Physical activity assessment – Questionnaire

Self-report methods are the most convenient and cheapest way to collect physical activity data from a large number of people in a short time. Physical activity questionnaires (PAQs) are the most widely used self-report instrument to assess physical activity and have been used extensively in research.  Self-report measures include self or interviewer administered (Matthews, 2002):

Questionnires vary greatly in their detail. Recall questionnaires typically contain 5-15 items and aim to stratify the population into broad categories of physical activity. Some may also try to estimate energy expenditure. The reference time frame varies depending on the objectives of the study but typically tends to be between a week and a month. Examples are the short version of the IPAQ and the Baecke questionnaire.

Quantitative history questionnaires are usually 15 to 60 items long.  Their objective is to capture information in various domains according to several dimensions of physical activity.  They provide an insight to the patterns of physical activity and may provide an estimate of energy expenditure using the Compendium of Physical Activity (Ainsworth et al, 2000). The time frame is highly variable, from a week to a lifetime. Examples are the second EPIC questionnaire EPAQ2 (Wareham et al, 2002) and the long version of the IPAQ.

Global self report instruments contain 1-4 items and aim to stratify population into main categories of activity.  They tend to inquire about activity over a long time frame (e.g. one year) which reduces systematic bias due to season and day of the week. An example of a global self-report tool is the shortened EPIC questionnaire (Wareham et al, 2003) and The Stanford Brief Activity Survey (Taylor-Piliae, 2006)

Self report instruments may be completed on paper and mailed back or completed on the internet.  Interviews may be conducted either face-to-face or via phone. 

There is a plethora of PAQs available and some of the best known or well validated tools are included in the ‘Examples and Links’ section.  There have been several excellent reviews of PAQ which provide details of testing for reliability and validity.  In 1997, a comprehensive journal supplement was published (Pereira et al, 1997).  This collection includes details of many instruments including results for reliability and validity. Among the questionnaires included are:

  • Baecke Physical Activity Questionnaire
  • Godin Shepard Leisure Time Questionnaire
  • Paffenbarger Physical Activity Questionnaire
  • Bouchard’s Activity Diary
  • Previous Day Recall (PDR) (Sallis et al, 1985; Sallis et al, 1993).

Other reviews of the validity and reliability of PAQs have been carried out for children (Kohl et al, 2000) and preschoolers (Oliver et al, 2007). 

Since the publication of the 1997 supplement (Pereira et al, 1997), other important questionnaires have been developed which merit particular mention. 

International Physical Activity Questionnaire (IPAQ)
This questionnaire is available in a short form for surveillance, and in a longer form when more detailed physical activity information is required, both forms are available in a number of languages. The questionnaire was rigorously tested for reliability and validity (Craig et al, 2003) and this has been replicated in a number of countries; this questionnaire is not designed to provide a detailed assessment of physical activity in all domains

Global Physical Activity Questionnaire (GPAQ)
This questionnaire was developed under the auspices of the World Health Organisation and it collects information on participation in physical activity in three domains: activity at work, travel to and from places and recreational activities.

Both the IPAQ and the GPAQ were developed for surveillance studies and the GPAQ more specifically for surveillance studies in developing countries. These instruments are not recommended for other purposes.

Other questionnaires aiming at measuring dose of physical activity by domains have been developed for investigative purposes. One of them is the EPIC Physical Activity Questionnaire 2 (EPAQ2), which has been developed and validated in England (Wareham et al, 2002).  The Recent Physical Activity Questionnaire (RPAQ) developed from the EPAQ2 with a shorter time frame of one month instead of one year is currently under validation with promising results (Besson et al, 2006).

Self reports for specific populations
Cognitive immaturity or degeneration make self-report of physical activity difficult in the young and elderly.  Children’s activity is unique in that it is characterised by short bouts rather than more sustained periods of activity (Pate, 1993; Kohl et al, 2000; Sirard & Pate, 2001).  For this reason specific recommendations of levels of desirable activity have been made for this age group (Welk et al 2000).  Self-report is not viable in the young, (Sallis 1991) and previous day’s recall has been suggested as the most appropriate method for children aged 10-11 years (Sallis et al, 1993).  Decisions then need to be made on how many days and on which days activity will be measured (Welk et al, 2000).  It is necessary to rely on proxy reports for children e.g. parents is difficult as a child gets older and becomes more independent (Pate, 1993).

