Intestinal Parasite Infections in Symptomatic Children Attending Hospital in Siem Reap, Cambodia

Background Infections with helminths and other intestinal parasites are an important but neglected problem in children in developing countries. Accurate surveys of intestinal parasites in children inform empirical treatment regimens and can assess the impact of school based drug treatment programmes. There is limited information on this topic in Cambodia. Methods In a prospective study of intestinal parasites in symptomatic children attending Angkor Hospital for Children, Siem Reap, Cambodia, April-June 2012, samples were examined by microscopy of a direct and concentrated fecal sample. Two culture methods for hookworm and Strongyloides stercoralis were employed when sufficient sample was received. Demographic, clinical and epidemiological data were collected. Principal Findings We studied 970 samples from 865 children. The median (inter-quartile range) age of the children was 5.4 (1.9-9.2) years, 54% were male. The proportion of children with abdominal pain was 66.8%, diarrhea 34.9%, anemia 12.7% and malnutrition 7.4%. 458 parasitic infections were detected in 340 (39.3%) children. The most common parasites using all methods of detection were hookworm (14.3%), Strongyloides stercoralis (11.6%) and Giardia lamblia (11.2%). Giardia lamblia was most common in children aged 1-5 years, hookworm and Strongyloides stercoralis were more common with increasing age. Hookworm, Strongloides stercoralis and Giardia lamblia were more common in children living outside of Siem Reap town. In a multivariate logistic regression increasing age was associated with all three infections, defecating in the forest for hookworm infection, the presence of cattle for S. stercoralis and not using soap for handwashing for G. lamblia. Conclusions/Significance This study confirms the importance of intestinal parasitic infections in symptomatic Cambodian children and the need for adequate facilities for laboratory diagnosis together with education to improve personal hygiene and sanitation.

Introduction a single fecal sample. Here we report a prospective study of fecal parasites in symptomatic Cambodian children attending AHC using direct microscopy, a formol-petrol concentration technique (FC), and culture methods [24]. We additionally identified factors associated with infection with common parasites detected.

Study setting
This was a prospective study of symptomatic children attending Angkor Hospital for Children (AHC) in Siem Reap, North-Western Cambodia, between 3 rd April 2012 and 29 th June 2012. AHC is a 50-bedded, charitably-funded pediatric hospital, providing free intensive, surgical and general medical care to children <16 years of age from Siem Reap and surrounding provinces. It has approximately 125,000 attendances and 4,000 admissions each year. About 5,000 stool samples are processed each year by direct microscopy for intestinal parasites [22]. Patients attend AHC from all over the country, some travelling a long distance to reach the hospital.

Ethical consideration and treatment
The study was approved by the Institutional Review Board at AHC and the Oxford Tropical Research Ethics Committee (OXTREC  and conducted in compliance with the STROBE initiative. Children in the out-patient clinic or admitted to hospital were eligible if the treating doctor had requested a fecal parasite examination because of diarrhoea, abdominal pain, clinical evidence of anemia or malnutrition. The study was explained to patients and their caregivers, and informed consent was confirmed by the caregiver's signature or a witnessed thumbprint if they were illiterate. Because the study was set in a busy outpatient clinic consent was missed in some patients where a sample was examined using the additional parasite diagnostic methods. For patients who did not provide consent, the results were used to treat the patient but no patient information was collected and the results were not included in this study. All patients (consented or not) found to be infected with a parasite not covered by the initial treatment were contacted by telephone and encouraged to return to hospital for additional treatment appropriate for the microbiological diagnosis.

Clinical, demographic and epidemiological data collection
The attending physician recorded clinical and demographic information on a standard form including age, gender, the presence of diarrhoea (duration if positive), abdominal pain (together with duration), anemia and malnutrition, weight and chronic medical conditions. Potential risk factors for infection were also recorded including the number of people living in the household (the number <2 years old, those 2-15 years and adults), the presence of domestic pets (specifically: cat, dog, bird or other), the presence of livestock (chickens, pigs, cows or other), the availability of water at the house, the type of water routinely used (city water, river, rain, well, pond or bottled water), whether soap was used to wash hands (always, sometimes or never), where the family passed their stools (in a toilet, in the forest, farm, outside the house or in the river), whether the patient attended school and whether they wore shoes.

