Background
The preventive effect of physical activity (PA) on risk of colorectal cancer is well-established [
1‐
3]. However, an increasing number of studies also examine beneficial effects of PA during cancer treatment as well as in the posttreatment period [
4‐
16], such as decreased all-cause mortality, increased disease-free survival, improved physical function and quality of life [
5,
6,
11‐
13,
17,
18]. Moreover, reduced sedentary time, such as sitting during daytime, may be associated with reduced mortality and lower risk of recurrence in cancer patients [
19‐
22].
The recommendations of PA for cancer patients and survivors provided by the American Cancer Society [
23] emphasize that exercise is safe and feasible during cancer treatment, and improves outcomes such as physical function, fatigue and completion of chemotherapy [
23]. The American Cancer Society, the World Health Organization and others [
24‐
27] recommend at least 150 min of moderate intensity PA (MPA) or 75 min of vigorous intensity PA (VPA) per week or an equivalent combination. In 2011, the Norwegian Directorate of Health published the Norwegian Food-Based Dietary Guidelines (FBDG) which also includes similar recommendations on PA as well as for sedentary time [
3].
In Norway, colorectal cancer (CRC) is the third most common cancer type, and the incidence is among the highest in Europe [
28]. Implementing the recommendations of PA and incorporating specific exercises in the clinical care may improve the health outcomes of CRC patients [
1‐
3,
24].
In order to estimate adherence to PA recommendations according to the Norwegian FBDG in a Norwegian CRC population, a valid and accurate physical assessment tool is needed. Importantly, assessment of adherence to the PA recommendations is required in counselling and when evaluating effectiveness of intervention studies. The use of objective monitors to record PA has increased during recent decades and gives valid and reliable data on intensity of PA and energy expenditure [
29]. However, these activity monitors are expensive and time consuming for the clinician and researcher, particularly when recording PA in larger populations. Therefore, less expensive and easier methods are required to measure adherence to PA recommendations.
The most common self-reporting method to assess PA is the use of questionnaires [
30,
31]. Over the past 2 or 3 decades, more than 30 PA questionnaires have been developed and validated [
32]. Long questionnaires are challenging to complete for cancer patients often experiencing treatment and disease related side-effects such as fatigue and functional decline [
33‐
36]. Questionnaires which contains few and well-defined questions regarding the different intensities of PA may be more suitable for this group of patients [
37].
Although many previous questionnaires have been successfully used to assess PA, there is no questionnaire specifically designed to assess adherence to the PA recommendations as defined in Norwegian FBDG. For a clinical trial in colorectal cancer patients [
38], a new short semi-quantitative questionnaire (NORDIET-FFQ) was developed to measure adherence to the Norwegian FBDG. The NORDIET-FFQ includes two questions on PA related to intensity levels similar to MPA and VPA. Another short questionnaire, the HUNT-PAQ [
39] has been used in large healthy populations in Norway; however this questionnaire has not previously been validated in a CRC population.
Thus, the primary aim of the present study was to validate the two short questionnaires and their ability to estimate adherence to the PA recommendations according to the Norwegian FBDG. Secondarily, self-reported sedentary time from the HUNT-PAQ was also evaluated.
Discussion
In the present study, we evaluated the ability of the questionnaires, NORDIET-FFQ and HUNT-PAQ, to estimate adherence to PA recommendations among CRC patients participating in the ongoing intervention, CRC-NORDIET study [
38].
