Ethics

The study was approved by the Clinical Research Ethics Committee of the Chinese University of Hong Kong and by the corresponding ethics committees of individual participating hospitals.

Sample

Children <5 years old were eligible for recruitment if they had been admitted to a participating hospital with a primary diagnosis of diarrhea within the previous 24 h and if their families could be contacted by telephone. Children were excluded if they subsequently did not pass any stool or were discharged within 48 h of admission without any bowel evacuation. After a pilot study was done to refine the questionnaires, participants were recruited from 1 March 2001 through 4 March 2003. The etiological organism responsible for the children's diarrhea was not known at the time of recruitment, and, to ensure that there was a sufficient number of patients with rotavirus-positive diarrhea available for analysis, we increased patient recruitment during the peak season for rotavirus disease (November 2002 through March 2003).

Data collection

Participating families signed informed consent forms. An initial questionnaire, which included demographic and clinical details, was completed at recruitment. Information regarding the clinical treatment of participants was obtained from hospital case notes by use of a structured questionnaire. Details of the costs related to each child's diarrheal episode were obtained from the families, and families were given a symptom diary to complete during the 10 days after recruitment (while the child was hospitalized and after discharge). The research assistant demonstrated how to complete the symptom diary, which was intended to be a memory aid, to document all costs related to the child's diarrheal episode. The families were encouraged to return the diary in a prestamped, preaddressed envelope after 10 days. Information recorded in the symptom diary included the length of the hospital stay, visits to other health care providers, extra travel costs, costs of additional diapers used, other miscellaneous costs (e.g., food supplements or nonprescription remedies), time off work required by family members responsible for the care of the child as a result of the child's illness, and the respective estimated hourly, daily, weekly, or monthly salary lost. The families were telephoned 5 days after the date of admission if the child had been discharged from hospital and, then, again 10 days after the date of admission. At these contacts, the details of the various costs were recorded, together with other information related to the child's illness. If the child still had diarrhea at the day-10 contact, a further contact would be made 5 days later.

Data analysis

Hong Kong has a dual, public, private health care system, but it has been estimated that 90% of inpatient care occurs within the public system at nominal cost to families [5]. The use of day care facilities for young children is relatively uncommon, because, even when both parents work, it is likely that the child will be cared for within the home by a maid or a member of the extended family. The cost analysis incorporated both monetary and time cost whenever applicable. Costs were estimated from 2 perspectives—social costs and private costs—as follows:

• Social costs = total direct costs + total indirect costs;

• Total direct costs = total HA costs + total family expenditure — family HA costs;

• Private cost = total indirect costs + total family expenditure.

Each of these aforementioned costs was calculated as follows:

• Total HA costs = HA inpatient cost + HA outpatient cost + HA ward follow-up cost;

• HA inpatient cost = (total number of days in HA hospital) × (cost of a general pediatric HA hospital bed/day);

• HA outpatient cost = (number of government outpatient department [GOPD] visits × official cost/GOPD visit) + (number of Maternal Child Health Clinic [MCH] visits × official cost/MCH visit) + (number of Accident & Emergency [A&E] visits × official cost/A&E visit), which includes all visits, occurring either before or after hospitalization, that were related to the same episode of gastroenteritis that required admission; and

• HA ward follow-up [WFU] cost = (number of WFU visits) × (official cost/WFU).

The HA calculates bed costs by using full cost estimates. Each hospital makes separate calculations for individual hospitals on an approximately yearly basis. For the Prince of Wales Hospital, the pediatric costs are calculated for 4 separate categories: neonatal intensive care unit bed, neonatal special care unit bed, pediatric oncology unit bed, and the remainder of beds. The latter figure was used to calculate the HA inpatient cost and was estimated by escalating 2001–2002 cost data to 2002–2003 price levels (table 1). All costs, including medications and diagnostic tests, are included in the per diem cost. The family expenditure included all payments made by the family to the government hospitals or clinics, to private doctors, for travel to and from the hospital, and for all incidental expenses related to the child's illness (e.g., extra diapers) and was calculated as total family expenditure = family HA costs + total family costs, where family HA costs = (total days in HA hospital) × (official cost of a pediatric HA bed per day to family) + (number of WFU visits) × (official cost per WFU to family) + (number of GOPD visits × official cost per GOPD visit to family]) + (number of A&E visits × official cost per A&E visit to family]). Each of those costs was calculated as follows:

• Total family costs = family outpatient cost + family “other provider” cost + family travel costs + family “other related” costs, where family outpatient costs = actual cost of all visits to outpatient private general practitioners and pediatrician + actual cost of all outpatient prescription medications; where

• family “other provider” costs = actual cost of any other provider costs, including cost of private hospital treatment (bed, consultation fees, and medications);

• family travel costs = actual cost of bus fare, taxi, or mileage to visit hospitals (HA or private) or outpatient service (HA service or private) because of the child's illness; and where

• family “other related” costs = actual costs to family for food supplements and nonprescription remedies + actual costs to family for additional day care + actual costs to family for extra use of diapers + actual costs to family for miscellaneous items (families were shown how to record these various costs in the symptom diary at the time of recruitment).

Total indirect costs were calculated as mother's time cost + father's time cost + other's time cost, where mother's time cost = (hours/day off work) × (mother's estimated hourly/daily salary); where father's time cost = (hours/day off work) × (father's estimated hourly/daily salary); and where other's time cost = (hours/day off work) × (other's estimated hourly/daily salary). Parents were asked to record in the symptom diary all time taken off from paid employment.

Statistical analysis

Cost data are presented as mean (SD), unless otherwise specified. The costs were calculated for the following etiological groups: any rotavirus positive, any rotavirus negative, any culture positive, any culture negative, only rotavirus positive, only culture positive, and confirmed rotavirus and culture negative. Average costs between these groups were compared using the bootstrap method, because the data were highly skewed. The method of calculation used involved the selection of a random sample (i.e., the bootstrap sample) from the original data set, with replacement. The process was stratified by etiological groups and was repeated 5000 times. The bootstrap mean cost differences were then calculated and compared on the basis of the 5000 bootstrap samples generated. The bootstrap BCa method was used to estimate the corresponding 95% confidence intervals of cost difference [6].