Abstract
Objective: to investigate how to interpret changes on the CASP-19 quality of life scale for older people, and whether it discriminates between, and is responsive to, relevant differences or changes in participants' circumstances.
Methods: analysis of data from the English Longitudinal Study of Ageing for those completing CASP-19 in both Wave 1 and Wave 2 (n = 6,482). Cross-sectional and longitudinal comparisons, using multiple linear regression, of CASP-19 scores with respect to eight anchor variables.
Results: cross-sectional comparisons found differences in mean CASP-19 scores at Wave 1 between categories of anchor variables varied from 1.9 for living alone to 8.0 for being able to walk ¼ mile with difficulty. Longitudinal comparisons of changes in CASP-19 found that subjects that had moved between categories of the anchor variables over 28 months, had changed their mean CASP-19 score by about 1 unit in the expected direction, compared with the unchanged category. These changes were statistically significant for six of the eight anchors.
Conclusions: the cross-sectional comparisons help interpret differences and indicate CASP-19 has discriminatory power. The longitudinal changes show that CASP-19 is responsive to changes in most anchor variables that reflect some aspects of quality of life.
Keywords: quality of life, older persons, responsiveness, discriminatory power, anchor-based methods; longitudinal studies, elderly
Introduction
In recent years, quality of life (QoL) measures have increased in use [1, 2]. Studies in older people have shown that QoL is perceived to extend beyond health, and any measure should also cover social circumstances and functional limitations [3]. For instance, an injury can affect mental and physical health, functional ability and social activities. Any intervention may affect one or more these, not necessarily all in the same direction. It can be very useful to gauge the impact of the intervention by a QoL measure, but relatively few measures have been developed for older people. CASP-19 is a wide-ranging theoretically derived measure of well-being developed for older people, spanning four domains of control, autonomy, self-realisation and pleasure. It is meant to reflect the ‘Third Age’ (characterised by freedom from work and family constraints) rather than just the Fourth Age (decline and dependence) [4]. It has been used in some major observational studies in Europe and USA [5–8].
However, before using this measure, we should consider how to interpret changes on this scale and whether it discriminates between, and is responsive to, relevant differences or changes in circumstances of the subjects involved. Interpretability has been defined as ‘the degree to which one can assign qualitative meaning to an instrument's quantitative scores' [9]. Discriminatory power is the ability to distinguish between subgroups differing in health or socio-economic position at a cross-sectional comparison [10]. Responsiveness assesses how well an instrument can detect longitudinal differences in outcomes that are of practical importance [11]. A single analysis cannot establish whether a measure meets these criteria: this becomes apparent as evidence accumulates.
In this paper, the interpretability, discriminatory power and responsiveness of CASP are investigated by anchor-based methods [12, 13]. These examine the relationship between the measure of interest and an independent measure that is interpretable and at least moderately correlated with the instrument (e.g. presence of symptoms or job loss). The data have been used to investigate cross-sectional and longitudinal differences in CASP scores for a number of explanatory variables related to the dimensions spanned by the measure and will be used as anchors. The analyses in this paper use data from the English Longitudinal Study of Ageing (ELSA).
Methods
Subjects
The ELSA cohort collects data on many aspects of economic, health-related and social circumstances from a nationally representative sample of non-institutionalised adults aged 50 years and over in England. The analyses in this paper use data from Wave 1 and Wave 2 downloaded from the UK Data Archive. Wave 1 ELSA interviews took place in 2002, and Wave 2 interviews in 2004–05: the average time between them was 28 months. Technical details and results are available in published reports [6, 14]. A total of 9,300 people completed CASP in Wave 1 and 6,820 in Wave 2. The analyses in this paper are restricted to 6,182 participants who completed CASP on both occasions. This restriction to complete cases allows a clear comparison between the cross-sectional and longitudinal differences.
