Associations between lifestyle behaviours and optimal wellbeing in a diverse sample of New Zealand adults

Data for this cross-sectional study were obtained from the Sovereign Wellbeing Index
(Round 1); a survey on the health and wellbeing of a large, demographically diverse
sample of New Zealand adults 16]. A web-based survey design was employed to collect data during September and October,
2012. Ethical approval to conduct the study was granted by the Auckland University
of Technology Ethics Committee on 23 August, 2012 (AUTEC: 12/201).

The web-based survey design was chosen as it offered a number of advantages over traditional
data collection modes (i.e. door-to-door or computer assisted telephone interviews).
These advantages include the relative cost-effectiveness of the approach, the ability
to overcome geographical constraints, and the minimisation of errors associated with
data entry 17]. Recent reports indicate the proportion of New Zealand households with access to
the internet (80 %) and landline telephones (85 %) is similar 18], 19].

Participants

A commercial market research company (TNS Global, New Zealand office) was contracted
to administer the web-based survey. Participants were recruited from the SmileCity
database; the largest commercially available database in New Zealand. The database
comprises 247,675 active members recruited through both offline (51 %) and online
(49 %) sources 20].

The target sample size for the current study was 10,000 participants. The sample size
was determined partly by financial constraints, and partly to obtain a reasonable
precision of estimates. Eligible individuals included SmileCity database members aged
over 18 years who had not participated in a survey within the last 7-days. There were
no further exclusion criteria.

Email invites—with a link to the survey—were sent to 38,439 individuals randomly selected
from the 229,032 eligible individuals. The survey was open to potential participants
for 7-days. No follow-up invites were sent to individuals who did not complete the
survey within the specified timeframe. All participants provided informed consent
prior to entering the survey.

Variables

The web-based survey included 134 questions on wellbeing, health and lifestyle, and
socio-demographics. To enable international and national comparisons, the wellbeing
component primarily comprised questions drawn from the European Social Survey (Round
6) 21] whilst the health and lifestyle component comprised questions primarily from the
New Zealand Health Survey (2006) 22]. Measures specific to the current study only are discussed in detail below.

Optimal wellbeing

Optimal wellbeing was treated as a binary variable. The ten items (refer to Table 1) to measure optimal wellbeing were drawn from the European Social Survey (Round 6)
21]. A modified version of Huppert and So’s scale, reflecting changes made to two items
between Rounds 3 and 6 of the European Social Survey, was used to calculate optimal
wellbeing 5], 23], 24]. The two items which differed from the original scale were ‘I love learning new things’ and ‘There are people in my life who really care about me’. These items were replaced with ‘To what extent do you learn new things in your life’ and ‘To what extent do you receive help and support from people you are close to when you
need it’, respectively 23], 24]. Hone et al. recently demonstrated moderate to strong agreement between the modified
version of Huppert and So’s measure and other measures of optimal wellbeing 24].

Table 1. Constructs, features, items and thresholds used to calculate optimal wellbeing

The ten items used to measure optimal wellbeing combined both hedonic (feelings) and
eudaimonic (functioning) aspects of wellbeing 5]. The items were rated on 4-point to 11-point Likert scales. All items were phrased
in a positive direction except for the item measuring resilience, which was reverse
coded. Optimal wellbeing was determined as meeting the thresholds for positive emotion
(happiness???8); and four out of five features of positive characteristics (vitality???3, optimism???4, resilience???4, self-esteem???4, emotional stability???2); and three out of four features of positive functioning (engagement???5, meaning???4, competence???4, positive relationships???4) 5], 24]. Table 1 provides a summary of the constructs, features, items, and thresholds used to calculate
optimal wellbeing.

