https://orcid.org/0000-0002-3390-9590
Abstract
Physical activity (PA) levels are low in patients with advanced chronic kidney disease (CKD), and associate with increased morbidity and mortality. Reliable tools to assess PA in CKD are scarce. We aimed to develop and validate a novel PA questionnaire for use in CKD (CKD-PAQ).
In Phase 1, a prototype questionnaire was developed based on the validated recent PAQ (RPAQ). Structured feedback on item relevance and clarity was obtained from 40 CKD patients. In Phase 2, the questionnaire was refined in three iterations in a total of 226 CKD patients against 7-day accelerometer and RPAQ measurements. In Phase 3, the definitive CKD-PAQ was compared with RPAQ in 523 CKD patients.
In the final iteration of Phase 2, CKD-PAQ data were compared with accelerometer-derived and RPAQ data in 60 patients. Mean daily metabolic equivalent of task (MET) and total energy expenditure (TEE) levels were similar by all methods. Intraclass correlation coefficients showed fair (MET) and good (TEE) agreement between accelerometry and both CKD-PAQ and RPAQ. Agreement between questionnaires was excellent. The mean [standard deviation (SD)] daily MET bias was 0.035 (0.312) for CKD-PAQ and 0.018 (0.326) for RPAQ. The mean (SD) TEE bias was 91 (518) for CKD-PAQ and 44 (548) kcal for RPAQ. Limits of agreement (LOA) were wide for both parameters, with less dispersion of CKD-PAQ values. In Phase 3, agreement between questionnaires was good (MET) and excellent (TEE). Bias of CKD-PAQ-derived mean (SD) daily MET from RPAQ-derived values was 0.031 (0.193), with 95% LOA −0.346 to 0.409. Corresponding mean (SD) values for TEE were 48 (325) and −588 to 685 kcal/day. CKD-PAQ appeared to improve discrimination between low activity groups.
CKD-PAQ performs comparably to the RPAQ though it is shorter, easier to complete, and may better capture low-level activity and improve discrimination between low activity groups.
KEY LEARNING POINTS
What is already known about the subject?
physical activity (PA) levels are substantially low in chronic kidney disease (CKD) patients. Reduced physical functioning capacity and increased fatigue levels contribute to low PA levels in these patients. Estimating PA is vital to nutritional management of CKD patients;
except for the recently developed Low Physical Activity Questionnaire (LoPAQ), none of the existing PA questionnaires (PAQ) was developed specifically in individuals with CKD. However, LoPAQ was developed only in dialysis patients and did not include non-dialysis CKD and transplant patients; and
a disease-specific PA measurement tool will increase accuracy and applicability of such tools for nutritional and dialysis management.
What this study adds?
this is the first study to develop a CKD-specific PA measurement tool across the spectrum of patients with CKD;
the CKD-PAQ questionnaire performs comparably to, if not slightly better than, the existing validated questionnaires, but has some advantages over the existing ones; and
this novel questionnaire is shorter when compared with the existing questionnaires, easier for patients to complete, focuses on activities commonly performed by this patient population and may better capture low-level activity commonly prevalent in CKD patients.
What impact this may have on practice or policy?
when validated in other larger cohorts, this novel CKD-PAQ questionnaire can be easily used in every day clinical practice for nutritional management of CKD patients; and
total energy expenditure estimated from the novel questionnaire can also be used in conjunction with other clinical data for appropriate dialysis management.
INTRODUCTION
Physical activity (PA) levels have been shown to be low in individuals with chronic kidney disease (CKD) and in those receiving kidney replacement therapy [1–4]. There are clear benefits, including improved survival and quality of life, in patients with higher PA levels [5–7]. Successful implementation of any programme aiming to encourage PA in this patient population depends on reliable tools to assess PA consistently.
The use of doubly labelled water or accelerometers for PA measurement is not feasible in routine clinical practice. Self-report activity questionnaires are a practical alternative. None of the PA questionnaires (PAQ) used in studies involving patients with CKD has been derived from this patient population. Most of these questionnaires are derived from young healthy adults and as such may not be applicable to specific patient groups. CKD is predominantly a disease of the elderly and these CKD questionnaires may not be valid in this patient population. A study examining the validity of 10 PAQ in elderly individuals in general population against doubly labelled water found that few questionnaires were reliable for use in the elderly [8]. Moreover, the individual variability was high for all the questionnaires, which limits their use in these individuals.
Most of the existing questionnaires focus on moderate to vigorous PA and were not designed for studying PA levels in populations with low-level PA. There is evidence to suggest that PA levels in dialysis patients are lower than healthy age-matched controls with no regular PA [4]. There is a dearth of reliable tools to measure PA level in populations such as CKD patients, who are predominantly elderly with low-level PA.
