Study design, setting and participants

We used data from the Mediators of Atherosclerosis in South Asians Living in America (MASALA) study, which is a community based cohort of South Asians residing in the U.S. The detailed study information can be found elsewhere [17]. Briefly, 906 middle- to older- aged South Asians with no existing CVD were recruited at baseline using surname-based recruitment methods between 2010 and 2013 in the San Francisco Bay Area and the greater Chicago area. Baseline data were collected through interviews and clinical examinations. All surviving participants were re-contacted between 2014 and 2018 for an in-person ancillary study on social networks, and 771 participants completed the social networks visit. Interviewers also administered a Food Frequency Questionnaire (FFQ) at the social network visit. We excluded individuals who had incomplete information on dietary intakes (n = 1) and had implausible energy intake (< 800 or > 4200 kcal/day for men, < 500 or > 3500 kcal/day for women, n = 14), leaving a total of 756 participants who were included in these analyses.

Measurement of personal social network characteristics

Data were collected from MASALA participants (herein called egos) using a social network module which included standard network elicitation as in previous work [18] and name generator instruments to collect egocentric network data. The surveys were administered in English, Hindi, or Urdu by trained interviewers. These network modules included a name generator that asked the names of people in the participants’ social network (herein called alters). Participants were asked to list up to 10 alters with whom they discuss “important matters” using a name generator which was previous used by the General Social Survey [19] and the National Social Life, Health, and Aging Project’s social networks module [20]. Network size was determined by the number of alters named by each participant in response to the name generator question. This name generator was designed to elicit network members who are close confidants (people with whom important information can be shared) and who may be socially influential. Name interpreter items followed and elicited information on the characteristics of alters, the relationship between egos and alters, and the egos’ perceptions of alters’ behaviors. For the first five alters, additional name interpreter questions obtained information about the alters’ socio-demographics (e.g. race/ethnicity and how many years living in the U.S.), type of relationship (e.g. spouse, child, friend), duration and strength of relationship, and frequency of communication. Name interpreter questions were limited to the first five alters to reduce ego burden which is a standard approach used in social network surveys [19, 20].

Egos were also asked about the dietary behaviors of their first five alters, including how often (‘never, a few times a year, monthly, weekly, daily’), over the 12 months, the alter consumed South Asian foods, fruits, vegetables, brown rice/quinoa, processed meat, sugar sweetened beverages, diet drinks, South Asian sweets, fried foods, fast foods, processed or packaged food. Egos also had the option of selecting “Don’t know”, and selected this response between 0.36% (12 respondents, for the question on consumption of South Asian foods) and 4.9% (161 respondents, for the question on consumption of brown rice/quinoa) of the time.

Organizational affiliation

South Asians in the U.S. have developed organizational structures that may also exert a strong influence on health behaviors. We developed a roster of South Asian organizations (religious, social, cultural, community-based organizations) in Chicago and the San Francisco Bay area using key informant input and an iterative approach (described previously [21]). Egos were asked to look through the pre-defined list and select organization(s) that they visited within the prior 12 months and the frequency with which they visited the organization in the prior 12 months. Respondents could choose multiple organizations if applicable, and they were also given the option to add an organization if it was not on the roster. There was no limit on the number of organizations a person could affiliate with.

Measurement of dietary intake and derivation of dietary patterns

The ego’s dietary intake at the social network visit was assessed with the Study of Health Assessment and Risk in Ethnic groups (SHARE) FFQ, which was developed for dietary assessment of South Asians in North America, has been previously validated among South Asians in Canada [22] and was used at baseline to assess dietary intakes of the cohort. A total of 161 food items was included in the SHARE FFQ with 61 food items specific for South Asian diet, and food items were additionally classified into 29 food groups, including added fat, alcohol, butter/pure ghee, coffee, eggs, fish, fried snacks, fruits, fruit juice, high-fat dairy products, high-sugar drinks, legumes/daal/tofu, low-fat dairy products, low-sugar drink, margarine, nuts and peanut butter, pasta, pizza, potatoes, poultry, red meat, refined grains, rice, snacks, Indian sweets and non-Indian desserts, tea, vegetable oil, vegetables, and whole grains. The SHARE FFQ assessed the dietary intakes including frequency and serving size of participants in the past year. Egos could choose from three options for serving size, which were small, medium and large, and small and large serving sizes were converted to medium serving size scale by multiplying by 0.5 and 1.5, respectively, in the analysis. Energy and nutrient intakes were derived using the ESHA Food Processor nutrient analysis software version 6·11 (1996).

