Defining a High-Quality Myalgic Encephalomyelitis/Chronic Fatigue Syndrome cohort in UK Biobank

Gemma L. Samms; Chris P. Ponting

doi:10.3310/nihropenres.13956.1

Home Browse Defining a High-Quality Myalgic Encephalomyelitis/Chronic Fatigue...

ALL Metrics

-

Views

66

Downloads

Get PDF

Get XML

Export

▬

✚

Research Article

Defining a High-Quality Myalgic Encephalomyelitis/Chronic Fatigue Syndrome cohort in UK Biobank

[version 1; peer review: 2 approved]

Gemma L. Samms¹, Chris P. Ponting ¹

PUBLISHED 28 Apr 2025

Author details Author details

¹ Institute of Genetics and Cancer, The University of Edinburgh MRC Human Genetics Unit, Edinburgh, Scotland, UK

Gemma L. Samms
Roles: Data Curation, Formal Analysis, Investigation, Writing – Original Draft Preparation, Writing – Review & Editing

Chris P. Ponting
Roles: Conceptualization, Funding Acquisition, Investigation, Supervision, Writing – Review & Editing

OPEN PEER REVIEW

REVIEWER STATUS

Abstract

Background

Progress in Myalgic Encephalomyelitis/Chronic Fatigue Syndrome (ME/CFS) research is being slowed by the relatively small-scale studies being performed whose results are often not replicated. Progress could be accelerated by analyses of large population-scale projects, such as UK Biobank (UKB), which provide extensive phenotype and genotype data linked to both ME/CFS cases and controls.

Methods

Here, we analysed the overlap and discordance among four UKB-defined ME/CFS cohorts, and additional questionnaire data when available.

Results

A total of 5,354 UKB individuals were linked to at least one piece of evidence of MECFS, a higher proportion (1.1%) than most prevalence estimates. Only a third (36%; n=1,922) had 2 or more pieces of evidence for MECFS, in part due to data missingness. For the same UKB participant, ME/CFS status defined by ICD-10 (International Classification of Diseases, Tenth Revision) code G93.3 (Post-viral fatigue syndrome) was most likely to be supported by another data type (72%); ME/CFS status defined by Pain Questionnaire responses is least likely to be supported (43%), in part due to data missingness.

Conclusions

We conclude that ME/CFS status in UKB, and potentially other biobanks, is best supported by multiple, and not single, lines of evidence. Finally, we raise the estimated ME/CFS prevalence in the UK to 410,000 using the most consistent evidence for ME/CFS status, and accounting for those who had no opportunity to participate in UKB due to being bed- or house-bound.

Plain Language Summary

Plain English summary

Myalgic Encephalomyelitis / Chronic Fatigue Syndrome (ME/CFS) is a highly debilitating and relatively common disease whose causes are unknown. Most of its research compares features (symptoms, molecules, cells or genes, for example) from small numbers of people with ME/CFS, against these features from others without the disease. Studies using small sample sizes have not yielded replicated discoveries that would quicken the pace of ME/CFS research towards effective therapies.

Alternatively, ME/CFS research could take advantage of population-scale bioresources, such as the UK Biobank, which holds diverse genetic, molecular, cellular, imaging and questionnaire data on nearly 500,000 individuals. This has the added advantage of being cheaper than recruiting a new cohort, and generating new data.

For ME/CFS, this approach raises the difficulty of how to best categorise people with this disease, for example by questionnaire response or through linked electronic health records. In the UK Biobank, there are four ways to categorise people with ME/CFS.

This study’s results show that just over 1% of the UK Biobank participants could be categorised as having a ME/CFS medical diagnosis. However, not all evidence for their ME/CFS diagnosis is consistent.

By cross-referencing different data in UK Biobank, we show that a participant’s ME/CFS diagnosis is best supported by two or more lines of evidence. Using the most consistent evidence, and accounting for those who – due to their illness – could not participate in the UK Biobank, we estimate that the UK’s prevalence of ME/CFS is 410,000.

Keywords

Myalgic Encephalomyelitis/Chronic Fatigue Syndrome prevalence; Population bioresource; Electronic health records; Post-viral fatigue syndrome

Corresponding authors: Gemma L. Samms, Chris P. Ponting

Competing interests: No competing interests were disclosed.

Grant information: This project is funded by the National Institute for Health and Care Research (NIHR) and Medical Research Council (MRC), grant number MC_PC_20005. GLS was funded by PhD-level research funding from ME Research UK. The views expressed are those of the author(s) and not necessarily those of the NIHR or the Department of Health and Social Care.
The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Copyright: © Crown copyright, 2025 Samms GL and Ponting CP. This open access work is licensed under the Open Government Licence v3.0

How to cite: Samms GL and Ponting CP. Defining a High-Quality Myalgic Encephalomyelitis/Chronic Fatigue Syndrome cohort in UK Biobank [version 1; peer review: 2 approved]. NIHR Open Res 2025, 5:39 (https://doi.org/10.3310/nihropenres.13956.1) First published: 28 Apr 2025, 5:39 (https://doi.org/10.3310/nihropenres.13956.1) Latest published: 28 Apr 2025, 5:39 (https://doi.org/10.3310/nihropenres.13956.1)

Introduction

Pathomechanisms of myalgic encephalomyelitis (ME; also known as chronic fatigue syndrome, CFS) are unknown. ME/CFS is a highly disabling, multi-system disorder affecting about 65 million people worldwide¹, although this is rising steeply due to symptoms of many people with Long Covid meeting ME/CFS diagnostic criteria². Approximately two-thirds of ME/CFS cases report an infection at onset of their symptoms^3–5. The hallmark symptom of ME/CFS is post-exertional malaise, when new symptoms arise, or previous ones worsen, following even minimal exertion that are disproportionate to the level of exertion, prolonged in recovery and not immediately alleviated by rest⁶. Additional symptoms are fatigue that does not go away with rest, pain including headache, cognitive dysfunction and multiple sensitivities⁷. ME/CFS is a female-biased disease with about five-times more women affected than men⁷. The annual economic impact of ME/CFS is $36–51 billion in the USA alone⁸.

Among patients’ and clinicians’ priorities for future ME/CFS research is the identification of ME/CFS diagnostic biomarkers and risk factors⁹. Little progress on this has been made, with biomarker and genetic studies yielding results that are mostly not replicated subsequently^10,11. One recent strategy for investigating ME/CFS pathomechanisms involved an in-depth analysis of 17 ME/CFS cases, a project costing $8 million¹². An alterative approach is to develop national cohorts of well-phenotyped and genotyped ME/CFS cases, such as DecodeME in the UK, a study costing £3.2 million¹³. A quicker and cheaper approach, however, is to analyse pre-existing data from national biobanks that have recruited hundreds-of-thousands of participants¹⁴. For this, data access costs can be modest, for example £9,000 for 3-year access to UK Biobank (UKB) data on approximately 500,000 people.

UKB’s open access strategy for health-related data has already yielded an academic dividend of 10,000 scientific papers, with many new discoveries being made annually^15,16. Its diverse data include whole-genome sequences, brain, heart and full-body magnetic resonance imaging, blood biochemistry markers, activity monitors and online questionnaires¹⁷. UKB data was acquired using standard protocols, and has been subjected to detailed quality control procedures. Disadvantages of UKB are its healthy participation and sociodemographic biases¹⁸.

UKB has been used to identify ME/CFS genetic risk factors, using data from approximately 2,000–2,500 ME/CFS participants. However, these studies revealed few genome-wide significant findings and these failed to replicate across analyses^11,19–22. This lack of discovery could be explained by low predictive power from too few cases, and/or phenotypic misclassification arising from an overly permissive case definition.

As we show below, 5,354 individuals in the UKB are linked to 1 or more pieces of evidence of ME/CFS. Future UKB studies may thus wish to draw upon data from these additional individuals, and to refine ME/CFS case definition, in order to identify genetic, molecular, imaging, or activity data separating ME/CFS cases from population controls.

Here, our aim was to aid the definition of ME/CFS caseness in UKB analyses by investigating the consistency, or otherwise, of UKB participants’ ME/CFS status, defined in four different ways. We evaluate each of the 4 cohorts in turn, in relation to their concordance or discordance with the 3 other cohorts, before interpreting which of their cohort combinations might provide most scientific value in future studies.

Methods

Public and Patient Involvement

Public consultations on biobanking began in 2000, before funding for UK Biobank was obtained, and have continued since. UK Biobank actively engages with its participants through questionnaires, annual newsletters, follow-up studies and participation in other research activities. UK Biobank employed focus groups and a lay Advisory Panel/Community Advisory Group to advise them on study design and recruitment. No patients/participants – other than those in UK Biobank – were directly involved in this study.

Data acquisition

Data was downloaded onto the UK Biobank’s Research Analyst Platform (UKB-RAP) on 23^rd March 2023. This project was refreshed on 31^st August 2023 and all analyses were performed on 20^th February 2024, except for the analysis of the Experience of Pain questionnaire symptom/comorbidity responses which was conducted between 22^nd–29^th April 2024 when cohort numbers remained unaltered from the previous analyses. UK Biobank field identifiers are provided in Extended Data.

Baseline measurements

On the same day, at their visit to the UKB assessment centre, yet prior to the verbal interview, participants completed a touchscreen questionnaire featuring a question about their overall health: “In general how would you rate your overall health?”. Participants were able to choose the following response options: ‘Poor’, ‘Fair’, ‘Good’, ‘Excellent’, ‘Do not know’, and ‘Prefer not to Answer’.

In a touchscreen questionnaire, participants were asked “Have you been told by a doctor that you have other serious illnesses or disabilities?”²³. Responses to these questions were provided to the interviewer, a nurse. Participants were then asked “You selected that you have been told by a doctor that you have other serious illnesses or disabilities, could you tell me what they are?”²⁴. One such response option was ‘Chronic fatigue syndrome’; neither ME nor ME/CFS responses were available.

Pain Questionnaire (PQ)

In 2019–20, 167,184 UKB participants (94,988 females, 72,196 males) who completed the PQ, answered the question “Have you ever been told by a doctor that you have ME/CFS?”²⁵ Here, we chose 15 additional PQ questions that are relevant to ME/CFS symptomatology. Questions 1–12 had no missing data. These were 6 questions with possible responses ‘yes’ (considered as ‘affected’), or ‘no’, ‘do not know’, ‘prefer not to answer’ (assumed to be ‘unaffected’ for this analysis): (1) “Have you ever been told by a doctor you have had Fibromyalgia syndrome?” (or “Ever had fibromyalgia”); (2) “Have you ever been told by a doctor that you have had migraine?” (“Ever had migraine”); (3) “Are you troubled by pain or discomfort, either all the time or on and off, that has been present for more than 3 months?” (“Troubled by pain or discomfort for >3 months?”); (4) “During the past 6 months have you had headaches?” (“Headache in past six months”); (5) “Do you have persistent or recurrent tiredness, weariness or fatigue that has lasted for at least 6 months?” (“Persistent fatigue for at least 6 months?”); and, (6) “Have you suffered from fatigue or exhaustion in the last week?” (“Fatigue or exhaustion in last week?”).

The next 6 questions were on symptom severity with possible responses ‘mild’, ‘moderate’ or ‘severe’ (considered as ‘affected’), or ‘no problem’ or ‘prefer not to answer’ (considered as ‘unaffected’); or on symptom frequency with possible responses ‘several days’, ‘more than half the days’ and ‘nearly every day’ (considered ‘affected’), and ‘not at all’ or ‘prefer not to say’ (considered ‘unaffected’): (7) “Indicate the level of severity over the past week of waking unrefreshed” (“Waking unrefreshed severity over the past week”); (8) “Indicate the level of severity over the past week of cognitive symptoms. For example, problems with memory, thinking, skills and/or concentration” (“Cognitive symptoms severity over the past week”); (9) “Please click the ONE box that best described your health TODAY: Mobility” (“Mobility problems today”); (10) “Please click the ONE box that best describes your health TODAY: Self-care” (“Self-care problems today”); (11) “Please click the ONE box that best describes your health TODAY: Usual activities” (“Problems doing usual activities”); and, (12) “Over the last 2 weeks, how often have you been bothered by feeling tired or having little energy?” (“Recent feelings of tiredness or low energy”).

The final 3 questions are related to post-exertional malaise, with possible responses: ‘yes’ (considered ‘affected’), ‘no’, ‘do not know’ or ‘prefer not to answer’ (all considered ‘unaffected’). These questions were only asked to participants who answered ‘yes’ to question (5) above: ‘Do you have persistent or recurrent tiredness, weariness or fatigue that has lasted for at least 6 months?’ including to 1,876 (69%) of those in Cohort 2. These 3 questions were: (13) “Do you get tired after minimal physical or mental exertion?” (“Tired after minimal physical or mental exertion”); (14) “Does this tiredness, weariness or fatigue go away when you rest?” (“Fatigue goes away when resting”); and, “Is this tiredness, weariness or fatigue happening only because you have been exercising and/or working too much?” (“Fatigue only because of exercising or working too much”).

UKB was given access to hospital inpatient records²⁶ for 446,974 individuals in their cohort (88%; 244,747 females, 202,227 males). Of 1,407 with a G93.3 ICD-10 code, few (6%; 87) had G93.3 listed as the main reason for admission with nearly all (97%; 1,361) having G93.3 listed under a secondary code; 41 (3%) have G93.3 listed as both a main and secondary code. Also, the UKB was granted access to primary care records²⁷ for 230,055 of its 502,364 participants (46%). Primary care codes are split into CTV3 codes and ReadV2 codes depending on the GP computer system supplier (UK Biobank 2019). In total, 1,818 individuals in the UKB have been linked to both CTV3 and ReadV2 codes (1,021 females, 797 males); these duplicate individuals are only counted once. ME/CFS referral codes were not included because referrals may not necessarily lead to a diagnosis.

Results

The 4 ME/CFS cohorts investigated were:

Cohort 1 (C₁): ‘Self-reported CFS’ – those who self-reported a clinical diagnosis of CFS in the verbal interview at a visit to the UKB assessment centre.

Cohort 2 (C₂): ‘ME/CFS (PQ)’ – those who self-reported clinical diagnosis of ME/CFS in the ‘Experience of Pain’ questionnaire (PQ²⁵; 2019–20).

Cohort 3 (C₃): ‘ICD-10:G93.3’ – those with this International Classification of Diseases tenth revision (ICD-10) ‘Post-viral fatigue syndrome’ (PVFS) code in hospital inpatient records. Notably, post-exertional malaise is not required for individuals to receive this code. Further, people with ME/CFS without an infection prior to disease onset will not be given this code. Consequently, PVFS is similar to but not equivalent to ME.

Cohort 4 (C₄): ‘GP code for ME/CFS’ – those with a ME/CFS code in their primary care (General Practice) records.

For UKB data fields used in these analyses, see Extended Data.

Lack of overlap between cohorts could be due to inconsistencies in evidence of an ME/CFS diagnosis. It may also be due to missing data: only subsets of UKB participants (i) completed the PQ, (ii) have undergone a hospital admission (and consequently have hospital inpatient data) and/or (iii) have available primary care records. Inconsistencies may also be due to asynchrony between data sources: (i) an ME/CFS diagnosis might not have been captured in ICD-10 codes as these date back to 1981–1998 depending on health service provider, and diagnosis may have preceded this date; (ii) recent ME/CFS diagnoses might not be present in primary care codes due to these being extracted for UKB in 2016–2017; and, (iii) a participant’s EHR may record a ME/CFS diagnosis despite them not previously self-reporting ME/CFS, when their diagnosis postdates their last self-reporting opportunity.

We also investigated apparent inconsistencies between evidence of self-reported CFS diagnosis and each participant’s self-reported overall health at entry into UKB. Our reasoning is that those with evidence of a diagnosis and yet who report ‘good’ or ‘excellent’ overall health might have misreported or been miscoded as self-reported CFS, or were recovered, misdiagnosed or in remission.

Cohort 1: Self-reported CFS diagnosis

A total of 2,312 (0.46%) of 502,364 UKB participants (273,297 female, 229,067 male), self-reported a clinical diagnosis of CFS in a verbal interview at one or more visit to the UKB assessment centre. This cohort (C₁) has 1,685 females (0.62%), and 627 males (0.27%), a female bias of 2.29:1. Surprisingly, 44%-69% of these failed to re-report CFS at subsequent visits to the UKB. As expected, those who self-reported a CFS diagnosis are significantly less likely to report ‘excellent’ or ‘good’ overall health, and significantly more likely to report ‘poor’ or ‘fair’ overall health, when compared to those who did not self-report a diagnosis of CFS (p<10^-5; Table 1). Most of C₁ reported ’poor’ or ‘fair’ overall health. Unexpectedly, however, 28% reported ‘good’ or ‘excellent’ overall health when they first volunteered their CFS diagnosis (Table 1).

Table 1. Self-reported overall health ratings in C₁.

Numbers and percentages of those who self-reported CFS at visits to the UKB assessment centre, stratified by gender (top). Percentages of those who did not self-report CFS at their first visit to the UK Biobank centre are also shown (bottom). Overall health ratings are as reported at the time an individual first reported CFS. Ambiguous responses ‘Do not know’ and ‘Prefer not to answer’ (n=22) are excluded. A small number of UKB participants failed to complete this question.

	Overall Health Rating	Both sexes (n=2,289)	Female (n=1,667)	Male (n=622)
Self-Reported CFS (at any visit)	Poor	771 (33%)	540 (32%)	231 (36%)
	Fair	857 (37%)	646 (38%)	211 (33%)
	Good	590 (25%)	432 (25%)	158 (25%)
	Excellent	71 (3%)	49 (3%)	22 (4%)
Did not Self- Report CFS at Visit 1	Poor	22,040 (4%)	10,249 (4%)	11,791 (5%)
	Fair	104,524 (21%)	52,286 (19%)	52,238 (23%)
	Good	288,402 (58%)	161,111 (59%)	127,291 (56%)
	Excellent	81,770 (16%)	46,266 (17%)	35,504 (16%)

C₁ has variable evidence of ME/CFS from other sources, when available. Among those who completed the PQ, most (89%, n=850) provided a concordant response, answering ‘yes’ to having a clinical diagnosis of ME/CFS. However, only one-third of C₁ with hospital inpatient records linked in UKB had an ICD-10:G93.3 code: 714 of 2,148 (33%; 531 females, 183 males). Further, only half (55%; n=617) of C₁ have a ME/CFS code in their UKB primary care records: 431 females and 186 males. In all, over one-third (36%) of C₁ have no further evidence of ME/CFS (602 females, 220 males), although this is in part due to data missingness. Thirty-nine percent of C₁ (894 people) have 1 further piece of evidence, 22% (501) have 2, and 4% (95) have all 3.

Cohort 2: ME/CFS defined in the Pain Questionnaire

Among 167,184 participants who completed the PQ, 2,720 (1.63%) are in cohort 2 (C₂), having answered ‘Yes’ to “Have you ever been told by a doctor that you have ME/CFS?” 1,929 of 94,988 were female (2.0%) and 791 of 72,196 (1.1%) were male, a female-bias of 1.85:1. Among those saying ‘Yes’ are 850 who are also present in C₁ (see above) leaving 1,870 cases newly-reported in C₂ (1,299 female, 571 male).

Among those in C₂, UKB has access to hospital inpatient data for 2,527 (92%) and primary care code records for 1,315 (48%). Of those with available data, only 459 (18%) have an ICD-10:G93.3 code and 441 (34%) have a ME/CFS code in their primary care records. Just over half (57%, n=1,556) of C₂ have no further evidence of ME/CFS (i.e., are not in C₁, C₃ or C₄), in part due to missingness in the data. A quarter (673; 25%) have 1 further piece of evidence, 15% (396) have 2 further pieces of evidence, and 3% (95) have all 3.

Cohort C₂ were significantly more likely than those in neither C₁ nor C₂ (‘controls’; C₁^C ∩ C₂^C) to have: been diagnosed with fibromyalgia (16-fold higher), been diagnosed with migraine (2.0-fold higher), been troubled by pain or discomfort for >3 months (1.4-fold), experienced headaches in the past 6-months (1.7-fold), experienced persistent or recurrent tiredness, weariness or fatigue that has lasted over 6-months (3.5-fold), and experienced fatigue or exhaustion in the past week (2.3-fold); all p<10^-5 (Figure 1A). ME/CFS cohort C₂ was also significantly more likely than controls to have experienced mild, moderate or severe problems with: unrefreshing sleep in the past week (1.6-fold higher), cognitive difficulties in the past week (1.9-fold), mobility problems today (1.9-fold), self-care problems today (3.0-fold), problems with usual activities today (2.0-fold), and frequent problems with tiredness or little energy in the last 2 weeks (1.7-fold; all p<10^-5) (Figure 1B).

Figure 1. ‘Yes’ responses to ME/CFS-relevant symptom and comorbidity questions as asked in the PQ.

Higher proportions of 6 comorbidities (A) and 6 symptoms (B) in a ME/CFS cohort than among controls. Cohort 2 (C₂, ME/CFS, PQ; blue) versus those answering ‘No’ in the PQ and who never self-reported CFS in the verbal interviews (‘controls’, C₁^C ∩ C₂^C; orange). Note that 2,066 individuals who answered 'I don't know' and 19 who answered 'Prefer not to answer' to the ME/CFS question in the PQ are not included among controls due to their ambiguous response. (A) C₂ individuals were significantly more likely than controls to have been diagnosed with fibromyalgia (21.8% versus 1.3%); migraine (37.5% versus 18.9%); to have been troubled by pain or discomfort for >3 months (77.2% versus 55.5%); experienced headaches in the past 6 months (56.4% versus 33.2%); experienced persistent or recurrent tiredness, weariness or fatigue that has lasted over 6 months (68.9% versus 19.9%); and experienced fatigue or exhaustion in the past week (73.9% versus 32.0%); all p<10^-5 (χ²-test). (B) C₂ individuals were significantly more likely than controls to have experienced mild, moderate or severe problems with: unrefreshing sleep in the past week (86.0% versus 54.3%), cognitive difficulties in the past week (69.2% versus 36.4%), mobility problems today (52.2% versus 28.2%), self-care problems today (25.4% versus 8.5%), problems with usual activities today (61.3% versus 31.2%), and were more likely to experience feeling tired or having little energy in the past 2 weeks (86.1% versus 50.5%) (all p<10^-5; χ²-test).

Among cohort C₂ were 1,876 (69.0%) who answered ‘yes’ to ‘Do you have persistent or recurrent tiredness, weariness or fatigue that has lasted for at least 6 months?’ These UKB participants then answered 3 questions that are relevant to post-exertional malaise. They were, as expected, significantly more likely than those in (C₁^C ∩ C₂^C) controls to report getting tired after minimal physical or mental exertion (1.6-fold higher), and significantly less likely to report that their tiredness goes away when resting (0.53-fold lower), or that their tiredness was only due to over-exertion (0.42-fold lower; all p<10^-5; Figure 2). Just under two-thirds (394 of 617; 64%) of those in both C₁ and C₂ answered these 3 questions in a manner consistent with post-exertional malaise (i.e., ‘Yes’, ‘No’, ‘No’; Figure 2), higher than the 57% (1,073 of 1,876) in C₂ who answered these in the same way. These questions are thus valid for further refining ME/CFS caseness.

Figure 2. ‘Yes’ responses as a percentage of each cohort for post-exertional malaise-related questions.

Questions were asked in the UK Biobank Experience of Pain Questionnaire (PQ). The ME/CFS (PQ) cohort C₂ (dark blue) was significantly more likely than C₁^C ∩ C₂^Ccontrols (orange) to report becoming tired after minimal physical or mental exertion (76.1% versus 48.2%, respectively), significantly less likely to report that their tiredness goes away when resting (25.2% versus 47.6%, respectively), and significantly less likely to report that their tiredness is only due to over-exertion (6.9% versus 16.6%, respectively); all p<10^-5. Individuals in both cohorts 1+2 (concordant cohort: C₁ and C₂; C₁ ∩ C₂; grey) more often reported feeling tired after minimal physical or mental exertion, than those in cohort 2 but not 1 (C₂ ∩ C₁^C; yellow; 80.6% versus 74.0%, respectively) or in cohort 1 but not 2 (C₁ ∩ C₂^C; light blue; 80.6% versus 65.0%, respectively). Similarly, those in C₁ ∩ C₂ (grey) were less likely than those in (C₂ ∩ C₁^C), or (C₁ ∩ C₂^C) (20.4% versus 27.5% or 31.8%, respectively) to report that their tiredness went away when resting. The discordant cohort (C₁ not C₂: C₁ ∩ C₂^C; light blue; i.e., self-reported CFS in verbal interview but answered ‘No’ or 'do not know' or 'prefer not to answer' in the PQ) was significantly more likely to report that their tiredness was due only to over-exertion than the concordant cohort (C₁ ∩ C₂; grey): 12.7% versus 5.8%, respectively. Asterisks denote significant differences between cohorts (χ² test; p<10^-5 ***; p<0.01 **; p<0.05 *; n/s, not significant at p<0.05).

Cohort 3: Hospital Inpatient Data: ICD-10 G93.3

In total, 446,974 individuals (244,747 female, 202,227 male) in the UKB underwent 1 or more hospital admission and so have linked hospital inpatient codes. Of these, 1,407 (0.31%) have an ICD-10 code for ‘Post-viral fatigue syndrome’ (G93.3); 1,048 are female (0.43%) and 359 male (0.18%), giving a female-to-male ratio of 2.4:1.

Among those in cohort 3 (C₃), 51% (714) are also in C₁ (‘self-reported CFS’); 88% (459) are in C₂ (Pain Questionnaire), and 60% (412) have a Primary Care code for ME/CFS (i.e., are in cohort C₄; see below), when data is available. Altogether, among 1,407 in C₃ (with an ICD-10:G93.3 code), 398 (28%) lack further evidence of ME/CFS, 528 (38%) have 1 further piece of evidence, 386 (27%) have 2 further pieces of evidence and 95 (7%) have 3 further pieces of evidence.

Cohort 4: Primary care records

Among all UKB participants, 230,055 have linked primary care records of whom 1,575 (0.68%; cohort C₄) have a ME/CFS code: 1,071 have a CTV3 code, 505 have a Readv2 code and 1 has both types of code. C₄ contains 1,109 females (0.88%) and 466 males (0.45%), giving a female-to-male ratio of 1.96:1.

The clinical diagnoses of people in C₄ allowed us to assess the diagnostic validity of C_1-3. We defined concordances between cohorts using only those individuals for whom all relevant data was available. Concordance with C₄ was highest for C₂ (71%), followed by C₁ (39%), and lowest for C₃ (28%). Despite their known clinical diagnosis in primary care, 61%, 29% and 72% of UKB participants in C₁, C₂ and C₃, respectively, thus failed to self-report CFS in the verbal interview, failed to self-report ME/CFS in the PQ or failed to have linked hospital inpatient data, respectively. Among 958 individuals in C₄ who did not self-report CFS in the verbal interview, 32% (302) have additional evidence of ME/CFS from the PQ or ICD-10:G93.3 code.

In all, 656 (42%) of those in C₄ have no further evidence of ME/CFS, 463 (29%) have 1 further piece of evidence, 361 (23%) have 2 further pieces of evidence and 95 (6%) have 3 further pieces of evidence.

ME/CFS status in 4 cohorts

The prevalence of ME/CFS in C_1-4 varied between 0.31% and 1.63% for both sexes combined, 0.18%-1.1% for males, and 0.43%-2.0% for females (Table 2). In the UKB, there are 5,354 individuals with 1 or more pieces of evidence for ME/CFS: 3,432 (64%) have 1 piece of evidence, 1,279 (24%) have 2 pieces of evidence, 548 (10%) have 3 pieces of evidence and 95 (2%) have all 4 pieces of evidence (Table 3).

Table 2. Prevalence and female-to-male (F/M) ratio of ME/CFS across different sources of ME/CFS data in UKB.

UKB Cohort	Subset	Prevalence	F/M ratio
C₁: Self-report CFS	Both sexes	0.46%	2.29:1
	Female	0.62%
	Male	0.27%
C₂: ME/CFS Pain Questionnaire	Both sexes	1.63%	1.85:1
	Female	2.03%
	Male	1.10%
C₃: ICD-10: G93.3 code	Both sexes	0.31%	2.38:1
	Female	0.43%
	Male	0.18%
C₄: GP code	Both sexes	0.68%	1.96:1
	Female	0.88%
	Male	0.45%

Table 3. Numbers of UK participants with 1, 2, 3 or 4 pieces of evidence for ME/CFS.

Among 5,354 individuals in the UKB with at least 1 piece of ME/CFS evidence: 3,432 have 1 piece of evidence, 1,279 have 2 pieces of evidence, 548 have 3 pieces of evidence and 95 have all 4 pieces of evidence. Each of the 5,354 participants is counted once only in this table.

Source of Data	n
1 piece of ME/CFS evidence:
Self-reported CFS (C₁) only	822
ME/CFS in PQ (C₂) only	1,556
ICD-10:G93.3 (C₃) only	398
Primary Care code for ME/CFS (C₄) only	656
2 pieces of ME/CFS evidence:
Self-reported CFS & ME/CFS in PQ only (C₁ & C₂ only)	406
Self-reported CFS & ICD-10:G93.3 only (C₁ & C₃ only)	280
Self-reported CFS & Primary Care code for ME/CFS only (C₁ & C₄ only)	208
ME/CFS in PQ & ICD-10:G93.3 only (C₂ & C₃ only)	130
ME/CFS in PQ & Primary Care code for ME/CFS only (C₂ & C₄ only)	137
ICD-10:G93.3 & Primary Care code for ME/CFS only (C₃ & C₄ only)	118
3 pieces of evidence:
Self-reported CFS & ME/CFS in PQ & ICD-10:G93.3 only (C₁ & C₂ & C₃ only)	187
Self-reported CFS & ICD-10:G93.3 & Primary Care code for ME/CFS only (C₁ & C₃ & C₄ only)	152
ME/CFS in PQ & ICD-10:G93.3 & Primary Care code for ME/CFS only (C₂ & C₃ & C₄ only)	47
Self-reported CFS & Primary care code for ME/CFS & ME/CFS in PQ only (C₁ & C₂ & C₄ only)	162
4 pieces of ME/CFS evidence:
Self-reported CFS & ME/CFS in PQ & ICD-10:G93.3 & Primary Care code for ME/CFS (C₁& C₂ & C₃ & C₄)	95

Discussion

Of 5,354 individuals in the UKB with evidence of a ME/CFS diagnosis, two-thirds (3,432; 64%) have only 1 piece of evidence. Also, each of the four pieces of evidence for ME/CFS is not strongly supported by any other (Figure 3). The exceptions are the ME/CFS-PQ cohort (C₂), whose evidence supports membership of each of the 3 other cohorts at >70%, when complete data is available, and the ICD10:G93.3 cohort (C₃) of whom 72% have supporting evidence from other data sources. Conversely, membership of each of C₁, C₃ or C₄ poorly supports C₂ membership. Here we speculate why there are concordances and discordances among the 4 cohorts including with respect to likely biases in self-reported and electronic health record data.

Figure 3. Concordance of ME/CFS evidence across data sources for each of the 4 UKB cohorts.

This considers only individuals with available data (i.e., no missingness). Each bar indicates the percentage of the cohort (e.g. C₁ – left) who are also members of a second cohort (e.g. C₂ – light blue), accounting for missing data (i.e., only considering those with linked data required for specifying C₁ and C₂ membership). A total of 850 are members of both C₁ and C₂ cohorts. Among those for whom data was available, this is 31% of C₁ members, and 89% of all C₂ members, a discrepancy that could reflect participation bias in completion of the Pain Questionnaire and/or that its completion was approximately 10 years after the baseline questionnaire. Unlike other values, the ‘No further data’ values (dark blue) do not account for data missingness. Significance testing was not employed due to data overlap and missingness between cohorts.

Membership of cohorts’ C_1-4 will be inflated when ‘Chronic fatigue’ has been wrongly conflated with ‘Chronic fatigue syndrome’ by clinicians or UKB participants and nurses. ‘Chronic Fatigue’ occurs at much higher prevalence than ‘Chronic Fatigue Syndrome’^28,29 yet, importantly, does not require hallmark ME/CFS symptoms such as post-exertional malaise. This conflation may explain why, in one study, only about one-third of GP-diagnosed ME/CFS cases met a more stringent ME/CFS case definition³⁰. Membership will also be inflated, relative to the general UK population, due to UKB’s sociodemographic biases including age, ethnicity and gender^16,18 which all favour increased rates of ME/CFS diagnosis³¹. Cohort 2 membership was also inflated due to participation bias: those in C₁ were significantly more likely to complete the subsequent optional online PQ than others (41% versus 33%, respectively; p<10^-5).

Cohort C₁ (n=2,312), without further filtering, does not have strong validity for ME/CFS research: it has unexpectedly high levels of individuals reporting ‘good’ or ‘excellent’ overall health (28%), many (44%–69%) fail to re-report CFS at subsequent visits to the UKB, 11% (n=105) who also completed the PQ provided a discordant response when asked about a clinical ME/CFS diagnosis, and there were high levels of discordance with hospital inpatient and primary care data. It is thus plausible that some within this cohort have mis-reported a diagnosis of CFS, been self-diagnosed (rather than receiving a clinical diagnosis), been diagnosed with ‘Chronic Fatigue’ rather than ME/CFS, been misdiagnosed with ME/CFS previously and since received an alternative appropriate diagnosis, or CFS has been erroneously recorded during the verbal interview. Previous analyses that used C₁ are thus called into question. For example, genetic studies of ME/CFS, which to be well-powered require a homogeneous population, have not yielded consistent findings using this cohort^20,32–36. These studies also used up to 3,042 UKB participants as controls who nevertheless have evidence of ME/CFS (i.e., those in C_2–4, but not C₁). These individuals would have been included as controls in previous ME/CFS analyses, likely weakening their power to detect true associations. We recommend filtering C₁ cases by removing (i) those with ‘good’ or ‘excellent’ overall health ratings, and (ii) those who provided a discordant response in the PQ. This strategy was employed in a separate analysis³⁷ which resulted in a larger yield of ME/CFS-associated blood traits.

Cohort C₂ (n=2,720) might be considered to have strong validity for ME/CFS research because it is defined by the explicit question “Have you ever been told by a doctor that you have ME/CFS?” Also, those in C₂ reported levels of fatigue, muscle pain, headaches, unrefreshed sleep and cognitive difficulties, and a comorbid fibromyalgia diagnosis, similar to those reported in a separate large-scale UK study covering adults of all ages⁴. Nevertheless, for two reasons we caution against C₂ being used for ME/CFS research without further refinement. First, nearly one-third (844; 31.0%) of C₂ did not report ‘persistent or recurrent tiredness, weariness or fatigue that has lasted over 6 months’ (p120114, Extended Data), which is a core symptom for ME/CFS diagnosis⁶. Second, not everyone in C₂ appears to experience post-exertional malaise: 448 (24%) of this cohort do not report ‘feeling tired after minimal physical or mental exertion’ (p120117, Extended Data), 472 (25%) report that their ‘tiredness goes away when resting’ (p120115, Extended Data) and 130 (6.9%) report their ‘tiredness is only due to over-exertion’ (p120116, Extended Data). Overall, less than half (43%) of those in C₂ are also in one or more of the other 3 cohorts. Thus, one possible strategy for filtering C₂ is to only consider those who provided responses consistent with post-exertional malaise: namely the 1,073 individuals who answered ‘yes’, ‘no’, ‘no’ to p120117, p120115, p120116, respectively. Nevertheless, these three questions may not effectively define post-exertional malaise because they only capture ‘tiredness’, as opposed to a flare up of ‘all symptoms’, they overlook ‘new symptoms’ which are characteristic of post-exertional malaise⁶, and they do not account for people with ME/CFS pacing (e.g., ‘resting’ and not ‘over-exerting’). Otherwise, a more permissive approach would be to include only those who answered ‘yes’ to p120117 (n=1,428).

The largest intersection between cohorts is between C₁ and C₂ (n=850; Table 3). These individuals are slightly more likely to report ME/CFS symptoms, relative to those in C₂ yet who did not self-report CFS at baseline (i.e., not in C₁): they are significantly more likely to be tired after minimal physical or mental exertion (81% vs 74%), and also less likely that their tiredness goes away after rest (20% vs 27%; Figure 2). Further requiring participants to have an ICD-10:G93.3 or ME/CFS primary care code reduces their number considerably, to 187 or 162, respectively. Whilst this provides cohorts with support from health record data, their numbers dwindle to low case numbers (approximately 20–40) when further intersected with UKB participants with plasma protein, or brain image or accelerometry data, for example^38–40, which would substantially weaken statistical power in such studies. Mandating participants to have at least 1 piece of self-report data (C₁ or C₂) and at least 1 piece of EHR data (C₃ or C₄) boosts case number to 1,398, a quarter (26%) of all UKB ME/CFS (C_1–4) cases.