Recent PAQ (RPAQ) has been validated in individuals with CKD using doubly labelled water measurements [9]. However, it showed that the questionnaire was not reliable in capturing low intensity and sedentary activities. This reinforces the need for developing a novel PAQ for better measurement of PA in CKD.
Our aim in this study was to develop and validate a novel PAQ specifically designed for individuals with CKD (CKD-PAQ) using accelerometer-derived PA measurements.
MATERIALS AND METHODS
Ethical review
The study was approved by the National Research Ethics service. All subjects gave written informed consent to take part.
Subjects
Patients aged >18 years with CKD Stages 1–5 including those receiving dialysis and those with functioning kidney transplant were recruited. Those who, in the judgement of the clinical team, had insufficient capacity or insufficient understanding of English to allow valid consent, were not approached for inclusion in the study by the study team.
Study protocol
The study was carried out in three phases: (i) an initial qualitative phase consisting of structured patients interviews, (ii) a development phase in which the questionnaire was modified sequentially to improve reliability and accuracy of energy expenditure estimation in comparison with accelerometer estimates and (iii) a final phase to compare energy estimates from the novel questionnaire against existing validated PAQ. A flowchart depicting the study design is shown in Figure 1. There was no overlap of study participants across the different phases of the study.
Flowchart depicting the study design.
Development of the questionnaire
A novel PAQ (CKD-PAQ) was developed based on the RPAQ. In the initial phase, 40 patients with CKD including dialysis and transplant patients were recruited to complete the first prototype questionnaire. Structured feedback was obtained through one-to-one interviews with each of the participant focusing on the clarity of questionnaire items, ease of completion and on the breadth of activities captured by the questionnaire. This feedback was then used to develop the first iteration of the questionnaire to be tested against accelerometer measured PA.
The second phase of the development of the questionnaire was conducted through three stages. The questionnaire was iterated at the end of each of the first two stages to improve capture of different levels of activity compared with measured PA from an accelerometer. The questionnaire items were modified to achieve this objective in the first two stages. The initial two versions of the questionnaire included an exhaustive list of leisure and work activities. However, on review of participant responses and the contribution of some of the activities to the final model for energy expenditure estimation compared with the accelerometer measures, some of the activities were removed from subsequent iterations. Some of the questionnaire items were also modified to improve clarity and for ease of analysis. The final version of the questionnaire (Supplementary Materials) at the end of the third stage was then employed in the final phase in a different cohort of patients for comparison of energy estimates against the validated RPAQ questionnaire.
CKD-PAQ and RPAQ both enquire regarding activities performed at home, at work and recreational activities over the preceding 4 weeks. However, CKD-PAQ is much shorter (27 items) compared with RPAQ (55 items). For haemodialysis (HD) patients, CKD-PAQ has an additional five items to collect information regarding their dialysis sessions. CKD-PAQ focuses on simple range of recreational activities compared with RPAQ, which contains a comprehensive list of high-intensity activities that are not commonly carried out by individuals with CKD.
Data collection
The following data were collected on all participants.
Demographic and anthropometric data including height and weight, and residence postcode. The English Index of Multiple Deprivation was calculated using the participants’ postcode.
Comorbidity data, which was used to calculate Charlson comorbidity index (CCI).
PA assessment using questionnaires and accelerometer.
Measurement of PA
PA was measured using a wrist-worn accelerometer (GT9X Link, ActiGraph LLC, FL, USA). Participants were advised to wear the accelerometer on the non-dominant wrist for 24 h a day for seven consecutive days. At the end of the measurement period, the accelerometer data were retrieved through the ActiLife software for analysis. The data included total vector magnitude counts, steps per minute and mean daily metabolic equivalent of task (MET) among other raw movement-related measures. The rate of energy spent during any PA is expressed as MET value. One MET is the energy spent sitting at rest and is approximately equal to 1 kcal/h/kg of body weight. The daily MET from the measured activity was used to calculate total energy expenditure (TEE) as described below.
PA assessment
At the end of the 7-day study period, subjects completed two PAQ—RPAQ and CKD-PAQ. RPAQ is a validated questionnaire that enquires about various activities performed at home, work and at leisure time, and the time spent in each of those activities over the preceding 4 weeks [10]. RPAQ has been validated in CKD patients for energy expenditure estimation using doubly labelled water [9].
Estimation of TEE
Accelerometery
PA questionnaires
The unaccounted time from the questionnaire was assigned a MET of 1.3 as previously published [9]. TEE was then calculated using the relationship depicted in Equation (1). Energy expenditure estimation from CKD-PAQ was carried out in the same manner as that used for RPAQ.
