Assessing Sleep, Diet and Physical Activity in Survey Research: Measurement Methods and Potential Issues |
|
| Chair | Ms Heidi Guyer (Survey Research Center, University of Michigan ) |
Data on sleep duration is of great interest to researchers, government, and health organizations, as sleep can impact important health and social outcomes. One common data collection method is time diaries such as the American Time Use Survey (ATUS). In the ATUS, interviewers use a set of scripted, open-ended questions to walk respondents chronologically through their activities during the prior 24-hour day, including sleep. In contrast, other surveys use stylized questions that ask people about the “average, normal, or typical” amount of sleep they get in a given timeframe. Although in theory diary and stylized questions should measure the same construct, a sleep gap has been observed where diary measures tend to exceed stylized measures of sleep. In 2014, the ATUS reported that adults 18 and older spend an average of 8.7 hours per day sleeping. In contrast, stylized questions, such as those used in the National Health Interview Survey (NHIS), reported that U.S. adults get an average of approximately 7.1 hours of sleep per day, a 1.6 hour difference. We describe multi-method research investigating possible reasons for the sleep gap. First we conducted cognitive interviews to better understand the response process for both diary and stylized sleep measures. Participants completed a time-diary and answered stylized questions about their sleep. They were debriefed about their interpretation of the questions and how they arrived at their answers. We found that how participants defined sleep, the retrieval and estimation strategies they used, and social desirability concerns seemed to affect reports of sleep duration. The findings also suggested that some features of diary and stylized sleep measures may lead to different types of measurement error. Although the cognitive interviews provided rich insights into the response process surrounding diary and stylized sleep measures, we were unable to determine whether one measure was more accurate than the other. In an attempt to obtain an objective measure of sleep, we conducted a validation study using a wearable device that tracked participants’ sleep duration over a one-week period. Participants completed two interviews about one week apart. At visit one, participants answered basic demographic questions and then were given instructions to wear the device at all times for one week. At visit two, participants completed a time-diary and answered stylized questions about their sleep. We then compared participants’ self-reported sleep duration from the time diary and stylized measures to the device-recorded sleep data. During debriefing, we asked participants about their response process across the diary and stylized measures, and probed about any discrepancies from the device-recorded sleep data. We discuss participants’ perceptions of the accuracy of the device-recorded sleep data, and how in some instances the data served as a useful memory cue for participants in recalling their sleep and wake times more precisely. We summarize the findings, implications, and logistics involved in conducting sleep validation research using wearable devices, and how these methodologies enabled us to gain a deeper understanding of measurement error.
We measure physical activity in the Netherlands, United States, and England, both by accelerometry and by self-reports. The self-reports include a global self-report on physical activity; a report on the frequency of vigorous, moderate, and mild activity, and a report on time spent on sedentary activity on a week day and a weekend day. The self-reports show only minor differences across countries and across groups within countries (such as different age groups or working versus non-working respondents). The accelerometer data on the other hand show dramatic differences; the Dutch appear to be much more physically active than Americans, with the English in between. Also, accelerometer data show a sharp decline of physical activity with age, while no such pattern is observed in the self-reports. The differences between objective measures and self-reports occur for all three types of self-reports.
Nutrition plays a pivotal role in the health status of children. Furthermore, in developed countries child and adolescent obesity runs and affects later health, educational attainment and quality of life (http://www.obesityday.worldobesity.org/about). Therefore, it represents an important social and political issue.
The eating habits of the Italian children and their parents have not however been extensively investigated. We know that obesity runs in families, with children of obese parents at greater risk of developing obesity than children of thin parents. However, we do not know the relationship among parents’ socio-cultural characteristics and their children’s nutrition style. In particular, may educational attainment and occupational status of parents affect the eating habits of whole family?
The aims of this study are to: 1) describe the eating habits of Italian children and compare them with those of their parents, focusing mainly on those families with almost one member afflicted with obesity (but also underweight) problems; 2) explore whether and how children’s dietary intake are affected by the cultural and socio-economic status of their parents; 3) focus on the relationship between children dietary intake and places where the main meal is usually consumed (home, school canteen, fast food, restaurant, etc.).
The design and methods imply a retrospective population study with data from the 2010 face-to-face multi-purpose survey by ISTAT (Italian National Statistics Institute). This survey focused on collecting data about eating habits and nutritional aspects, and was conducted through face-to-face guided interviews with a pre-defined questionnaire. The survey aimed at Italian families; each family was extracted with random criterion by the municipal registry lists, according to a statistically representative sample of the population residing in Italy. All members of the family were interviewed. Among 48,336 interviewees (21,091 families) of all ages, we will focus on participants with 4-13 years old children (4,491 interviewees) and their parents.
