Physical activity in idiopathic pulmonary fibrosis: Longitudinal change and minimal clinically important difference

Abstract

Background and objective

Reference values of physical activity to interpret longitudinal changes are not available in patients with idiopathic pulmonary fibrosis (IPF). This study aimed to define the minimal clinical important difference (MCID) of longitudinal changes in physical activity in patients with IPF.

Methods

Using accelerometry, physical activity (steps per day) was measured and compared at baseline and 6-months follow-up in patients with IPF. We calculated MCID of daily step count using multiple anchor-based and distribution-based methods. Forced vital capacity and 6-minute walk distance were applied as anchors in anchor-based methods. Effect size and standard error of measurement were used to calculate MCID in distribution-based methods.

Results

One-hundred and five patients were enrolled in the study (mean age: 68.5 ± 7.5 years). Step count significantly decreased from baseline to 6-months follow-up (−461 ± 2402, p = .031). MCID calculated by anchor-based and distribution-based methods ranged from 570-1358 steps.

Conclusion

Daily step count significantly declined over 6-months in patients with IPF. MCID calculated by multiple anchor-based and distribution-based methods was 570 to 1358 steps/day. These findings contribute to interpretation of the longitudinal changes of physical activity that will assist its use as a clinical and research outcome in patients with IPF.

Introduction

Idiopathic pulmonary fibrosis (IPF) is a progressive disease with a median survival of only 2-3 years, ¹ and its clinical course among individual patients is highly variable. ² IPF is manifested by deterioration in pulmonary function, ³ exercise capacity, ⁴ physical activity ⁵ and health-related quality of life (HRQOL). ⁶ To date, longitudinal studies have shown that decline of these clinical measurements are associated with higher mortality rates in patients with IPF.^3,7–9 Therefore, evaluation of longitudinal changes in these clinical outcomes is important for refinement of clinical decision making.

Minimal clinically important difference (MCID) is widely used to interpret longitudinal changes of outcomes clinically and in research trials. It is defined as the smallest difference in a measure that is interpreted to be either beneficial or harmful from the patient’s and/or clinician’s perspective. ¹⁰ The MCID helps determine whether a particular intervention or overall management has been effective. ¹⁰ Several studies have reported the MCID of pulmonary function, ¹¹ HRQOL, ¹² functional exercise capacity^13–16 and extent of lung fibrosis ¹⁷ in patients with IPF. Although the MCID of physical activity has been demonstrated to be 600–1100 steps/day ¹⁸ and 350–1100 steps/day ¹⁹ in chronic obstructive pulmonary disease (COPD) patients, this has not yet been determined in IPF patients. Thus, it is difficult for clinicians and researchers to judge whether a patient’s longitudinal change in physical activity is clinically relevant in determining the decline, improvement, and management of their condition.

Common approaches of MCID are anchor-based and distribution-based methods. Anchor-based methods establish the MCID of the variable of interest to be “anchored” to the minimal change of a well-defined clinical variable. In contrast, distribution-based methods use statistical descriptors, including the effect size and standard error, to estimate the MCID. Because distribution-based methods do not integrate comparison to a well-defined clinical variable, anchor-based methods are usually recommended as the primary means to establish estimates of MCID. ²⁰

Accelerometry, the most commonly used motion sensor, can quantify physical activity more accurately than questionnaires. ²¹ Physical activity questionnaires can be susceptible to recall bias in low-intensity activity for a longer survey period,^22,23 as shown by patients with COPD who substantially overestimated their physical activity.^21,24,25 Therefore, accelerometry is a more valid measure of physical activity than physical activity questionnaires, especially for outcomes in longitudinal studies. Moreover, step count is deemed to be the most sensitive outcome to detect longitudinal changes ²⁶ and is readily understood by clinicians and patients. A previous investigation ²⁷ reported MCID of physical activity in interstitial lung disease (ILD) patients, however, the study aim was to validate the MCID of an activity questionnaire against accelerometry for moderate to vigorous physical activity and 65% of the sample had an ILD diagnosis besides IPF. Further, there was a lack of consistency of the MCID derived from the anchor-based compared to the distribution-based approaches. Taken together, the estimated MCID in this study has limited applicability to the IPF population over a broader range of exertion that includes lighter physical activity. ²⁷

This study aimed to investigate longitudinal change in physical activity, specifically daily step count measured by accelerometer, and to define the MCID of physical activity using anchor-based and distribution-based methods in patients with IPF.

