The validity of the International Physical Activity Questionnaire (IPAQ) for adults with progressive muscle diseases

Abstract

Purpose

Measuring the physical activity of adults with progressive muscle diseases is important to inform clinical practice, for activity recommendations and for outcomes meaningful to participants in clinical trials. Despite its wide use, the measurement properties of the International Physical Activity Questionnaire (IPAQ) have not been established in a muscle disease population.

Materials and methods

The sample of 103 adults with progressive muscle diseases included independently mobile participants and wheelchair users. Their home-based activity measured by the IPAQ was compared to simultaneous weeks of accelerometer activity data collected remotely in a longitudinal, measure evaluation study. Validity, reliability, and responsiveness were evaluated for the IPAQ alone, and for the IPAQ used in conjunction with a smart activity monitor.

Results

The IPAQ did not demonstrate satisfactory criterion validity, reliability or responsiveness and it systematically overestimated moderate and vigorous physical activity time by 161 minutes per week. Measurement properties of the IPAQ were improved when it was used in combination with a smart activity monitor.

Conclusions

The IPAQ did not have satisfactory measurement properties compared to accelerometry in adults with progressive muscle disease. Combining self-report and objective activity measures might improve the accuracy of physical activity assessment in this and other comparable populations.

Implications for Rehabilitation

• Physical activity is a meaningful health outcome for adults with progressive muscle diseases, for whom precise activity quantification is important because of the potential for activity-related disease exacerbation.

• The International Physical Activity Questionnaire (IPAQ) had unsatisfactory measurement properties compared to accelerometry; however, these were improved by adjunctive smart activity monitoring.

• Objective or combined physical activity measurement is recommended over self-report alone for clinical assessment of physical activity as part of rehabilitation and self-management programmes.

Keywords:

• Physical activity
• measurement
• validity
• reliability
• responsiveness
• muscular dystrophy

Introduction

Physical activity is defined as “any bodily movement produced by skeletal muscles that requires energy expenditure”; insufficient physical activity has been linked to poor health outcomes [1]. Measuring the physical activity of adults with progressive muscle diseases is important to inform clinical practice, activity recommendations and for outcomes, meaningful to participants, in clinical trials [2]. In muscle disease, activity has been linked to improved muscle strength, cardiorespiratory fitness, fatigue management, quality of life and protection against comorbidities [3–8].

However, concerns remain over potential activity-related exacerbations of muscle disease [4,9]. World Health Organisation (WHO) activity recommendations seem generally applicable and safe [1] for many other neuromuscular diseases. For example, those originating from neuromuscular junction pathology, such as Myasthenia Gravis [10], or nervous system pathologies, such as peripheral neuropathies [11,12] and Spinal and Bulbar Muscular Atrophy [12]. However, there is less evidence regarding safe activity dosage for adults with muscle disease [3]. The reciprocal relationship between activity and muscular gene expression [13] means that, in muscle disease, greater understanding and precision is required for measurement, and evidence-based prescription, of physical activity throughout disease progression [3,14].

Progressive muscle diseases, including muscular dystrophies, myopathies and inclusion body myositis, are characterised by pathological changes in skeletal muscles [15,16], resulting in a clinical course of slowly progressive weakness [17,18]. Disease symptoms are variable, for some there is a mobility decline that may require wheelchair use [19]. This functional heterogeneity makes activity measurement challenging. Multiple approaches have been used but there is insufficient evidence to recommend any activity measurement tool for use with adults who have progressive muscle diseases [20,21].

Despite not being designed originally for use in small samples for clinical studies [22], IPAQ is more widely used in neuromuscular disease research than objective measures or questionnaires tailored for disabled populations [21], such as the Physical Activity Scale for Individuals with Physical Disabilities [23]. Disability and wheelchair-specific activity questionnaires have received criticisms relating to the questionable accuracy of self-report [20] and may be suited to activity measurement only for those at later stages of disease progression. Continuity of activity measurement is preferrable using a single measure inclusive of early and late stages of disease progression.

