# Validating patient-reported outcomes measures (PROMs) in adults with chronic kidney disease (CKD) at primary care setting

Validating patient-reported outcomes measures (PROMs) in adults with chronic kidney disease (CKD) at primary care setting

236

30 September 2016

01 November 2013

31 September 2016

34 Months

EQ-5D-5L, CAPABILTY-A, Quality of Life VAS, construct validity

- Dr Yayling Yang, Senior researcher in Health Economics, Nuffield Department of Primary Care Health Sciences, University of Oxford

- Professor Richard Hobbs, Professor of Primary Care Health Sciences Nuffield Department of Primary Care health Sciences, University of Oxford
- Professor Rafael Perera-Salazar. Oxfor Biomedical Research Centre, University of Oxford
- Daniel Lasserson, University of Oxford at time of project initiation)
- Dr Nathan Hill, University of Oxford at time of project initiation)
- Dr Ben Thompson, University of Oxford at time of project initiation)
- Mary Selwood, University of Oxford at time of project initiation)

- Professor Jose Valderas, Professor of Health Services and Policy Research, Medical School University of Exeter

## Project objectives

To validate three patient-reported outcome measures (EQ-5D-5L, ICECAP-A and QoL VAS )

among adults with chronic kidney disease in primary care.

## Brief summary

This project is a bolt-on to three ongoing CKD studies in the Nuffield Department of Primary Care Health Science at Oxford University. It was planned to be a 20-month project completed by the end of July 2015. There have been delays due to applications for ethical amendments for the studies. More substantially, all three studies encountered huge difficulties in recruitment. As a result, the FORM-2C study decided to reduce its sample size from 1200 to 750 and has completed baseline recruitment 2016. Unfortunately, recruitment is still ongoing for both OxRen and BARACK-D. Given the delays, a non-cost extension was applied and granted and data was provided by the research team until the end of September 2016. Only limited follow up data have been collected therefore longitudinal data analysis was postponed to the future to avoid potentially biased or misleading results. This report summarizes some preliminary results based on analysis of the baseline data from the three studies.

#### Methods

Using available baseline cross-sectional data from the three studies, statistical analyses were conducted to examine feasibility (rate of missing data), and the distribution of responses (e.g. the absence of ceiling or floor effects to measure full spectrum of a concept) of the three PROMs.

The EQ-5D utility index was calculated using the value algorithm and syntaxes for England (https://www.ohe.org/publications/valuing-health-related-quality-life-eq-5d-5l-value-set-england) recommended by the EuroQol group (http://www.euroqol.org/about-eq-5d/valuation-of-eq-5d/eq-5d-5l-value-sets.html). The capability index was calculated following recommendations of the ICECAP-A developers and used the syntaxes provided (http://www.birmingham.ac.uk/research/activity/mds/projects/HaPS/HE/ICECAP/ICECAP-A/index.aspx).

Construct validity of the overall indices was examined in two ways where possible. The extent to which measures correlate with each other were tested using correlation coefficients or regression analysis. The ‘known group’ method will be used to further investigate the ability of the measures to differentiate sub-groups which are expected to differ on the construct being measured. The respondents were grouped on the basis of their disease severities (e.g. defined by CKD stages) and on the basis of QoL VAS to indicate different levels and trend of well-being. The ability of EQ-5D-5L, EQ-VAS and ICECAP-A reflects the patterns in disease severity or well-being can be assessed using statistical tests (e.g t-test, ANOVA).

#### Results

*OxRen:*

In terms of feasibility, for each item of the EQ5D5L, the rate of missing data ranged from 2.3% (10/493) for the pain/discomfort dimension to 2.84% (14/493) for the mobility dimension and 3% for the EQVAS. The missing data was around 0.85% (4/473) for the 5 items of ICECAP-A, and 0 for qolVAS. This demonstrated good acceptance and feasibility as all PROMs given the low missing data.

The responses to 4 dimensions of EQ5D5L covered the full range of concept with no missing for the least and most severity levels, apart from the self-care dimension. The responses tent to skew to the less severity levels (i.e. no or slight problems) with 78% , 95%, 85%, 73% and 91% of responses allocated to the two least severe levels for the 5 EQ5D dimensions. On the other hand, only 3.4%, 0.2%, 1.6%, 4.0%, and 1.4% of responses allocated to the two most severe levels for the 5 EQ5D dimensions. This is similar for the ICECAP-A responses where majorities of responses demonstrated a good level of capabilities.

