Limits...
Social participation in older adults with joint pain and comorbidity; testing the measurement properties of the Dutch Keele Assessment of Participation.

Hermsen LA, Terwee CB, Leone SS, van der Zwaard B, Smalbrugge M, Dekker J, van der Horst HE, Wilkie R - BMJ Open (2013)

Bottom Line: The primary outcome was person-perceived participation, as measured with the KAP.KAPd2 lacks sufficient measurement properties for application in studies, although items may be used as single items.Further development of the concept 'participation' may help the development and validation of instruments to measure participation.

View Article: PubMed Central - PubMed

Affiliation: Department of General Practice and Elderly Care Medicine, EMGO Institute for Health and Care Research, VU University Medical Center, Amsterdam, The Netherlands.

ABSTRACT

Objective: The Keele Assessment of Participation (KAP) questionnaire measures person-perceived participation in 11 aspects of life. Participation allows fulfilment of valued life activities and social roles, which are important to older adults. Since we aimed to use the KAP in a larger Dutch cohort, we examined the measurement properties of KAP in a Dutch sample of older adults with joint pain and comorbidity.

Design: Cohort study.

Setting: A community-based sample in Amsterdam, the Netherlands and North Staffordshire, UK.

Participants: Participants were aged 65 years and over, had at least two chronic diseases (identified through general practice consultation) and reported joint pain on most days (questionnaire). The Dutch cohort provided baseline data (n=407), follow-up data at 6 months (n=364) and test-retest data 2 weeks after 6 months (n=122). The UK cohort provided comparable data (n=404).

Outcome measures: The primary outcome was person-perceived participation, as measured with the KAP. The measurement properties examined were the following: structural validity (factor analysis), internal consistency (Cronbach's α), reliability (intraclass correlation coefficients; ICC), construct validity (hypothesis testing), responsiveness (hypothesis testing and area under the curve) and cross-cultural validity (differential item functioning; DIF).

Results: Factor analysis revealed two domains: KAPd1: 'participation in basic activities' and KAPd2: 'participation in complex activities', with Cronbach's α of 0.74 and 0.57 and moderate test-retest reliability: ICC of 0.63 and 0.57, respectively. Further analyses of KAPd1 showed poor construct validity and responsiveness. Despite the uniform DIF in item 'interpersonal relations', the total KAPd1 score seemed comparable between the Dutch and UK sample.

Conclusions: Only KAP domain 'participation in basic activities' showed good internal consistency and sufficient reliability. KAPd2 lacks sufficient measurement properties for application in studies, although items may be used as single items. Further development of the concept 'participation' may help the development and validation of instruments to measure participation.

No MeSH data available.


Related in: MedlinePlus

The left plot is a true score (Item Response Theory score) of the Dutch and UK sample on differential item functioning (DIF) item 7 interpersonal interaction. The right plot shows the item characteristic curves for item 7. The curves show the probability of endorsing a particular item response (0=all, 1=most, 2=some, 3–4=little or none of the time) as a function of the DIF-free scale score and country.
© Copyright Policy - open-access
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC3753519&req=5

BMJOPEN2013003181F2: The left plot is a true score (Item Response Theory score) of the Dutch and UK sample on differential item functioning (DIF) item 7 interpersonal interaction. The right plot shows the item characteristic curves for item 7. The curves show the probability of endorsing a particular item response (0=all, 1=most, 2=some, 3–4=little or none of the time) as a function of the DIF-free scale score and country.

Mentions: In the Dutch sample, there were only three participants with incomplete KAP data at baseline. However, the UK sample had incomplete data in 124 participants (6.9%). Compared with the group with complete UK data (n=1661), the group with incomplete data (n=124) was significantly younger and more likely to live alone, but were no different for gender, number of chronic diseases, number of joint pain sites and level of physical functioning (data not shown). For optimal comparison, we matched the Dutch and UK sample based on age (four categories) and gender, which provided a final sample of 404 Dutch and 404 UK participants. The proportion of women was 62%. The proportion of participants in the four age categories was as follow: 15.6% between 65 and 70 years, 25.5% between 70 and 75 years, 26.5% between 75 and 80 years and 32.4% ≥80 years. The distribution of the trait scores of both countries showed that, in general, the Dutch participants scored better on KAPd1 (less restrictions), compared with the UK participants. The lordif method detected one item with DIF, that is, item 7 ‘interpersonal relations’. The R2 difference between models 1 and 3 was 0.0381, which indicated overall DIF. As the R2 difference between models 2 and 3 was 0.0001, item 7 showed no non-uniform DIF, which indicated only the presence of uniform DIF (figure 2). The item characteristic curve showed that the Dutch participants scored easier on this item, thus higher values, especially between the categories 0–1 and 1–2 (figure 2). This indicates that the Dutch participants had a higher chance to score restrictions on this item, given a similar trait level. Although item 7 showed uniform DIF, this DIF only had limited impact on the total domain score. Figure 3 displays the TCC and indicates no substantial impact of DIF in item 7 on the total domain score, because of the similar slopes and overlap in scorings between the two samples.