The International Physical Activity Questionnaire (IPAQ) was developed to standardise population-level activity surveillance worldwide [24,25]. Modified versions of the IPAQ have also been designed to evaluate people with functional limitations [26,27]. It is an easy to administer, self-report, 7-day recall questionnaire (or interview). It collects information about time spent in vigorous, moderate, walking, and sedentary activities. The overall score estimates metabolic expenditure and was designed to categorise people into low, moderate, or high activity. Early studies reported satisfactory reliability and validity in the general population [25,28,29]. However, subsequent systematic reviews indicated the IPAQ considerably overestimated activity and had only low to moderate validity when compared to objective measures [30–32]. IPAQ overestimation might be attributed to social desirability and recall biases [33] and systematic reviews in neuromuscular diseases have also questioned the accuracy of self-report physical activity measures [20,21].

However, if the IPAQ was valid for ambulant and non-ambulant adults with progressive muscle diseases, it would provide a suitable activity measurement tool to chart physical activity throughout the stages of muscle disease progression, confirm the accuracy of existing physical activity research in muscle disease, and allow comparisons with studies in the general population. To date, the IPAQ (modified or otherwise) has never been validated for people with progressive muscle diseases.

The objective of this study was to test the validity, reliability, and responsiveness of the IPAQ in adults with progressive muscle diseases. The secondary objective was to discover whether the measurement properties of the self-report questionnaire were improved by adjunctive smart activity monitoring. The hypothesis was that wearing a smart activity monitor would improve the accuracy of self-reported IPAQ physical activity data by reducing overestimation.

Materials and methods

Study design

Ethical approval for this longitudinal, measure evaluation study was granted by King’s College London Research Ethics Committee (LRS-18/19-10909). Recruitment started in April 2019. The setting was home-based as the study was conducted remotely. Baseline cross-sectional IPAQ and accelerometer data were collected from April to October 2019 and at longitudinal follow-up from February to August 2020. They were compared to evaluate validity, reliability, and responsiveness of the questionnaire. Measurement properties for the IPAQ alone were compared to those for IPAQ data collected after a week of adjunctive smart activity monitoring.

Participants

People with progressive muscle diseases were invited to participate via national muscle disease registries (The John Walton Muscular Dystrophy Research Centre, Newcastle University) and the charity, Muscular Dystrophy UK (https://www.musculardystrophyuk.org/) advertised via their website and newsletter. Participants were included if they were UK resident adults with a confirmed diagnosis of Inclusion Body Myositis, Myotonic Dystrophy or Muscular Dystrophy (including Facioscapulohumeral Dystrophy, Limb Girdle Muscular Dystrophy, Dysferlinopathy, Dystrophinopathy (including manifesting female carriers) or specific congenital myopathies lasting into adulthood). Participants were excluded if they were aged <18 years, cognitively impaired, unable to wear an accelerometer or did not have a confirmed diagnosis of muscle disease. After volunteers made contact with the research team, screening was carried out by telephone appointment. A physiotherapist (SRL) clinically assessed eligibility criteria and gathered supporting evidence (including diagnosis confirmation and cognitive history). Any eligibility discrepancies were resolved on a case-by-case basis by a consultant neurologist (MR). For eligible participants, informed consent and data were collected by email and post.

Outcomes

The primary outcome was concurrent validity of the IPAQ overall score correlated with overall activity intensity in mean accelerations per minute measured by accelerometer. Secondary outcomes included convergent, divergent, and discriminative validity according to demographic characteristics, test–retest reliability over 2 weeks and responsiveness to change from baseline to follow up. Measurement error of IPAQ activity minutes compared to weekly moderate and vigorous minutes measured by accelerometry was also tested. Each measurement property was also evaluated for the IPAQ plus adjunctive smart activity monitoring.

Measures

The self-report, short-form IPAQ [25], consists of three questions about days and time of vigorous, moderate, and walking activity in bouts of ≥10 min, with one question about daily sedentary time in the preceding seven days. In this study, the question descriptors were modified to incorporate the range of functional ability in the sample; changes were similar to previous modified versions [26,27]. Modifications included extra vigorous and moderate activity examples, including wheelchair activities. The walking question was modified to include “time spent walking, self-propelled wheeling or equivalent light activities.” “Sedentary” time was changed to “inactive” time. The questionnaire was preceded by completion instructions and three questions about the types of vigorous, moderate, and light activity undertaken [34]. It was followed by three questions about bedtime, waketime and sleep hours. Weekly activity times were calculated from days and minutes of vigorous, moderate, and walking/light activity. Total score in metabolic equivalents (METs) per week was calculated by multiplying weekly active time by intensity specific metabolic values as per IPAQ scoring instructions [25].

