Identify performance based and questionnaire based instruments which have been evaluated for their utility in the assessment of key domains of activity in obese adults

Assess the measurement properties of performance based and questionnaire based tools used to assess key aspects of activity in obese adults

A search of the computerized bibliographic databases available via the OVID interface, namely EMBASE (1980 to present) and MEDLINE (1950-present) and via the EBSCO interface and AMED (1980 to present) and SPORTDiscus (1985 to present) will be undertaken.

The search will be limited to English language publications (because of resource constraints) but with no time limit restrictions. Reference lists of identified studies and reviews will be hand searched to identify further relevant papers. Authors will be contacted to further clarify study design and results, if appropriate.

Any quantitative study design will be included if the study reports on at least one measurement property of an instrument purporting to measure the relevant constructs in the condition and population of interest. Studies which use the measurement tool but which do not report on any aspect of the measurement properties of that tool will be excluded, that is, studies will only be included if they are explicitly designed to develop tools or to evaluate the measurement properties of tools

Studies which report on perfomance based or self report measures of one or more of the following key domains of activity in obese adults will be included: functional capability, activity limitations, participation, beliefs and attitudes, barriers and facilitators.

Obese adults over 18 years will be included. Studies which include only data on obese children or adolescents will be excluded or where obese children or adolescents constitute more than 10% of the study population. Studies in sporting cohorts or those in disease specific groups will be excluded

Not applicable.

Not applicable.

any setting

Primary outcomes

Measurement properties of performance based or self report instruments including % of patients completing the measure; validity (content validity; construct validity including structural validity, hypothesis testing; criterion validity); reliability ( internal consistency; test-retest, intra-rater, interrater reliability; measurement error); responsiveness; interpretation (minimum clinically important difference). The COSMIN taxonomy will be used to define the measurement properties.

Not applicable.

Secondary outcomes

None

Not applicable.

The electronic database search will be undertaken by one investigator. All identified abstracts will be examined by two investigators independently to determine studies to be included. Differences in opinion will be resolved by a third investigator. For all included studies, full manuscripts will be reviewed by two investigators and data extracted independently. Disagreements between assessors will be resolved by a third investigator. The dates of the searches, the databases and key terms used in the searches, the output of the searches, the titles, abstracts and sources of retrieved papers and the data extracted from each study will be recorded contemporaneously. Endnote will be used to store information on abstracts retrieved from the searches. Key decisions on the inclusion and exclusion of papers, with reasons, will also be recorded.

For each study, information on definition of condition, participant characteristics (age, gender, BMI range, mean/median), study setting (e.g. clinic, research laboratory, population), study design, the construct evaluated, the type of instrument (e.g observed performance, self completed questionnaire, interview based questionnaire), content of the instrument and the reported measurement properties (as above) and their results.

Methodological quality (risk of bias) will be assessed for each measurement property within each study using the COSMIN checklist. This checklist includes a separate section for each measurement property relevant to this review. Each property is scored on a 4 point rating scale. The COSMIN checklist can be used to determine an overall score which can be categorised as good, fair or poor. Assessors will be trained together in data extraction tools and methodological issues and undertake a pilot exercise when inter-rater reliability will be informally assessed.

This review is essentially exploratory to inform the design of future studies of the evaluation of instruments. In addition, the characteristics of populations and instruments are likely to be heterogeneous. Hence, it is not anticipated that formal meta-analysis will be performed. A narrative approach to analysis will be undertaken for each instrument, the first stage will be to classify the instruments by review construct. The construct measured in the study will be assessed independently by two investigators to determine the proportion of items of each measure deemed to fall into each of the construct categories or none. These will be compared and any discrepancies which cannot be resolved by discussion, will be resolved by a third party. A measure, or subscale of a measure, will be assigned to a construct when at least two thirds of the items are felt to belong to that construct. Then for each instrument, the key data measurement properties for that instrument will be agreed, using the COSMIN taxonomy to classify the different properties. For each relevant measurement property the results of the different studies will be tabulated for each instrument with key information provided on study limitation for that property. This will be repeated when study quality is categorised into good, fair and poor using the criteria set out in the COSMIN checklist. Finally, a profile for each instrument within a particular construct will be generated including for each measurement property: the number of studies reporting on each measurement property and if more than one study, the evidence on methodological quality for each measurement property across the studies and the consistency in findings between studies for each measurement property; for each instrument: the coherence between measurement properties for the same instrument and the completeness of information across relevant measurement properties.

When possible, analysis will be undertaken for different BMI groups ( e.g < 35,35 to 40, 40 to 50, 50+) and by study setting e.g clinic, research laboratory and population

The findings of this study will be disseminated to the National Obesity Observatory, through clinical and professional networks to those managing weight management services, through the Association for the Study of Obesity and conference presentations and peer reviewed journals.

Paola Dey

School of Postgraduate Medical and Dental Education

Greenbank Building

University of Central Lancashire

Adelphi Street

Preston

PR1 2HE

MPDey@uclan.ac.uk

None

Professor Paola Dey, University of Central Lancashire
Professor John Wilding, University of Liverpool
Dr Paul Norman, University of Sheffield
Professor James Selfe, University of Central Lancashire
Dr Chris Sutton, University of Central Lancashire
Mr Neil Cook, University of Central Lancashire

none

02 July 2012

29 March 2013

Liverpool Primary Care Trust Flexibility and Sustainability Fund

None known

None

English

England

Subject indexing assigned by CRD

Exercise; Humans; Obesity; Program Evaluation; Weight Loss

None

None

14 August 2012

14 August 2012