Measures
The QoL measure used is CASP [15, 16], which was developed for older people. It is a self-completion questionnaire and spans four derived dimensions of control, autonomy, self-realisation and pleasure based on Likert scaled items. Examples of items are ‘My health stops me from doing the things I want to do’ and ‘Shortage of money stops me from doing the things I want to do’. CASP has both a 12-item and the more commonly used 19-item version. The scale developers recommended CASP-12, particularly if researchers want to use the subscales [16] (NB an earlier version of CASP-12 was used in SHARE [5]). However, there is some evidence of a ceiling effect in CASP-12, which is less of a problem for CASP-19, so CASP-19 is used in the analyses in this paper (Supplementary data available in Age and Ageing online, Figure 1, where CASP-12 and CASP-19 scores are plotted against each other, for the two successive waves of data collection). CASP-19 has an overall summary measure on a 0–57 scale: high scores correspond to greater well-being.
Eight of the variables collected in ELSA have been used to characterise participants' circumstances. They were chosen to reflect different aspects of well-being included in the CASP-19 items and had a high response rate (≥98% in those who completed CASP-19 in both waves). They comprised four health indicators and three socio-economic indicators: whether or not the respondent has a limiting chronic illness, suffers from depression (CES-D 8 score>3 [17]), is often troubled by pain, has difficulty walking ¼ mile, lives alone, is in employment and has access to a car. A self-assessed socio-economic variable was also included, their ‘self-perceived social status on a ladder’ (0–100 scale, 100 being highest status).
The changes in their circumstances between the Wave 1 and Wave 2 were characterised as ‘Same’, ‘Better’ or ‘Worse’ for the binary variables: this allows consideration of whether the direction of change is important in describing meaningful change [12]. ‘Worse’ describes those whose health-related condition had deteriorated, and, arbitrarily, those who had changed to living alone or were no longer in employment in Wave 2 were also categorised as ‘Worse’. In addition to calculating the difference between the ‘position on social ladder’ collected at Wave 1 and Wave 2 (the prospective approach), participants had also been asked at Wave 2 whether their ‘position on social ladder’ was ‘Same’, ‘Better’ or ‘Worse’ (the retrospective approach). A number of studies have found differences between retrospective and prospective assessment of change [18, 19]. Warnings have been given about using retrospective assessments, since patients can have difficulty recalling previous states and are unduly influenced by their current state when describing any change [18]. Given these problems, only the results of the prospective change in social ladder score have been reported.
Covariates which did not change over time were also used in the analyses: age at Wave 1, whether the respondent was educated to A-level standard (completion of high school education) and gender, as these have been shown to be associated with CASP-19 scores [20].
Statistical analysis
The interpretation of changes of scores on the CASP-19 scale has been addressed by both cross-sectional and longitudinal anchor-based approaches. In the cross-sectional analyses, mean levels of CASP-19 at Wave 1 were compared between subjects who differed on the eight anchor variables chosen for the analysis. This was done as eight separate linear regressions looking at the relationship between the CASP-19 score at Wave 1 (the outcome variable) and the separate health and socio-economic anchor variables while adjusting for age as a quadratic function, gender and education (the non-varying covariates). Social ladder position was included as five categories (and therefore four dummy variables), as the relationship between CASP-19 at Wave 1 and this variable was non-linear. The strength of association of the ‘social ladder’ (0–100 scale) with CASP-19 was measured by a Spearman correlation coefficient.
A similar approach was taken to look at longitudinal changes, where the responsiveness of CASP-19 was explored using regression models [21]. Here, the change in CASP-19 (Wave 2–Wave 1) was the outcome variable and was regressed on the changes in the anchor variables between the two waves. The change in the prospective measurement of social ladder position was included as a linear term, since the relationship with the change in CASP-19 scores appeared to be linear. The separate regressions for each anchor variable all also contained age (as a quadratic function), education, gender and time in months between data collection for the two Waves.