Adjunctive smart activity monitoring was by Fitbit Inspire HR (Fitbit Inc., San Francisco, CA). It is a tri-axial accelerometer and continuous optical heart rate monitor. It syncs with the Fitbit smart phone app and yields physical activity metrics including daily active minutes, metabolic expenditure, steps, and sleep. In the adjunctive study, activity monitoring was used for a week preceding each IPAQ completion. Participants regularly synced the Fitbit with their smart phone app. Thus, participants had seen Fitbit activity metrics before completing the questionnaire at the end of the week.

The comparator measure was a GENEActiv tri-accelerometer (ActivInsights Ltd., Kimbolton, UK). It has been validated for wrist-wear [35] and ankle-wear in adults with progressive muscle diseases [36]. Participants wore it on their non-dominant wrists unless ankle-wear was indicated because of work or using crutches (which can interfere with wrist-worn activity monitors). The sampling frequency was 10 Hz. Data were processed using the GGIR package in R (version 3.6.0) [37] in 1-min epochs of milli-gravitational units (milli-g) with gravitational correction [38]. Weekly overall activity intensity was calculated by mean accelerations (milli-g) per minute. Minutes of sleep, inactivity, light activity, and bouts of ≥10 min of moderate and vigorous activity were calculated using the following cut-points: light ≥30 milli-g/min, moderate ≥100 milli-g/min and vigorous activity ≥400/milli-g/min [39].

The moderate to vigorous physical activity (MVPA) threshold of 100 milli-g/min has been widely used for healthy adults and people with long-term conditions [40,41]. Lower thresholds were considered for this cohort, whose muscle weakness might mean greater exertion produces less activity compared to healthy controls [36]. However, muscle strength is not a direct determinant of limb acceleration [36] and a <100 milli-g/min threshold would have been misleading for more active individuals, including wheelchair users, whose activity and exertion are equivalent to people with other long-term conditions [42].

Procedure

Before baseline, demographic information was collected; these data underpinned evaluation of IPAQ construct validity (discriminative, convergent, and divergent). Demographics included condition, age, gender, anthropometrics, handedness, employment status, mobility, quality of life and self-perception. Quality of life was measured using the Individualised Neurological Quality of Life (INQoL), a 45-item questionnaire designed and validated for people with neurological diseases [43]. Items are Likert rated 0–6, yielding an indexed score of 0–100 from best to worst quality of life. Subscales include activities, independence, relationships, emotions, and appearance. Self-perceived activity level was measured using the Physical Self-Description Questionnaire activity subscale (PSDQ-activity) [44]. The 40-item, Likert-rated PSDQ was validated for disabled adults with neuromuscular diseases [45]. The PSDQ-activity subscale of four items yields a mean score of 1–6 from least to most activity.

At baseline weeks 1 and 2, test–retest reliability was tested by repeated administration of the IPAQ and testing for functional stability each week using the Health Assessment Questionnaire (HAQ) [46]. The 24-item HAQ has been validated in electronic and paper format [47]. Activities of daily living are rated 0–3 according to difficulty, yielding an indexed total score of 0–3 from least to most disabled. At baseline week 3, concurrent validity and measurement error were evaluated by comparison of IPAQ data with GENEActiv accelerometry data. The GENEActiv was posted to participants, who wore it for 7 days, only removing it to wash, then returned it by post. The third baseline week was used to ensure that accelerometry did not interfere with reliability testing of the questionnaire in baseline weeks 1 and 2.

At follow up, participants were sent the accelerometer to wear for another week and questionnaires were repeated. Responsiveness was examined by comparing changes in IPAQ and accelerometry over time.

In the adjunctive study, 3 months after baseline, participants were invited to wear a smart activity monitor and complete the IPAQ again. Those who opted in were sent a wrist-worn Fitbit. Participants used the Fitbit for 2 weeks at time points 1 and 2 and during the follow up week. They completed the IPAQ and HAQ each week to test adjunctive IPAQ test–retest reliability. Participants who wore their Fitbit at follow up provided data to test adjunctive IPAQ concurrent validity (compared to GENEActiv accelerometry) and construct validity (relationships with demographics re-tested at follow up) (see Figure 1).

Figure 1. Schematic of study recruitment, screening, data collection and study design. PSDQ: Physical Self-Description Questionnaire; INQoL: Individualised Neurological Quality of Life; IPAQ: International Physical Activity Questionnaire; HAQ: Health Assessment Questionnaire.