Statistical analysis
Statistical analysis was carried out using SPSS® version 26 (SPSS Software, IBM Corporation, Armonk, NY, USA) and PRISM 9 (Graphpad Software LLC). Based on previous data on correlation of energy estimation from RPAQ and measured TEE, a sample size of 40 in each phase was considered to be sufficient to establish significant intragroup correlations, assuming α = 0.05 and for power of 0.8. A sample size of 400 for the final phase was considered to provide sufficient power. Normally distributed data are presented as mean ± standard deviation (SD) and non-normally distributed data as median (interquartile range). The significance of differences between means was determined using Student’s t-test and of medians by Mann–Whitney U-test. For Phase 2 data, comparison was made between MET and TEE values derived from accelerometry (METACC and TEEACC) and those derived from RPAQ (METRPAQ and TEERPAQ) and CKD-PAQ (METCKD and TEECKD) by calculating the relevant intraclass correlation coefficient (ICC) and by Bland–Altman analysis. For Phase 3 data, similar comparisons were made between the questionnaire-derived parameters. A P < 0.05 was considered significant.
RESULTS
A total of 266 patients were recruited in the development phases, 40 in the initial qualitative phase and 226 in the remaining three stages of the development. The number of participants in each stage was 89, 77 and 60, respectively. The results presented are from the final iteration involving 60 patients. This version was used for the final phase involving 523 patients, 394 of whom completed both questionnaires. Demographic and biochemical characteristics for both cohorts are shown in Table 1.
Demographic characteristics of study participants
| Parameters | Development phase (n = 60) | Final phase (n = 523) |
|---|---|---|
| Age, years | 58.3 ± 15.1 | 60.8 ± 16.1 |
| Males, % | 57.6 | 63.7 |
| Weight, kg | 86.6 ± 22.1 | 76.5 ± 18.9 |
| Height, cm | 169.6 ± 9.5 | 168.2 ± 10.6 |
| BSA, m2 | 2.03 ± 0.3 | 1.89 ± 0.27 |
| BMI, kg/m2 | 30.0 ± 6.7 | 27.0 ± 6.2 |
| CCI | 4.2 ± 2.5 | 4.9 ± 2.4 |
| REE, kcal/day | 1706 ± 283 | 1577 ± 258 |
| CKD | 20 | 24 |
| ICHD | 24 | 436 |
| HHD | 1 | 20 |
| PD | 0 | 9 |
| Transplant | 15 | 34 |
| Parameters | Development phase (n = 60) | Final phase (n = 523) |
|---|---|---|
| Age, years | 58.3 ± 15.1 | 60.8 ± 16.1 |
| Males, % | 57.6 | 63.7 |
| Weight, kg | 86.6 ± 22.1 | 76.5 ± 18.9 |
| Height, cm | 169.6 ± 9.5 | 168.2 ± 10.6 |
| BSA, m2 | 2.03 ± 0.3 | 1.89 ± 0.27 |
| BMI, kg/m2 | 30.0 ± 6.7 | 27.0 ± 6.2 |
| CCI | 4.2 ± 2.5 | 4.9 ± 2.4 |
| REE, kcal/day | 1706 ± 283 | 1577 ± 258 |
| CKD | 20 | 24 |
| ICHD | 24 | 436 |
| HHD | 1 | 20 |
| PD | 0 | 9 |
| Transplant | 15 | 34 |
Data are presented as mean ± SD, % or n. BMI, body mass index; PD, peritoneal dialysis.
Demographic characteristics of study participants
| Parameters | Development phase (n = 60) | Final phase (n = 523) |
|---|---|---|
| Age, years | 58.3 ± 15.1 | 60.8 ± 16.1 |
| Males, % | 57.6 | 63.7 |
| Weight, kg | 86.6 ± 22.1 | 76.5 ± 18.9 |
| Height, cm | 169.6 ± 9.5 | 168.2 ± 10.6 |
| BSA, m2 | 2.03 ± 0.3 | 1.89 ± 0.27 |
| BMI, kg/m2 | 30.0 ± 6.7 | 27.0 ± 6.2 |
| CCI | 4.2 ± 2.5 | 4.9 ± 2.4 |
| REE, kcal/day | 1706 ± 283 | 1577 ± 258 |
| CKD | 20 | 24 |
| ICHD | 24 | 436 |
| HHD | 1 | 20 |
| PD | 0 | 9 |
| Transplant | 15 | 34 |
| Parameters | Development phase (n = 60) | Final phase (n = 523) |
|---|---|---|
| Age, years | 58.3 ± 15.1 | 60.8 ± 16.1 |
| Males, % | 57.6 | 63.7 |
| Weight, kg | 86.6 ± 22.1 | 76.5 ± 18.9 |
| Height, cm | 169.6 ± 9.5 | 168.2 ± 10.6 |
| BSA, m2 | 2.03 ± 0.3 | 1.89 ± 0.27 |
| BMI, kg/m2 | 30.0 ± 6.7 | 27.0 ± 6.2 |
| CCI | 4.2 ± 2.5 | 4.9 ± 2.4 |
| REE, kcal/day | 1706 ± 283 | 1577 ± 258 |
| CKD | 20 | 24 |
| ICHD | 24 | 436 |
| HHD | 1 | 20 |
| PD | 0 | 9 |
| Transplant | 15 | 34 |
Data are presented as mean ± SD, % or n. BMI, body mass index; PD, peritoneal dialysis.