Socio-demographic variables

Self-reported socio-demographic variables including gender, date of birth, ethnicity,
and household income were collected as part of the web-based survey. In accordance
with Statistics New Zealand’s Statistical Standard for Ethnicity, respondents were
provided with the option of selecting multiple ethnic response categories 25]. Responses were coded into three independent categories (European/Other, Maori/Pacific,
and Asian) using Statistics New Zealand Level 1 prioritised ethnic classifications
25]. Date of birth was used to calculate age with the survey start date as the reference.
Continuous age was recoded into 10-yearly groupings according to Statistics New Zealand’s
Statistical Standard for Age 26]. Finally, household income was stratified into tertiles to reflect low (? $40,000),
moderate ($40,000-$90,000), and high (? $90,001) incomes.

Lifestyle behaviours

Ten lifestyle behaviours were included in the analysis including breakfast consumption,
sugary drink consumption, fruit intake, vegetable intake, smoking, alcohol consumption,
exercise, sedentary behaviour, sleep quality, and body mass index (BMI).

Questions to measure breakfast consumption, sugary drink consumption, fruit intake,
vegetable intake, smoking, and alcohol consumption were drawn from the New Zealand
Health Survey (2006), an annual door-to-door survey conducted by the Ministry of Health
22]. Respondents were asked to indicate how many days during the past week they had breakfast
(never, 1-2 days, 3-4 days, 5-6 days, 7 days); how often during the past week they drank sugary beverages (I don’t drink sugary drinks, less than once, 1-2 times, 3-4 times, 5-6 times, ?7 times); on average how many servings of fruit they had over the past week (I don’t eat fruit, 1 serving/day, 1 serving/day; 2 servings/day; 3 servings/day,
? 4 servings/day); and on average how many servings of vegetables they had over the past week (I don’t eat vegetables, 1 serving/day, 1 serving/day; 2 servings/day; 3 servings/day,
? 4 servings/day) 22]. For smoking, respondents were asked if they smoke cigarettes regularly (yes, no) 22]. Alcohol consumption was assessed by asking respondents to indicate how often they
have a drink containing alcohol (I don’t drink alcohol, monthly or less, up to four times/month, up to three times/week,
? 4 times/week)22].

Exercise was measured using a single item exercise frequency question which asked
participants to report how often during the past week they exercised (I didn’t exercise, 1-2 times, 3-4 times, 5-6 times, ? 7 times) 14], 27]. Sedentary behaviour was measured using a single item sitting question 28]. Response options were adapted from their original format (never, seldom, sometimes, often, always) 28] to reflect the response scales used throughout the web-based survey (none or almost none of the time, a little of the time, some of the time, most of the
time, all or almost all of the time).

Sleep quality was assessed using a question drawn from the European Social Survey
(Round 6) Survey 29]. The question originates from the Center for Epidemiologic Studies Depression Scale
30] and has been used to measure restless sleep elsewhere 31], 32]. Respondents were asked to indicate how much of the time during the past week their
sleep was restless (none or almost none of the time, some of the time, most of the time, all or almost
all of the time).

Body mass index was derived using self-reported height and weight measures and was
calculated as weight
_kg
/(height
_m²
). World Health Organization thresholds were used to categorise BMI as: underweight
(?18.4), normal weight (18.5–24.9), overweight (25.0–29.9), and obese (? 30.0) 33].

With the exception of BMI, all lifestyle variables are reported as per their original
response scales.

Data analysis

Optimal wellbeing was treated as the dependent variable. Participants’ data were,
therefore, only included in the final analyses if a response was provided for each
of the ten items used to calculate optimal wellbeing. Binary logistic regression analysis
was used to determine associations between both demographic factors and lifestyle
behaviours and optimal wellbeing (IBM SPSS Statistics version 19 for Windows). Crude,
partially adjusted (adjusted for age, gender, ethnicity, and household income), and
fully adjusted (adjusted for all socio-demographic and lifestyle variables concurrently)
odds ratios were calculated. Bootstrapped 95 % confidence intervals (CI) were calculated
using 1000 samples. The alpha was set at 0.05 to determine statistical significance.
Missing data for lifestyle behaviours and socio-demographic variables were excluded
pairwise.