Cohort C₃ has most support from other cohorts (72%). Nevertheless, using C₃ without further filtering is not advised as some may have ‘Post viral Fatigue Syndrome’, which does not require post-exertional malaise, rather than ME/CFS which does. Thus, we recommend restricting C₃ to the 1,009 individuals who also fall within one of the other cohorts: C₁, C₂ or C₄. A further caveat to consider is that C₃ will likely not include the one-third of people with ME/CFS who do not have an infection prior to the onset of their symptoms⁷. It is also the smallest cohort: 26% of all individuals in C_1–4 (Table 2); this fraction reduces to 19% when the recommended filters are applied.

As a clinically-defined cohort, C₄ has strong validity and numbers are relatively high (n=1,575). Nevertheless, we recommend removing those with discordant PQ data, specifically ~29% of participants (n=177) who completed the PQ yet failed to answer ‘Yes’ to the question: “Have you ever been told by a doctor you have ME/CFS?” Doing so, reduces this cohort to about 1,120 (of 230,055 with GP records), or a UKB prevalence of 0.49%. Extrapolating this percentage to the UK population⁴¹ predicts 330,000 people with a life-time risk of an ME/CFS diagnosis. This is lower than an estimate of 404,000 that assumed equal access to diagnosis, which is far from being achieved³¹.

We recommend that control cohorts exclude C_1-4 members and remove those providing an ambiguous response in the PQ or who have ME/CFS referral codes in their primary care records.

This analysis benefits from the diversity of data types being compared for the same population-scale cohort, UKB, rather than for other biobanks, such as FinnGen and the Estonian Biobank, which link to ME via ICD-10 codes only^42,43. The UKB ME cohort is the largest globally. The All of Us CFS cohort⁴⁴ might appear to be larger (at 8,600 participants) but 93% of this cohort are linked to chronic fatigue, unspecified (R53.82 ICD-10CM) rather than to ME or PVFS. This study’s weaknesses are that missing UKB data is assumed to be missing completely at random, and that the healthy participation bias inherent in the UKB¹⁸ will result in disproportionately few people participating in UK who were severely- or even moderately-affected by ME/CFS. Accounting for the estimated one-quarter of people with ME/CFS who had no opportunity to participate in UKB due to being bed- or house-bound⁴⁵, our estimated ME/CFS prevalence in the UK then rises from 330,000 to 410,000 (0.6%).

Ethical approval

UK Biobank has approval from the North West Multi-centre Research Ethics Committee (MREC) as a Research Tissue Bank (RTB) approval (2011, renewed 2016 and 2021). This approval means that researchers do not require separate ethical clearance and can operate under the RTB approval.

Consent

The basis for consent in UK Biobank rests on participants' explicit and informed consent. UK Biobank uses “legitimate interests” as the primary lawful basis on which to process personal data under the UK GDPR.

Data availability

UK Biobank data is available upon application via their website⁴⁶ https://www.ukbiobank.ac.uk/enable-your-research/apply-for-access. UK Biobank research fields are provided in the Main Text and in Extended Data Tables 1–2, which are freely available from the Open Science Framework⁴⁷: https://osf.io/rp5ca.

Author contributions

Gemma Louise Samms: Conceptualization, Data curation, Investigation, Writing – original draft, Writing – review & editing. Chris P. Ponting: Conceptualization, Funding acquisition, Supervision, Writing – original draft, Writing – review & editing.

Acknowledgements

This research has been conducted using the UK Biobank Resource under Application Number 76173. This work uses data provided by patients and collected by the NHS as part of their care and support. Our sincere thanks to the participants and researchers of the UK Biobank who made this effort possible. GLS was funded by a PhD studentship from ME Research UK. For this project, CPP was jointly funded by NIHR and MRC (grant number MC_PC_20005).

Faculty Opinions recommended

References

1. Hanson MR, Germain A: Letter to the editor of Metabolites. Metabolites. 2020; 10(5): 216. PubMed Abstract | Publisher Full Text | Free Full Text
2. Annesley SJ, Missailidis D, Heng B, et al.: Unravelling shared mechanisms: insights from recent ME/CFS research to illuminate long COVID pathologies. Trends Mol Med. 2024; 30(5): 443–458. PubMed Abstract | Publisher Full Text
3. Salit IE: Precipitating factors for the Chronic Fatigue Syndrome. J Psychiatr Res. 1997; 31(1): 59–65. PubMed Abstract | Publisher Full Text
4. Bretherick AD, McGrath SJ, Devereux-Cooke A, et al.: Typing myalgic encephalomyelitis by infection at onset: a decodeme study [version 1; peer review: 1 approved with reservations, 1 not approved]. NIHR Open Res. 2023; 3: 20. PubMed Abstract | Publisher Full Text | Free Full Text
5. Choutka J, Jansari V, Hornig M, et al.: Unexplained Post-Acute Infection Syndromes. Nat Med. 2022; 28(5): 911–923. PubMed Abstract | Publisher Full Text
6. Carruthers BM, Jain AK, de Meirleir KL, et al.: Myalgic Encephalomyelitis/Chronic Fatigue Syndrome: clinical working case definition, diagnostic and treatment protocols. J Chronic Fatigue Syndr. 2003; 11(1): 7–115. Publisher Full Text
7. Bretherick AD, McGrath SJ, Devereux-Cooke A, et al.: Typing Myalgic Encephalomyelitis by infection at onset: a decodeme study [version 4; peer review: 2 approved]. NIHR Open Res. 2023; 3: 20. PubMed Abstract | Publisher Full Text | Free Full Text
8. Mirin AA, Dimmock ME, Jason LA: Updated ME/CFS prevalence estimates reflecting post-COVID increases and associated economic costs and funding implications. Fatigue. 2022; 10(2): 83–93. Publisher Full Text
9. Tyson S, Stanley K, Gronlund TA, et al.: Research priorities for Myalgic Encephalomyelitis/Chronic Fatigue Syndrome (ME/CFS): the results of a James Lind alliance priority setting exercise. Fatigue. 2022; 10(4): 200–211. Publisher Full Text
10. Maksoud R, Magawa C, Eaton-Fitch N, et al.: Biomarkers for Myalgic Encephalomyelitis/Chronic Fatigue Syndrome (ME/CFS): a systematic review. BMC Med. 2023; 21(1): 189. PubMed Abstract | Publisher Full Text | Free Full Text
11. Dibble JJ, McGrath SJ, Ponting CP: Genetic risk factors of ME/CFS: a critical review. Hum Mol Genet. 2020; 29(R1): R117–R124. PubMed Abstract | Publisher Full Text | Free Full Text
12. Walitt B, Singh K, LaMunion SR, et al.: Deep phenotyping of post-infectious Myalgic Encephalomyelitis/Chronic Fatigue Syndrome. Nat Commun. 2024; 15(1): 907. PubMed Abstract | Publisher Full Text | Free Full Text
13. Devereux-Cooke A, Leary S, McGrath SJ, et al.: DecodeME: community recruitment for a large genetics study of Myalgic Encephalomyelitis / Chronic Fatigue Syndrome. BMC Neurol. 2022; 22(1): 269. PubMed Abstract | Publisher Full Text | Free Full Text
14. Gallagher CS, Ginsburg GS, Musick A: Biobanking with genetics shapes precision medicine and global health. Nat Rev Genet. 2024: 26(3); 191–202. PubMed Abstract | Publisher Full Text
15. Conroy M, Sellors J, Effingham M, et al.: The advantages of UK Biobank’s open-access strategy for health research. J Intern Med. 2019; 286(4): 389–397. PubMed Abstract | Publisher Full Text | Free Full Text
16. Bycroft C, Freeman C, Petkova D, et al.: The UK Biobank resource with deep phenotyping and genomic data. Nature. 2018; 562(7726): 203–209. PubMed Abstract | Publisher Full Text | Free Full Text
17. UK Biobank: About our data. 2024. Reference Source
18. Fry A, Littlejohns TJ, Sudlow C, et al.: Comparison of sociodemographic and health-related characteristics of UK Biobank participants with those of the general population. Am J Epidemiol. 2017; 186(9): 1026–1034. PubMed Abstract | Publisher Full Text | Free Full Text
19. Wang Q, Dhindsa RS, Carss K, et al.: Rare variant contribution to human disease in 281,104 UK Biobank exomes. Nature. 2021; 597(7877): 527–532. PubMed Abstract | Publisher Full Text | Free Full Text
20. Hajdarevic R, Lande A, Mehlsen J, et al.: Genetic association study in Myalgic Encephalomyelitis/Chronic Fatigue Syndrome (ME/CFS) identifies several potential risk loci. Brain Behav Immun. 2022; 102: 362–369. PubMed Abstract | Publisher Full Text
21. Canela-Xandri O, Rawlik K, Tenesa A: An atlas of genetic associations in UK Biobank. Nat Genet. 2018; 50(11): 1593–1599. PubMed Abstract | Publisher Full Text | Free Full Text
22. Neale Lab: UK Biobank. 2018. Reference Source
23. UK Biobank: The verbal interview within ACE centres. Reference Source
24. UK Biobank: UK Biobank verbal interview stage. 2012; 1–17. Reference Source
25. UK Biobank: Experience of Pain Questionnaire. Reference Source
26. UK Biobank: Resource 138483 hospital episode statistics. Reference Source
27. UK Biobank: Resource 591 - primary care linked data. Reference Source
28. Cho HJ, Menezes PR, Hotopf M, et al.: Comparative epidemiology of Chronic Fatigue Syndrome in Brazilian and British primary care: prevalence and recognition. Br J Psychiatry. 2009; 194(2): 117–122. PubMed Abstract | Publisher Full Text
29. Wessely S, Chalder T, Hirsch S, et al.: The prevalence and morbidity of chronic fatigue and Chronic Fatigue Syndrome: a prospective primary care study. Am J Public Health. 1997; 87(9): 1449–1455. PubMed Abstract | Publisher Full Text | Free Full Text
30. Nacul LC, Lacerda EM, Pheby D, et al.: Prevalence of Myalgic Encephalomyelitis/Chronic Fatigue Syndrome (ME/CFS) in three regions of England: a repeated cross-sectional study in primary care. BMC Med. 2011; 9: 91. PubMed Abstract | Publisher Full Text | Free Full Text
31. Samms GL, Ponting CP: Unequal access to diagnosis of myalgic encephalomyelitis in England. BMC Public Health. in press.
32. Tanigawa Y, Li J, Justesen JM, et al.: Components of genetic associations across 2,138 phenotypes in the UK Biobank highlight adipocyte biology. Nat Commun. 2019; 10(1): 4064. PubMed Abstract | Publisher Full Text | Free Full Text
33. Neale Lab UK Biobank GWAS round 2 results. 2018. Reference Source
34. Zhou W, Nielsen JB, Fritsche LG, et al.: Efficiently controlling for case-control imbalance and sample relatedness in large-scale genetic association studies. Nat Genet. 2018; 50(9): 1335–1341. PubMed Abstract | Publisher Full Text | Free Full Text
35. Pan UK Biobank Team. Reference Source
36. Canela-Xandri O, Rawlik K, Tenesa A: An atlas of genetic associations in UK Biobank. Nat Genet. 2018; 50(11): 1593–1599. PubMed Abstract | Publisher Full Text | Free Full Text
37. Beentjes SV, Kaczmarczyk J, Cassar A, et al.: Replicated Blood-based biomarkers for Myalgic Encephalomyelitis not explicable by inactivity. medRxiv. 2024. Publisher Full Text
38. UK Biobank: Category 1008 physical activity measurement. Reference Source
39. Sun BB, Chiou J, Traylor M, et al.: Plasma proteomic associations with genetics and health in the UK Biobank. Nature. 2023; 622(7982): 329–338. PubMed Abstract | Publisher Full Text | Free Full Text
40. UK Biobank: Imaging study updates. 2024. Reference Source
41. Office for National Statistics: Population estimates for the UK, England, Wales, Scotland, and Northern Ireland: mid-2022. Reference Source
42. Leitsalu L, Haller T, Esko T, et al.: Cohort profile: Estonian Biobank of the Estonian Genome Center, University of Tartu. Int J Epidemiol. 2015; 44(4): 1137–47. PubMed Abstract | Publisher Full Text
43. Kurki MI, Karjalainen J, Palta P, et al.: FinnGen provides genetic insights from a well-phenotyped isolated population. Nature. 2023; 613(7944): 508–518. PubMed Abstract | Publisher Full Text | Free Full Text
44. All of us data browser - Chronic Fatigue Syndrome. Reference Source
45. Pendergrast T, Brown A, Sunnquist M, et al.: Housebound versus nonhousebound patients with Myalgic Encephalomyelitis and Chronic Fatigue Syndrome. Chronic Illn. 2016; 12(4): 292–307. PubMed Abstract | Publisher Full Text | Free Full Text
46. UK Biobank: Apply for access website. https://www.ukbiobank.ac.uk/enable-your-research/apply-for-access
47. Extended Data Table 1. UK Biobank Field IDs used in analyses. Extended Data Table 2. Definition of ME/CFS codes in Primary care records. https://osf.io/rp5ca

