Setting and study population
Supported by Chinese Ministry of Health and UNICEF, a rural primary health care survey in 45 counties of west China’s 10 provinces----Xinjiang, Inner Mongolia, Qinghai, Gansu, Ningxia, Sichuan, Chongqing, Guizhou, Jiangxi and Guangxi was conducted from June to August 2005. These 45 counties were pre-determined but the much smaller sampling units as townships and villages were sampled through a multi-stage probability-proportion-to-size sampling (PPS) method. Five townships out of each county and four villages out of each sampled township were randomly selected. In the household-sampled process, a completely random sampling method was adopted to extract sixteen households from each village. If a village had more than 16 households, 16 households were selected randomly; if a village had fewer than 16 households, all the households were determined and the rest were selected out of the neighboring villages. In every sampled household, only one child under 36 months old was selected randomly and the caretaker of the selected child was interviewed.
Data collection
All data in the study were collected by means of pre-coded structured family and village clinic interview questionnaires. First, we had a face-to-face interview with all the caretakers involved in the survey about their families, their children and themselves after they had signed the informed consent form. All socio-demographic information in the study was included in the family questionnaire. If their children had suffered diarrheal episode in the previous two weeks (Diarrhea was defined as the passage of 3 or more loose or watery stools in the proceding 24 hours), we further interviewed them about the recent diarrheal episode in detail, including recognition of 7 dangerous symptoms of diarrhea (frequent watery stools in the proceeding one or two hours, blood in stools, repeated vomiting, high fever, extreme thirst, no desire to drink and refusal to eat), their care-seeking behaviors and ORS use in different care locations. Meanwhile, we also asked them whether they had received educational materials including the basic prevention or care knowledge of common childhood diseases which were specific in the rural primary health care program. Then, we collected the data of village clinics from village health personnel through a village clinic questionnaire about the basic information of village doctors, their medication and retail pharmacy distribution in the village. To make the collected data available, the chief of the investigation team must review every sheet of questionnaire and verify its appropriateness carefully before all questionnaires were accepted. The study was approved by the Ethics Committee of College of Medicine of Xi’an Jiaotong University.
Variables
The outcome variables of interest included the ORS use in home-based care (HBC) and village-level care (VLC). If a child had received care at home, or in a village clinic, a township hospital or a county-level-or-above hospital in a last diarrheal episode during the previous two weeks, he/she was identified as receiving HBC, VLC, township-level care (TLC) or county-level-or-above care (CLC) respectively. HBC indicated that the caretakers gave some special care or treatment, such as increasing the frequency of feeding, giving ORS, increasing fluid intake or medical care, and so on. If a sick child was not given the above-mentioned special care at home or some care at health facilities, he/she was regarded as receiving no care (NC). In rural China, all caretakers did not have medical background. Village doctors, many of whom did not receive formal medical education, were engaged both in healthcare and farming. Therefore, we identified HBC or VLC as a low-level care. If a child with diarrhea had received ORS packets in a recent diarrheal episode in the previous two weeks and all ORS packets were administered at the corresponding care location, he/she was identified as using ORS at the care location. The caretakers’ capacity of judging the danger signs of childhood diarrhea was assessed through the number of the dangerous symptoms of diarrhea they could recognize out of the 7 ones. The Demographic and Health Survey (DHS) wealth index generated with the five variables (type of vehicle, water supply, income resource, texture of pot and type of television) was used to assess the socioeconomic status of the families [14, 15]. According to the tertiles of the DHS wealth index, the economic statuses of the families fell into three: poor, medium and rich [14, 15].
Data analysis
The data from the qualified questionnaires was entered in Epidata 3.1 by double entry. SPSS version 17 (SPSS Inc, Chicago, IL, USA) was employed to make the statistical analysis. The chi-square test was adopted to compare the proportions. Generalized estimated equation (GEE) logistic regression models were used to predict the determinants of ORS use in HBC and VLC respectively while controlling for the possible correlation of ORS use in the same village. The level of the significance of analysis was set at 0.05.