All participants were asked about their eating and drinking habits. The administered questionnaire included queries on how often certain foods and drinks were consumed (daily, weekly, occasionally, never). Foods included bread, pasta, rice, meat, fish, eggs, dairy products, cold cuts, savoury snacks, fruits, and vegetables. Beverages included alcoholic and non-alcoholic ones. Although the quantities were not recorded (except for drinks, where the number of glasses consumed was collected), we are able to develop dietary patterns and particularly to observe situations where certain foods are never - or rarely - eaten. Additionally, the collection of height and weight allowed us to calculate the Body Mass Index (BMI=weight [kg]/height2 [m2]), which is a widely adopted and simple proxy of nutritional status of adults and, with specific adjustment, of children.
A complex study of 300 adults aged 65 and older took place in Austin, Texas took place between fall 2016 and spring 2017. Households were first called by a telephone interviewer and screened to assess eligibility. Eligible respondents were transferred to a local field interviewer who called to schedule an in-person interview. The 90-minute in-person interview included assessments of cognitive and physical health as well as familial and social support networks. After completing the in-person interview, respondents were then asked to wear a number of mobile devices for the following five days including an Actigraph watch and a mobile phone that delivered surveys every 3 hours throughout the day. The five days following the survey included two weekdays and two weekend days. In order to set the times to launch the mobile surveys on the phone, the interviewer asked respondents the time that they usually wake up and go to sleep on weekends and weekdays (4 separate questions) at the time of scheduling the appointment. Thus, sleep and wake times were assessed in three different ways: the interviewer asked at the time of scheduling the in-person interview appointment and entered the information in the sample management system, the respondent answered questions in the mobile survey on the phone on the five days following the in-person interview, the Actical watch was worn 24 hours a day over five days and registered movement allowing for the detection of movement at the start and end of day.
This paper will assess participation rates in each of these three measures as well as variance in responses between the three methods. First, the proportion of respondents willing to provide this information at the time of scheduling the appointment will be assessed. The survey team did not have a priori estimates of the proportion of respondents that would be willing to answer information that could be considered personal or private. Additionally, older respondents may be particularly concerned about the invasiveness or purpose of such questions. Next, the proportion of respondents who responded to the mobile survey questions will be assessed. Given the complexity of navigating the mobile app and the potential lack of familiarity with mobile devices in this population, such as the lower rate of cell phone ownership and use, response rates on this component are likely lower than via self-report. Finally, compliance with wearing the actigraph watch over the five day period will be assessed. Variation in the wake and sleep times reported by the respondent via the two methods and recorded by the actigraph will be assessed at the individual level, separately for weekdays and weekends. This paper will provide insights into new methods of assessing sleep and wake times among survey participants as well measurement variation based on the method of assessment.
Background: The known health benefits of diets of higher quality, characterised by high consumption of fruit and vegetables, and low consumption of energy-dense micronutrient-poor foods, underpin current dietary recommendations. Whilst researching the determinants of diet quality may be key to developing successful strategies to promote health in the future, this requires short, simple assessment methods for use in population studies, appropriate for different age-groups and accessible to all individuals. A number of a priori-defined diet quality indexes have been developed. However, many use scoring systems based on estimates of food and nutrient intake, requiring detailed assessment and analysis of dietary data, that are both burdensome for participants and expensive. We have therefore developed a set of short food-based questionnaires to assess diet quality in our studies of children, young women and older adults. We describe our approach and methods, and comment on their use.
Methods: The diets of children (aged 3 years) and young women in the Southampton Women’s Survey, and older adults in the Hertfordshire Cohort Study, were assessed using administered, age-specific, validated food frequency questionnaires (FFQ). We used principal component analysis (PCA) of the FFQ data to identify dietary patterns. In each group, the first component described a ‘prudent’ dietary pattern (PDP), characterised by high consumption of fruit and vegetables, oily fish and wholemeal cereals. A PDP score was calculated for every participant, based on the PCA coefficients and the participant’s reported frequencies of food consumption; the score indicated compliance with the pattern, and was interpreted as a marker of their diet quality. The foods that characterise the PDP (with largest absolute value of the coefficients) contribute most to pattern scores. We therefore developed short FFQs, including around 20 of these indicator foods, to provide an index of diet quality in each age group. Here, we describe PDP scores from the short and long FFQs, and for the adult groups, comparison with nutrient biomarkers.
Findings: In each group, PDP scores calculated from the short list of foods (n~20) were highly correlated with scores from the full FFQ (r>0.9). In follow-up studies using the short FFQ, PDP scores showed good agreement with scores based on full dietary assessments at baseline (children: r=0.68 assessed after 12-20 weeks; older men: r = 0.70, older women: r = 0.66 assessed after 10 years (all comparisons, P<0.001)). Biomarker concentrations in the young women (red cell folate) and the older adults (serum vitamin C) were positively associated with PDP scores from both short and long FFQs.
Conclusions: Short FFQs can be used to describe diet quality in children, young and older adults. They are simple to administer, do not require nutrient analysis, and therefore have potential to be of value to non-specialist researchers. Simple assessment tools, that enable routine collection of relevant dietary data, could make an important contribution to understanding the role of diet quality as an influence on the health across the population.