Laboratory methods
Fecal samples were examined using a variety of methods depending on the volume received. All samples had a routine direct microscopy examination of a wet preparation of feces within one hour of receipt [25] and most underwent FC using the Evergreen fecal parasite concentrator (Evergreen Scientific, Los Angeles, California). Ether could not be obtained in Cambodia and this was replaced with locally available petrol [26,27]. The presence of Cryptosporidium species was determined using Ziehl-Neelson staining of a smear prepared from the concentrated deposit [28].
The agar plate culture (APC) [29,30] and charcoal coproculture method [31] were used to detect the presence of hookworm and S. stercoralis. Samples were processed for culture within 4 hours. Plates were incubated at room temperature for seven days and examined daily using a Kyowa Iroscope (Tokyo) plate microscope. When tracks appeared on the agar plates or larvae were observed using the coproculture method, the larvae were removed from culture and examined under the light microscope for morphological identification.
An intensive one-week training course was conducted for all laboratory staff involved prior to commencement of the study by an experienced UK parasitologist.

Statistical analysis
Based on previous data using direct microscopy we expected 3% of samples to be positive for S. stercoralis [22]. Direct microscopy may miss up to 70% of S.stercoralis infections. We planned to study approximately 1,000 disease episodes which we therefore estimated would give approximately 30 S. stercoralis-positive episodes by direct microscopy, 70 episodes using concentration and 100 episodes by culture.
Statistical analyses were performed using STATA version 13.1 (Stata Corporation; College Station, TX, USA). Study participants were subdivided into five age groups for analysis: neonates (28 days); infants (29 days-<1 year); 1-5 years; 6-10 years and 11-16 years. A chisquared test was used to examine the association between categorical variables. Risk factors were examined for a statistically significant association with the detection of the most common parasites, hookworm, S. stercoralis and G. lamblia, were analysed using univariate and multivariate logistic regression using an odds ratio with 95% confidence intervals. A logistic regression model was utilised to determine the prevalence of hookworm, S. stercoralis and G. lamblia in the samples.

Data collection
During the study period 1,248 stool samples were examined from 1,138 patients. Of these, seven samples were received from over-age patients, 40 samples were received from other health care facilities in Siem Reap and consent was missed for 231 samples. None of the remaining patients or family members refused consent to participate (Fig 1). There were no significant differences in the age, sex and range of parasites detected between the patients in whom consent was or was not obtained.
A total of 865 patients submitted 970 samples; 787 patients provided a single sample and 78 patients provided multiple samples (183 multiple samples). All repeat samples were received within a two-month period and were considered to be part of a single disease episode for each patient. The first negative sample (if all subsequent samples were negative) or one positive sample (if all other samples were negative) was included in the analysis. If different pathogens were detected in different samples by any method for the same patient, the results were combined.
ten. There were 39 patients (50%) with at least one positive sample, S1 Table. Nine patients had two samples positive with identical parasites detected in five but with differences in four.
Demographic, clinical and epidemiological data (Table 1) The median (interquartile range (IQR)) age of study children was 5.4 (1.9-9.2) years and 54% were male. There were no significant differences in sex between different age groups (p = 0.75), S2 Table. About half of the children were outpatients and these were older than inpatient children (Table 1 and S2 Table). All of the neonates and 65.1% of the infants were inpatients (p<0.0001). Overall 24.2% of patients lived in Siem Reap town. Diarrhoea was present in onethird of patients and was most common among infants (p<0.0001). Two-thirds of patients reported having abdominal pain and this was significantly more common in older children (p<0.0001). Malnutrition and wasting were most common in the infants (p<0.0001 and p = 0.057 respectively). Kwashiorkor was only present in the 1-5 year age group (seven patients, 5.6% of patients aged 1-5 years).
Dogs were recorded as present in 84.6% of households. Sixty five percent of patients had livestock (water buffalo, chickens, pigs, cattle or ducks) living in their household with chicken ownership most common at 87.7%. A garden well was the main source of water for 77.5% of patients, although this was often supplemented by city water and bottled water for drinking. Families with younger children were more likely to use water from a pond as their main water source (p = 0.02).   Percentage of disease episode by age (total number of disease episodes/number in age group*100). *331 samples examined for Cryptosporicium species and Cyclospora cayetanesis using ZN stains and **Four samples were culture positive, however the parasite species could not be determined Parasite infections ( Table 2) A total of 459 parasitic infections were detected in 339 (39.2%) children (Table 2). No parasitic infections were detected in neonates. Parasitic infections were detected in 4.0% of infants, 40.3% of children 1-5 years, 54.4% of children 6-10 years and 44.5% of children aged 11-16 years. Hookworm (14.3% of episodes), S. stercoralis (11.6% of episodes) and G. lamblia (11.2% of episodes) were the most common parasites detected. Children aged 1-5 years had the highest proportion of G. lamblia with 27.7% positive. The proportion of children positive for both hookworm and S. stercoralis increased with increasing age (p<0.001).