There were a total of 475 observations available for EQ5D5L index, 478 for EQ-VAS, 470 for ICECAP, and 422 for quality of life VAS (qolVAS). The mean EQ5D index was 0.856(SD 0.142, ranged from -0.0004 to 1). The mean EQ-VAS score was 79.780 (SD 14.983, ranged from 6 to 100). The mean ICECAP score was 0.901 (SD 0.113, ranged from 0.147 to 1). The mean qolVAS score was 77.509 (SD 21.028, ranged from 5 to 100).

The PROMs achieved moderate correlations with each other with the correlation coefficients were 0.53 for EQ5D and eqVAS scores, 0.53 for EQ5D and ICECAP scores, 0.56 for eqVAS and qolVAS, 0.43 for EQ5D and qolVAS, 0.43 for eqVAS and ICECAP, and 0.37 for qolVAS and ICECAP. The correlations are lower than expected particularly that both EQ5D5L and eqVAS are intended to measure health status, and both ICECAP and qolVAS are intended to measure broader quality of life.

*BARACK-D*

In terms of feasibility, for each item of the EQ5D5L, the rate of missing data was very low as to a maximum of 1.4% (13/950) for the pain/discomfort dimension. The missing data was around 1.5% (5/344) for the 5 items of ICECAP-A. Data missing rates were 2.3% (22/950) for eqVAS and 1.8% (6/329) for qolVAS. This demonstrated good acceptance and feasibility as all PROMs given the low missing data.

The responses to all five dimensions of EQ5D5L covered the full range of concept with no missing for the least and most severity levels. The responses tent to skew to the less severity levels (i.e. no or slight problems) with 70% , 95%, 80%, 73% and 69% of responses allocated to the two least severe levels for the 5 EQ5D dimensions. On the other hand, only 7.5%, 1.1%, 5.7%, 9.7%, and 0.8% of responses allocated to the two most severe levels for the 5 EQ5D dimensions. This is similar for the ICECAP-A responses where majorities of responses demonstrated a good level of capabilities with very few responses indicated limited capabilities.

In the dataset, There were a total of 1041 observations available for EQ5D5L index, 950for EQ-VAS, 339 for ICECAP, and 329 for quality of life VAS (qolVAS). The mean EQ5D index was 0.846(SD 0.187, ranged from -0.152 to 1). The mean EQ-VAS score was 75.76 (SD 17.72, ranged from 8 to 100). The mean ICECAP score was 0.887 (SD 0.133, ranged from 0.221 to 1). The mean qolVAS score was 72.68 (SD 25.22, ranged from 0 to 100).

The PROMs achieved moderate correlations with each other with the correlation coefficients were 0.61 for EQ5D and eqVAS scores, 0.61 for EQ5D and ICECAP scores, 0.67 for eqVAS and qolVAS, 0.48 for EQ5D and qolVAS, 0.59 for eqVAS and ICECAP, and 0.56 for qolVAS and ICECAP. The correlations are lower than expected especially between EQ5D5L and qolVAS.

A total of 977 respondents were grouped into CKD stages according to their eGFR values. The majority of them were in CKD stage 3A (n=364, 37%) and 3B (n=552, 57%) while a small number were in a more severe CKD stage 4(n=35, 3.6%) and a less severe CKD stage 2(n=26, 2.7%). Mean EQ5D values were 0.83 (SD 0.27), 0.84 (SD 0.19), 0.84 (SD 0.18), and 0.79 (SD 0.25) for CKD groups stage2, 3a, 3b and 4 group respectively. ANOVA analysis showed no significant differences of EQ5D values between CKD severity groups (p=0.43). Mean ICECAP values were 0.88 (SD 0.15), 0.87 (SD 0.13), 0.89 (SD 0.13) and 0.85 (SD 0.18) for the four CKD groups and no significant difference detected between the groups (p=0.90). Mean eqVAS scores were 77.2 (SD 17.4), 75.8 (SD 18.3), 76 (SD 17.3) and 74.2 for CKD groups, and no significant differences were found using ANOVA (p=0.93). Mean qolVAS scores were 73 (SD 16.9), 71.1 (SD 28.2), 74.2 (SD 23.1), 75 (SD 26.6), and no significant differences were found between CKD severity groups (p=0.78).

*FORM-2C*

In terms of feasibility, for each item of the EQ5D5L, the rate of missing data was very low as to a maximum of 0.82% (6/736) for the anxiety and depression dimension. The missing data was also very low as to the maximum of 0.41% (3/736) for the 5 items of ICECAP-A. Data missing rates were 0.54% (4/734) for eqVAS and 0.27% (2/734) for qolVAS. This demonstrated good acceptance and feasibility as all PROMs given the low missing data.

