Sampling
This study used data from Malaysia’s National Health and Morbidity Survey (NHMS) 2016. The NHMS is a series of nationwide surveys conducted on a regular basis to support the Ministry of Health in monitoring and evaluating the health of the Nation. This survey was designed to capture a nationally representative sample of residents in non-Institutional living quarters in both urban and rural localities in all 13 states and 2 Federal Territories in Malaysia. The sample of children under five was selected based on a sampling frame of live births provided by the National Registration Department (NRD). Details of the Survey methodology was also mention in another published article [10].
As NHMS covered broad topics under Maternal and Child Health which includes immunization as the main topic and diarrhoea as one of sub topic. Thus, the sample size was calculated based on an 50% estimated prevalence of immunization and design effect of 1.5. The final sample size was inflated to 17,330, which were then allocated to the states in equal proportions to compensate for a 50% expected non-response rate [11].
The Questionnaire for this study was based on the WHO Multiple Indicator Cluster Survey (MICS), adapted with permission to suit Malaysian requirements [13]. Permission for its use was authorised by UNICEF. This questionnaire was administered in two languages and anonymous. Data collection started on 17 February 2016 in Sabah and Sarawak, and on 25 February 2016 in Peninsular Malaysia and was completed on 26 May 2016.
Bilingual questionnaire; English and Bahasa Malaysia languages were used for this survey. Questionnaire in Bahasa Malaysia version was translated from English by content experts with English proficiency then back-translated into English by English speaking experts with Bahasa Malaysia proficiency. A pilot test was done using the Bahasa Malaysia version involving 30 individuals from Malaysia’s major ethnicities; Malays, Chinese and Indians, and from both low and highly educated individuals. Cognitive testing which looks into understanding of the content and terms used was found to be satisfactory. The questionnaire was conducted by face-to-face interview using mobile devices, with the parent or caregiver of children aged 0–59 months who responded to this module.
Variable definitions
In this study, diarrhoea was defined by a question “In the last 2 weeks, has …… (child’s name) had diarrhoea” on the passage of three or more loose/watery stools per day with/without blood in the stools in the past 2 weeks prior to data collection, as perceived by the mother/caregiver. The response options were either ‘YES’ or ‘NO’. Water sources were also inquired into by the question “What is the main source of drinking water for this household?”, and for analysis it was further categorised into - either Treated (Boil, add bleach/chlorine, use water filter) or Untreated water (others or nothing has been done). Method of garbage disposal was categorised into groups of sanitary or unsanitary waste disposal. Sanitary waste disposal included those collected by a local authority/management regularly, collected by local authority/management irregularly and buried outside the house. Unsanitary waste disposal consisted of open burning, throw into the drain, river or sea or anywhere. For latrines, the options were regrouped into either sanitary latrine (child used toilet, put/rinsed into the toilet) or unsanitary latrines (thrown into drain, ditch, river or sea, thrown in garbage, left open and others).
Locality was classified into Urban and Rural based on Department of Statistics Malaysia definitions: Urban area was classified as an area with a population above 10,000 and at least 60% of its population aged above 15 years involved in non-agricultural activities. Rural area was classified for areas that did not meet those criteria.
In this study, five major ethnic groups were identified. Three main ethnicities include Malay, Chinese and Indian descent with another two comprising ‘Other Bumiputera’ (local ethnic groups of the East Malaysian states of Sabah and Sarawak) and ‘Others’ (foreigners, immigrants, both legal and illegal, residing in Malaysia). For children, their ethnicity was based on paternal ethnicity. Citizenship was categorised as either Malaysian citizens or non-Malaysian.
The Malaysian education system was used to further categorise both parents’ education level. No formal education means the respondent had not attended any formal schooling while primary education applied to those who completed up to six years of primary school. Secondary education applied to those who finished 11 years of formal schooling and, finally, respondents with a minimum qualification of a diploma and/or higher were coded as having tertiary education.
Occupation of parents was clustered into either working in public sector, private sector, self-employed or housewife/unemployed. The pooled income of family members was further divided into quantiles to categorise household income with the first quintile representing the lowest 20% and the fifth quintile representing the highest 20% of income.
The survey or final weight was calculated based on design weight and non-response weight and plan for analysis was created. To ensure that the data can be inferred to the Malaysian population, particularly for ethnicity, post stratification adjustment was made.
To obtain the association of diarrhoea among children with factors such as sociodemographic profile, treated water, waste and latrine disposal, we used multiple logistic regression. The adjusted odd ratio (aOR) of having diarrhoea, with 95% confidence intervals (CIs), and P value < 0.05 were used to describe associations.
First, we did univariate analysis to determine the associations between diarrhoea and other associated factors using chi square and binary logistic regression. Those with a P value less than 0.05 were included in the final multivariable model. Taking into consideration the complexity of the sampling design, analysis was done using complex sample module in the IBM Statistical Package of Social Sciences (SPSS) for Windows version 23 (IBM Corp., Armonk, NY, USA),.