All analyses were done using Stata version 9 [22], using sample weights to allow for the complex survey structure.
Results
The mean CASP-19 score at Wave 1 for the 6,182 participants who had completed the score at both Waves was 43.3. The baseline distribution of the anchor variables and other covariates are shown in Table 1.
Table 1.
Baseline distribution of variables at ELSA Wave 1
| Variable | % | n |
|---|---|---|
| Male | 46 | 6,182 |
| Age group | ||
| 50–59 | 43 | 6,182 |
| 60–69 | 32 | |
| 70+ | 25 | |
| Educated to A level or above | 35 | 6,179 |
| Limiting chronic illness | 30 | 6,182 |
| Depression (CESD >3) | 26 | 6,120 |
| Often troubled by pain | 36 | 6,181 |
| Difficulty walking ¼ mile | 21 | 6,177 |
| Lives alone | 21 | 6,182 |
| Access to car | 89 | 6,180 |
| In employment | 38 | 6,155 |
| Self-perceived position on social ladder (100 = highest) | ||
| 0–20 | 3 | 6,058 |
| 21–40 | 15 | |
| 41–60 | 44 | |
| 61–80 | 33 | |
| 81–100 | 5 | |
Table 2 shows the unadjusted mean CASP-19 scores at Wave 1 for different subgroups. It can be seen that the differences in means between the subgroups (yes/no) vary between 8.0 for difficulty walking ¼ mile, and 1.9 for living alone: adjusting for age, gender and education made very little difference to these estimates. The mean CASP-19 score varied considerably along the social ladder subgroups. The 95% confidence intervals indicate that all these changes are statistically significant at the 5% level, indicating that CASP-19 has adequate discriminatory power between subgroups varying in factors reflecting some relevant aspects of QoL.
Table 2.
Distribution of CASP-19 at ELSA Wave 1—by characteristics at Wave 1
| Separate regressions |
||||
|---|---|---|---|---|
| Variable | Unadjusted meana | Adjusted difference in meansb | 95% CI | Sample size |
| Limiting chronic illness | ||||
| No | 45.4 | 4,351 | ||
| Yes | 38.0 | −7.2 | −7.6 to −6.7 | 1,831 |
| Depression (CESD >3) | ||||
| No | 45.3 | 4,515 | ||
| Yes | 37.6 | −7.6 | −8.1 to −7.2 | 1,605 |
| Often troubled by pain | ||||
| No | 45.2 | 3,959 | ||
| Yes | 39.6 | −5.5 | −5.9 to −5.1 | 2,222 |
| Difficulty walking ¼ mile | ||||
| No | 44.9 | 4,862 | ||
| Yes | 36.9 | −8.0 | −8.5 to −7.5 | 1,315 |
| Lives alone | ||||
| No | 43.6 | 4,887 | ||
| Yes | 41.5 | −1.9 | −2.4 to −1.3 | 1,295 |
| Access to car | ||||
| No | 39.5 | 694 | ||
| Yes | 43.7 | 3.7 | 3.0–4.3 | 5,486 |
| In employment | ||||
| No | 42.4 | 3,819 | ||
| Yes | 44.5 | 3.0 | 2.5 to 3.6 | 2,336 |
| Position on social ladder | ||||
| 0–20 | 32.0 | −16.5 | −17.8 to −15.1 | 177 |
| 21–40 | 37.9 | −10.6 | −11.6 to −9.6 | 908 |
| 41–60 | 43.3 | −5.3 | −6.2 to −4.4 | 2,671 |
| 61–80 | 46.2 | −2.4 | −3.3 to −1.5 | 2,007 |
| 81–100 | 48.6 | 295 | ||
aUsing survey weighting.
bAdjusted differences in means (95% CI) arise from separate regressions for each explanatory variable adjusted for age, age-squared, education and gender only.
