Can a brief screening scale accurately detect ME/CFS?

Key messages

A tool such as the DePaul symptom Questionnaire (DSQ) which assesses ME/CFS symptoms in a standardised way would be beneficial to researchers and clinicians.
The initial version of the DSQ (the DSQ-1) is long and includes 99 questions, 54 of which relate to the frequency and severity of symptoms.
The DSQ-Short Form (DSQ-SF) version is shorter with 14 questions, but could an even shorter version with just 4 questions – the DSQ-Brief – be used?
This study collected information from 3,427 participants, including 2,313 with ME/CFS, 192 with post-polio syndrome, 268 with multiple sclerosis, 299 with long COVID and 355 healthy participants.
The researchers used this information to compare how accurately the DSQ-1, DSQ-SF and DSQ-Brief could distinguish cases of ME/CFS from healthy controls, and from those with multiple sclerosis, post-polio syndrome and long COVID.
While all three versions of the DSQ could accurately distinguish those with ME/CFS from healthy controls, compared with the DSQ-1 and DSQ-SF, the DSQ-Brief was less accurate at differentiating cases of ME/CFS from the other chronic illnesses.
The DSQ-Brief could be used as a screening tool to detect likely cases of ME/CFS, but tools such as the DSQ-1 and DSQ-SF would still be needed to differentiate them from other illnesses that cause fatigue.
More research is needed to assess whether these results could be replicated in more diverse samples of people with ME/CFS.

Introduction

Without an operationalised diagnostic biomarker for ME/CFS, the illness is identified based on the presence of certain symptoms – in the absence of another disease that would explain them – using case definitions such as the Canadian Consensus Criteria (CCC), Fukuda criteria and the International Consensus Criteria (ICC) for ME/CFS, CFS and ME, respectively.

It is thought that over 20 case definitions for ME/CFS exist. The symptoms they require for a diagnosis vary, are inconsistently defined, and are sometimes even conflicting. For example, while the CCC includes ‘post exertional malaise’”’ (PEM) as a symptom that is essential for a diagnosis to be made, the ICC requires ‘post-exertional neuroimmune exhaustion’ which, although a similar concept to PEM, is defined differently, and the Fukuda criteria does not require PEM at all.

Using the different diagnostic criteria in research leads to complications, particularly when the results from several studies need to be considered as a whole to further the understanding of a specific topic, or to estimate the number of people with ME/CFS (prevalence). This is because individuals diagnosed using different methods may not be directly comparable. For example, a study funded by ME Research UK found that the ICC for ME/CFS identifies individuals with more severe symptomatology compared with the Fukuda criteria for CFS.

Therefore, the use of methods which assess symptoms of ME/CFS in the same way – even where participants with ME/CFS have been identified using different diagnostic criteria – would enable better comparisons to be made between studies, and help ME/CFS research progress.

The DePaul Symptom Questionnaire

In 2010, Professor Leonard Jason and colleagues at DePaul University, Chicago, created a tool which aims to provide a uniform way of assessing ME/CFS symptoms – the DePaul Symptom Questionnaire (DSQ). The original DSQ (the DSQ-1) asked 99 questions overall, 54 of which relate to symptoms of ME/CFS from both the CCC and Fukuda criteria.

Assessing how well a diagnostic tool works

When a new questionnaire or diagnostic test is developed, it is important to consider how well it works based on whether it is able to identify a positive result for those who truly have the disease (true positive), and the absence of the disease in someone who does not have it (true negative).

When a person takes a diagnostic test, there are four possible outcomes:

True negative	False negative	False positive	True negative
Do not have the disease *Negative* test result	Do have the disease *Negative* test result	*Do not* have the disease *Positive* test result	Do have the disease *Positive* test result

Based on these results, three key measures are then used to assess how well the test has worked. These are:

Sensitivity: The ability of the test to detect true positives.
Specificity: How well the test detects true negatives.
Accuracy: The number of true positives and true negatives divided by the total number of people asked to complete the test (both the true positives and negatives, and the false positives and negatives added together).

In addition to having high values of sensitivity, specificity and accuracy, a test also needs to be user friendly – for both the person administering the test, and for the one undergoing it. It also needs to be cost-effective.

New research

While the DSQ-1 has been found to work well, with studies showing high sensitivity, specificity and accuracy, the large number of questions asked makes it difficult to use in practice.

With this in mind, in 2019 the team at DePaul University developed a shorter version of the DSQ, the DSQ-short form (or DSQ-SF). This was based on the DSQ-1 but asked 4 questions relating to a smaller number of key symptoms of ME/CFS. Again, this questionnaire has been found to work well, but as the DSQ is completed by the participants themselves, an even shorter version would be useful should it achieve similar levels of accuracy.