Comments on this article Comments (0)

Version 1

VERSION 1 PUBLISHED 28 Apr 2025

Author details Author details

¹ Institute of Genetics and Cancer, The University of Edinburgh MRC Human Genetics Unit, Edinburgh, Scotland, UK

Gemma L. Samms
Roles: Data Curation, Formal Analysis, Investigation, Writing – Original Draft Preparation, Writing – Review & Editing

Chris P. Ponting
Roles: Conceptualization, Funding Acquisition, Investigation, Supervision, Writing – Review & Editing

Competing interests

No competing interests were disclosed.

Grant information

This project is funded by the National Institute for Health and Care Research (NIHR) and Medical Research Council (MRC), grant number MC_PC_20005. GLS was funded by PhD-level research funding from ME Research UK. The views expressed are those of the author(s) and not necessarily those of the NIHR or the Department of Health and Social Care.
The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Article Versions (1)

version 1

Published: 28 Apr 2025, 5:39

https://doi.org/10.3310/nihropenres.13956.1

Copyright

Download

Export To

metrics

VIEWS

493

downloads

66

Citations

0

SEE MORE DETAILS

CITE

how to cite this article

Samms GL and Ponting CP. Defining a High-Quality Myalgic Encephalomyelitis/Chronic Fatigue Syndrome cohort in UK Biobank [version 1; peer review: 2 approved]. NIHR Open Res 2025, 5:39 (https://doi.org/10.3310/nihropenres.13956.1)

NOTE: If applicable, it is important to ensure the information in square brackets after the title is included in all citations of this article.

track

receive updates on this article

Track an article to receive email alerts on any updates to this article.

Open Peer Review

Version 1

VERSION 1

PUBLISHED 28 Apr 2025

Views

8

Reviewer Report 29 May 2025

Matthias Wielscher, Medical University of Vienna, Vienna, Austria

Leonardo Vincenzi, Department of Dermatology, Medical University of Vienna, Vienna, Vienna, Austria

Approved

https://doi.org/10.3310/nihropenres.15168.r35434

The study by Samms and Ponting presents a comprehensive approach to defining ME/CFS cohorts within the UK Biobank using a combination of self-reported questionnaire responses and linked healthcare records. They delineate four cohorts: C1, based on self-reported ME/CFS at baseline; ... Continue reading

The study by Samms and Ponting presents a comprehensive approach to defining ME/CFS cohorts within the UK Biobank using a combination of self-reported questionnaire responses and linked healthcare records. They delineate four cohorts: C1, based on self-reported ME/CFS at baseline; C2, derived from responses to the UK Biobank pain questionnaire (PQ); C3, defined by the presence of the ICD-10 G93.3 code in hospital episode data; and C4, identified via ME/CFS-related Read V2 and CTV-3 primary care codes. The authors provide a detailed comparative analysis of these cohorts, highlighting both overlaps and discrepancies in case identification across data sources.

This is a timely and important contribution given the diagnostic ambiguity and under-recognition surrounding ME/CFS. By demonstrating how to refine case definitions using available UK Biobank data, the study offers a valuable framework for researchers seeking to identify valid ME/CFS cases for downstream analyses—including genetic, proteomic, and imaging data. The authors also underscore the critical need for careful cohort curation to avoid misclassification and improve study power in ME/CFS research.

Minor comments:

1) Figure 3: I believe the intention is to illustrate the overlap between cohorts C1 to C4, which is highly valuable. However, as a bar plot, it is somewhat challenging to interpret the intersections clearly. I would kindly suggest considering an UpSet plot to more intuitively convey the cohort overlaps, including the number of participants exclusive to each group and those shared across combinations (e.g., https://r-graph-gallery.com/upset-plot.html). Additionally, it might be helpful to present this visualization earlier in the results section, to provide readers with a high-level overview before delving into cohort-specific analyses. A well-constructed UpSet plot could also consolidate and visually represent the key information currently found in Table 3.

2) In the Discussion, you make a compelling case for refining the cohorts to improve the validity of ME/CFS case identification—specifically, by excluding individuals in C1 who report “very good” or “excellent” health, and by filtering C2 for those who also exhibit post-exertional malaise (PEM). I fully agree that these strategies represent a sound approach to making the most of the available data. Given this, I was curious why these suggested filters were not applied when analyzing the cohort properties throughout the manuscript. While I appreciate that providing unfiltered summary statistics is useful for transparency and to caution others against potential misclassification, it would also be highly informative to include at least some analysis of a more stringently defined ME/CFS cohort—essentially, the best approximation of a “valid” ME/CFS group from a public biobank.

In particular, this would add context to Figure 1, where symptoms are analyzed across broad cohort definitions for C2. If certain participants are later acknowledged as likely misclassified, it might be more insightful to see symptom profiles after applying your own recommended filters. This could strengthen the manuscript by showing not only where the issues lie, but also how they might be addressed in practice.

3) Have you considered using repeated diagnoses within the ICD-10 or Read V3 data as an additional factor to increase confidence in case identification? For example, how many individuals in C3 have multiple entries of the G93.3 ICD-10 code, and how many in C4 have multiple ME/CFS-related Read V3 code entries? Repeated entries could potentially strengthen the case for a true clinical diagnosis and help distinguish between transient or tentative codes and more persistent conditions.

Additionally, when you mention that only about one-third of participants have two or more “pieces of evidence,” could you clarify whether this includes repeated instances of the same evidence type (e.g., multiple EHR codes), or only distinct evidence types across cohorts (e.g., one from a questionnaire and one from hospital data)? This distinction would be helpful in understanding how evidence is being aggregated to support cohort membership.