As has been done previously in the cohort [23], foods on the FFQ were categorized into 29 food groups based on their similarity, nutrient composition, and culinary use in South Asian diets. Principal component analyses (PCA) with varimax rotation was then performed on the correlation matrix of the 29 food groups to derive dietary patterns. A 3-factor solution which explained 23% of the total variance of food intake was selected after evaluation of factor solutions with eigenvalues > 1 and examination of scree plots. Participants were assigned a factor score for each dietary pattern based on the linear combination of his or her FFQ data with the factor loadings in the 3 prevalent patterns. Food groups with factor loadings > 0.2 were considered a significant contributor to the dietary pattern, and the three dietary patterns were named based on food groups with highest factor loadings.

Covariates

Covariates were selected based on previous literature and univariate analysis and included egos’ demographic characteristics such as age, sex, marital status, smoking status, annual income (≥$75,000 vs. <$75,000), education (≥ Bachelor’s degree vs. < Bachelor’s degree) all assessed by structured interview. Self-rated health was analyzed on a continuous scale ranging from 1 to 10, where 1 indicates poor health and 10 indicates excellent health. Participants were classified as having strong, moderate or weak traditional South Asian cultural beliefs according to a traditional cultural beliefs scale previously developed and validated in the cohort [24]. Daily energy intake was estimated from FFQ and analyzed on a continuous scale. BMI was calculated from measured weight in kilograms divided by square of height in meters. Information on social network size was derived using data from social network interview, and because detailed information was only obtained from first 5 alters, we adjusted the social network size with the maximum of 5. For individuals born outside the U.S., length of residency in the U.S. was asked and the percentage of time that living in the U.S. was calculated. Intentional exercise was assessed by the Typical Week’s Physical Activity Questionnaire, including activities such as walking for exercise, dance, conditional activities and sports, and total metabolic equivalent minutes per week were used for analysis [25]. While age, self-rated health, daily energy intakes and social network size were from the social network visit; sex, marital status, smoking status, BMI, income, education, study site, and traditional South Asian cultural beliefs were from the baseline visit.

Statistical analysis

Socio-demographic characteristics were reported with means and standard deviation (SD) or percentage. The most basic measure of personal network structure was degree: the number of network members reported. The remaining network variables were calculated using information on the first five alters listed in response to the name generator question. The proportion of the social network members with each dietary behavior was calculated and used for analysis. We also calculated the average number of organizational affiliations for each participant.

We used residual diagnostics to assess the linearity of the relationship between dietary pattern scores and perceived dietary intake of network members. Partial correlations were used to assess the correlation between dietary pattern factor score and social network characteristics adjusting for ego age, gender, study site, education, income, marital status, traditional cultural beliefs, caloric intake, social network size, self-rated health, intentional exercise and percent life in the U.S.. When calculating correlations, network characteristics and dietary pattern factor scores were analyzed in their original (continuous) scale in order to better preserve relationships between dietary patterns and network characteristics and also to avoid the loss of statistical power that would be the result of collapsing continuous data into discrete categories.

All statistical tests were performed using two-sided tests with α = 0.05 and were conducted using SAS, version 9.4 (SAS Institute; Cary, NC).

Social network characteristics are correlated with dietary patterns among middle aged and older South Asians living in the United States (U.S.) | BMC Nutrition

Post navigation


Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.

%d bloggers like this: