Nine people with DCM were asked to rate the survey scales; all were members of the RECODE-DCM steering committee. RECODE-DCM is an international community working to accelerate knowledge discovery and translation that can improve outcomes in DCM. This original project was delivered in partnership with AO Spine, a global nonprofit organization for spine surgeons. RECODE-DCM now sits within Myelopathy.org, a global DCM charity. People with DCM were purposively recruited to participate in the RECODE-DCM steering committee from within the professional network of the RECODE-DCM principal investigators. As such, participants and researchers knew each other for at least 1 year prior to completing the survey. Expertise in people with DCM was defined as having a diagnosis of DCM while also serving on DCM consensus workgroups, managing support groups, engaging in research, being involved in the Myelopathy.org charity, and/or other patient advocacy efforts for DCM. Notably, the first author of this paper was a survey rater. His summary scores largely agreed with the prevailing views of the other raters. Scores including his data were highly correlated with scores excluding his data (r=0.98). Of the 9 people with DCM who rated the scales, 6 provided demographic information. Of these 6 individuals, 4 were female, 3 resided in the United Kingdom, and 3 resided in the United States. Their mean Modified Japanese Orthopaedic Association score was 11.00 (SD 4.05; range 7-17 out of 18 points), and the mean age was 53.38 (SD 12.14; range 33.17-69.98) years. The survey was administered between March 1 and 14, 2022.

The methods specific to this study have been previously reported elsewhere [14]. In brief, we previously established the core outcome set (COS) and core measurement set (CMS) for DCM. Between establishing the COS, which determined that fatigue, mental health, pain, and economic impact were “core” outcomes, and subsequently establishing the CMS, a systematic review was undertaken to identify questionnaire instruments related to fatigue used in research related to DCM [15]. As no suitable scale was identified, the systematic review was expanded to conceptually similar conditions to DCM that affect the nervous system (ie, traumatic spinal cord injury [SCI], fibromyalgia, and stroke) [15]. This was performed primarily to determine whether any scale was suitable for inclusion in the CMS. This resulted in a list of 19 fatigue scales potentially suitable for consideration. These scales were presented to people with DCM involved in the RECODE-DCM steering committee for their consideration of the suitability of each scale.

People with DCM were queried regarding their perceptions of the relevance, comprehensiveness, and clarity of the fatigue scales (Multimedia Appendix 1). Question items and responses were adapted from the COSMIN manual. Perceptions were scored on 4-point Likert scales ranging from 1 (low intensity of agreement) to 4 (high intensity of agreement). For the relevance item, 1 implied “no items are relevant,” and 4 indicated “all items are relevant.” Responses to the comprehensiveness question ranged from 1 (“key aspects are missing”) to 4 (“no key aspects are missing”). For clarity, 1 indicated “not clear,” and 4 indicated “very clear.” Operationally, we defined a mean score of 3.5 or more across raters as a “good” score for each domain. Additionally, people with DCM were asked to estimate how long (in minutes) each instrument would require for completion. This was considered of relevance to the project’s principal purpose of informing a minimum dataset for DCM research where, given a wide range of core outcomes, the potential burden on participants to complete all assessments would need to be balanced against its broader value for DCM research. Finally, a free-response question was posed to people with DCM to elicit comments on the perceived qualities of each instrument. This was intended to provide preliminary evidence on the perspectives of people with DCM with respect to the potential fatigue scales listed below. This evidence will be used to streamline future mixed methods studies. This cohort of individuals was derived largely from the United States and the United Kingdom. Following the scoring of each domain (eg, relevance, comprehensiveness, and clarity), scores across domains were averaged to obtain a summary score from each person for each instrument.

In total, people with DCM were asked to evaluate and provide scores for 19 fatigue-related tools. These tools were the visual analog scale for fatigue [20], Fatigue Assessment Instrument [21], Modified Fatigue Impact Scale for Spinal Cord Injury (MFIS-SCI) [22], 36-item Short Form Health Survey (SF-36) fatigue subscale [23], Functional Assessment of Chronic Illness Therapy–Fatigue Scale (FACIT-F) [24], Multidimensional Assessment of Fatigue (MAF) [25], Chronic Fatigue Syndrome Activities and Participation Questionnaire [26], Fatigue Symptom Inventory (FSI) [27], Checklist Individual Strength (CIS) [28], FibroFatigue Scale (FFS) [29], Bristol Rheumatoid Arthritis Fatigue Multidimensional Questionnaire [30], Multidimensional Daily Diary of Fatigue–Fibromyalgia–17 (MDF-Fibro-17) [31], Fatigue Assessment Scale [32], numeric rating scale for fatigue [33], Chalder fatigue scale [33], Brief Fatigue Inventory (BFI) [34], Fatigue Impact Scale (FIS) [35], Multidimensional Fatigue Inventory (MFI) [36], and Piper Fatigue Scale [37].

