Cohort design and population

The aim of the cohort was to sample children 12–24 months of age covering all three residential sectors across the country (urban, rural and the estate consisting of plantations), with longitudinal data being collected retrospectively from the prepregnancy stage. The cohort design intended to capture growth parameters and associated factors at defined time points within the first 1000 days of life, from conception up to 2 years.

In the Sri Lankan state preventive health system, pregnant women and children are registered with the Medical Officer of Health (MOH) for the provision of clinic based and domiciliary MCH services. Around 90% of the respective target populations (89.8% of pregnant women, 90.8% of infants) obtain preventive healthcare through this service.21 Public health midwives (PHMs) are the frontline healthcare providers in the MOH areas who provide field level MCH services. The PHM maintains registers and records relevant to MCH care, such as Register of Pregnant Mothers, Birth and Immunization Register and personal health records such as pregnancy record (containing details of the mother before and during pregnancy) and the Child Health and Development Record (CHDR) (containing birth details, immunisation, health and development screening, micronutrient supplementation and growth monitoring data). During the first 2 years of life, weight is measured every month, and length at 4, 9, 12, 18 and 24 months.

Recruitment of participants for the present cohort was carried out during October and November 2023. Mother–child pairs of term babies (born at 37-week gestation or later) currently between 12 and 24 months of age were selected from the Birth and Immunization Registers. Exclusion of preterm babies was due to the fact that their growth pattern is different from that of term babies.22 23 The children with less than 4 wt recordings in their CHDR during the first 12 months and those with chronic medical conditions and structural deformities causing feeding difficulties were excluded.

Cohort sampling

A total of 1875 mother–child pairs were recruited using a two-stage stratified cluster sampling method. Stratification was carried out at the level of residential sector as defined by the Department of Census and Statistics24 with approximately 600 mother–child pairs selected from each sector, namely urban, rural and estates. In the first stage, three districts each from the urban (Colombo, Gampaha and Galle) and rural sectors (Kilinochchi, Polonnaruwa and Ampara) and two districts from the estate sector (Nuwaraeliya and Badulla) were purposely selected as the most appropriate districts to represent the respective sector. An MOH area was considered a cluster. In the second stage, clusters (MOH areas) were sampled from a list comprising all MOH divisions in the selected district. MOH areas with the highest percentage of urban population were selected for the urban stratum. Similarly, the MOH areas with the highest percentage of rural and estate populations were selected for the rural and estate strata, respectively. 45 clusters were selected, and from each cluster 40 mother–child pairs were sampled systematically according to the Birth and Immunization Registers of the PHMs. Figure 1 illustrates the MOH areas selected for the cohort.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Map of Sri Lanka showing the Medical Officer of Health areas selected for the cohort.

Measurement taken for outcome assessment

Weights and lengths recorded serially in the CHDR were extracted in a longitudinal manner. Weight at birth and monthly intervals were available with the date of weighing in the B-portion of the CHDR. Length at birth and subsequent time points were also available with measurement date in the same record. On the date of the interview, the weight and the length of the child were measured by the trained enumerators adhering to the standard protocol. Visual inspection of the weight-for-age chart, based on WHO multicountry growth reference curves25 in the local CHDR of the child, was carried out by the enumerators, and any presence of any growth faltering (ie, a downward deviation, flattening or dip in the growth curve) was recorded.

Measurements taken for exposure assessment

Detailed information on exposure status was collected using a life course approach. Pregnancy-related information, including illness before and during pregnancy, use of micronutrients, blood haemoglobin level, prepregnancy/booking visit BMI and weight gain during pregnancy, was gathered from the pregnancy record. Data on haemoglobin and weight were obtained at booking visit (first trimester), second trimester and before the delivery (third trimester). Baseline characteristics of the child, including the date of birth, sex, parity, singleton or multiple pregnancy, gestational age at birth and measurements recorded at birth such as birth weight, length and head circumference, were extracted from the CHDR. The mode of delivery, and the reason for caesarian delivery if performed, was collected through interviews.