Generally, self-reported measures tend to over-report both duration and level of PA compared to objective methods [
55], but under-reporting has also frequently been documented [
55,
56] which may have several different explanations. A review of studies focusing on the comparison of objective measures versus self-reporting of PA was performed by Prince et al. [
55]. They found that self-reported measures of PA were higher than the objective measure when accelerometers were used. However, in the present study MVPA (only HUNT-PAQ) and MPA were under-reported with the questionnaires compared to SWA. This may be for several reasons; firstly, the intensity level of MPA was defined as activities resulting in slight breathlessness. Cancer patients undergoing disease-related treatment and in a recovery phase post-surgery might experience breathlessness at lighter intensity than before, due to treatment effects and comorbidities such as anaemia, chronic obstructive pulmonary disease, and physical deconditioning [
57,
58]. Breathlessness may result in over-reporting of higher intensity (VPA) and under-reporting of MPA. Slightly reduced physical function, measured by handgrip-strength and 30-s sit-to-stand test, was observed in the CRC patients participating in the present study as compared to healthy individuals in Norway (Table
1) [
59].
Secondly, the under-reporting of MPA might also be explained by the different techniques in recording physical activities used by the two methods. All activities are recorded by the SWA within a 24 h day, whereas the questionnaires rely on the participant’s memory and subjective evaluation of activity while responding to just a few questions [
60].
Thirdly, the degree of under-reporting of MVPA and MPA was higher with the HUNT-PAQ than with the NORDIET-FFQ. This might be due to the restricted opportunity for the participants to report both MPA and VPA in the HUNT-PAQ, which is possible with the NORDIET-FFQ. Moreover, under-reporting may also be explained by the different reporting intervals of frequencies in the responses; the NORDIET-FFQ contained responses for activities lasting both less than and above 10-min intervals, while the HUNT-PAQ only asked for activities lasting more than 10-min intervals. Therefore, increased accuracy in reporting of intensities was possible with the NORDIET-FFQ compared to the HUNT-PAQ, since intensities performed for less than 10 min were not recorded with the HUNT-PAQ.
Bias in reporting of intensity seems to be influenced by the amount of questions for a specific activity within a questionnaire, i.e. whether it contains a single-item question or domain-item questions [
61‐
64]. The self-reported sedentary time in the present study was based on a single-item question and was greatly under-reported by the HUNT-PAQ compared to SWA, an effect supported by other studies [
61,
64]. Since the HUNT-PAQ asked for sedentary time during day-time, a general definition of a day in the SWA was performed by removing night-hours between midnight and 6 am. Consequently, sedentary time during day-time recorded by the SWA was calculated from 6 am to midnight. However, this definition may be challenged in cancer patients facing several disease- and treatment side-effects influencing sleeping pattern due to increased need for resting time [
65]. A diary report from each participant would probably improve the definition of night-time resulting in higher precision in reporting sedentary time during day-time.
Vassbakk- Brovold et al. [
34] documented an over-reporting of 366% of MVPA recommendation with the short form International Physical Activity Questionnaire (IPAQ-sf) compared to the SWA among cancer patients undergoing chemotherapy. The IPAQ-sf contains 9 questions on PA [
66], whereas the NORDIET-FFQ and HUNT-PAQ contains 2 and 4 questions on PA, respectively. Both questionnaires in the present study contained few detailed question about type of PA activities. Thus, under-reporting of the activities may be due to decreased precision in reporting different kinds of activities during a day. However, the number of questions depends on the rationale of the questionnaire. In the present study, the aim was to estimate adherence to the PA recommendations based on the Norwegian FBDG. In clinical practice as well as intervention studies, it is advantageous to have a short and easy PA assessment tool to be used when monitoring adherence to the PA recommendations.
A small mean difference of only 4% was revealed for MVPA by the NORDIET-FFQ, whereas HUNT-PAQ under-estimated by 58% compared to the SWA. This is comparable with previous studies, which have reported mean differences around 44% (ranging from − 78% to 500%) [
55]. Evenly distributed differences above and below the mean difference in the Bland Altman plots indicated no systematic bias of activities in any of the questionnaires. However, linear regression revealed a systematic bias as shown by the significant negative slope for both questionnaires, indicating a trend towards more under-reporting with increased amount of PA. As can be seen from the Bland Altman plots, this negative trend seems to be accounted for by intensities higher than 250 and 200 min/week with the NORDIET-FFQ and HUNT-PAQ, respectively.