Development phase
Median values for METACC, METRPAQ and METCKD were similar [1.35 (0.26), 1.26 (0.27) and 1.31 (0.33), respectively]. There were no significant differences between METACC and either METRPAQ (P = 0.08) or METCKD (P = 0.084), nor between METRPAQ and METCKD (P = 0.287). Likewise, mean values for TEEACC, TEERPAQ and TEECKD were similar (2379 ± 630, 2413 ± 873 and 2361 ± 827 kcal, respectively), and there were no differences between TEEACC and either TEERPAQ (P = 0.561) or TEECKD (P = 0.203), nor between TEERPAQ and TEECKD (P = 0.598).
There was fair agreement between METACC and both METRPAQ [ICC = 0.441 (0.031–0.677); P = 0.019)] and METCKD [ICC = 0.455 (0.059–0.685); P = 0.015] and excellent agreement between METRPAQ and METCKD [ICC = 0.905 (0.836–0.944); P < 0.001]. Agreement was good between TEEACC and both TEERPAQ [ICC = 0.789 (0.636–0.878); P < 0.001] and TEECKD [ICC = 0.751 (0.572–0.855); P < 0.001] and excellent between TEERPAQ and TEECKD [ICC = 0.917 (0.857–0.951); P < 0.001].
Table 2 shows the results of Bland–Altman analysis for comparisons of mean daily MET and TEE from questionnaires and from accelerometry. Bias for both parameters was small and slightly lower for RPAQ-derived parameters. However, both the SD of the bias and the 95% limits of agreement (LOA) showed less dispersion for CKD-PAQ than for RPAQ. Figure 2 shows the Bland–Altman plot of TEE derived from accelerometry (TEEACC) and that from CKD-PAQ (TEECKD). Bland–Altman comparisons of METCKD and METRPAQ, and TEECKD and TEERPAQ showed minimal bias and even less dispersion (Table 2).
Bland–Altman plot showing bias and LOA between TEEACC and TEECKD. Difference between TEE measured by accelerometer and CKD-PAQ plotted against the mean of the two measurements. A negative sign indicates an overestimation and a positive sign indicates an underestimation by the questionnaire.
Bland–Altman Comparisons between PA measures derived from accelerometry and CKD-PAQ and RPAQ questionnaires
| Parameters | CKD-PAQ derived | RPAQ derived |
|---|---|---|
| Bland–Altman comparisons with accelerometer | ||
| Mean daily MET | ||
| Bias | 0.035 | 0.018 |
| SD bias | 0.312 | 0.326 |
| 95% LOA | −0.646 to 0.577 | −0.656 to 0.621 |
| TEE | ||
| Bias | 91 | 44 |
| SD bias | 518 | 548 |
| 95% LOA | −925 to 1108 | −1030 to 1117 |
| Bland–Altman comparisons with RPAQ | ||
| Mean daily MET | ||
| Bias | 0.008 | |
| SD bias | 0.265 | |
| 95% LOA | −0.512 to 0.527 | |
| TEE | ||
| Bias | 34 | |
| SD bias | 481 | |
| 95% LOA | −909 to 978 | |
| Parameters | CKD-PAQ derived | RPAQ derived |
|---|---|---|
| Bland–Altman comparisons with accelerometer | ||
| Mean daily MET | ||
| Bias | 0.035 | 0.018 |
| SD bias | 0.312 | 0.326 |
| 95% LOA | −0.646 to 0.577 | −0.656 to 0.621 |
| TEE | ||
| Bias | 91 | 44 |
| SD bias | 518 | 548 |
| 95% LOA | −925 to 1108 | −1030 to 1117 |
| Bland–Altman comparisons with RPAQ | ||
| Mean daily MET | ||
| Bias | 0.008 | |
| SD bias | 0.265 | |
| 95% LOA | −0.512 to 0.527 | |
| TEE | ||
| Bias | 34 | |
| SD bias | 481 | |
| 95% LOA | −909 to 978 | |
Bland–Altman Comparisons between PA measures derived from accelerometry and CKD-PAQ and RPAQ questionnaires
| Parameters | CKD-PAQ derived | RPAQ derived |
|---|---|---|
| Bland–Altman comparisons with accelerometer | ||
| Mean daily MET | ||
| Bias | 0.035 | 0.018 |
| SD bias | 0.312 | 0.326 |
| 95% LOA | −0.646 to 0.577 | −0.656 to 0.621 |
| TEE | ||
| Bias | 91 | 44 |
| SD bias | 518 | 548 |
| 95% LOA | −925 to 1108 | −1030 to 1117 |