In addition to providing subjective scoring, respondents were asked to provide free-text responses via an online survey to discuss what they did or did not like about each tool. The survey was sent to participants via email by one investigator (BMD), a male trainee physician trained in qualitative methods. Data were collected virtually. Responses were collated in Microsoft Excel. Responses were read through and coded and recoded after a period of 1 month to arrive at an accurate and consistent representation of the data. Analyses were conducted by one investigator (TFB), a male PhD scientist with DSM trained in qualitative methods. Thematic analysis was used to inductively identify codes and themes in the written responses from participants. For example, comments about requiring naps during the day were coded as “nap,” and comments about disturbed sleep were coded as “sleep.” Subsequently, themes were identified from underlying codes and stored in Microsoft Excel. As such, codes broadly related to the length, layout, or clarity of the instrument were themed as “structure.” Codes including words such as “psychosocial” and “socioemotional” were assigned to the theme “psychosocial.” Codes related to the length of recall were themed “recall.” Filler words such as “liked,” “mentions,” and “perhaps” were discarded. All the free responses were then entered into word cloud–generating software that extracted the top 10 most frequent words, which is a validated method [38,39]. The first author analyzing the results continuously reflected on whether results were due to his own biases, including that he is a person with DCM, and he discussed underlying responses, codes, and thematic results with the second and last authors. The results were compared to the published COS for DCM. Participants were invited to comment on the results and were included as authors if they desired. This study was intended to inform future qualitative focus groups and larger-scale qualitative survey studies.

Given the qualitative-to-semiquantitative nature of this study, no formal hypothesis testing was undertaken. Nevertheless, descriptive and summary statistics were calculated in Microsoft Excel.

Ethics approval was obtained from the University of Cambridge (HBREC2019.14) in accordance with the Declaration of Helsinki.

When pooled across all scores from all participants for all instruments, the median summary score was 3.67 (IQR 2.83-4), indicating that raters scored all instruments rather well. Four instruments had a between-participant average of more than 3.67 (MFIS-SCI, FACIT-F, MAF, and Fatigue Assessment Instrument). Four of the scales received a summary score below 3, specifically the FFS (summary score=2.48, SD 0.60), FSI (summary score=2.89, SD 0.91), MDF-Fibro-17 (summary score=2.93, SD 0.76), and CIS (summary score=2.93, SD 0.76). Generally, across scales, comprehensiveness was scored lower than relevance and clarity. Scores for the various scales are shown in Figure 1. The frequency of themes with respect to various scales is visually presented in Multimedia Appendix 2.

Of the top-scored scales, the mean times to complete were approximately 6 minutes (MFIS-SCI: mean 6.00, SD 4.03 minutes; FACIT-F: mean 7.56, SD 3.97 minutes; MAF: mean 6.67, SD 3.46 minutes). Thus, most scales could easily be administered without adding substantial testing burden in many clinical trials.

In total, 26 codes were generated from responses that were mapped to 7 themes, as can be observed in Table 1. One prevailing theme was an insufficient discussion of sleep. This was noted in regard to the FIS, MFIS-SCI, MAF, BFI, and MFI. Regarding the FIS some specific comments included the following:

Another respondent suggested that questions be added to the FIS and MFIS-SCI on requiring naps, such as “Do you require a nap during the day?” (participant 9). Regarding the MFIS-SCI, one responder said the following:

Another comment noted that the BFI “doesn’t include pain or sleep” (participant 7). On the other hand, the FACIT-F was commended for discussing sleep:

Other common themes were related to the effects of pain on fatigue and of fatigue on employment status. For example, the MAF, MDF-Fibro-17, MFI, CIS, and FFS were criticized for not accounting for the effects of pain on fatigue and/or of fatigue on employment status. The FACIT-F and MFIS-SCI, however, were commended for mentioning pain. Regarding the MFIS-SCI, it was said that “it discusses physical discomfort which is good” (participant 7).

Similarly, mood and psychological state was also a topic frequently commented on. The FACIT-F and SF-36 were commended for discussing “socioemotional impact.” Other scales, such as the Fatigue Assessment Scale and CIS, were criticized on this topic.

