The main finding of this study was that all disability measurements were highly significantly correlated: the physical functioning scale of the SF-36, the number of steps on an activity meter, and the percentage peak VO2 relative to a reference value of healthy controls. Few studies have been performed to validate disability measurements in ME/CFS patients, especially with the use of more objective measures other than history taking. To the best of our knowledge this is the first study to have measured all three in determining activity levels in ME/CFS patients.
In comparison to values reported in other ME/CFS populations, our study participants had similar steps per day [21], and %peak VO2 values [22, 27, 30]. The reported median or mean physical functioning scale of the SF-36 questionnaire vary widely between studies ranging between 17 and 59% [13–18]. This large variation is probably due to patient selection. Inclusion of more severely affected patients (with lower physical functioning scales) reduces the mean of the data. This is shown in the study of Pendergrast et al. [18] where the housebound patients had a mean physical functioning scale of 17% while the non-housebound patients had a score of 42%. In the present study the mean physical functioning scale was 46%, with a range between 0 and 95%, indicating a large variation in ME/CSF severity. This strengthens the generalizability of the observed relations between VO2 and activity measurements and the physical function scale.
All three types of measurements in this study are related to the activity level of patients. Activity in women is partially determined by age, race, menopausal status, educational level, body mass index, depressive symptoms, smoking, chronic medical conditions, and pain [42]. The peak VO2 is influenced by genetics, gender, age, training status, exercise mode, bedrest, altitude, body composition, medication, the capacity of the respiratory and circulatory systems to take up and transport oxygen, and the capacity of the working muscles to receive and use oxygen. In ME/CFS patients the degree of fatigue/exhaustion, post-exertional malaise, underlying metabolic abnormalities, fibromyalgic pain, kinesiophobia and the use of medication may further influence physical activities.
A large number of studies have examined the validity of the SF-36 questionnaire, showing that the physical functioning scale discriminates between various diseases and healthy controls [43]. In ME/CFS patients Jason et al. reviewed the ability of the different subscales of the SF-36 questionnaire to discriminate CFS patients from healthy controls for the goal of operationalizing the measurement of “substantial reductions in previous levels of occupational, educational, social, or personal activities” [44]. The authors found that the physical functioning subscale was slightly less than optimal to discriminate between patients and healthy controls, using an area-under-the-curve (AUC) cut-off value of > 0.90 for optimal discrimination. In the community-based sample the AUC of the physical functioning scale was 0.84 and in the tertiary care sample 0.87. However, another study found an AUC for assessing substantial reductions of the physical functioning scale of 0.91 [45], suggesting that the use of the physical functioning score is valid with an acceptable sensitivity and specificity.
Several studies have shown a decreased peak VO2 in ME/CFS patients [46]. However, only one study determined the relation between the peak VO2 and accelerometer data in female ME/CFS patients: higher peak VO2 values were related to a higher physical activity time, physical activity energy expenditure, and a mean energy expenditure [47]. In the present study a significant linear relation was found between the peak VO2 and the number of steps as assessed by the Sensewear™ meter. Given the above mentioned large number of influencing factors on peak VO2 and physical activity in combination with the variation of steps on consecutive days, it is not surprising that, although a very significant relation between peak VO2 and the number of steps exists, the prediction intervals are wide.
There are also limited data available on the relation between peak VO2 and activity questionnaires. One study found a moderate association between exercise capacity and, using the Chronic Fatigue Syndrome Activities and Participation Questionnaire, activity limitations/participation restrictions in patients with ME/CFS [48]. Our data on the relation between peak VO2 and the physical function scale are consistent with their observations.
With respect to the relation between the physical functioning score and the activity meter, a previous study failed to demonstrate a difference in activity, as measured by an actometer, between two CFS groups, one with a mean physical functioning score of 54% and another with a mean score of 33% [49]. One case study noted a discrepancy between the psychometrically assessed improvement in function after graded exercise therapy versus the decrease in steps after therapy [50]. In the present study, we found a significant relation between the physical function scale and the number of steps; this was the strongest correlation in our study. The close correlation is not unexpected in ME/CFS patients as a previous study showed that ME/CFS patients are more aware of their daily physical activities compared to healthy controls [51]. Nevertheless, there is considerable variation between individual patients at specific levels of the physical functioning scale. For example, at the level of 30% of the physical functioning scale, the number of steps of individual patients varied between 1558 and 4266. At a physical functioning scale of 60%, the number of steps varied between 6277 and 9641, reflecting quite different function. The same holds true for the peak VO2 versus the number of steps taken. For a peak VO2 between 50 and 60% of normal the number of steps of patients ranged between 1135 and 4683.