| Bland–Altman comparisons with RPAQ | ||
| Mean daily MET | ||
| Bias | 0.008 | |
| SD bias | 0.265 | |
| 95% LOA | −0.512 to 0.527 | |
| TEE | ||
| Bias | 34 | |
| SD bias | 481 | |
| 95% LOA | −909 to 978 | |
| Parameters | CKD-PAQ derived | RPAQ derived |
|---|---|---|
| Bland–Altman comparisons with accelerometer | ||
| Mean daily MET | ||
| Bias | 0.035 | 0.018 |
| SD bias | 0.312 | 0.326 |
| 95% LOA | −0.646 to 0.577 | −0.656 to 0.621 |
| TEE | ||
| Bias | 91 | 44 |
| SD bias | 518 | 548 |
| 95% LOA | −925 to 1108 | −1030 to 1117 |
| Bland–Altman comparisons with RPAQ | ||
| Mean daily MET | ||
| Bias | 0.008 | |
| SD bias | 0.265 | |
| 95% LOA | −0.512 to 0.527 | |
| TEE | ||
| Bias | 34 | |
| SD bias | 481 | |
| 95% LOA | −909 to 978 | |
The relationship between TEEACC and tertiles of TEECKD and TEERPAQ, respectively, are shown in Figure 3. There was a significant difference in TEEACC levels between both middle and upper TEECKD tertiles compared with the lowest tertile. For TEERPAQ, the only significant difference in TEEACC was between the lowest and highest TEERPAQ tertile.
A plot of accelerometer measured TEE against (A) CKD-PAQ TEE tertiles and (B) RPAQ TEE tertiles.
In a multivariable regression model of TEEACC (Table 3), METCKD was a significant predictor after adjustment for age, sex, body surface area (BSA) and comorbidity (adjusted R2 = 0.719). Substituting METRPAQ for METCKD in the model gave similar results (adjusted R2 = 0.703).
Multivariate linear regression model of independent predictors of accelerometer-measured TEE
| Variables | Unstandardized coefficients | Standardized coefficients | t | P-value | |
|---|---|---|---|---|---|
| B | Std error | Beta | |||
| (Constant) | −1048.66 | 387.63 | – | −2.705 | 0.009 |
| Age ≥65 years | −453.37 | 96.79 | −0.354 | −4.684 | <0.001 |
| Sex | −207.34 | 98.77 | −0.164 | −2.099 | 0.041 |
| BSA (m2) | 1708.45 | 181.05 | 0.761 | 9.436 | <0.001 |
| Log METCKD | 909.73 | 258.59 | 0.266 | 3.518 | 0.001 |
| Variables | Unstandardized coefficients | Standardized coefficients | t | P-value | |
|---|---|---|---|---|---|
| B | Std error | Beta | |||
| (Constant) | −1048.66 | 387.63 | – | −2.705 | 0.009 |
| Age ≥65 years | −453.37 | 96.79 | −0.354 | −4.684 | <0.001 |
| Sex | −207.34 | 98.77 | −0.164 | −2.099 | 0.041 |
| BSA (m2) | 1708.45 | 181.05 | 0.761 | 9.436 | <0.001 |
| Log METCKD | 909.73 | 258.59 | 0.266 | 3.518 | 0.001 |
Adjusted R2 = 0.719. METCKD, mean daily MET from CKD-PAQ.
Multivariate linear regression model of independent predictors of accelerometer-measured TEE
| Variables | Unstandardized coefficients | Standardized coefficients | t | P-value | |
|---|---|---|---|---|---|
| B | Std error | Beta | |||
| (Constant) | −1048.66 | 387.63 | – | −2.705 | 0.009 |
| Age ≥65 years | −453.37 | 96.79 | −0.354 | −4.684 | <0.001 |
| Sex | −207.34 | 98.77 | −0.164 | −2.099 | 0.041 |
| BSA (m2) | 1708.45 | 181.05 | 0.761 | 9.436 | <0.001 |
| Log METCKD | 909.73 | 258.59 | 0.266 | 3.518 | 0.001 |
| Variables | Unstandardized coefficients | Standardized coefficients | t | P-value | |
|---|---|---|---|---|---|
| B | Std error | Beta | |||
| (Constant) | −1048.66 | 387.63 | – | −2.705 | 0.009 |
| Age ≥65 years | −453.37 | 96.79 | −0.354 | −4.684 | <0.001 |
| Sex | −207.34 | 98.77 | −0.164 | −2.099 | 0.041 |
| BSA (m2) | 1708.45 | 181.05 | 0.761 | 9.436 | <0.001 |
| Log METCKD | 909.73 | 258.59 | 0.266 | 3.518 | 0.001 |
Adjusted R2 = 0.719. METCKD, mean daily MET from CKD-PAQ.