The changes in CASP-19 for individuals between Wave 1 and 2 varied between −23 and 24 with a mean of −0.3 (SD: 4.3). So considering the scale ranges from 0 to 57, the CASP-19 score of most participants changed relatively little over a period that averaged 28 months (SD = 2 months). Of the total, 20–25% of subjects reported changes (for better or worse) in limiting chronic illness, depression and pain categories. Changes in being able to walk ¼ mile and in were seen in around 15% of subjects, but changes in living alone, having access to a car and being in employment only occurred in 4–9% of subjects. Changes in multiple categories were not very common: 32% reported no changes in any of the eight anchor variables, but 34% reported changes in one, 21% in two, 9% in three and only 3% in 4 or more.
Table 3 shows the results of the linear regressions of the change in CASP-19 scores associated with changes in each of the anchor variables. The regression coefficients estimate the difference in the change in CASP-19 score for those where that category improved compared with those where it was unchanged, and similarly between those whose situation worsened against those unchanged. The mean changes in CASP-19 show that they were positive when a variable improved, and negative when the variable worsened (as would be expected). In contrast to the cross-sectional analyses, these mean changes were smaller, none being greater than 2 units. For most anchor variables, the mean change in CASP-19 was higher by about 1 unit when a category improved and lower by about 1 unit if it worsened. The analysis of the change in social ladder score provides an estimate of the mean change in CASP-19 for a 10-unit change on the ladder scale. The 95% confidence intervals indicate that these changes are statistically significant at the 5% level for all anchors, except for living alone and being in employment, indicating that CASP-19 is responsive to changes in many anchors reflecting some relevant aspects of QoL.
Table 3.
Change in CASP-19 between ELSA Waves 1 and 2—by changes in anchor variables between Waves 1 and 2
| Separate regressions |
||||
|---|---|---|---|---|
| Variable | Sample size | Unadjusted meana | Adjusted difference in meansb | 95% CI |
| Limiting chronic illness | ||||
| Same | 4,993 | −0.4 | ||
| Better | 504 | 0.7 | 1.2 | 0.6 to 1.7 |
| Worse | 683 | −1.9 | −1.4 | −1.9 to −0.9 |
| Depression | ||||
| Same | 4,576 | −0.4 | ||
| Better | 712 | 0.9 | 1.4 | 0.9 to 1.9 |
| Worse | 787 | −2.0 | −1.4 | −1.9 to −1.0 |
| Often troubled by pain | ||||
| Same | 4,729 | −0.4 | ||
| Better | 720 | 0.1 | 0.5 | 0.0 to 1.0 |
| Worse | 732 | −1.6 | −1.2 | −1.7 to −0.7 |
| Difficulty walking ¼ mile | ||||
| Same | 5,400 | −0.4 | ||
| Better | 285 | 0.9 | 1.4 | 0.6 to 2.1 |
| Worse | 489 | −2.5 | −1.9 | −2.5 to −1.4 |
| Lives alone | ||||
| Same | 5,957 | −0.5 | ||
| Better | 59 | 0.3 | 0.8 | −0.8 to 2.4 |
| Worse | 166 | 0.1 | 0.7 | −0.2 to 1.6 |
| Access to car | ||||
| Same | 5,747 | −0.4 | ||
| Better | 149 | 0.6 | 1.3 | 0.3 to 2.2 |
| Worse | 284 | −1.6 | −0.8 | −1.6 to −0.1 |
| In employment | ||||
| Same | 5,550 | −0.5 | ||
| Better | 162 | −0.4 | −0.2 | −1.1 to 0.8 |
| Worse | 429 | −0.2 | 0.0 | −0.6 to 0.6 |
| Position on social ladderc | βd | |||
| 0.5 | 0.4 to 0.6 | |||
aUsing survey weighting.
bAdjusted differences in means (95% CI) from separate regressions for each explanatory variable adjusted for age, age-squared, education and gender only.
cNumeric difference between scores at Wave 1 and Wave 2.
dRegression coefficient per 10 unit increase.