Although the physical functioning scale is adequate to make a distinction between a group of diseased and non-diseased individuals, the physical functioning scale is less useful to describe the degree of disability in individual patients, given the variation of the number of steps for a certain value of the physical functioning scale. An integrated approach of more than one type of measurement is therefore needed for the purposes of research study outcomes, disability determination and individual patient management.
Our observation of the relation between the peak VO2, the number of steps and the physical functioning scale is further strengthened by the observation that 20 patients who were re-evaluated because of worsening of symptoms had a significant reduction in peak VO2, the number of steps and the physical functioning scale (Fig. 2). It is interesting to note that 1 of the 20 patients who presented with worsening of symptoms actually had an improvement in the number of steps/day and an unchanged physical function scale (individual data not shown). This stresses the notion that symptomatology needs to be confirmed by objective measurements.
It can be argued that the peak VO2 is the limit for a certain amount of activity: the lower the peak VO2 the less activity can performed or steps taken. Ideally, nomograms of the relation between peak VO2, number of steps per day and the physical functioning scale should be available for ME/CFS patients, allowing clinicians to determine whether patients have gross over- or under-performance (in terms of number of steps) and explore the reasons for this over- and underperformance. Similarly, an imbalance between the physical function scale and the number of steps could be explored. This could also be beneficial for patient activity management. However, from the present data the construction of nomograms is not possible and future studies are needed.
Our data highlight a discrepancy between recommendations for CPET. In studies measuring %peak VO2, guidelines have advocated using the RER for assigning a test as maximal effort using an RER > = 1.1 [52]. In our subgroup analysis, 54% of the patients reached a peak RER > = 1.1 while 46% reached a peak RER < 1.1. Despite this difference (by definition P < 0.0001), the %peak VO2, the number of steps/day and the physical functioning scale of the SF-36 were not different between patients reaching a peak RER > = 1.1 and patients with a RER < 1.1. Moreover, the relations between the three measurements: %peak VO2, number of steps per day and the physical functioning scale were not different between patients with and without a RER > = 1.1. In ME/CFS patients a number of studies have shown that metabolic skeletal muscle abnormalities are present [5, 53, 54]. These skeletal muscles abnormalities may limit the maximal performance with RER values above 1.1. As Mezzani stated: “It must be borne in mind, however, that patients with severely impaired exercise tolerance can attain skeletal muscle strength exhaustion even earlier than central hemodynamic and ventilatory factors become limiting, interrupting exercise at peak respiratory exchange ratio values even lower than 1.00” [41]. Our data therefore suggest that an RER < 1.1 should not be used as an exclusion criterion for future studies of exercise performance in ME/CFS patients. Further support for this position comes from examining the fibromyalgia subgroup. In the present study 59% of the ME/CFS patients had a concomitant diagnosis of fibromyalgia. In all these patients, the reason for exercise termination was leg muscle pain in combination with leg muscle strength exhaustion. In these fibromyalgia patients the RER was significantly lower than in the ME/CFS patients without fibromyalgia. The low RER in fibromyalgia patients can be explained by the earlier exercise termination because of muscle pain. The %peak VO2, the number of steps and the physical functioning scale were not different between patients with and without fibromyalgia. This observation again argues against the use of a low RER as exclusion criterion.
Limitations
This was a retrospective analysis taking data from patients with a maximum interval of 3 month between the 3 different measurements. We made the assumption that the clinical course would be stable over this period of time, but the retrospective nature of the study did not allow us to confirm this. A prospective study would be needed to evaluate the variability in measurements over time. Because of the limited number of patients, we did not correct for all the confounding factors as mentioned above for the three measurements. The Sensewear™ activity meter is not available anymore, but the present commercial actographs and smart watches have step measurements included. Only patients who were evaluated because of worsening of symptoms were re-analysed. We have no repeat data on patients who were stable or who improved.