Therefore, in a recent paper, Jason and colleagues set out to investigate how well a new version of the DSQ – the DSQ-Brief, which has only 4 questions – identified cases of ME/CFS compared with the DSQ-1 and DSQ-SF.

What did the study do?

The researchers considered the sensitivity, specificity and accuracy of the DSQ-Brief, and looked at how well it could distinguish those with ME/CFS from healthy controls, and from those with other long-term illnesses: multiple sclerosis (MS), long COVID and post-polio syndrome.

The study included 3,427 participants: 2,313 with ME/CFS, 192 with post-polio syndrome, 268 with MS, 299 with long COVID and 355 healthy participants, all identified from multiple, international datasets.

The methods used to recruit and diagnose participants in each disease group were as follows:

ME/CFS: Some were defined through self-report, while others received a diagnosis based on medical history and physical examination by an experienced healthcare professional. No specific diagnostic criteria were reported to be used.
Long COVID: Recruited through online support forums relating to the experience of ongoing symptoms following a COVD infection. There was no diagnosis or assessment of symptoms by a health professional.
MS: Recruited as part of another study investigating chronic illnesses. It was not specified how those with MS were diagnosed.
Post-polio syndrome: Recruited through social media, email and online support groups. Again, there was no mention of how they were diagnosed.

The information obtained on participants from the existing datasets included sex, age, marital status and level of education, alongside their responses to three versions of the DSQ: the DSQ-1, DSQ-SF and DSQ-Brief.

Results

All three questionnaires were able to distinguish those with ME/CFS from healthy controls with high accuracy, sensitivity and specificity.

DSQ-1: Accuracy 99%, sensitivity 98%, specificity 99%.
DSQ-SF: Accuracy 98%, sensitivity 97%, specificity 99%.
DSQ-Brief: Accuracy 96%, sensitivity 94%, specificity 98%.

Both the DSQ-1 and DSQ-SF were able to distinguish those with ME/CFS from those with other chronic illnesses (post-polio syndrome, MS and long COVID considered together) with high accuracy, sensitivity and specificity. However, the DSQ-Brief did not perform so well.

DSQ-1: Accuracy 94%, sensitivity 95%, specificity 94%.
DSQ-SF: Accuracy 93%, sensitivity 93%, specificity 94%.
DSQ-Brief: Accuracy 84%, sensitivity 85%, specificity 81%.

The DSQ-Brief also had reduced accuracy, sensitivity and specificity compared with the DSQ-1 and DSQ-SF, when considering how well the questionnaires could identify people with ME/CFS from those with post-polio syndrome, MS and long COVID separately.

Discussion

The DSQ-1 and DSQ-SF were both able to accurately distinguish cases of ME/CFS from healthy controls, and those in the other chronic illness groups considered. In general, the accuracy scores were very similar for the two questionnaires, despite the difference in the numbers of questions asked. Although the DSQ-Brief achieved high accuracy when comparing those with ME/CFS with healthy controls (96%), the accuracy was reduced when comparing those with ME/CFS with other chronic illnesses (92%, 88%, and 82% compared with post-polio-syndrome, MS and long COVID, respectively). These findings suggest that, while the DSQ-Brief could be used as a screening tool to detect likely cases of ME/CFS, tools such as the DSQ-1 or DSQ-SF would still be needed to differentiate them from other illnesses that cause fatigue.

Interestingly, despite the observed similarities between symptoms of ME/CFS and long COVID, both the DSQ-1 and DSQ-SF – which capture more detailed symptoms including autonomic dysfunction and gastrointestinal symptoms – were able to accurately distinguish the two disease groups: 99% and 98% accuracy for the DSQ-1 and DSQ-SF, respectively. For the DSQ-Brief, which captures information on fewer symptoms, the accuracy was decreased at 82%. When considering this finding, it is important to note that those with long COVID were recruited from social media groups, rather than through a confirmed diagnosis. More research is needed to establish whether these findings could be replicated in those with a formal diagnosis.

Overall, the findings from this study are important, and indicate that the DSQ-Brief may provide a useful tool to screen for ME/CFS in research. However, caution must be applied when interpreting the results due to the differing recruitment methods used. In addition, the study did not consider how accurately the versions of the DSQ could distinguish cases of ME/CFS in different severity groups, or for different duration of illness. Another point to consider is that the majority of the participants in this study were middle-aged, white females. This means that the findings of the study may not be applicable to other age groups, males with ME/CFS, or to those with ME/CFS in other ethnic groups – limitations that are not unique to this study. More research is needed to assess whether these results could be replicated in more diverse samples of people with ME/CFS.