Statistical analysis

A sample size of 100 (plus 10% to allow for attrition) was planned based on recommendations for evaluating measurement properties [48] and a power calculation [49]. Statistical analyses were carried out using R (version 3.6.0). Alpha was set at 0.05. Data were tested for normality using Shapiro–Wilk [50] and nonparametric equivalents were used when data were not normally distributed. MVPA was summed moderate and vigorous activity time recorded using IPAQ/GENEActiv. Concurrent validity between IPAQ and GENEActiv data was tested by Spearman’s correlation; rho ≥0.70 was considered satisfactory [48]. However, lower correlations of 0.09–0.39 [31], have been reported and ≥0.50 is sometimes used as a satisfactory correlation cut-point [31]. Based on a predicted correlation of 0.50–0.70, the study had 99–100% power [49]. Convergent and divergent validity were tested by correlation with related constructs (PSDQ-activity, INQoL, and HAQ) and unrelated constructs (height, handedness, and age) respectively. Satisfactory construct validity cut-points were ≥0.30 and ≤0.20, respectively [48]. Discriminative validity tested the difference in IPAQ score between wheelchair users and ambulant participants by the Mann–Whitney U test. Measurement error of activity time reported in the IPAQ was examined by absolute measurement error and Bland–Altman plots. Reliability was tested by intra-class correlation coefficient (ICC) between repeated IPAQ scores one week apart. An ICC of ≥0.75 was considered satisfactory [51]. Responsiveness of the IPAQ to change overtime was tested by area under the curve (AUC) analysis of receiver operating characteristic curves for GENEActiv change. An AUC of ≥0.70 was considered satisfactory [48]. Questionnaires were scored using available items only; questionnaires with >10% of items missing were excluded from analyses. For accelerometry, missing data of ≤10 min were included in daily means, but days with <23 h monitored were excluded from analyses. Participants with missing datasets or lost to follow up were excluded from analyses.

Results

Figure 1 summarises recruitment. One hundred and three completed baseline data, 100 returned follow up data There was a mean of 9.4 months between baseline and follow up. Demographics at baseline are summarised in Table 1.

Table 1. Demographics and baseline information (N = 103).

Characteristic	All	Independent	Assisted	Wheelchair users
Sample size	103	45	22	36
Diagnoses	33 LGMD, 25 FSHD, 16 IBM, 11 MD, 10 DM, 8 CM	11 LGMD, 16 FSHD, 5 IBM, 5 MD, 7 DM, 1 CM	7 LGMD, 4 FSHD, 3 IBM, 2 MD, 10 DM, 4 CM	15 LGMD, 5 FSHD, 8 IBM, 4 MD, 1 DM, 3 CM
Gender	48 Male,	19 Male,	12 Male,	17 Male,
	55 Female	26 Female	10 Female	19 Female
Employment status	40 FT, 17 PT,	26 FT, 6 PT,	6 FT, 6 PT,	8 FT, 5 PT,
	46 NW	12 NW	12 NW	22 NW
Handedness	93 Right, 10 Left	42 Right, 3 Left	18 Right, 4 Left	33 Right, 3 Left
Mobility	36 WC, 22 AM, 45 IM	45 IM	22AM	36 WC
Age (years) mean (SD)	47.8 (15.6)	43.0 (14.7)	49.4 (16.1)	51.3 (15.8)
Height (cm) mean (SD)	170.3 (10.3)	170.7 (11.1)	171.8 (8.6)	169.4 (10.3)
Weight (kg) mean (SD)	74.9 (17.6)	70.3 (13.7)	79.6 (19.1)	75.2 (18.9)
Body Mass Index (mean (SD))	25.8 (6.0)	24.2 (4.8)	26.8 (6.3)	26.1 (6.1)
INQoL (median (IQR))	42.2 (36.7–49.4)	37.8 (26.7–41.1)	48.9 (44.8–52.1)	48.9 (45.3–56.0)
PSDQ-activity (median (IQR))^a	1.75 (1.13–2.75)	2.25 (1.50–3.50)	1.38 (1.00–2.00)	1.25 (1.00–2.06)
HAQ (DI) (median (IQR))	1.75 (1.06–2.50)	0.88 (0.50–1.50)	1.75 (2.09–2.28)	2.63 (2.30–2.88)
IPAQ (median (IQR))
IPAQ total score	2604 (1317–4814)	4484 (2920–6523)	1985 (852–3432)	1628 (679–2678)
IPAQ MVPA minutes per week	107 (15–235)	210 (120–360)	50 (12–128)	25 (0–105)
Accelerometer (mean (SD))
Mean accelerations per minute	27.0 (9.1)	34.1 (6.5)	25.5 (7.4)	18.4 (4.8)
GENEActiv MVPA mins/week	6.2 (10.8)	12.3 (13.5)	7.7 (16.0)	1.7 (4.6)

