-gastrointestinal manifestations reported is shown in Fig. 2.

Subgroup analysis

Table 3 shows the pooled estimates for the prevalence of post-Cryptosporidium sequelae by time period post-infection. With the exception of eye pain and headache, all sequelae were more frequently reported within 6 months of Cryptosporidium infection.

In all eight studies included in the quantitative analysis, species identification of Cryptosporidium had been performed. Four were outbreak cohort follow-up studies so contained only one species (three contained C. hominis cases exclusively and one contained C. parvum exclusively). The other four studies contained both species; one of these four also contained a small number of other species (17/271 cases), but because of the low numbers, these have not been considered here. Figure 3 shows the pooled estimates for the prevalence of post-Cryptosporidium sequelae by Cryptosporidium species. Overall, long-term sequelae were more prevalent following infection with C. hominis, with only weight loss and blood in stool being more prevalent following infection with C. parvum. IBS was reported in 11% of cases, however, it should be noted that data for this outcome were only available from 2 studies, one of which only studied C. parvum cases.

Sequelae risk

Five of the 8 qualitative synthesis studies included a control group. A limited evaluation of risk of individual sequelae using the five case-control studies available was undertaken [4, 15, 17, 19, 20]. Data were available for 10 sequelae (Table 4).

Individuals were 6 times more likely to report chronic diarrhoea and weight loss up to 28 months after a Cryptosporidium infection than controls. Long-term abdominal pain, loss of appetite, fatigue, vomiting, joint pain, headache and eye pain were also 2–3 times more likely following a Cryptosporidium infection (Fig. 4).

Fig. 4
figure 4

Pooled risk ratio of individual post-Cryptosporidium sequelae showing 95% confidence intervals

Full size image

To view the PRISMA checklist relating to this work, please see Additional file 3.

Discussion

Of the 15 studies investigating long-term sequelae, just over half were set in industrialised countries. In contrast to those in non-industrialised settings, these involved mainly adult cases, with the inclusion of some children. Half were outbreak cohort studies, with the rest involving sporadic community cases. Studies from non-industrialised countries involved exclusively children, reflecting the greater clinical importance and recognition of paediatric infection in such settings. In industrialised countries there is more focus on detecting sporadic community cases of cryptosporidiosis in all age groups, partly in order to facilitate early detection of community outbreaks, for example from drinking water, swimming pools, or other environmental sources. The studies in non-industrialised countries also differed in that the children were recruited and tested as part of the specific studies, whilst the studies in industrialised countries relied on cases initially diagnosed routinely.

The eight studies suitable for inclusion in the quantitative analysis were all carried out in just three countries in Europe (UK, Sweden and the Netherlands), where species data are routinely generated and all except one [11] were relatively recent, dated between 2013–2019. In many non-industrialised countries, or in earlier European studies, species identification would not be routinely performed, and this is reflected in the study data. The geographical reach of the eight studies is somewhat limited, since they were all located in northwest Europe. Only five were case-control studies, and of these, only two included both C. hominis and C. parvum, with the other three limited to studying C. hominis alone following outbreaks. Since the bulk of the cryptosporidiosis burden is found in low-income countries, there is a need in future to conduct similar quantitative evaluations using data from developing countries, where obtaining suitable data may be more challenging.

There were some limitations to this review. The role of genotype in long-term outcomes could not be explored. Typing was undertaken by gp60 sequencing in three of the studies but was either not analysed with symptoms data [16], or was an outbreak where all had the same subtype [4, 18]. There were insufficient data to compare between studies. Another limitation was that since not all cases were necessarily tested for all gastrointestinal pathogens, or the results of such tests were not stated, long-term sequelae identified cannot be proven to be Cryptosporidium-specific and not due to other infectious agents.

Most of the studies examined quantitatively were concentrated on adult individuals, whereas cryptosporidiosis is commonest in young children. This over-representation of adults results from the fact that several of the studies followed large waterborne outbreaks involving many adults, rather than sporadic cases. Identifying and defining sometimes rather non-specific sequelae is more difficult in very young children. However, a study by Carter et al. [21] of sporadic cases did include children, and in fact this study found that the proportion developing IBS or IBS-like symptoms was higher in children than in adults, with 78% reporting it among 5–17 years-old and 63% at 6 months to 4 years-old.

The results indicate that sequelae are frequently reported after cryptosporidiosis lasting up to at least 2 years. Only one study investigated cases for longer, up to 36 months [11]. For both main infecting species, sequelae occur, but there are differences in the frequency of each depending on the species. Following the publication of the first study in 2004 [15], the evidence base surrounding post-Cryptosporidium infection sequelae has continued to expand [16,17,18,19,20,21]. Gastrointestinal sequelae such as continuing diarrhoea, nausea and abdominal pain appear particularly common, each reported by around a quarter of cases up to 36 months post-infection, with analysis of the case-control studies finding that persistent diarrhoea is around six times more likely than in controls and weight loss over three times more likely over 28 months. Fatigue and headache were also commonly reported and occurred in the case-control studies between two-three times more commonly in cases than controls over the same time period. Overall, the most commonly reported long-term sequelae were diarrhoea (25%), abdominal pain (25%), nausea (24%), fatigue (24%) and headache (21%). Where it was investigated, there was evidence that symptoms meeting the definition for IBS were described just over 10% of cases up to 36 months.

Conclusions

This is the first systematic review of the long-term sequelae of cryptosporidiosis. The proportion of cases self-reporting sequelae post-infection has been estimated and estimates of risk of specific sequelae presented. Risk factors for sequelae were less well identified. A better understanding of the long-term outcomes of cryptosporidiosis is valuable to inform the expectations of clinicians and their patients and public health policy makers regarding the control and prevention of this infection.