4.2.3.1 Anthropometric measures in the SLLCC
Changes in adiposity over time can be based on the change in BMI, or the proportional (percentage) change in BMI, or the change in BMI Z-score or centile (168). In 2006, the WHO launched new growth standards for children irrespective of ethnicity, socio-economic status and feeding mode (169). By April 2011, at least 125 countries, representing 75% of the world’s under-5 population, had adopted the standards and were at varying stages of their
implementation (170). It is important to note that WHO standards depict how children should grow, on average, in all countries, when properly fed and cared for, rather than merely describing how they grew at a particular time and place such as use of as national/local growth references.
BMI for sex- and age-specific percentiles had been used in Singapore for ease of
communicating concerns of severe underweight, overweight and severe overweight (See Chapter One for percentile classifications). However, when required to assess longitudinal growth in children, the use of percentiles is less than ideal due to inherent limitations such as same increments at different percentile levels could correspond to different changes in both Z-scores and absolute measures and it does not allow for quantifying the change in percentile values near the extremes of the reference distribution. It has been suggested that percentiles should not be used to assess change in status over time, while change in BMI Z-scores is a better measure for such research (171).
There are several advantages of using BMI z-scores over percentiles as the former are calculated based on the distribution of the reference population (both the mean and the standard deviation), thus, they reflect the reference distribution. Secondly, as standardised measures,
Z-scores are comparable across age, sex and measure. Thirdly, a group of z-Z-scores can be subject to summary statistics such as mean and standard deviation (SD) and can be studied as a continuous variable thereby allowing quantification of the growth status of children outside of the normal percentile ranges (171).
Therefore, for the purposes of the Singapore Longitudinal and Life Course Cohort (SLLCC), sex-specific BMI-for-age z-scores were standardised for all students using the new Stata command “zanthro” (172). Briefly, this extension converted child anthropometric data to Z-scores using the LMS method and the reference data available from the 2000 CDC Growth Reference, the British 1990 Growth Reference, the WHO Child Growth Standards, the WHO Reference 2007, the UK-WHO Preterm Growth Reference, and the UK-WHO Term Growth Reference. In the SLLCC, standardised BMI-for-age Z-scores were derived from WHO child growth standards so that results can be internationally more comparable.
In addition, students were categorised as normal weight, overweight or obese using BMI categories that correspond to equivalent adult BMI cut-off points endorsed by the World Health Organization: BMI<25 kg/m2 for normal weight, BMI 25–29.99 kg/m2 for overweight, and BMI
>30 kg/m2 for obesity (Table 33).
Table 33 WHO classification of body mass index (BMI) Value Grade/Label BMI range at 18 years -3 Grade 3 thinness <16
-2 Grade 2 thinness 16 to <17 -1 Grade 1 thinness 17 to <18.5 0 Normal weight 18.5 to <25
1 Overweight 25 to <30
2 Obese 30+
4.2.3.2 Race, gender and social economic position
Gender and race had been reported to have significant early life influences for class membership in a study that identified developmental trajectories of overweight in children and adolescents (173). A cross-sectional study investigating BMI of Chinese, Malays and Indians in Singapore has consistently shown that for males there are few ethnic differences, however, for females, Malays and Indians are significantly more obese than Chinese (156), consistent with the National Health Survey 2010 findings. Findings from the Singapore Cardiovascular Cohort Study (156) show that Indians had a three fold increased relative risks of incident CHD (RR=3.1, CI: 2.0–4.8) compared with Chinese and Malays, after adjusting for age, ethnic group and other risk factors (LDL-Cholesterol, HDL-Cholesterol, Triglycerides, BMI, smoking, diabetes, hypertension and alcohol use).
It is unlikely that BMI trajectories will not be influenced by social disparities due to differential childhood social and economic status/positions (SES/SEP), commonly defined by highest educational level of the child’s parent (76) or classification of parent’s occupation (77) or income levels or type of housing. In the Singapore Malay Eye Study, lower SES, defined by categories of education and income were associated with higher prevalence of
overweight/obesity in Malay women. In contrast, higher SES was associated with higher
prevalence of overweight/obesity in Malay men (78). A prospective follow-up study of the 1998 National Health Survey on the socio-demographic determinants of changes in body weight and waist circumference in Singapore adopted highest education level, housing type and employment status as proxy measures for SES (79). In a life-course study on SES and obesity in older
Singaporean Chinese men and women, childhood SES was based on the participant’s
(self-reported) family financial status while growing up; adulthood SES was based on highest education attained; and older adulthood SES was based on type of housing one resides in (i.e., private or public), and within public housing, the number of rooms (ranging from 1 to 5) in the house (166). Nevertheless, there is not yet a standard index of SES or SEP in Singapore.
The SLLCC (Master Dataset A) included data on gender and race groups for all its subjects, namely Chinese, Malays, Indians, Eurasian, Indonesians and Others. Housing type data was recorded only from TAF 2005 to 2011 datasets only and thus was not included in the final SLLCC dataset but available separately. Postal codes of subjects’ residence were also available for additional linkage but had not yet been geo-referenced to determine housing type (Table 27).
4.2.3.3 School-level factors
Academic performance in schools has the potential to alter the trajectory of adult weight gain and subsequent midlife outcomes. Previous studies have reported that the odds of being persistently overweight were significantly reduced among those with a higher average grade in high school (174). The evidence also suggests that individual resources and other collective social capital operating in the school setting can attenuate the risks for obesity and overweight even among those from a lower social economic status (313).
At the time of constructing the SLLCC, there was no obvious access to individual level academic performance in schools in Singapore. One possible indirect school level factor predictor that could be the ranking of schools in Singapore based on overall academic performance.
4.2.3.4 Walkability Index
Understanding why some people exercise more than others is a complex social and psychological challenge. Having knowledge about the general prevalence of physical activity in the population is essential in order to support interventions aimed at increasing the level of physical activity at the community or neighbourhood level. It is also necessary to help understand of the relationships between social determinants of health, physical activity and where people live.
To this end, a validated Walkability Index has been constructed in Singapore based on my Master’s thesis research at the HPB in 2010. In constructing the Walkability Index for adults, 5 distinct domains driving physical activity were postulated: public transport, sports facilities, food, park connectors and community centres. The features of each domain were selected based on the ease of availability of data and its relevance to influence choice of physical activity. Two measures of social economic status (Populations living in public housing and resident working persons aged 15 years and above requiring public transportation to work) were also postulated to be predictors for walkability and included in the Index. A high score for the Walkability Index predicted the ‘walkability’ of a community or the extent to which characteristics of the built environment and land use may or may not be conducive to residents in the area walking for either leisure, exercise or recreation, to access services, or to travel to work. This could be used in future research as a proxy for the propensity of the school environment to encourage students to walk and serve as a potential predictor for subsequent studies.
In the SLLCC, each student record was linked to their school code at the time of health examination. In addition, each school in the dataset could be assigned a score of the Walkability Index based on the postal code of the location of the school residing within one of the 55
Development Guiding Plans (geographical zones) in Singapore. Classifications of school types (public/private and other categorical measures) and the school neighbourhood’s walkability index could be determined separately as this had not yet been done in the current SLLCC datasets.
4.2.4 Data preparation
Annual records of SHS and TAF were provided to an external third party data vendor to be de-identified by replacing the students’ NRIC numbers with a unique record identifier string number (see Chapter Three). Figure 3 is an illustration of the age-period-cohort schematic for SLLCC data.
Figure 3 Age-period-cohort data structure of the Singapore Longitudinal and life Course Cohort (SLLCC)