Discussion
The ELSA cohort data set has allowed consideration of the interpretation of a QoL measure aimed at those in early old age. When anchor-based approaches are taken to interpreting values or differences on a QoL scale, the anchors need to be both interpretable in themselves and be associated with the QoL scale. The health-related and binary socio-economic variables were clearly interpretable, and were related to one or more of the Likert scales on which CASP-19 was based. The numerical social ladder score was moderately correlated with CASP-19: the Spearman correlation coefficient was 0.41.
It was found that the difference in mean CASP-19 scores at Wave 1 between binary categories varied between 1.9 for living alone or not, to 8.0 for being able to walk ¼ mile with difficulty or not. These sizes of differences correspond to standardised effect sizes [21] (with respect to the Wave 1 standard deviation of CASP-19 scores) of between 0.2 and 1.0. These effect sizes would usually be interpreted as small to large effects, and since all the differences were statistically significant, this indicates CASP-19 has adequate discriminatory power. Although the anchor variables were chosen to reflect different aspects of well-being, they are correlated to some extent, e.g. those who had difficulty walking ¼ mile were more often troubled by pain. Therefore, the difference in mean CASP-19 score between categories of one anchor variable may also reflect the fact that they have a tendency to be affected by other anchor variables.
There has been little work on trying to interpret values on the CASP scale. Blane et al. [23] used the Wave 1 data set from ELSA to investigate the association between social position and CASP-19 scores and reported that ‘as a yardstick against which to judge differences in CASP-19 scores, the differences between not having a long-standing illness and having a limiting long-standing illness … is 8.2’. This is slightly larger than the unadjusted differences reported here, but probably reflects the fact that they used all those who had completed CASP-19 in Wave 1, not the smaller subgroup those who had completed CASP-19 in both waves used in this analysis. However, the results from this analysis further aid the interpretation of CASP-19 by providing estimates of differences due to extra factors related to the CASP-19 items. As reported earlier, CASP-19 was not completed by all members of the cohort. It is difficult to estimate whether this has biased the results in this analysis. However, a previous study found that there was no difference between the analyses of CASP-19 scores in Wave 1 of ELSA using either subjects who had completed CASP-19 or using multiple imputation to replace missing records [20].
Note that this analysis has not shed any light on the minimal important difference—usually defined as the smallest difference in score that patients perceive as either beneficial or harmful and that would lead the patient or clinician to consider a change in management [24, 25]. It is not possible to say whether changes in CASP-19 score across any of the factors investigated should be regarded as of practical importance to overall QoL.
When the longitudinal changes in CASP-19 were investigated, it was found that subgroups who had changed (for better or worse) over approximately 28 months on the chosen binary variables had changed their mean CASP-19 score by about 1 unit, in the expected direction, compared with those where the category remained unchanged: the exception was in the employment subgroups. There is considerable debate about the most appropriate way to measure responsiveness without using anchor methods [26]. To allow some comparison between the cross-sectional and longitudinal results, a difference of approximately 1 unit in means of change scores of subgroups could be expressed as an effect size of 0.2–0.3 (with respect to the SD of changes in CASP-19), which would usually be interpreted as ‘small’. Whereas the Wave 1 differences in CASP-19 between levels of an anchor variable showed some variation between the anchors in the cross-sectional analysis, there was far less variation in the longitudinal changes between change categories of these anchors. Nonetheless, these mean changes over time were statistically significant at the 5% level for six out of the eight anchor variables, indicating that CASP-19 is responsive to changes in anchors that reflect some aspects of QoL.
Conclusion
This analysis has made a contribution to the interpretation and performance of the CASP-19 score, and to an understanding of the factors that are associated with cross-sectional and longitudinal changes in QoL in older people within the general population in England.
Key points.
Cross-sectional comparison of CASP-19 scores found small differences in those who differed in indicators of socio-economic position and larger differences in those with differing health states.