Other frequent comments related to the topic of aesthetics and setup or structure of the survey. This included comments such as “Far too detailed” (participant 3), “Feels like duplicate questions” (participant 5), and “It would take me [a long] time to recollect all the information it asks for” (participant 7) for the FSI and “A bit long winded for a person with DCM to fill out” for the BFI. This topic also included comments that the responder was “distracted/bothered by the layout or dark highlighted bars” (participant 4) for the BFI.

Another common topic of comment was on the duration of recall required from each scale. For example, scales that asked about symptoms within a 24-hour period or “right now” were viewed unfavorably compared to those that asked about symptoms within a 2- to 4-week period. For example, regarding the BFI, participant 8 commented that “24 hours seem like such a short time frame.” Another participant wrote the following regarding the BFI:

This sentiment was related to views that DCM is variable and that too brief a recall period would fail to capture the worst days. For example, regarding the Chalder fatigue scale, one person wrote the following:

In analyzing the criticisms of the bottom 3 scales in terms of performance, length and detail, format, and extraneous concepts were frequently noted for the FSI, CIS, and FFS. For example, criticisms of the FFS frequently included its long, poorly designed layout. Comments on the CIS included “Not as easy as the other formats” (participant 2) and “confusing—the scoring” (participant 4). Similarly, comments on the FSI included “Misses out on the physical fatiguability possible in DCM” (participant 1), “far too detailed” (participant 3), “would dwell on these” (participant 4), and “It would take me a long time to recollect all the information it asks for” (participant 7). Conversely, despite sometimes being criticized for being long, the top-scoring scales were described as more “clear” but also “relevant,” “comprehensive,” and “to the point.” Regarding the MFIS-SCI, comments included “Nicely broken down and helpful” (participant 4), “clearer questions and less bothersome” (participant 5), and “I like this scale—it seems comprehensive” (participant 7). Similarly, on the FACIT-F, one person responded that “this one seems more well-rounded” (participant 8). Regarding the MAF, one person responded that “I found this instrument a bit more to the point” (participant 5).

The primary purpose of this study was to provide preliminary content validation of candidate fatigue scales existing in the current literature, and our primary finding was that the MFIS-SCI, MAF, and FACIT-F demonstrated good content validity. These scales could represent starting points to implement a fatigue scale into use in research in DCM. Our results further suggest that any future development of fatigue scales specific to DCM should incorporate assessments of sleep, effects of pain on fatigue, psychosocial factors such as stress due to having a chronic condition, and impact of fatigue on employment status. Alternatively, additional assessment scales for pain (eg, Neck Disability Index, Brief Pain Inventory, and numeric rating scale for pain), employment status, and mental health could supplement deficiencies in assessments of fatigue on these factors. These are in addition to assessments of lower-extremity fatigability (such as the 6-minute walk test) that have been shown to differ in healthy controls [40]. Furthermore, future studies examining the impact of symptomatic fatigue and neuromuscular fatigability on quality of life are needed.

As noted above, fatigue was identified as a core outcome of DCM; however, we have little understanding of its biological basis in DCM. This presents several issues related to measurement validity, and we were previously unable to identify any scales with sufficient scientific evidence to be included in the CMS [13,15]. People with DCM have identified sleep, pain, employment status, and mental health as common critical components of fatigue in DCM. While the overall relationship between many of these and fatigue is poorly understood (and this includes the causal direction), we discuss them subsequently and present a hypothesized formative model of poor sleep, pain, and psychosocial factors driving fatigue, with employment status likely a consequence of fatigue.

The MFIS-SCI was the least surprising scale to be relevant to people with DCM as it is drawn from a highly similar clinical population (eg, SCI vs DCM) rather than chronic fatigue syndromes. These factors suggest that the MFIS-SCI is the most likely to effectively translate to a DCM population. The Modified FIS is also validated in multiple sclerosis and is designed to capture the “impact” of fatigue [41,42]. In comparison, the MAF was also designed to capture the “impact” of fatigue but structures questions as they relate to specific activities. The MAF does not cover cognitive aspects of fatigue (which the MFIS-SCI does); therefore, it is possible that people with DCM viewed the MAF as incomplete [25]. Additionally, the FACIT-F contains questions about feeling ill, feeling nauseous, side effects of treatment, and worries about death, which may have resulted in views that it is less relevant than the MAF or MFIS-SCI [24]. The FACIT-F, however, included more items directed at psychosocial factors. It is possible that the length of the FACIT-F (40 items in total) and the items that may be viewed as less relevant offset the effects of including more psychosocial items. This phenomenon may relate to the respective origins of the MAF (rheumatoid arthritis) and FACIT-F (patients with cancer).