Final phase
The final version of CKD-PAQ questionnaire was compared against RPAQ in a larger cohort of 523 CKD patients (Table 1). Median unaccounted time was lower with CKD-PAQ than RPAQ (1.0 versus 9.8 h; P < 0.001). Mean METCKD was slightly lower than mean METRPAQ (1.24 ± 0.28 versus 1.27 ± 0.23; P = 0.001). ICC for the comparison showed good agreement [0.839 (0.802–0.868); P < 0.001]. Bias of METCKD from METRPAQ was 0.031 (SD 0.193) with 95% LOA −0.346 to 0.409. Mean TEECKD was lower than TEERPAQ (1964 ± 643 versus 2012 ± 580 kcal/day; P < 0.001). ICC for the comparison showed excellent agreement [0.923 (0.905–0.937); P < 0.001]. Bias was 48 (SD 325) with LOA −588 to 685 kcal/day.
Women had a lower METRPAQ (1.23 versus 1.29; P = 0.001), but not METCKD, than men. Younger patients (<65 years) had higher TEECKD (2183 versus 1689 kcal/day; P < 0.001) and TEERPAQ (2238 versus 1738 kcal/day; P < 0.001) than older counterparts. CCI correlated negatively with TEECKD (r = −0.269, P < 0.001) but not TEERPAQ. Both METCKD (ρ = 0.232, P < 0.001) and METRPAQ RPAQ (ρ = 0.180, P < 0.001) correlated with deprivation index and both TEECKD (1864 versus 2046 kcal/day; P = 0.002) and TEERPAQ (1921 versus 2091 kcal/day; P = 0.002) were lower in participants living in the most deprived areas (deprivation index < median).
There were significant differences in both mean daily MET and TEE levels between modality groups (Table 4). Compared with in-centre HD (ICHD) patients both these parameters from both questionnaires were higher in CKD and transplant patients. CKD-PAQ-derived, but not RPAQ-derived, levels of both parameters were higher in home than in ICHD patients—this was significant for TEE. Transplant patients were younger and had lower Charlson scores than ICHD patients. CKD patients had lower Charlson scores and were less deprived than ICHD patients. Home HD (HHD) patients were also less deprived (Supplementary data, Table S1).
Mean daily MET and TEE derived from CKD-PAQ and RPAQ questionnaires in different modalities
| Parameters | Mean daily MET | TEE | ||||
|---|---|---|---|---|---|---|
| N | Mean ± SD | P-value | N | Mean ± SD | P-value | |
| CKD-PAQ | ||||||
| ICHD | 384 | 1.18 ± 0.23 | Reference | 383 | 1860 ± 514 | Reference |
| PD | 9 | 1.18 ± 0.15 | 1.00 | 9 | 1774 ± 357 | 1.00 |
| HHD | 19 | 1.35 ± 0.27 | 0.079 | 19 | 2299 ± 648 | 0.014 |
| Transplant | 33 | 1.48 ± 0.28 | <0.001 | 32 | 2535 ± 674 | <0.001 |
| CKD | 24 | 1.46 ± 0.56 | <0.001 | 16 | 2691 ± 1422 | <0.001 |
| RPAQ | ||||||
| ICHD | 360 | 1.24 ± 0.20 | Reference | 359 | 1940 ± 491 | Reference |
| PD | 9 | 1.21 ± 0.10 | 1.00 | 9 | 1823 ± 335 | 1.00 |
| HHD | 18 | 1.30 ± 0.20 | 1.00 | 18 | 2152 ± 540 | 1.00 |
| Transplant | 30 | 1.44 ± 0.21 | <0.001 | 29 | 2494 ± 565 | <0.001 |
| CKD | 22 | 1.46 ± 0.57 | <0.001 | 14 | 2657 ± 1490 | <0.001 |
| Parameters | Mean daily MET | TEE | ||||
|---|---|---|---|---|---|---|
| N | Mean ± SD | P-value | N | Mean ± SD | P-value | |
| CKD-PAQ | ||||||
| ICHD | 384 | 1.18 ± 0.23 | Reference | 383 | 1860 ± 514 | Reference |
| PD | 9 | 1.18 ± 0.15 | 1.00 | 9 | 1774 ± 357 | 1.00 |
| HHD | 19 | 1.35 ± 0.27 | 0.079 | 19 | 2299 ± 648 | 0.014 |
| Transplant | 33 | 1.48 ± 0.28 | <0.001 | 32 | 2535 ± 674 | <0.001 |
| CKD | 24 | 1.46 ± 0.56 | <0.001 | 16 | 2691 ± 1422 | <0.001 |
| RPAQ | ||||||
| ICHD | 360 | 1.24 ± 0.20 | Reference | 359 | 1940 ± 491 | Reference |
| PD | 9 | 1.21 ± 0.10 | 1.00 | 9 | 1823 ± 335 | 1.00 |
| HHD | 18 | 1.30 ± 0.20 | 1.00 | 18 | 2152 ± 540 | 1.00 |
| Transplant | 30 | 1.44 ± 0.21 | <0.001 | 29 | 2494 ± 565 | <0.001 |
| CKD | 22 | 1.46 ± 0.57 | <0.001 | 14 | 2657 ± 1490 | <0.001 |
Transplant, patients with functioning kidney transplant, patients with Stages 3–5 CKD. P-values indicate significance of differences of mean values of parameters in other modalities from mean levels in patients receiving ICHD using one-way analysis of variance with Bonferroni method of post hoc testing.