All participants had at least five days of useable accelerometry data at baseline and follow up. All participants, except seven, wore the accelerometer on their wrists. Outputs did not differ significantly by position, so all accelerometer data were analysed together. Most participants completed electronic questionnaires. Only five at baseline and one at follow up, completed paper questionnaires. SD: standard deviation; IQR: interquartile range; MVPA: moderate and vigorous activity; HAQ (DI): health Assessment Questionnaire (disability Index); INQoL: Individualized Neurological Quality of Life questionnaire; LGMD: Limb Girdle Muscular Dystrophy (types 1A, 2A, 2B, 2L, 2I and Myoshi Myopathy); FSHD: facioscapulohumeral dystrophy; IBM: inclusion body myositis (sporadic and hereditary); DM: Myotonic Dystrophy (types 1 and 2); MD: Duchene and Becker’s Muscular Dystrophy (including 4 manifesting female carriers); CM: congenital myopathies (including Centronuclear Myopathy, Emery Dryfuss, SEPN1-related and Ullrich and Bethlem Myopathy); FT: full time working; PT: part-time working (or volunteering (2) or studying (3)); NW: not working or retired (28); WC: wheelchair users or bedbound (2); AM: assisted mobility, ambulant with a walking aid or occasional wheelchair use outdoors for long distances; IM: independent mobility.

^aMissing data information (HAQ: 4 participants missed 1/24 items, all cases included; INQoL: 11 participants missed 1/45 items and 2 participants missed 2/45 items, all cases included; PSDQ-activity: 1 participant missed 1/4 items, 1 case excluded; accelerometry: all cases included as every participant had at least 4 days of data).

The concurrent validity between IPAQ score and GENEActiv overall intensity did not meet acceptability criteria, although there was a significant moderate correlation (rho = 0.49, p < 0.0001, N = 103). Convergent validity of IPAQ score was demonstrated by significant low to moderate correlations with the HAQ Disability Index and PSDQ-Activity subscale (rho = −0.35 to 0.42). Divergent validity of IPAQ score was demonstrated by lack of significant correlation with height, handedness, employment status and age (see Table 1). Discriminative validity was demonstrated by significant differences between wheelchair users and independently mobile participants for activity measured by IPAQ and GENEActiv (see Tables 1 and 2). Test–retest reliability of the IPAQ was just below the acceptability threshold (ICC = 0.73, confidence interval = 0.66–0.81). There was no significant change in HAQ between reliability testing weeks. Responsiveness of the IPAQ to activity change over time was not satisfactory (AUC = 0.64). Using the IPAQ with the smart activity monitor (Fitbit) improved the validity and reliability of the questionnaire. However, criterion validity was still unsatisfactory (see Table 2).

Table 2. Validity and reliability statistics for IPAQ total score.

	IPAQ (N = 103^a)		IPAQ plus Fitbit (N = 90^b)
Measurement property	Statistic	p Value	Statistic	p Value
Concurrent validity with IPAQ score
GENEActiv	rho = 0.49	<0.0001	rho = 0.57	<0.0001
Convergent validity with IPAQ score
PSDQ-Activity subscale	rho = 0.42	<0.0001^a	rho = 0.57	<0.0001
HAQ Disability Index	rho = −0.35	0.0002	rho = −0.37	<0.0001
INQoL-Independence	rho = −0.29	0.0032	rho = −0.39	<0.0001
Weight	rho = −0.19	0.0438	rho = −0.20	0.0321
Divergent validity with IPAQ score
Handedness	rho = −0.13	0.1971	rho = −0.03	0.7694
Height	rho = −0.13	0.2075	rho = −0.08	0.4449
Employment status	rho = 0.10	0.3011	rho = 0.09	0.2671
Age	rho = 0.02	0.8445	rho = 0.00	0.9686
Gender	rho = 0.19	0.0615	rho = 0.14	0.1591
BMI	rho = −0.18	0.0743	rho = −0.18	0.0622
Discriminative validity between extreme groups (wheelchair users and independently mobile)
IPAQ overall score	U = 1316	<0.0001	U = 1444	<0.0001^b
GENEActiv overall intensity (milli-g/min)	U = 1589	<0.0001	U = 1616	<0.0001^b
Reliability of IPAQ score (between weeks 1 and 2 and timepoints 1 and 2, respectively)
Test re-test	ICC = 0.73	<0.0001	ICC = 0.79	<0.0001

IPAQ: International Physical Activity Questionnaire; HAQ: Health Assessment Questionnaire; INQoL: Individualized Neurological Quality of Life questionnaire; PSDQ: Physical Self-Description Questionnaire; ICC: intra-class correlation coefficient.