Methods

Study design

This study was a retrospective chart review of patients with IPF that had physical activity assessed at baseline and a 6-months follow-up from May 2007 to March 2017 at Tosei General Hospital (Aichi, Japan).

Subjects

Inclusion criteria were: (1) diagnosis of IPF made by a multidisciplinary team. ¹ Patients diagnosed with IPF between 2007 and 2011 had their diagnosis reconfirmed using the 2011 criteria in September 2016; (2) had undergone evaluation of pulmonary function, arterial blood gases, St George’s Respiratory Questionnaire (SGRQ), ²⁸ modified Medical Research Council (mMRC) dyspnea scale, ²⁹ 6-minute walk test (6MWT) and (3) had their physical activity measured at baseline and 6-months follow-up with a minimum accelerometer wear time of 8 h per day for 7 days. ³⁰ Exclusion criteria were: (1) significant comorbidities that limited physical activity (i.e., cardiovascular, musculoskeletal or neurological impairment); (2) undergone any intervention to promote physical activity (i.e., pulmonary rehabilitation, counseling or motivational sessions) during study period or within 6-month prior to study enrollment; (3) initiation of long-term oxygen therapy during the study period.

Ethics approval

This study was approved by the ethics committee of Tosei General Hospital (IRB number: 742) and was conducted in accordance with principles of the Declaration of Helsinki. Written informed consent was waived because this study involved a retrospective chart review of patients’ records. Opt-out methods on the Web site were applied to obtain patient consent.

Data collection

Clinical data were derived from medical records.

Pulmonary function and arterial blood gases

Pulmonary function was measured by spirometry (CEHSTAC-55 V, Chest, Tokyo, Japan) and single-breath diffusion capacity for carbon monoxide (DLco) according to recent guidelines.^31,32 Arterial blood gases were sampled at rest breathing room air in the sitting or supine position.

Health-related quality of life, dyspnea and functional exercise capacity

HRQOL was assessed using the SGRQ. ²⁸ Functional limitation due to dyspnea was evaluated using the mMRC dyspnea scale. ²⁹ The 6MWT was performed according to published recomendations, ³³ and distance walked (6-min walk distance [6MWD]) was used as a measure of functional exercise capacity.

Physical activity

Physical activity was measured over seven consecutive days using a uniaxial accelerometer (Lifecorder GS, Suzuken, Nagoya, Japan), which has been validated for chronic respiratory disease. ³⁴ Patients were asked to wear the activity monitor on their right waist belts for 24 h/day except for bathing and sleeping, and to maintain usual levels of physical activity. The average value of step count (steps per day) over the seven-day period was used as the measure of physical activity. A minimum of eight hours of wear time (during waking time) on each of seven days was required for data to be used in the analysis. Details of physical activity measurement according to reporting reccomendations^35,36 are shown in supplementary table 2.

Calculation of minimal clinical important difference

The MCID of daily step count was established using anchor-based and distribution-based approaches.

For the anchor-based approach, forced vital capacity (FVC), SGRQ and 6MWD were selected as potential anchors because they: (1) have been established MCID in IPF;^3,12,16 (2) are key variables that enable tracking of clinical status and research trials; (3) have previously shown significant correlations to step count in IPF.^37,38 An anchor was considered valid to determine MCID of step count, if its change showed a significant correlation coefficient >0.3 to the change in step count over the 6 month period. If this level of significance was shown, MCID of step count was determined using three anchor-based approaches by comparisons to: (1) stratified of levels of change - changes were stratified to three levels “unchanged”, “changed minimally”, “changed more than minimally” according to previous studies^3,12,13,16 (see Table 1 for these levels). The MCID was estimated by weighted averages of change in step count for subjects who categorized “changed minimally”; (2) linear regressions were used to define changes in step count corresponding to the MCID of the anchors; ³⁹ (3) receiver operating characteristic (ROC) curves were used to estimate MCID of step count. ³⁹ In doing so, the cut-off value of changes in step count was calculated that could discriminate between those who demonstrated a decline of their anchor by the respective MCID and those who did not.