Mean daily MET and TEE derived from CKD-PAQ and RPAQ questionnaires in different modalities
| Parameters | Mean daily MET | TEE | ||||
|---|---|---|---|---|---|---|
| N | Mean ± SD | P-value | N | Mean ± SD | P-value | |
| CKD-PAQ | ||||||
| ICHD | 384 | 1.18 ± 0.23 | Reference | 383 | 1860 ± 514 | Reference |
| PD | 9 | 1.18 ± 0.15 | 1.00 | 9 | 1774 ± 357 | 1.00 |
| HHD | 19 | 1.35 ± 0.27 | 0.079 | 19 | 2299 ± 648 | 0.014 |
| Transplant | 33 | 1.48 ± 0.28 | <0.001 | 32 | 2535 ± 674 | <0.001 |
| CKD | 24 | 1.46 ± 0.56 | <0.001 | 16 | 2691 ± 1422 | <0.001 |
| RPAQ | ||||||
| ICHD | 360 | 1.24 ± 0.20 | Reference | 359 | 1940 ± 491 | Reference |
| PD | 9 | 1.21 ± 0.10 | 1.00 | 9 | 1823 ± 335 | 1.00 |
| HHD | 18 | 1.30 ± 0.20 | 1.00 | 18 | 2152 ± 540 | 1.00 |
| Transplant | 30 | 1.44 ± 0.21 | <0.001 | 29 | 2494 ± 565 | <0.001 |
| CKD | 22 | 1.46 ± 0.57 | <0.001 | 14 | 2657 ± 1490 | <0.001 |
| Parameters | Mean daily MET | TEE | ||||
|---|---|---|---|---|---|---|
| N | Mean ± SD | P-value | N | Mean ± SD | P-value | |
| CKD-PAQ | ||||||
| ICHD | 384 | 1.18 ± 0.23 | Reference | 383 | 1860 ± 514 | Reference |
| PD | 9 | 1.18 ± 0.15 | 1.00 | 9 | 1774 ± 357 | 1.00 |
| HHD | 19 | 1.35 ± 0.27 | 0.079 | 19 | 2299 ± 648 | 0.014 |
| Transplant | 33 | 1.48 ± 0.28 | <0.001 | 32 | 2535 ± 674 | <0.001 |
| CKD | 24 | 1.46 ± 0.56 | <0.001 | 16 | 2691 ± 1422 | <0.001 |
| RPAQ | ||||||
| ICHD | 360 | 1.24 ± 0.20 | Reference | 359 | 1940 ± 491 | Reference |
| PD | 9 | 1.21 ± 0.10 | 1.00 | 9 | 1823 ± 335 | 1.00 |
| HHD | 18 | 1.30 ± 0.20 | 1.00 | 18 | 2152 ± 540 | 1.00 |
| Transplant | 30 | 1.44 ± 0.21 | <0.001 | 29 | 2494 ± 565 | <0.001 |
| CKD | 22 | 1.46 ± 0.57 | <0.001 | 14 | 2657 ± 1490 | <0.001 |
Transplant, patients with functioning kidney transplant, patients with Stages 3–5 CKD. P-values indicate significance of differences of mean values of parameters in other modalities from mean levels in patients receiving ICHD using one-way analysis of variance with Bonferroni method of post hoc testing.
DISCUSSION
The primary aim of this study was to develop and validate a novel PAQ specifically in individuals with CKD. The study showed that the energy estimates from the novel questionnaire, CKD-PAQ, provided acceptable estimations of accelerometer-based parameters and performed similarly to the existing validated RPAQ questionnaire. The novel CKD-PAQ questionnaire also has the advantage of being substantially shorter and simpler to complete than RPAQ.
Routine use of accelerometers is not practical in day-to-day clinical practice. PAQ are useful alternatives for estimation of PA and energy expenditure. Developing a PAQ in a CKD population poses some challenges. Patients with CKD are more likely to be elderly and have higher comorbidity, and hence questionnaires developed in younger people may not be reliable when used in a CKD population [12, 13]. As CKD patients are more likely to have low levels of PA, it is vital that any disease-specific PAQ is able to capture low-intensity activity to enable accurate assessment. Besides a recently developed Low PAQ (LoPAQ) questionnaire in dialysis patients [14, 15], none of the existing PAQ has been developed in a CKD patient population.