The responses to three dimensions of EQ5D5L covered the full range of concept with no missing for the least and most severity levels but responses to the other two dimensions missed the most severe level. The responses tent to extremely skew to the less severity levels (i.e. no or slight problems) with only 3.8%, 0.14%, 2.3%, 4.3% and 0.14% responses allocated to the two least severe levels for the 5 EQ5D dimensions. On the other hand, only 4.9%, 3.1%, 3.0%, 6.1%, and 4.5% of responses allocated to the two most severe levels for the 5 EQ5D dimensions. This is similar for the ICECAP-A responses where majorities of responses demonstrated a good level of capabilities with very few responses indicated limited capabilities.

In thedataset, There were a total of 749 observations available for EQ5D5L index, 730 for EQ-VAS, 749 for ICECAP, and 732 for quality of life VAS (qolVAS). The mean EQ5D index was 0.882 (SD 0.137, ranged from 0.108 to 1). The mean EQ-VAS score was 81.43 (SD 14.81, ranged from 20 to 100). The mean ICECAP score was 0.906 (SD 0.152, ranged from 0 to 1). The mean qolVAS score was 82.851 (SD 15.53, ranged from 3 to 100).

The majority of PROMs achieved moderate correlations with each other with the correlation coefficients were 0.61 for EQ5D and eqVAS scores, 0.67 for eqVAS and qolVAS, 0.515 for EQ5D and qolVAS, 0.50 for eqVAS and ICECAP, and 0.63 for qolVAS and ICECAP. However the correlation coefficient between EQ5D and ICECAP scores was 0.21 which is surprisingly low.

A total of 749 respondents were grouped into CKD stages according to their eGFR values. The majority of them were in CKD stage 1 (n=375, 50%) and 2 (n=293, 39%) while a small number were in a more severe CKD stage 3A (n=62, 8.3%) and 3B (n=19, 2.5%). Mean EQ5D values were 0.88 (SD 0.15), 0.89 (SD 0.12), 0.88 (SD 0.12), and 0.81 (SD 0.16) for CKD groups stage1, 2, 3a and 3b group respectively. ANOVA analysis showed significant differences of EQ5D values between CKD severity groups (p=0.049). Mean ICECAP values were 0.90 (SD 0.17), 0.91 (SD 0.13), 0.90 (SD 0.14) AND 0.88 (SD 0.12) for the four CKD groups and no significant difference detected between the groups (p=0.60). Mean eqVAS scores were 80.6 (SD 15.5), 83.2 (SD 13.5), 79.4 (SD 16.5) and 77.1 (SD 11.3) for CKD groups, and significant differences were found using ANOVA (p=0.046). Mean qolVAS scores were 82.3 (SD 15.8), 84.2 (SD 14.9), 81.0 (SD 16.2), 78.6 (SD 16.4), and no significant differences were found between CKD severity groups (p=0.16).

#### Conclusions

The three PROMs (i.e. EQ5D5L, ICECAP-A and quality of life VAS) are feasible for use among the CKD patients. The correlations between the measures are moderate which reflects different but related concepts they intend to capture: health, capacity and quality of life. The measures failed to differentiate CKD groups defined by clinical biomarkers.

## Plain English summary

**Background**: There has been increasing interest to use Patient-reported outcome measures (PROMs) in both clinical research and daily clinical practice after sufficient development and validation. Research is required to further validate PROMs for use among adults with Chronic Kidney Disease (CKD).

**Aims and objectives**: This project aims to validate three PROMs - EQ-5D-5L, ICECAP-A (ICEpop CAPability measure for Adults), and QoL VAS (Quality of Life Visual Analogue Scale) for use among individuals with CKD at primary care. Data collection: The study will obtain PROMs data from three major studies – BARACK D (RCT), OxRen and FORM-2C (observational cohort studies). **Data analysis**: Statistical analyses will be conducted to assess psychometric performance of the three PROMS of interests using available baseline data. The criteria include feasibility and acceptability (e.g. rate of missing data), precise (e.g. absence of ceiling or floor effects), construct validity (e.g. convergent validity and known group test).

**Results and conclusion**: The three PROMs (i.e. EQ5D5L, ICECAP-A and quality of life VAS) are feasible for use among the CKD patients. The correlations between the measures are moderate which reflects different but related concepts they intend to capture: health, capacity and quality of life. The measures failed to differentiate CKD groups defined by clinical biomarkers.

## Dissemination

## Public involvement

None

## Impact

This research will provide evidence for health states and quality of life for the CKD population, and inform chose of PROMs for use in the CKD population in the future.

This project was funded by the National Institute for Health Research School for Primary Care Research (project number 236 )

### Department of Health Disclaimer

The views and opinions expressed therein are those of the authors and do not necessarily reflect those of the NIHR School for Primary Care Research, NIHR, NHS or the Department of Health.