Table 1.

Stratification of Anchor Scores and established MCID.^3,12,13,16

Measure	Stratification of anchor			MCID
	Unchanged	Changed minimally	Changed more than minimally
FVC	<7%	7%–12%	12%	10%
SGRQ	<5 points	5-10 points	>10 points	7 points
6MWD	<24 m	24–45 m	>45 m	35 m

6MWD, 6-minute walk distance; FVC, forced vital capacity; MCID, minimal clinically important difference; SGRQ, St. George’s Respiratory Questionnaire.

Two distribution-based approaches were applied to estimate MCID: (1) 0.3 times the effect size (calculated as the change in step count divided the standard deviation [SD] of baseline step count), as previously described^20,40 and (2) 1-standard error of measurement (SEM)^40,41 of step count using following equation:

$$\text{SEM}=\text{SD} \text{of} \text{baseline} \text{step} \text{count}\times \text{square} \text{root} \left(1-\text{intraclass} \text{correlation} \text{coefficient} \text{of} \text{step} \text{count}\right)$$

Statistical analysis

Continuous data were expressed as mean and SD. Categorical data were presented as frequencies and percentages. Paired t test or Wilcoxon ranked sign test were performed to compare outcomes baseline and 6-months follow-up. Pearson’s correlation or Spearman rank correlation were used to investigate the associations between change in step count and change in clinical outcomes. One-way analysis of variance was conducted to compare the change in step count for subjects within each anchor change categories.

To determine MCID for anchor-based methods, a correlation coefficient >0.3 between external anchors and outcomes of interest is required.^20,40 Therefore, the sample size was calculated to detect a correlation coefficient >0.3 between change in step count and FVC, SGRQ and 6MWD with 0.8 statistical power at the significance level set at 0.05, ⁴² and it required minimum sample of 89 participants. All analyses were performed using SPSS version 25.0 (IBM, Chicago, USA). Statistical significance was considered at p < .05.

Results

One hundred and five patients were enrolled in the study. The baseline characteristics are shown Table 2.

Table 2.

Baseline characteristics of the 105 patients with IPF.

Age, years	68.5 ± 7.5
Gender, male	87 (82.8)
BMI, kg/m2	23.2 ± 3.0
Arterial blood gas values
PaO2, mmHg	82.0 ± 11.0
PaCO2, mmHg	41.1 ± 3.6
Pulmonary function
FVC, L	2.57 ± 0.69
FVC, % predicted	82.3 ± 18.6
FEV1, % predicted	94.2 ± 21.3
FEV1/FVC, %	83.5 ± 7.9
DLco, % predicted	62.6 ± 20.1
Dyspnea
mMRC grade (0/1/2/3/4)	28/51/20/6/0
HRQOL
SGRQ total score	30.8 ± 17.2
Functional exercise capacity
6MWD, m	561 ± 96
Lowest SpO2 during 6MWT, %	82.4 ± 9.2
Physical activity
Steps per day	6090 ± 4034
Anti-fibrotic
Pirfenidone	19 (18.1)
Nintedanib	11 (10.5)
LTOT	10 (9.5)

Data are presented as mean ± SD or number (%). 6MWD, 6-minute walk distance; 6MWT, 6-minute walk test; BMI, body mass index; DLco, diffusion capacity for carbon monoxide; FEV₁, forced expiratory volume in one second; FVC, forced vital capacity; HRQOL, health-related quality of life; IPF, idiopathic pulmonary fibrosis; LTOT, long-term oxygen therapy; mMRC, modified Medical Research Council; PaCO₂, partial pressure of carbon dioxide; PaO₂, partial pressure of oxygen; SD, standard deviation; SGRQ, St. George’s Respiratory Questionnaire; SpO₂, oxygen saturation measured via pulse oximetry.