CASP-19 can discriminate between subgroups differing in health or socio-economic position.
CASP-19 is responsive to changes in factors that reflect aspects of QoL.
Conflicts of interest
None declared.
Supplementary data
Supplementary data mentioned in the text is available to subscribers in Age and Ageing online.
Acknowledgements
Data from the English Longitudinal Study of Ageing were used with permission from the UK Data Archive, University of Essex.
References
- 1.Higginson I, Carr A. Using quality of life measures in the clinical setting. BMJ. 2001;322:1297–300. doi: 10.1136/bmj.322.7297.1297. doi:10.1136/bmj.322.7297.1297. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Testa M, Simonson D. Assessment of quality-of-life outcomes. JAMA. 1996;334:835–40. doi: 10.1056/NEJM199603283341306. [DOI] [PubMed] [Google Scholar]
- 3.Netuveli G, Blane D. Quality of life in older ages. Br Med Bull. 2008;85:113–26. doi: 10.1093/bmb/ldn003. doi:10.1093/bmb/ldn003. [DOI] [PubMed] [Google Scholar]
- 4.Laslett P. A Fresh Map of Life. 2nd edition. London: Macmillan; 1996. [Google Scholar]
- 5.Borsch-Supan A, Hank K, Jurges H. A new comprehensive and international view on aging: introducing the Survey of Health, Ageing and Retirement in Europe. Eur J Ageing. 2005;2:245–53. doi: 10.1007/s10433-005-0014-9. doi:10.1007/s10433-005-0014-9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Marmot M, Banks J, Blundell R, Lessof C, Nazroo J. Health, Wealth and Lifestyles of the Older Population in England : the 2002 English Longitudinal Study of Ageing. London: Institute for Fiscal Studies; 2002. [Google Scholar]
- 7.National Institute on Aging. Growing Older in America: the Health and Retirement Study. Washington, DC: US Dept of Health and Human Services; 2007. [Google Scholar]
- 8.Peasey A, Bobak M, Kubinova R, et al. Determinants of cardiovascular disease and other non-communicable diseases in Central and Eastern Europe: rationale and design of HAPIEE study. BMC Public Health. 2006;6:255. doi: 10.1186/1471-2458-6-255. doi:10.1186/1471-2458-6-255. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Lohr KN, Aaronson NK, Alonso J, et al. Evaluating quality-of -life and health status instruments: development of scientific review criteria. Clin Ther. 1996;18:979–92. doi: 10.1016/s0149-2918(96)80054-3. doi:10.1016/S0149-2918(96)80054-3. [DOI] [PubMed] [Google Scholar]
- 10.Macran S, Weatherly H, Kind P. Measuring population health: a comparison of three generic health status measures. Med Care. 2003;41:218–31. doi: 10.1097/01.MLR.0000044901.57067.19. [DOI] [PubMed] [Google Scholar]
- 11.Guyatt GH, Kirshner B, Jaescke R. Measuring health status: what are the necessary measurement properties. J Clin Epidemiol. 1992;45:1341–5. doi: 10.1016/0895-4356(92)90194-r. doi:10.1016/0895-4356(92)90194-R. [DOI] [PubMed] [Google Scholar]
- 12.Crosby RD, Kolotkin RL, Williams GR. Defining clinically meaningful change in health-related quality of life. J Clin Epidemiol. 2003;56:395–407. doi: 10.1016/s0895-4356(03)00044-1. doi:10.1016/S0895-4356(03)00044-1. [DOI] [PubMed] [Google Scholar]
- 13.Guyatt GH, Osoba D, Wu AW, Wyrwich KW, Norman GR. Methods to explain the clinical significance of health status measures. Mayo Clinic Proc. 2002;77:371–83. doi: 10.4065/77.4.371. doi:10.4065/77.4.371. [DOI] [PubMed] [Google Scholar]