Analysis of standard correlations is not an appropriate method to assess the agreement between methods [16–18]. In the first instance, we calculated ICC levels. These showed fair agreement between mean daily MET from both CKD-PAQ and from RPAQ with accelerometer measures. For TEE, agreement in both cases was good. Agreement between CKD-PAQ- and RPAQ-derived parameters was excellent in Phase 2 and performed similarly in the Phase 3 cohort. CKD-PAQ-derived MET adjusted for age, sex and BSA was an independent predictor of measured TEE—comparable to RPAQ-derived MET (Table 3). We also used the Bland–Altman technique to compare mean daily MET and TEE estimated from CKD-PAQ and RPAQ with that measured by accelerometry. Mean bias between the accelerometer and both questionnaires was small, though slightly lower for RPAQ. However, the LOA was quite wide—though slightly less for CKD-PAQ—and perhaps related to differences in the assessment time periods: 7 days in the accelerometer study and 4 weeks for both questionnaires [19]. All these findings suggest comparable performance of the CKD-PAQ and RPAQ questionnaires.
In addition, we found better discrimination between lower and middle accelerometer-derived TEE tertiles with CKD-PAQ-derived TEE than with RPAQ-derived TEE. We also found significant differences in energy expenditure levels between ICHD and HHD patients with CKD-PAQ-derived values but not with RPAQ values. Both these findings suggest that CKD-PAQ-derived energy expenditure values capture low-intensity activities better than RPAQ-derived values. There may be a number of reasons for this, including that CKD-PAQ is simpler to complete than RPAQ and thus may be more likely to be completed accurately. However, another major difference is the length of unaccounted time, i.e. the number of hours in a day that are not captured by the questionnaires, which is significantly lower with CKD-PAQ. In the final phase of the study, the median unaccounted time with CKD-PAQ was 1 h/day compared with >9 h with RPAQ. This demonstrates much more complete activity data capture with the novel questionnaire. As data capture is more complete, there is less risk of overestimating PA level with CKD-PAQ, as evidenced by significantly lower MET and TEE with CKD-PAQ compared with RPAQ. Hence there may be better discrimination between assessed PA levels in groups, such as those across the range of CKD, with low activity levels.
This study has its limitations. As with any questionnaire-based study, recall bias may have been a confounding factor. Although this has been minimized to some extent by enquiring about specific activities, some activities, especially the low-intensity ones, may not have been accurately reported by participants. The use of accelerometry in CKD patient population with low levels of PA may also limit the accuracy of the accelerometer data as these are designed predominantly to measure PA levels and not sedentary lifestyle. However, the gold standard doubly labelled water method of measuring TEE is not always feasible due to high costs and clinical problems with water turnover in CKD (especially dialysis) patients, leaving accelerometery as the most practical alternative ‘gold standard’.
There are a number of potential benefits in deploying CKD-PAQ in routine clinical practice. PA measurement in patients with CKD could be useful in their nutritional management by providing energy expenditure estimation and help identify patients with declining physical functioning. As CKD-PAQ focuses on common and routine physical activities, it can provide insight into potential target areas to increase PA levels and physical functioning, and can also be used to assess response to PA-related interventions.
In conclusion, this is the first study to have developed a PAQ in individuals across the range of CKD. This study has shown that CKD-PAQ is a valid tool for assessment of PA in CKD patients and performs comparably to the existing validated RPAQ questionnaire. CKD-PAQ may capture low-level PA more completely and better discriminate between groups with habitually low PA—such as the CKD population. The novel CKD-PAQ questionnaire needs further validation in larger cohort of patients with CKD.
SUPPLEMENTARY DATA
Supplementary data are available at ndt online.
FUNDING
This research was partly funded by a joint research grant from British Renal Society and Kidney Care UK.
AUTHORS’ CONTRIBUTIONS
S.S. was responsible for designing and setting up the study, collection and interpretation of data, drafting and revising the manuscript and approving the final version. E.V. was responsible for collection and interpretation of the data, and revised and approved the final version of the manuscript. S.R. was involved in collection and interpretation of the data, and revised and approved the final version of the manuscript. A.D. was involved in collection and interpretation of the data, revised and approved the final version of the manuscript, and provided intellectual content of critical importance to the article. K.F. was involved in concept of study, interpretation of the data, revised and approved the final version of the manuscript, and provided intellectual content of critical importance to the article.
CONFLICT OF INTEREST STATEMENT
None declared.
DATA AVAILABILITY STATEMENT
The data underlying this article will be shared on reasonable request to the corresponding author.
Comments