Social participation in older adults with joint pain and comorbidity; testing the measurement properties of the Dutch Keele Assessment of Participation.

Hermsen LA, Terwee CB, Leone SS, van der Zwaard B, Smalbrugge M, Dekker J, van der Horst HE, Wilkie R - BMJ Open (2013)

The left plot is a true score (Item Response Theory score) of the Dutch and UK sample on differential item functioning (DIF) item 7 interpersonal interaction. The right plot shows the item characteristic curves for item 7. The curves show the probability of endorsing a particular item response (0=all, 1=most, 2=some, 3–4=little or none of the time) as a function of the DIF-free scale score and country.
© Copyright Policy - open-access
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC3753519&req=5

BMJOPEN2013003181F2: The left plot is a true score (Item Response Theory score) of the Dutch and UK sample on differential item functioning (DIF) item 7 interpersonal interaction. The right plot shows the item characteristic curves for item 7. The curves show the probability of endorsing a particular item response (0=all, 1=most, 2=some, 3–4=little or none of the time) as a function of the DIF-free scale score and country.
Mentions: In the Dutch sample, there were only three participants with incomplete KAP data at baseline. However, the UK sample had incomplete data in 124 participants (6.9%). Compared with the group with complete UK data (n=1661), the group with incomplete data (n=124) was significantly younger and more likely to live alone, but were no different for gender, number of chronic diseases, number of joint pain sites and level of physical functioning (data not shown). For optimal comparison, we matched the Dutch and UK sample based on age (four categories) and gender, which provided a final sample of 404 Dutch and 404 UK participants. The proportion of women was 62%. The proportion of participants in the four age categories was as follow: 15.6% between 65 and 70 years, 25.5% between 70 and 75 years, 26.5% between 75 and 80 years and 32.4% ≥80 years. The distribution of the trait scores of both countries showed that, in general, the Dutch participants scored better on KAPd1 (less restrictions), compared with the UK participants. The lordif method detected one item with DIF, that is, item 7 ‘interpersonal relations’. The R2 difference between models 1 and 3 was 0.0381, which indicated overall DIF. As the R2 difference between models 2 and 3 was 0.0001, item 7 showed no non-uniform DIF, which indicated only the presence of uniform DIF (figure 2). The item characteristic curve showed that the Dutch participants scored easier on this item, thus higher values, especially between the categories 0–1 and 1–2 (figure 2). This indicates that the Dutch participants had a higher chance to score restrictions on this item, given a similar trait level. Although item 7 showed uniform DIF, this DIF only had limited impact on the total domain score. Figure 3 displays the TCC and indicates no substantial impact of DIF in item 7 on the total domain score, because of the similar slopes and overlap in scorings between the two samples.

Bottom Line: The primary outcome was person-perceived participation, as measured with the KAP.KAPd2 lacks sufficient measurement properties for application in studies, although items may be used as single items.Further development of the concept 'participation' may help the development and validation of instruments to measure participation.

View Article: PubMed Central - PubMed

Affiliation: Department of General Practice and Elderly Care Medicine, EMGO Institute for Health and Care Research, VU University Medical Center, Amsterdam, The Netherlands.

ABSTRACT

Objective: The Keele Assessment of Participation (KAP) questionnaire measures person-perceived participation in 11 aspects of life. Participation allows fulfilment of valued life activities and social roles, which are important to older adults. Since we aimed to use the KAP in a larger Dutch cohort, we examined the measurement properties of KAP in a Dutch sample of older adults with joint pain and comorbidity.

Design: Cohort study.

Setting: A community-based sample in Amsterdam, the Netherlands and North Staffordshire, UK.

Participants: Participants were aged 65 years and over, had at least two chronic diseases (identified through general practice consultation) and reported joint pain on most days (questionnaire). The Dutch cohort provided baseline data (n=407), follow-up data at 6 months (n=364) and test-retest data 2 weeks after 6 months (n=122). The UK cohort provided comparable data (n=404).

Outcome measures: The primary outcome was person-perceived participation, as measured with the KAP. The measurement properties examined were the following: structural validity (factor analysis), internal consistency (Cronbach's α), reliability (intraclass correlation coefficients; ICC), construct validity (hypothesis testing), responsiveness (hypothesis testing and area under the curve) and cross-cultural validity (differential item functioning; DIF).

Results: Factor analysis revealed two domains: KAPd1: 'participation in basic activities' and KAPd2: 'participation in complex activities', with Cronbach's α of 0.74 and 0.57 and moderate test-retest reliability: ICC of 0.63 and 0.57, respectively. Further analyses of KAPd1 showed poor construct validity and responsiveness. Despite the uniform DIF in item 'interpersonal relations', the total KAPd1 score seemed comparable between the Dutch and UK sample.

Conclusions: Only KAP domain 'participation in basic activities' showed good internal consistency and sufficient reliability. KAPd2 lacks sufficient measurement properties for application in studies, although items may be used as single items. Further development of the concept 'participation' may help the development and validation of instruments to measure participation.

No MeSH data available.


Related in: MedlinePlus