The children physical activity questionnaire (CPAQ) and the youth physical activity questionnaire (YPAQ) have recently been tested for their validity and reliability. Validation results suggested that these PAQs were unable to accurately estimate time spent at moderate and vigorous intensity physical activity and physical activity energy expenditure. However, they may rank individuals accurately (Corder et al, AJCN, in press).   

For the elderly, using physical activity questionnaires which have been designed and validated in younger populations is inappropriate (Washburn, 2000).  Four questionnaires have been specifically designed for this segment of the population (Washburn, 2000):

  • Modified Baecke Questionnaire
  • Zutphen Physical Activity Questionnaire
  • Yale Physical Activity Survey
  • Physical Activity Survey for the Elderly.

The elderly are a diverse population group in terms of physical and cognitive function and this is likely to be reflected in a wide range of activity levels and competence to self-report this activity.

Limitations of PAQs
The challenge of PAQs is to capture the multiple domains in which activity occurs i.e.

  • Occupation
  • Household
  • Transport
  • Leisure

and then the various characteristics of this activity i.e.

  • Frequency
  • Intensity
  • Duration

(Sallis and Saelens, 2002).

To date, many questionnaires have focused on the leisure time domain to the exclusion of other domains which will not provide a measure of habitual activity.   Many PAQs also make an estimate of physical activity energy expenditure by using the Compendium of Physical Activity (Ainsworth et al, 2000); a paediatric compendium has recently been published (Ridley et al, 2008).  These compendia provide a classification of specific activities by rate of energy expenditure and define the ratio of work metabolic rate to a standard metabolic rate (MET).  Most questionnaires are unable to do this with accuracy (Shepherd, 2003; MacFarlane et al, 2006).  Questionnaires are able to give estimates of time spent in activities of various levels of intensity, and are able to rank people in levels of reported activity.

Cognitive demands
PAQs are also subject to bias as a result of the cognitive demands of the recall process itself.  The sequential cognitive processes underlying the storage of memories have been described (Baranowski and Domel, 1994) along with models explaining their retrieval (Wills et al, 1991), illustrating the complexity of the task especially to report durations. Unfortunately, very few questionnaires have been designed taking cognitive demands into account.

Questionnaire design
A poorly constructed and inadequately piloted questionnaire will be badly completed and ultimately provide data of very limited quality.  Questionnaires should be well formatted and clearly structured to guide the individual. They should be comprised primarily of closed questions and be ordered chronologically, in a logical sequence and move from domain to domain.  Questions must be unambiguous and explain the intent of the question clearly to the respondent; for example the term ‘exercise’ may mean only activities which cause breathlessness to some, while a question about leisure time walking or cycling will not capture active commuting to work. Leisure time physical activity is a broad descriptor, and includes formal exercise programs, walking, hiking, gardening, sport and dance to name a few. Many questionnaires used in observational studies have focused on leisure time physical activity but there is a danger in focusing on one aspect of physical activity (Wareham, 2007).  This is particularly the case where the exact relationship between physical activity (e.g. energy expenditure or intensity) and health outcome such (e.g. diabetes prevention) is not known.  Additionally, most of the physical activity questionnaires which focus on participation in sports have been developed for men (Oguma & Shinoda-Tagauwa, 2004).

Validation (see Basic Concepts section)
Many PAQs claim to have proven validity but this should be examined carefully (Kohl et al, 2000).  PAQs should be compared to a criterion measure such as doubly labelled water, indirect calorimetry, observation or an objective method such as accelerometry. Moreover, in any given questionnaire usually a limited number of items and derived variables are valid. Deriving new variables from an existing questionnaire requires a new validation. Markers of physical fitness are unsuitable serving as criterion methods in validation studies of questionnaires as fitness and physical activity are not synonymous; the former is influenced by many factors (age, sex, training effect, genetics). These issues along with problems with reliability, validity and sensitivity have been comprehensively discussed (Rennie & Wareham, 1998; Shephard, 2003).

Self-reports are also culturally dependent. Validity results assessed in one population may not be directly applicable to other populations, ethnic groups or other geographical regions. Few questionnaires have been appropriately validated in developing countries.


Web design by Studio 24