In order to generate infant and young child feeding (IYCF) indicators, details on breastfeeding practices, complementary feeding, including the timing, frequency, decision-making on feeding and responsive feeding behaviours, were assessed by questioning the principal caregiver. Feeding of specific food items at each month was collected through a recall, which can also be used to assess dietary diversity at different ages. Longitudinal data on feeding different liquids and solid food items every month from birth up to the current age were inquired from the mother or caregiver. Additionally, micronutrient supplement use and food supplements given to the child were also gathered.

Illnesses of the child were recorded longitudinally at 3 monthly intervals during the first year and 6 monthly intervals during the second year of life. Respiratory illness, fever, diarrhoea, urine infection and other illness requiring hospitalisation were assessed. An episode of respiratory tract infection was defined as an illness with upper respiratory symptoms and signs, which lasted more than 48 hours with or without fever. An episode of diarrhoeal illness was defined as watery loose stools with more than three bouts per day lasting more than 48 hours, with frequent passage of formed or pasty stools seen in breastfed babies being excluded. Urinary tract infection and other illness requiring hospitalisation was considered if it was indicated in medical records of the child. Any change in the diet of the child during illness was gathered during the interview.

Several sociodemographic and family level factors were assessed using the questionnaire. The variables included education level and employment of both mother and father, income of the family, maternity leave, housing conditions, water and sanitation and indicators to assess the Household Wealth Index. Availability of social safety nets for the family and presence of home risk factors as per the guidelines of the Ministry of Health26 was also collected.

Tables 1 and 2 provide a description of the key variables with the time period relevant to the information and the source of information.

Data collection

Enumerators were trained to administer the questionnaire and to conduct anthropometric measurements. National and regional level clearance was obtained to conduct face-to-face interviews and to collect data from the relevant records. The selected participants were requested via telephone to come to a centre within their village or estate for data collection. If they did not respond for three separate attempts of contact, or when a participant did not have a means of communication, a home visit was made for this purpose. If all attempts failed, the next eligible child was sampled. It was explicitly mentioned that participation is voluntary and that all data would be confidential. All study material was in the local language (Sinhala/Tamil). The research team regularly monitored the data collection in the field.

Two enumerators were assigned to collect data from one mother–child pair. Photo images of the relevant sections of the pregnancy record, and CHDR including the growth charts (weight-for-age chart, length-for-age chart, weight-for-length chart) and the table for weight and length records, were obtained. The data was entered into a tailor-made online database in parallel to data collection during the field work. The current database contains password protected, deidentified data along with the images of the documents as a single PDF file under serial numbers.

Statistical analyses

Quality of data

In order to ensure the reliability and validity of collected information, investigators followed the best practices given below:

1. High standards were maintained in recruiting suitable persons with experience in field work, and adequate training was provided on interviewing techniques and the subject matter of the survey. All team members were given a 2-day classroom and a field day training programme, including training on anthropometric measurements of a child.

2. Interviews were conducted in a language conversant to the respondents. Data collection tools were available in all three languages. Interviews were conducted at a time most convenient to the respondents through an appointment system.

3. Interviewing etiquettes were strictly followed by the enumerators and survey supervisors closely monitored the process.

4. The interviewer assured the respondents of the confidentiality of information, indicated the time required for completing the interview and sought their approval to continue with the survey.

Best practices followed to ensure quality of data collection:

1. The survey consultants and data managers undertook random field visits to ensure surveys are conducted appropriately and acceptably.

2. Completed questionnaires were scrutinised first by the supervisors to ensure completeness and accuracy of the filled responses and incomplete questionnaires were returned to the field enumerator to complete the same. Entered data were randomly checked to ensure accuracy.

3. The survey teams conducted further quality checking before entering the data into the online database.

Patient and public involvement

At the stage of designing the study, three group discussions were held with the mothers of children under 2 years of age from the three residential sectors, and their views were considered in defining certain variables such as types of food given to children, responsive feeding behaviour and childhood illness according to their sociocultural context. It is expected to inform the relevant findings to participants during the second wave of assessment in this cohort after 1 year.