^aMissing data: N = 102 for PSDQ, as one participant was excluded from analyses due to and incomplete dataset and N = 81 for discriminative validity based on numbers in each extreme group compared (see Table 1).

^bN = 75 for discriminative validity groups who used Fitbit.

The validity and test–retest reliability for activity time derivations of the IPAQ, including MVPA, light activity, inactivity, and sleep time were similarly unsatisfactory to that of IPAQ total score. However, like IPAQ total score, they showed slightly improved measurement properties for the IPAQ plus Fitbit compared to IPAQ alone (see Table 3).

Table 3. Validity and reliability statistics for IPAQ activity, inactivity, and sleep time.

	IPAQ (N = 103)		IPAQ plus Fitbit (N = 90)
Measurement property	Statistic	p Value	Statistic	p Value
Concurrent validity IPAQ activity time vs. GENEActiv activity time
MVPA	rho = 0.51	<0.0001	rho = 0.56	<0.0001
Light Activity	rho = 0.37	0.0001	rho = 0.55	<0.0001
Inactivity	rho = 0.25	0.0105	rho = 0.38	0.0002
Sleep	rho = 0.22	0.0254	rho = 0.30	0.0037
Reliability IPAQ activity time (between weeks 1 and 2 and timepoints 1 and 2, respectively)
MVPA	ICC = 0.70	<0.0001	ICC = 0.75	<0.0001
Light Activity	ICC = 0.33	0.0002	ICC = 0.54	<0.0001
Inactivity	ICC = 0.42	<0.0001	ICC = 0.59	<0.0001
Sleep	ICC = 0.68	<0.0001	ICC = 0.76	<0.0001

IPAQ: International Physical Activity Questionnaire; ICC: intra-class correlation coefficient.

Measurement error was high for IPAQ derived MVPA time, compared to the GENEActiv accelerometer, as shown by the Bland–Altman plot (see Figure 2(a)). There were wide limits of agreement, high proportional error (0.86) and considerable absolute and systematic errors indicating overestimation of MVPA using IPAQ alone (172 and 161 min per week, respectively). Those who reported fitness or transportation activities tended to overestimate less than respondents who reported only domestic activities. However, the types of activity reported were predominantly domestic (e.g., shopping, housework, self-care, house maintenance and caring responsibilities). Self-reported MVPA time showed the greatest overestimation error and inactivity time the greatest underestimation error. Whereas light activity and sleep time reporting had more random error (see Figure 2(a,c,e,g)).

Figure 2. Bland–Altman plots of IPAQ measurement error in MVPA, light activity, inactive and sleep time. IPAQ: International Physical Activity Questionnaire; MVPA: moderate and vigorous activity.

IPAQ plus Fitbit MVPA showed the greatest improvement in self-report accuracy compared to IPAQ alone. MVPA measurement error reduced by 135 min per week and there was a 34% reduction in proportional bias (see Figure 2(b)). The effect was similar, but smaller, for light activity (see Figure 2(d)). However, there was little change in measurement error of inactivity or sleep time reporting between IPAQ alone and IPAQ plus Fitbit (see Figure 2(f,h)).

Discussion

The IPAQ did not have satisfactory validity compared to accelerometry for the assessment of physical activity in adults with progressive muscle disease. Despite satisfactory construct validity (convergent, divergent, and discriminative), the IPAQ did not demonstrate satisfactory criterion validity, reliability, or responsiveness. IPAQ measurement error was unacceptably high with a strong tendency for overestimation of activity. However, measurement properties and reporting accuracy of the IPAQ were slightly improved when it was used in combination with a smart activity monitor.