Multiple parasites (polyparasitism)
Of 330 patients positive for any parasite type: a single parasite type was present in 227 patients (68.8%); two parasite types in 80 patients (24.2%); three types in 22 patients (6.7%); and four types in a single patient (0.3%) ( Table 3). The most common multiple infections were combinations of G. lamblia, hookworm and S. stercoralis: G. lamblia together with S. stercoralis were detected in 14 (4.2%) patients, G. lamblia and hookworm were detected in 18 (5.5%) patients, hookworm and S. stercoralis were detected in 52 (15.8%) patients and all three parasites were detected in 7 (2.1%) patients. Due to an insufficient volume of sample in some patients, APC was performed in 533/865 (61.6%) and charcoal culture in 293/865 (33.8%). Some cultures were lost because of maggot infestation in 25/533 (4.7%) of APC and 8/293 (2.7%) in charcoal culture. APC was compared with microscopy in 508 samples. The additional yield of APC over microscopy for hookworm was only 4/82 (4.9%) but for the S. stercoralis was 49/83 (59.4%) ( Table 5). Both culture methods could be compared with microscopy in 263 samples. The additional absolute yield for APC was 9.5% (4/42) for hookworm and 14.6% (7/48) S. stercoralis and for charcoal culture was 11.9% (5/42) for hookworm and 18.8% (9/48) for S. stercoralis. There were no significant differences in the additional yield from APC compared with charcoal culture.

Risk factors for the detection of G. lamblia, hookworm and S. stercoralis
Hookworm and S. stercoralis were more commonly detected in older children (Table 6 and S3  Table; Table 4. Comparison of methods for presence of faecal parasites a (see Table 2  Using the toilet for defecation was protective for all three parasites (OR 0.26, 0.48 and 0.66 for hookworm, S. stercoralis and G. lamblia with 95% CIs 0.17-0.39, 0.31-0.73 and 0.43-1.0 respectively), using the city water was protective against hookworm infections (OR 0.34, 95%CI 0.14-0.86) and living in Siem Reap town rather than in the more rural area in the province was protective, particularly for hookworm infection (OR 0.24, 0.37 and 0.55, 95%CI 0.13-0.48, 0.19-0.70 and 0.31-0.98 respectively). The associations which remained significant for infection using multivariate logistic regression were an increasing age for hookworm and S. stercoralis, using the forest for defecating for hookworm (OR1.92, 95%CI 1.02-3.61), the presence of cattle for S. stercoralis (OR1.99, 95%CI 1.14-3.46) and not using soap for handwashing for G. lamblia (OR2.15, 95%CI 1.27-3.65).