- 14.Banks J, Breeze E, Lessof C, Nazroo J. Retirement, Health and Relationships of the Older Population in England: The 2004 English Longitudinal Study of Ageing (Wave 2) London: Institute for Fiscal Studies; 2006. [Google Scholar]
- 15.Hyde M, Wiggins RD, Higgs P, Blane DB. A measure of quality of life in early old age: the theory, development and properties of a needs satisfaction model (CASP-19) Aging Mental Health. 2003;7:186–94. doi: 10.1080/1360786031000101157. doi:10.1080/1360786031000101157. [DOI] [PubMed] [Google Scholar]
- 16.Wiggins RD, Netuveli G, Hyde M, Higgs P, Blane D. The evaluation of a self-enumerated scale of quality of life (CASP-19) in the context of research on ageing: a combination of exploratory and confirmatory approaches. Soc Indic Res. 2008;89:61–77. doi:10.1007/s11205-007-9220-5. [Google Scholar]
- 17.Radloff LS. The CES-D scale: a self-report depression scale for research in the general population. Appl Psych Meas. 1997;1:385–401. doi:10.1177/014662167700100306. [Google Scholar]
- 18.Guyatt GH, Norman GR, Juniper EF, Griffith LE. A critical look at transition ratings. J Clin Epidemiol. 2002;55:900–8. doi: 10.1016/s0895-4356(02)00435-3. doi:10.1016/S0895-4356(02)00435-3. [DOI] [PubMed] [Google Scholar]
- 19.McPhail S, Comans T, Haines T. Evidence of disagreement between patient perceived change and conventional longitudinal evaluation of change in health-related quality of life among older adults. Clin Rehabil. 2010;24:1036–44. doi: 10.1177/0269215510371422. doi:10.1177/0269215510371422. [DOI] [PubMed] [Google Scholar]
- 20.Netuveli G, Wiggins RD, Hildon Z, Montgomery SM, Blane D. Quality of life at older ages: evidence from the English longitudinal study of aging (wave 1) J Epidemiol Comm Health. 2006;60:357–63. doi: 10.1136/jech.2005.040071. doi:10.1136/jech.2005.040071. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Husted JA, Cook RJ, Farewell VT, Gladman DD. Methods for assessing responsiveness: a critical review and recommendations. J Clin Epidemiol. 2000;53:459–68. doi: 10.1016/s0895-4356(99)00206-1. doi:10.1016/S0895-4356(99)00206-1. [DOI] [PubMed] [Google Scholar]
- 22.StataCorp. Stata Statistical Software: Release 9. College Station, TX: Statacorp LP; 2005. [Google Scholar]
- 23.Blane DB, Netuveli G, Bartley M. Does quality of life at older ages vary with socio-economic position. Sociology. 2007;41:717–26. doi:10.1177/0038038507078927. [Google Scholar]
- 24.Brozek JL, Guyatt GH, Schunemann HJ. How a well grounded minimal important difference can enhance transparency of labelling claims and improve interpretation of patient reported outcome measure. Health Qual Life Outcomes. 2006;4:69. doi: 10.1186/1477-7525-4-69. doi:10.1186/1477-7525-4-69. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Revicki D, Hays RD, Cella D, Sloan J. Recommended methods for determining responsiveness and minimally important differences for patient-reported outcomes. J Clin Epidemiol. 2008;61:102–9. doi: 10.1016/j.jclinepi.2007.03.012. doi:10.1016/j.jclinepi.2007.03.012. [DOI] [PubMed] [Google Scholar]
- 26.Terwee CB, Dekker FW, Wiersinga WM, Prummel MF, Bossuyt PMM. On assessing responsiveness of health-related quality of life instruments: Guidelines for instrument evaluation. Qual Life Res. 2003;12:349–62. doi: 10.1023/a:1023499322593. doi:10.1023/A:1023499322593. [DOI] [PubMed] [Google Scholar]