Characteristics of the participants

The cohort covered three main residential sectors of the Sri Lankan population namely the urban, rural and estate, in nearly equal proportions (table 3). The majority of mothers of the children were less than 35 years of age (77.9%) and housewives (78.4%). The proportion of mothers who completed General Certificate of Education (ordinary level) was 69% compared to 63.7% of fathers. A great majority of fathers (94.4%) were engaged in some form of economic activity. Approximately 30% of mothers were overweight or obese, while 15.0% were underweight at the time of registering for pregnancy.

Of the children, 49.2% were girls, 42.5% were the first-born children in their family and 34.2% were born by caesarean section. Their ages ranged from 12 months to 24 months, and the number of children in the age 21–24 months and 12–14 months were somewhat higher than other age categories. This cohort includes only term-born children, and the mean birth weight was 2917 g (SD 406), with the proportion of low birth weight (less than 2500 g) being 13.0%.

Strengths

The SLCGC serves as a comprehensive cohort study with a countrywide distribution in Sri Lanka, covering the three main residential sectors namely the urban, rural and estate sectors. It captures data on a wide spectrum of individual, household and societal determinants of child growth commencing from the prepregnancy period up to the current time point. This is one of the few studies that harnessed a wealth of longitudinally recorded data during pregnancy and early childhood, supplemented with direct interviews and anthropometric assessments. The findings will fill-up evidence gaps that are needed to inform policy and programme decisions to improve child malnutrition in the country, such as the predictors of low birth weight, and timing and determinants of early growth faltering etc. This data enables investigation of the effects of single exposures on child growth, as well as, more complex epidemiological analyses on multiple simultaneous and time-varying exposures on the growth of the child. Furthermore, potential confounders and effect modifiers are available to model the outcome through advanced statistical modelling.

Measures and precautions to assure quality of data is one major strength of the SLCGC. Careful recruitment of the study team, adequate training and close monitoring process at different stages of the study enhanced the validity data.

All children in the cohort had completed at least 12 months of age at the time of enrolment. This enabled retrospective assessment of exposures during the first 12 months of life of all children.

Limitations

The SLCGC is not fully representative of the Sri Lankan child population with respect to socioeconomic diversity of the country. The study may have not adequately covered the upper socioeconomic strata of the population since those may not register for field-based services provided through the state health sector.

Secondary data were extracted from the pregnancy records and CHDR. Certain weight measurements and higher number of length measurements were not available in the CHDR B-portions. Anthropometric measurements extracted from the records had been taken for the purpose of service delivery, therefore may not be ideally suited for research purposes. Recall bias is a problem when collecting the details regarding breastfeeding and complementary feeding practices.

The cohort did not enrol children born preterm, that is before 37-week gestation, and this limits the interpretation of findings regarding growth of preterm babies. As the cohort did not include preterm births, the proportion of low birth weight children is an under-representation of the overall expected low birth weight prevalence.

It is noteworthy to understand that this cohort was followed-up in the backdrop of an ongoing economic crisis in the country. Children recruited to the cohort were those born between October 2021 and September 2022, thus their conception, gestational period, birth and postnatal period would have been exposed to economic crisis and its consequences. Further, there is no clear starting point or endpoint of the economic crisis and all that can be stated is that the study outcomes could have been influenced to a varying degree by the economic crisis. Thus, caution should be exercised when interpreting the results.

Further analysis of available data

Data analyses will be continued according to the aims of the cohort. Analysis will be extended to identify the determinants of low birth weight including the effect of weight gain and other parameters during pregnancy, using multivariate regression models. Growth faltering at different time points will be correlated with the concurrent events including IYCF practices, childhood illness, care practices and utilisation of health services. Furthermore, longitudinal data analysis techniques will be applied to explain different growth patterns. Current growth status in terms of stunting, wasting and underweight will be analysed against a range of sociodemographic and other predictors. Correlation of growth faltering with other measures of anthropometric deficits at different points in time would also be investigated using this data.

Continuation of the cohort

The next wave of assessment is expected to be done after 12 months of recruitment. SLCGC team recommends that this cohort be extended with clearly defined waves of assessment through their childhood and adolescence.