Improving our understanding of the pathophysiology of DCM is critical. Similarly, furthering our understanding of the pathophysiological mechanisms leading from spinal cord compression to symptoms such as fatigue will guide novel therapies to alleviate symptom burden in patients. As noted above, several of the themes identified by our expert panel of people with DCM are interrelated and consistent with other fatigue-related conditions. Themes such as sleepiness, pain, and mental health have been richly documented in fibromyalgia. It is not known, however, the extent to which the causal directionalities of these constructs are similar in DCM as they are in other conditions. Together, this cluster of symptoms is described using the biopsychosocial model (Figures 2 and 3), which has been applied to appreciate the interrelationship among pain, mental health, and fatigue in a variety of other conditions, including chronic fatigue syndrome, fibromyalgia, cancer, and traumatic brain injury. Importantly, there is preliminary evidence in people with DCM on a number of factors within the biopsychosocial model that contribute to fatigue (such as pain, inflammation, endurance, and mental health) [9,12,40,51,57].

In cancer, for example [58], fatigue was hypothesized to be caused by reductions in strength, endurance, cardiopulmonary fitness, and body composition as physiological factors. Similarly, fatigue was proposed to be directly caused by changes to immunological, metabolic, and endocrine functions and systemic inflammation. Fatigue was proposed to be caused via indirect pathways made up of psychological (eg, depression, anxiety, distress, and cognition), behavioral (eg, sleep quantity and quality and appetite), and social (eg, social interaction and positive reinforcement) factors. While this description from McNeely and Courneya [58] did not include pain, other relevant descriptions of biopsychosocial models of fatigue, such as for rheumatoid arthritis [59], have added pain to their model.

Fatigue tools did not feature in the final DCM minimum dataset [13]. While the COS included fatigue, mental health, pain, and employment, none of these were explicitly included in the CMS. The measurement set pragmatically focused on one tool per domain (related group of core outcomes) and prioritized tools with a high-quality evidence base for use in DCM. This was intended to prevent measurement burden in trials. As such, the SF-36 version 2 was selected to address core outcomes such as mental health and employment [13,15]. The minimum dataset is designed to be used in all DCM trials. It is notable in our analysis, however, that the SF-36 was reported to be less clear, relevant, and comprehensive than the MFIS-SCI, MAF, and FACIT-F. As such, for dedicated investigations of fatigue in DCM, additional measurement tools will be required.

This study examined the perceptions of people living with DCM regarding the appropriateness of various existing self-reported outcome scales assessing fatigue. One limitation of this study was the small sample size, potentially resulting in a failure to reach saturation. However, our sample of people with DCM was recruited from multiple countries across Europe and North America and consisted of seasoned advocates who served on the RECODE-DCM steering committee and led peer support groups educating and advocating for people with DCM. Given these backgrounds, our panelists were well suited to gauge the included fatigue surveys based on their own experiences as well as common questions and symptom concerns raised by others. Additionally, we were unable to capture data on potential comorbidities from our panelists as this study was a retrospective secondary analysis. A follow-up, true qualitative study using focus groups capturing comprehensive data on comorbidities that could impact perspectives on fatigue should be conducted based on the short-listed scales to recruit additional opinions and control for confounding comorbidities. Nevertheless, our results largely agree with our previously published COS based on a modified Delphi process relying on the opinions of 113 people with DCM. This COS defined fatigue, mental health, pain, and economic impact as core outcomes. Another potential limitation is related to cross-cultural adaptability [60]. All scales were tested in English, and the participants surveyed in this study were all native English speakers. While English makes nuanced distinctions among words such as “sleepy,” “tired,” and “fatigued,” other languages make no distinction between some of these terms. For this reason, our recommendations should only be taken for English-based studies unless there is a validated version in a research study’s target language (eg, the Modified FIS). Finally, establishing the validity of an instrument encompasses numerous components. Other aspects of validity (eg, structural validity, convergent validity, internal consistency, and repeatability) remain to be established in DCM specifically for these scales.

Fatigue is recognized as an important symptom of DCM, but it remains poorly understood. Existing measures are considered reasonable for use in DCM but need further field-testing. The MFIS-SCI, MAF, and FACIT-F were the preferred outcome measures according to the people with DCM who participated in this study. While these represent a starting point, panelists recognized limitations, and it is likely that dedicated measurement and/or additional measurement approaches are required to address fatigue in DCM.