Step count, FVC, SGRQ and 6MWD significantly worsened during the study period (Table 3). There were significant correlations that were greater than 0.3 between the change in step count and FVC and 6MWD, but not SGRQ (r _s = −0.260) (Table 3). Figure 1 shows scatter plots of change in step count relative to FVC, 6MWD, and SGRQ. Mean changes in step count significantly differed among categories of change in each of two anchors (FVC, p = .009; 6MWD, p = .003) (supplementary table 1). Figure 2 shows box plot of change in step count for subjects within each anchor change categories. Anchor-based and distribution-based MCID estimates ranged between 570-1358 steps (Table 4).

Table 3.

Longitudinal changes and correlation of potential anchors with physical activity.

	Baseline	6-months	Change	p-value a	r s	p-value b
Steps per day	6090 ± 4034	5629 ± 3697	−461 ± 2402 c	0.031	-	-
FVC, L	2.57 ± 0.69	2.50 ± 0.73	−2.8 ± 9.2 d	0.003	0.323	<0.001
SGRQ total score	30.8 ± 17.2	33.3 ± 18.4	2.6 ± 10.8 c	0.017	−0.260	0.007
6MWD, m	561 ± 96	537 ± 116	−24 ± 63 c	<0.001	0.403	<0.001

Data are presented as mean ± SD. 6MWD, 6-minute walk distance; SD, standard deviation; FVC, forced vital capacity; SGRQ, St. George’s Respiratory Questionnaire.

^ap-value for baseline to 6-months difference compared by Wilcoxon ranked sign test.

^bp-value of Spearman’s correlations between change in steps per day and change in potential anchors.

^cAbsolute change (6-months – baseline).

^dRelative change ([6-months – baseline]/baseline * 100).

Figure 1.

Scatter plots of change in step count and potential anchors (a), FVC; (b), SGRQ; (c), 6MWD. 6MWD, 6-minute walk distance; FVC, forced vital capacity; SGRQ, St. George’s Respiratory Questionnaire.

Figure 2.

Box plots of change in step count categorized by changes in FVC (a) and 6MWD (b). 6MWD, 6-minute walk distance; FVC, forced vital capacity.

Table 4.

MCID estimates for step count (steps per day).

	∆Anchor	ROC curve	Regression equations	ES = 0.3	1-SEM
FVC	570 a	742 b	Δstep = −220.5 + 90.6(ΔFVC) corresponding to 10% change in raw FVC, Δstep = 1126 (95% CI: 180-2073)	1210	1335
6MWD	1358 a	752 b	Δstep = −121.7 + 13.0(Δ6MWD) corresponding to 35 m change in 6MWD, Δstep = 602 (95% CI: 377-827)

6MWD, 6-minute walk distance; CI, confidence interval; ES, effect size; FVC, forced vital capacity; MCID, minimal clinical important difference; ROC, receiver operating characteristic; SEM, standard error of measurement.

^aWeighted mean of subjects whose anchor changed by the minimal (FVC = 7%–12%, 6MWD = 24–45 m).

^bCut-off value to discriminate subjects whose anchor changed over the minimal or not (FVC = 10%, 6MWD = 35 m).

Discussion

We found that daily step count, an easily interpreted assessment of physical activity, significantly declined over 6-months follow-up in patients with IPF. Moreover, anchor-based and distribution-based MCID estimates of physical activity showed decreases in the range of 570 to 1358 steps/day.

We determined three anchor-based methods each using two anchors (FVC and 6MWD), and two distribution-based methods. Anchor-based methods require comparisons with clinically relevant measures and secondly, that these measures show a significant correlation of rho greater than 0.3 with the outcome of interest.^20,40 We found that two of our clinically relevant measures, FVC and 6MWD, met these two criteria but SGRQ did not (Table 3). Consequently, FVC and 6MWD were further analysed as external anchors for anchor-based methods (Table 4).

Distribution-based methods, based on statistical variability, are recommended using as supportive information to anchor-based method. ²⁰ Moreover, combination of several anchor-based methods in multiple anchors and various distribution-based methods are important. ²⁰ Therefore, we used multiple anchor-based and distribution-based methods, and they provided a degree of reassurance on its validity. In the results, anchor-based and distribution-based MCID showed similar values, and provide further support of the robustness of MCID value determined in our study.