4) I found it a little confusing that the terms "PQ" and "C2" are used somewhat interchangeably throughout the manuscript. For clarity and consistency, it might be helpful to refer to this group exclusively as "C2" once it's been clearly defined. This would make the narrative easier to follow, especially for readers who are less familiar with the structure of the UK Biobank questionnaires.

5) Figure 2 is quite difficult to interpret at first glance. One suggestion to improve readability might be to consistently place the control group at the far left within each set of bars, which could help establish a visual baseline for comparison. I was also unclear about the rationale for comparing only certain subgroups with each other, while omitting others. For example, why are subgroups centered around C1 and C2, whereas C3 and C4 are not similarly explored?

Including a supplementary table that lists all possible pairwise comparisons—with summary statistics and significance levels—could be very helpful for readers who want a comprehensive view of the data.

6) Figure 1 would benefit from including the total number of cases and controls somewhere—either within the figure itself, in the legend, or in the caption. Providing this context would help readers better interpret the proportions being shown. Additionally, I would suggest moving the detailed percentage values currently listed in the figure caption to a supplementary table instead. This would help declutter the caption while still making all relevant information accessible for those interested in the exact figures.

7) Based on the data analyzed and presented, it would be very valuable to include a set of recommendations for defining a high-confidence ME/CFS cohort from among C1–C4. Given the nuanced discussion around misclassification and cohort overlap, a proposed strategy—perhaps combining multiple evidence sources or applying key filtering criteria—could greatly aid other researchers in selecting a more robust case set for downstream analyses.

Is the work clearly and accurately presented and does it cite the current literature?

Yes
Is the study design appropriate and is the work technically sound?

Yes
Are sufficient details of methods and analysis provided to allow replication by others?

Yes
If applicable, is the statistical analysis and its interpretation appropriate?

Yes
Are all the source data underlying the results available to ensure full reproducibility?

Yes
Are the conclusions drawn adequately supported by the results?

Yes

Competing Interests: No competing interests were disclosed.

Reviewer Expertise: Genetic Epidemiology

We confirm that we have read this submission and believe that we have an appropriate level of expertise to confirm that it is of an acceptable scientific standard.

CITE

Report a concern

Respond or Comment

Views

2

Reviewer Report 28 May 2025

Joanna Elson, Newcastle University, Newcastle upon Tyne, England, UK

Approved

https://doi.org/10.3310/nihropenres.15168.r35428

The paper considers the progress in Myalgic Encephalomyelitis/Chronic Fatigue Syndrome (ME/CFS) research which has been frustratingly slow. This is due to a number of factors one being relatively small-scale studies are common in the field and the results of these ... Continue reading

The paper considers the progress in Myalgic Encephalomyelitis/Chronic Fatigue Syndrome (ME/CFS) research which has been frustratingly slow. This is due to a number of factors one being relatively small-scale studies are common in the field and the results of these studies are frequently not replicated. Funding being another factor.

It is frequently suggested that progress could be improved by analyses using the UKBB. In the context of the replication of a study my lab conducted on ME/CFS and mtDNA variation the same suggestion was made. The UKBB provide extensive phenotype and genotype data linked to both ME/CFS cases and controls. This paper highlights some of the difficulties in this approach.

The paper took a systematic approach guided by many years of expertise in studying this complex phenotype. The authors analysed the overlap/ discordance among four UKBB-defined ME/CFS cohorts. Interestingly a total of 5,354 UKB individuals were linked to at least one piece of evidence of ME/CFS. But only a third had 2 or more pieces of evidence for ME/CFS.

The message in the paper concerning use of UKBB data in ME/CFS Research UKBB - ME/CFS cohort numbers are inflated due to frequent misclassification, primarily from conflating "chronic fatigue" with "chronic fatigue syndrome." Chronic fatigue is much more common (linked to many conditions, including primary mitochondrial disease) and does not require hallmark ME/CFS symptoms like post-exertional malaise. The paper highlights that only about one-third of GP-diagnosed ME/CFS cases met a stringent ME/CFS definition in one study. Accurate diagnostics is a challenge for clinicians and patients.

The paper also highlights weak concordance across evidence sources used in the assessment of those on the UKBB. The four evidence sources for ME/CFS in UKBB show poor overlap, with two-thirds of individuals having only one piece of evidence. An example of a cohort where support is stronger is ME/CFS-PQ (C2) and ICD10:G93.3 (C3) cohorts, which show higher internal concordance. This data indicates inconsistencies in diagnosis and reporting.

The various cohorts have strengths and weakness for those wanting to conduct studies in this complex phenotype. The paper gives a good appraisal of these factors which will be of use to other researchers. The overall view highlights that in the UKBB the use of weak case definitions, poor cohort overlap, and inappropriate control selection may undermine the validity of genetic and epidemiological studies conducted using this data, without a thorough understanding of the issues detailed in this paper. The paper concludes ME/CFS status is best supported by multiple lines of evidence. It is able to give an estimated ME/CFS prevalence in the UK to 410,000 (inc. of house bound patients).

Is the work clearly and accurately presented and does it cite the current literature?

Yes
Is the study design appropriate and is the work technically sound?

Yes
Are sufficient details of methods and analysis provided to allow replication by others?

Yes
If applicable, is the statistical analysis and its interpretation appropriate?

Yes
Are all the source data underlying the results available to ensure full reproducibility?

Yes
Are the conclusions drawn adequately supported by the results?

Yes

Competing Interests: No competing interests were disclosed.

Reviewer Expertise: My research explores the role of mitochondrial DNA (mtDNA) variation in complex traits as well as the impact of mtDNA haplogroup background in the diagnosis of primary mitochondrial diseases (PMD). My group's finding have shown that ME/CFS is not a PMD, as ME/CFS patients do not exhibit the pathogenic mtDNA mutations typically seen in those disorders. However, our work also indicates that common mtDNA variants present in the general population may influence ME/CFS.

I confirm that I have read this submission and believe that I have an appropriate level of expertise to confirm that it is of an acceptable scientific standard.

CITE

Report a concern

Respond or Comment

Comments on this article Comments (0)

Version 1

VERSION 1 PUBLISHED 28 Apr 2025

Open Peer Review

Reviewer Status

Reviewer Reports

	Invited Reviewers
	1	2
Version 1 28 Apr 25	read	read

Joanna Elson, Newcastle University, Newcastle upon Tyne, UK
Matthias Wielscher, Medical University of Vienna, Vienna, Austria

Leonardo Vincenzi, Medical University of Vienna, Vienna, Austria

Comments on this article

All Comments(0)

Add a comment

Sign up for content alerts

Back to all reports

Reviewer Report

8 Views

29 May 2025 | for Version 1

Matthias Wielscher, Medical University of Vienna, Vienna, Austria

Leonardo Vincenzi, Department of Dermatology, Medical University of Vienna, Vienna, Vienna, Austria

8 Views Cite this report Responses(0)

Approved

The study by Samms and Ponting presents a comprehensive approach to defining ME/CFS cohorts within the UK Biobank using a combination of self-reported questionnaire responses and linked healthcare records. They delineate four cohorts: C1, based on self-reported ME/CFS at baseline; C2, derived from responses to the UK Biobank pain questionnaire (PQ); C3, defined by the presence of the ICD-10 G93.3 code in hospital episode data; and C4, identified via ME/CFS-related Read V2 and CTV-3 primary care codes. The authors provide a detailed comparative analysis of these cohorts, highlighting both overlaps and discrepancies in case identification across data sources.

This is a timely and important contribution given the diagnostic ambiguity and under-recognition surrounding ME/CFS. By demonstrating how to refine case definitions using available UK Biobank data, the study offers a valuable framework for researchers seeking to identify valid ME/CFS cases for downstream analyses—including genetic, proteomic, and imaging data. The authors also underscore the critical need for careful cohort curation to avoid misclassification and improve study power in ME/CFS research.

Minor comments:

1) Figure 3: I believe the intention is to illustrate the overlap between cohorts C1 to C4, which is highly valuable. However, as a bar plot, it is somewhat challenging to interpret the intersections clearly. I would kindly suggest considering an UpSet plot to more intuitively convey the cohort overlaps, including the number of participants exclusive to each group and those shared across combinations (e.g., https://r-graph-gallery.com/upset-plot.html). Additionally, it might be helpful to present this visualization earlier in the results section, to provide readers with a high-level overview before delving into cohort-specific analyses. A well-constructed UpSet plot could also consolidate and visually represent the key information currently found in Table 3.

2) In the Discussion, you make a compelling case for refining the cohorts to improve the validity of ME/CFS case identification—specifically, by excluding individuals in C1 who report “very good” or “excellent” health, and by filtering C2 for those who also exhibit post-exertional malaise (PEM). I fully agree that these strategies represent a sound approach to making the most of the available data. Given this, I was curious why these suggested filters were not applied when analyzing the cohort properties throughout the manuscript. While I appreciate that providing unfiltered summary statistics is useful for transparency and to caution others against potential misclassification, it would also be highly informative to include at least some analysis of a more stringently defined ME/CFS cohort—essentially, the best approximation of a “valid” ME/CFS group from a public biobank.

In particular, this would add context to Figure 1, where symptoms are analyzed across broad cohort definitions for C2. If certain participants are later acknowledged as likely misclassified, it might be more insightful to see symptom profiles after applying your own recommended filters. This could strengthen the manuscript by showing not only where the issues lie, but also how they might be addressed in practice.