Discussion
In this prospective study of more than 800 symptomatic children attending hospital in Siem Reap, the most common intestinal parasites were G. lamblia, hookworm and S. stercoralis. G. lamblia infections were most common in children aged 1-5 years and hookworm and S. stercoralis infections in school-aged children. The range and relative proportions of the different parasites in this study were similar to a previous retrospective study of more than 5,000 children at this centre. With staff refresher training and the use of additional laboratory methods the proportion of G. lamblia positive episodes increased from 8.0% to 11.2%, hookworm from 5.1% to 14.3%, and S. stercoralis from 2.6% to 11.6% [22]. Table 5. Comparison of methods for presence of faecal parasites (see Table 2 for the overall numbers).  Recent regional studies using similar methods have reported the proportion of G. lamblia positive children to range between 2.9-4.2% [19,32,33], hookworm up to 76.8% [34] and S. stercoralis from 24.5-44.7% [12,13,16,23] The proportion of children in this study with Ascaris lumbricoides was low compared to other studies [19,25], but similar to recent studies conducted in Cambodia [16,19]. Multiple parasites were detected in one third of infected patients, most commonly combinations of G. lamblia, hookworm and S. stercoralis.
Children in Siem Reap province are treated twice a year (May and November) with mebendazole in the mass drug administration programme. The programme aims to reduce the prevalence of STH infections, and their associated morbidity, rather than eliminate them [35]. Although this could have potentially affected the numbers of infections that were observed in our study there was no evidence of a lower proportion of positives in May and June compared with April. Single dose mebendazole has limited efficacy against G. lamblia, hookworm and S. stercoralis [36][37][38]. The recent study by Khieu et al. found a low cure rate for hookworm infections when treated with mebendazole [16,19].
Direct microscopy detected parasites in 32.7% of patients. Fecal concentration increased the prevalence to 43.9% and this was particularly apparent for hookworm, B. hominis, S. stercoralis and G.lamblia. In a similar study by Moges et al, 50.3% were positive using direct microscopy and 79.1% and 73.6% positive using formol-ether and formol-acetone concentration methods respectively. [39]. In that study 3.9% of samples were positive for hookworm using the direct iodine method compared to 19.6% and 19.4% by the formol-ether and formol-acetone concentration methods. Parija et al. [40] also described an increase in positives from 34.7% positive by wet preparation to 65.3% using formol-ether concentration.
Culture methods increased the detection of S. stercoralis by 30-62.5% and hookworm by 9-21.4%. This was less than a recent study in Cambodian children and adults [16]. A limitation of culture methods is of the potential to underestimate hookworm infections. When mixed infections are cultured, S. stercoralis can multiply, overwhelm and mask hookworm. There was no clear difference between the two culture methods employed, but sample volumes were insufficient in many cases to perform both culture methods on all samples and as a result the comparison was under-powered to demonstrate a difference.
The study was limited as only a single stool sample was examined from most of the children [41]. Two studies in Cambodia which examined asymptomatic children 6-19 years of age found an increased yield of S. stercoralis of 4.1% in the second sample and 1.7% in the third sample using both the Koga agar plate and Baermann technique [16]. Another study in Cambodian schoolchildren found an increase in yield from 18.6% with one sample to 22.7% with two samples and 24.4% with three samples for S. stercoralis and an increase from 36.0% to 45.9% to 49.3% with one, two and three cultures respectively for hookworm [16]. These are similar increases to those seen elsewhere, e.g. in 1253 adults in Bangkok the first sample detected 74.6% of all parasites found, the second sample 18.5% and the third 6.9% [41]. A further limitation is the lack of molecular methods which could have increased the numbers of fecal parasites detected, improved detection of mixed infections and allowed for parasite speciation [42]. Recent studies in asymptomatic children in southern Cambodia using real-time PCR found a prevalence of hookworms of 34.9% and S. stercoralis of 17.4% [43] and S. stercoralis at a prevalence of 44.7% in a recent cross-sectional study of children and adults living in the Preah Vihear region [16].
Our findings do not imply causation of disease in this patient population, as we did not perform a case-control study comparing the prevalence of parasites in symptomatic children and asymptomatic children from similar demographic settings. A case-control study might also clarify the importance of risk and protective factors. However, other studies support a pathogenic role for most of these parasites in children in South Asia [13,16,18,23,44].
Hookworm, strongyloides and giardia infections were all more common in children with abdominal pain and in children regularly defecating anywhere apart from a toilet (such as the forest or farm). Hookworm and S. stercoralis infection were also more common among children living outside Siem Reap town. Hookworm infection occurred in older (one year or above) children, who drank pond water, in households with domestic animals (cats or dogs), and with cattle/water buffalo and those who defecated in the forest. An association between hookworm infection and living with water buffalo has been described in Thailand [45]. Specific additional risk factors associated with S. stercoralis infection included living with chickens or more significantly cattle (but not water buffalo) and for giardiasis using water from a well and not washing hands with soap. This is the first prospective systematic study to examine the presence of intestinal parasites in symptomatic Cambodian children. We found significant levels of infection with G. lamblia, hookworm and S. stercoralis, despite the national Mass Drug Administration (MDA) programme. The effectiveness and risks of MDA programmes have been debated recently [46]. Local data from studies such as this one are important to inform appropriate clinical management, the appropriate choice of diagnostic tests and to assess the true impact of public health initiatives such as the MDA programme [35].
Supporting Information S1