Our findings are similar to other studies which have reported unsatisfactory validity and reliability of the IPAQ in healthy [52] and clinical populations [53]. Conversely, some earlier studies reported IPAQ had satisfactory reliability and validity, however, they used lower acceptability thresholds [31]. Our findings of significant overestimations of IPAQ MVPA time are concordant with measurement error findings from other studies of self-reported physical activity measures [31,52,53]. Measurement error might be explained by recall and social desirability bias (which leads people to overestimate self-reported activity) [54].

Some researchers have suggested ways to improve the accuracy of self-report physical activity measurement [55]. It is possible that self-report bias might be tempered by adjunctive objective monitoring. This might explain the improved IPAQ plus Fitbit measurement properties detected in this study. However, some experts recommend preferential reliance on objective activity measures, only using self-report tools, like the IPAQ, alone, if objective measurement is unfeasible [52,53,56]. The IPAQ alone is unsuitable for clinical studies of adults with progressive muscle diseases, where activity represents a health outcome indicative of disease progression. Similarly, the IPAQ alone is unsuitable when the precise quantification of physical activity is crucial. For example, informing muscle disease activity prescription and guidelines, because inflated activity targets might lead to activity-related disease exacerbation [9] or activity disengagement. Insufficient activity might increase the detrimental health sequelae associated with sedentarism [57].

Self-report measures, like the IPAQ, are useful in population surveillance studies where activity categorisation, rather than exact activity quantification, is desirable [22,52]. Also, as additional activity outcomes to help interpret objective data and provide information about activity type and perceived intensity. Indeed, self-reported physical activity data provides insight into perceived activity intensity [30]. Self-reported MVPA overestimation, compared to accelerometry, might be attributable to the increased energy cost of physical activity experienced by some people with progressive muscle diseases [58,59]. Accelerometry is unable to differentiate the level of effort required for limb acceleration and some experts recommend combined activity measurement approaches encompassing movement and associated exertion [60].

In this study, the excessive activity overestimation of IPAQ alone might be explained by social desirability bias, recall bias and accelerometer insensitivity to exertion. Combined IPAQ plus adjunctive smart activity monitoring potentially represents a more accurate activity assessment, eliminating some self-report overestimation biases whilst also accounting for perceived energy cost of activity.

The improved accuracy of activity time reporting using the IPAQ plus smart activity monitoring might be accounted for because the monitoring app dashboard displayed active minutes. Thus, immediate availability of activity time information from smart activity monitoring might have facilitated more accurate MVPA (and light) activity self-report. However, self-report of inactivity time was consistently underestimated using IPAQ alone and unchanged with adjunctive smart activity monitoring. The monitoring app dashboard did not display inactivity data. This is a possible explanation for why smart activity monitoring did not improve awareness of inactivity. Thus, for assessment or interventions targeting inactivity, clinician-facilitated, or inactivity-focused, self-monitoring might be required to improve inactivity awareness and self-report.

The key strengths of this study were the thorough examination of multiple measurement properties and the remote design, allowing the inclusion of a wide range of participants from across England, Ireland, Scotland and Wales with varied functional ability and activity levels. Limitations include that the standard accelerometry cut points, used to differentiate vigorous, moderate, and light activity, might have been too severe for this population which included some highly disabled individuals. The measurement properties of the IPAQ may have been impacted by modifications to question descriptors or the administration method, because IPAQ was originally designed for telephone use which allows real-time response guidance (potentially reducing self-report errors) [22]. Responsiveness might have been deflated by the lack of clinically important change in activity of the whole sample at follow up after less than a year. However, responsiveness might also have been inflated by the reduced overestimation linked to Fitbit use at follow up. The study would have been strengthened by a larger sample size allowing sub-group analyses of measurement properties by muscle disease diagnosis.

In conclusion, the IPAQ, compared to accelerometry, was not sufficiently valid, reliable, or responsive to measure physical activity in adults with progressive muscle diseases. Adjunctive objective activity monitoring improved the measurement properties and accuracy of IPAQ self-reported physical activity. For accurate quantification of physical activity, we recommend using objective or combined approaches to physical activity measurement in preference to self-report alone. Given our diverse sample, this recommendation is likely to be generalisable to activity measurement in other long-term conditions and, even, healthy adults.

Acknowledgements

The authors thank The John Walton Muscular Dystrophy Research Centre at Newcastle University and the Muscular Dystrophy UK charity for their help with advertising the study.

Disclosure statement

The authors report no conflicts of interest.

Additional information

Funding

PhD studentship at King’s College London [MR/K50130X/1].