The limits of agreements were wide for both questionnaires, indicating weak ability to assess MVPA and MPA on an individual level. This has been supported by Ekelund et al. [
67] and by Vassbakk- Brovold et al. [
34], who validated the short form of the International Physical Activity Questionnaire (IPAQ-s) against an objective monitor among healthy Swedish adults and adult cancer patients, respectively. In the present study, limits of agreement were smaller at 150 min/week for MVPA and MPA for both questionnaires (about 500 min/week) than at higher levels of PA.
Hence, the NORDIET-FFQ was able to measure intensities up to about 250 min/week (i.e. including the PA recommendation of at least 150 min/week), but the HUNT-PAQ was less well suited to measure the corresponding intensities.
Studies including physical activities categorized in terms of different levels of exertion (light, moderate, vigorous) tend to result in more outliers, with VPA contributing the most outliers [
55]. The present study reported more outliers at higher levels of all intensities, of which the more extreme differences in reporting tended to be among males. Importantly, there were few observations with high amounts of PA, indicating high uncertainty and low interpretation of those data.
Previous studies differ in degrees of correlation between self-reported methods and objective measurements of PA [
55], with no specific trend. In the present study, there were poor correlations for all variables between the NORDIET-FFQ and SWA, whereas fair correlations were found between the HUNT-PAQ and SWA for MPA and sedentary time among women only. Ranking of individuals according to time in MPA and sedentary time were thus fairly good with the HUNT-PAQ.
NORDIET-FFQ identified 63% of individuals fulfilling the MVPA (specificity), but was not able to identify those in need of PA counselling (sensitivity). However, the HUNT-PAQ was able to identify 71% not fulfilling the MVPA and 36% of those who did. Hence, the NORDIET-FFQ provided a fairly specific measure of PA, but limited sensitivity to correctly classify individuals not fulfilling the MVPA. Thus, NORDIET-FFQ should be used with care in a clinical setting. In contrast, the HUNT-PAQ was able to identify those in need of PA counselling, but limited in identifying those who fulfilled the PA recommendations.
About 66% reported meeting the recommended level of MVPA with the NORDIET-FFQ (i.e. 150 min/week) whereas 55% actually met the MVPA according to the SWA. This is comparable with Vassbakk- Brovold et al. [
34] who also documented a higher proportion (i.e. 90%) of cancer patients perceiving themselves as meeting the MVPA recommendation of 150 min/week, while less than 50% actually met the PA recommendations recorded with SWA. This compares with a normal adult population in Norway, in which one in five met the national PA recommendations (i.e. 30 min/day) [
21]. Importantly, the physical activity assessment method used in the normal Norwegian population survey was different from the one used in the present study and the study of Vassbakk-Brovold et al. [
34]. Several barriers to meet PA recommendations among cancer survivors have been documented, of which treatment and disease-related factors are dominant [
68,
69]. Consequently, cancer patients may feel breathlessness at lighter intensities than normal, as abovementioned, resulting in over-reporting of PA. Thus, these considerations are important to bear in mind when using self-reported data on PA in cancer patients.
The main strength in the present study was the use of SWA as the objective reference method in evaluating self-reported PA from the two questionnaires. Additionally, there was high compliance with the protocols for both self-reporting PA and wearing time of SWA. The NORDIET-FFQ and the HUNT-PAQ asked for PA in recent weeks (i.e. the previous 1–2 months), whereas the SWA recorded PA the subsequent week. Since none of the patients in the present study underwent chemotherapy during the validation period (i.e. mean time since last treatment of 155 days), less variation due to treatment effect on physical activity was therefore assumed. The limitation in our study was the use of different cut-off points defining frequency and duration of PA, which might have caused misclassification into MVPA, MPA and VPA activities between the questionnaires and SWA. This is supported by other studies, of which one should be aware of the different qualities in measuring levels of PA between methods [
21,
66].