Assessment Using Self-Report

Until the development of movement sensors, such as pedometers and accelerometers, the assessment method of choice for physical activity has been self-report. Consequently, there are a large number of instruments, with varying degrees of formatting and development, with many aimed at assessment in young people. However, this approach to measurement is fraught with difficulties and consequently there are many poorly developed instruments alongside those that have a history of more robust design and development.

There are many good sources of discussion concerning issues in self-reported assessment of physical activity, in both youth and adults (e.g., [6–10]). Moreover, there are many methods that can be used, including questionnaires/surveys and diaries. Survey methods can require self or proxy completion, the latter, for example, by parents for young children.

In the analysis of self report surveys that follows, we focused on instruments that might be suitable for assessment in large-scale population surveillance, usually for the purpose of estimating the prevalence of physical activity. Prevalence estimates refer to the proportion of the population estimated within physical activity time-based categories (e.g., percentage meeting national guidelines, such as 60 minutes of MVPA each day of the week). This contrasts with possible outputs such as mean estimates of time spent in specific types of physical activity behaviours, or total physical activity behaviour, or 'dose'.

Such assessments are usually part of a wider public health surveillance system. The USA Centres for Disease Control [11] stated that "Public health surveillance systems should be evaluated periodically, and the evaluation should include recommendations for improving quality, efficiency, and usefulness." This will include an evaluation of the type of assessment (e.g., objective or self-report), and continued assessment of the nature of the instrument being used. To this end, Project ALPHA was established, with one aim being to provide guidance on the measurement of physical activity in young people using self-report instruments and assess their suitability for population surveillance monitoring, particularly in the European context (see http://sites.google.com/site/alphaprojectphysicalactivity/). That is, instruments relevant for populations of these types of countries.

In evaluating, as well as developing, self-report instruments for surveillance of physical activity in youth, consideration needs to be given to several key issues [12], including:

This paper addresses the assessment of physical activity for children and adolescents for population level surveillance using self-report instruments. With technological improvements and reductions in costs, population surveillance may, in future, routinely use more objective instruments, such as accelerometers. However, even if this is the case, concomitant assessment of some aspects of physical activity (e.g., type) will require self-report assessment, alongside objective methods.

The purpose of this paper, therefore, is to review existing self-report instruments purporting to assess physical activity in young people. Instruments will be appraised and a short list of measures that may be suitable for population surveillance, with the ability to provide suitable prevalence estimates, will be considered with the purpose of making recommendations on usage, particularly in a European context.

Search Strategy

Literature searches, using tailored search terms appropriate to each database, were conducted using Web of Science, Medline, PubMed, PsycINFO, SportDISCUS, and SIGLE. Typical search terms included physical activity, exercise, sport, children, adolescent, boys, girls, infants, measurement, surveillance, and survey. Search terms followed the same order of (1) behaviour terms, (2) population terms, and (3) method terms. The census date for searches was June 2008. We also searched the reference lists of key texts and reviews published since 2000 (e.g., [13–19]), plus the special issues on physical activity measurement in Medicine and Science in Sports and Exercise (1997, 29 (6) Supplement; 2000, 32 (2) Supplement) and Research Quarterly for Exercise and Sport (2000, 71 (2)).

We also contacted 42 people across 37 European countries asking them to identify any population based survey instruments that were used to measure physical activity in young people within their country. Ten people replied but no new instruments were identified through this process.

Inclusion criteria for instruments to be considered for further evaluation required that studies a) reported the use of a self-report measure of physical activity and b) targeted participants less than 19 years of age. Studies only assessing physical fitness or employing only objective measures were excluded.

The search strategy resulted in 24,190 titles which were initially screened for duplicates and potential relevance. After this initial screening, 1839 titles and abstracts were assessed against the inclusion/exclusion criteria. In total 437 papers were assessed as potentially relevant and full papers were obtained. From these a total of 89 physical activity measures were identified. The measures were grouped as activity-based instruments (n = 67), time-based instruments (n = 5), proxy instruments (n = 17) and observation instruments (n = 1). Activity-based instruments are structured around a list of activities with minimal reference to time of day. By contrast, time-based instruments divide the day into time blocks and respondents provide the dominant activity/energy expenditure for each time block. Proxy instruments are completed on behalf of the individual (often by a parent) and observational tools are used to record behaviour by an observer.

Data Synthesis

Information about each instrument was extracted and tabulated, including instrument name, indicative references, details of physical activity assessment (including physical activity dimensions assessed, recall period, method and structure of the instrument, number of items, other relevant information, and measurement format), age of participants, indicative countries in which it has been used, and information on reliability and validity. From the initial long table a short list of 20 activity-based instruments and 2 time-based instruments was constructed (Additional file 1, Table S1). Observational and proxy measures were not considered appropriate tools for population surveillance. The inclusion criterion for short-listing was that some level of reliability and/or validity had been demonstrated. After further consideration, based mainly on participant burden and the necessity for the instruments to be suitable for population level surveillance, the two time-based instruments were dropped from further analysis.

Expert Assessment

PAQ-C/PAQ-A (Physical Activity Questionnaire)

There are two versions of this scale, with one for children (C) aged 8-14 y [20], and one for adolescents (A) aged 14-20 y [21]. For example, the PAQ-C requests responses for the last 7 days by asking participants to check a list of activities for frequency of participation on a scale from 'no', 1-2 times per week, 3-4, 5-6, to 7 times or more. Questions are also asked about physical activity in PE lessons, recess, at lunch time, right after school, and evenings, as well as 'the last weekend'. A measure of frequency of participation is requested for each day, ranging from 'none' to 'very often'.

The PAQ-C is a self-administered 7-day recall questionnaire. It is intended to measure habitual moderate-to-vigorous physical activity in children and was developed for use in the Saskatchewan Pediatric Bone Mineral Accrual Study [18]. The PAQ-A is a slightly modified version for adolescents. The instruments were produced using appropriate development methods, moreover they are fairly short and easy to use, thus were viewed as suitable for use in surveillance and monitoring. Validity data exist, using objective assessment of physical activity. However, the PAQ is designed for assessment only during the school year, and there are no data showing use in Europe.

YRBS (Youth Risk Behaviour Surveillance Survey)

The YRBS was developed in the early 1990s and has been extensively used since, including for large scale surveillance across a good range of children and youth in the USA [22, 23] (see http://www.cdc.gov/HealthyYouth/yrbs/index.htm). Respondents can report physical activity for the past year or past week, and it comprises only five items for both moderate and vigorous physical activity. A measure of sedentary behaviour is also included, and this may prove useful as the focus on sedentary behaviour increases [24]. Validity data was good for this instrument, including convergent validity with accelerometry.

The Teen Health Survey

This is a 2-item instrument developed for adolescents. Items were shortened and adapted from the YRBS and tested alongside other instruments for possible use in primary health care [25]. Items also formed part of the CITY100 (Correlates of Indoor Tanning in Youth) survey [26]. The recall period is either the previous 7 days or typical week, and data exist showing moderate-to-good associations with accelerometry. Its applicability to younger children (i.e., pre-teens) has not been tested and may be questionable.

Other instruments that received some support from the expert group did so usually for reasons of ease of administration. These included the APARQ (but has weak validity), the instrument used in the large HBSC study (has limited validity against objective measures of physical activity), CLASS (has weak reliability), and, as stated earlier, SHAPES (meets many of the criteria but has too many items for surveillance), and the Finnish Twin Cohort Study (limited age range).