Two studies have investigated longitudinal declines of physical activity in patients with IPF.^7,43 Bahmer et al. ⁴³ reported that daily step count was declined 3017 steps over 3-years follow-up period in 26 patients. Badenes-Bonet et al. ⁷ observed 784 decline of step count during 12 months in 22 patients. These previously reported changes equate to 6-months step count declines of approximately 500 and 400 steps, respectively, similar to the decrease in our study (mean decrease of 461 steps). Compared to COPD patients, IPF patients appear to demonstrate a greater decline in physical activity as assessed by accelerometry. One report of COPD patients, revealed an annual decline of 451 steps although sit-to-stand and 6MWD remained stable. ⁴⁴ Teylan et al. ¹⁹ demonstrated that after a medical event, COPD patients decreased to a MCID of between 350-1100 steps/day in patients. Our results suggest that longitudinal decline of physical activity in patients with IPF may be twice that shown in COPD ⁴⁴ – 461 steps in 6-month. Moreover, the range of decline in IPF patients to a MCID of 570-1358 steps in our study appears more consistent with the range shown in COPD patients after a medical event. ¹⁹ Considering together that physical activity decline is associated with worse survival in COPD and IPF patients,^7,45 judicious evaluation of physical activity may be required to optimize clinical management in IPF patients.

Pulmonary rehabilitation and other interventions to promote physical activity may prove advantageous as reflected by an increase step counts per day following these interventions. Demeyer et al. ¹⁸ showed that distribution-based MCID of step count in a sample COPD patients undergoing pulmonary rehabilitation resulted in improvements ranging between 600-1100 steps/day. Of clinical significance, hospital admissions were reduced in patients that showed improvement more than 600 steps. ¹⁸ Accordingly, due to the potential greater downward trajectory of step count in IPF patients, interventions to mitigate decreased of physical activity might play more prominent roles in these patients.

To our knowledge, this is the first study to establish MCID of physical activity in patients with IPF. A major strength of this study is that step count is readily understood by clinicians and patients. Secondly, this study examined multiple anchor-based and distribution-based methods to define the MCID that derived similar values. A previous study reported the MCID of physical activity in patients with ILD using distribution-based methods, ²⁷ however, the measure was moderate-to-vigorous physical activity minutes. In addition, the majority of the sample of ILD patients (65%) were not IPF, ²⁷ which may limit the applicable of this MCID to the IPF population.

There were several limitations in this study. First, we did not consider seasonal effects on physical activity, however, data were collected across all seasons over several years. Secondly, because MCID reflected a decline of physical activity, it may not be sensitive to detect the effect of a physical activity intervention. Moreover, participants were not blinded to their step count on the accelerometer, which may have positively influenced their activity. Another consideration was that our sample consisted of patients during a relatively early stage of the IPF trajectory and there was a disproportionate dominance of males, which may limit generalizability of the MCID. Another concern may be the lack of comparability of the step counts obtained by the Lifecorder versus other pedometers but in fact a comparison the six pedometers (including Lifecorder) showed accuracy of ±1% of actual steps ⁴⁶ and the Lifecorder was accurate even during slow walking speeds. ⁴⁷

In conclusion, physical activity significantly declined over 6-months in patients with IPF. Additionally, MCID calculated by multiple anchor-based and distribution-based methods was 570 to 1358 steps/day. These findings provide a basis to interpret the longitudinal change of physical activity and to assist with evaluations of clinical and research outcomes in patients with IPF.

Abbreviations.

6MWD

6-min walk distance

6MWT

6-minute walk test

COPD

Chronic obstructive pulmonary disease

DLco

Diffusion capacity for carbon monoxide

FVC

Forced vital capacity

HRQOL

Health-related quality of life

ILD

Interstitial lung disease

IPF

Idiopathic pulmonary fibrosis

MCID

Minimal clinically important difference

mMRC

modified medical research council

MVPA

Moderate to vigorous physical activity

ROC

Receiver operating characteristic

SD

Standard deviation

SEM

Standard error of measurement

SGRQ

St. george’s respiratory questionnaire.

ORCID iDs

Kazuya Shingai https://orcid.org/0000-0001-9036-6636

W. Darlene Reid https://orcid.org/0000-0001-9980-8699

Ryo Kozu https://orcid.org/0000-0002-9310-185X

Data Availability Statement

Data are available on reasonable request.