3) Have you considered using repeated diagnoses within the ICD-10 or Read V3 data as an additional factor to increase confidence in case identification? For example, how many individuals in C3 have multiple entries of the G93.3 ICD-10 code, and how many in C4 have multiple ME/CFS-related Read V3 code entries? Repeated entries could potentially strengthen the case for a true clinical diagnosis and help distinguish between transient or tentative codes and more persistent conditions.

Additionally, when you mention that only about one-third of participants have two or more “pieces of evidence,” could you clarify whether this includes repeated instances of the same evidence type (e.g., multiple EHR codes), or only distinct evidence types across cohorts (e.g., one from a questionnaire and one from hospital data)? This distinction would be helpful in understanding how evidence is being aggregated to support cohort membership.

4) I found it a little confusing that the terms "PQ" and "C2" are used somewhat interchangeably throughout the manuscript. For clarity and consistency, it might be helpful to refer to this group exclusively as "C2" once it's been clearly defined. This would make the narrative easier to follow, especially for readers who are less familiar with the structure of the UK Biobank questionnaires.

5) Figure 2 is quite difficult to interpret at first glance. One suggestion to improve readability might be to consistently place the control group at the far left within each set of bars, which could help establish a visual baseline for comparison. I was also unclear about the rationale for comparing only certain subgroups with each other, while omitting others. For example, why are subgroups centered around C1 and C2, whereas C3 and C4 are not similarly explored?

Including a supplementary table that lists all possible pairwise comparisons—with summary statistics and significance levels—could be very helpful for readers who want a comprehensive view of the data.

6) Figure 1 would benefit from including the total number of cases and controls somewhere—either within the figure itself, in the legend, or in the caption. Providing this context would help readers better interpret the proportions being shown. Additionally, I would suggest moving the detailed percentage values currently listed in the figure caption to a supplementary table instead. This would help declutter the caption while still making all relevant information accessible for those interested in the exact figures.

7) Based on the data analyzed and presented, it would be very valuable to include a set of recommendations for defining a high-confidence ME/CFS cohort from among C1–C4. Given the nuanced discussion around misclassification and cohort overlap, a proposed strategy—perhaps combining multiple evidence sources or applying key filtering criteria—could greatly aid other researchers in selecting a more robust case set for downstream analyses.

Is the work clearly and accurately presented and does it cite the current literature?

Yes
Is the study design appropriate and is the work technically sound?

Yes
Are sufficient details of methods and analysis provided to allow replication by others?

Yes
If applicable, is the statistical analysis and its interpretation appropriate?

Yes
Are all the source data underlying the results available to ensure full reproducibility?

Yes
Are the conclusions drawn adequately supported by the results?

Yes

Competing Interests

No competing interests were disclosed.

Reviewer Expertise

Genetic Epidemiology

We confirm that we have read this submission and believe that we have an appropriate level of expertise to confirm that it is of an acceptable scientific standard.

Respond to this report

Responses (0)

Back to all reports

Reviewer Report

2 Views

28 May 2025 | for Version 1

Joanna Elson, Newcastle University, Newcastle upon Tyne, England, UK

2 Views Cite this report Responses(0)

Approved

The paper considers the progress in Myalgic Encephalomyelitis/Chronic Fatigue Syndrome (ME/CFS) research which has been frustratingly slow. This is due to a number of factors one being relatively small-scale studies are common in the field and the results of these studies are frequently not replicated. Funding being another factor.

It is frequently suggested that progress could be improved by analyses using the UKBB. In the context of the replication of a study my lab conducted on ME/CFS and mtDNA variation the same suggestion was made. The UKBB provide extensive phenotype and genotype data linked to both ME/CFS cases and controls. This paper highlights some of the difficulties in this approach.

The paper took a systematic approach guided by many years of expertise in studying this complex phenotype. The authors analysed the overlap/ discordance among four UKBB-defined ME/CFS cohorts. Interestingly a total of 5,354 UKB individuals were linked to at least one piece of evidence of ME/CFS. But only a third had 2 or more pieces of evidence for ME/CFS.

The message in the paper concerning use of UKBB data in ME/CFS Research UKBB - ME/CFS cohort numbers are inflated due to frequent misclassification, primarily from conflating "chronic fatigue" with "chronic fatigue syndrome." Chronic fatigue is much more common (linked to many conditions, including primary mitochondrial disease) and does not require hallmark ME/CFS symptoms like post-exertional malaise. The paper highlights that only about one-third of GP-diagnosed ME/CFS cases met a stringent ME/CFS definition in one study. Accurate diagnostics is a challenge for clinicians and patients.

The paper also highlights weak concordance across evidence sources used in the assessment of those on the UKBB. The four evidence sources for ME/CFS in UKBB show poor overlap, with two-thirds of individuals having only one piece of evidence. An example of a cohort where support is stronger is ME/CFS-PQ (C2) and ICD10:G93.3 (C3) cohorts, which show higher internal concordance. This data indicates inconsistencies in diagnosis and reporting.

The various cohorts have strengths and weakness for those wanting to conduct studies in this complex phenotype. The paper gives a good appraisal of these factors which will be of use to other researchers. The overall view highlights that in the UKBB the use of weak case definitions, poor cohort overlap, and inappropriate control selection may undermine the validity of genetic and epidemiological studies conducted using this data, without a thorough understanding of the issues detailed in this paper. The paper concludes ME/CFS status is best supported by multiple lines of evidence. It is able to give an estimated ME/CFS prevalence in the UK to 410,000 (inc. of house bound patients).

Is the work clearly and accurately presented and does it cite the current literature?

Yes
Is the study design appropriate and is the work technically sound?

Yes
Are sufficient details of methods and analysis provided to allow replication by others?

Yes
If applicable, is the statistical analysis and its interpretation appropriate?

Yes
Are all the source data underlying the results available to ensure full reproducibility?

Yes
Are the conclusions drawn adequately supported by the results?

Yes

Competing Interests

No competing interests were disclosed.

Reviewer Expertise

My research explores the role of mitochondrial DNA (mtDNA) variation in complex traits as well as the impact of mtDNA haplogroup background in the diagnosis of primary mitochondrial diseases (PMD). My group's finding have shown that ME/CFS is not a PMD, as ME/CFS patients do not exhibit the pathogenic mtDNA mutations typically seen in those disorders. However, our work also indicates that common mtDNA variants present in the general population may influence ME/CFS.

I confirm that I have read this submission and believe that I have an appropriate level of expertise to confirm that it is of an acceptable scientific standard.

Respond to this report

Responses (0)

[1] 1. Hanson MR, Germain A: Letter to the editor of Metabolites. Metabolites. 2020; 10(5): 216. PubMed Abstract | Publisher Full Text | Free Full Text

[2] 2. Annesley SJ, Missailidis D, Heng B, et al.: Unravelling shared mechanisms: insights from recent ME/CFS research to illuminate long COVID pathologies. Trends Mol Med. 2024; 30(5): 443–458. PubMed Abstract | Publisher Full Text

[3] 3. Salit IE: Precipitating factors for the Chronic Fatigue Syndrome. J Psychiatr Res. 1997; 31(1): 59–65. PubMed Abstract | Publisher Full Text

[4] 4. Bretherick AD, McGrath SJ, Devereux-Cooke A, et al.: Typing myalgic encephalomyelitis by infection at onset: a decodeme study [version 1; peer review: 1 approved with reservations, 1 not approved]. NIHR Open Res. 2023; 3: 20. PubMed Abstract | Publisher Full Text | Free Full Text

[5] 5. Choutka J, Jansari V, Hornig M, et al.: Unexplained Post-Acute Infection Syndromes. Nat Med. 2022; 28(5): 911–923. PubMed Abstract | Publisher Full Text

[6] 6. Carruthers BM, Jain AK, de Meirleir KL, et al.: Myalgic Encephalomyelitis/Chronic Fatigue Syndrome: clinical working case definition, diagnostic and treatment protocols. J Chronic Fatigue Syndr. 2003; 11(1): 7–115. Publisher Full Text

[7] 7. Bretherick AD, McGrath SJ, Devereux-Cooke A, et al.: Typing Myalgic Encephalomyelitis by infection at onset: a decodeme study [version 4; peer review: 2 approved]. NIHR Open Res. 2023; 3: 20. PubMed Abstract | Publisher Full Text | Free Full Text

[8] 8. Mirin AA, Dimmock ME, Jason LA: Updated ME/CFS prevalence estimates reflecting post-COVID increases and associated economic costs and funding implications. Fatigue. 2022; 10(2): 83–93. Publisher Full Text

[9] 9. Tyson S, Stanley K, Gronlund TA, et al.: Research priorities for Myalgic Encephalomyelitis/Chronic Fatigue Syndrome (ME/CFS): the results of a James Lind alliance priority setting exercise. Fatigue. 2022; 10(4): 200–211. Publisher Full Text

[10] 10. Maksoud R, Magawa C, Eaton-Fitch N, et al.: Biomarkers for Myalgic Encephalomyelitis/Chronic Fatigue Syndrome (ME/CFS): a systematic review. BMC Med. 2023; 21(1): 189. PubMed Abstract | Publisher Full Text | Free Full Text

[11] 11. Dibble JJ, McGrath SJ, Ponting CP: Genetic risk factors of ME/CFS: a critical review. Hum Mol Genet. 2020; 29(R1): R117–R124. PubMed Abstract | Publisher Full Text | Free Full Text

[12] 12. Walitt B, Singh K, LaMunion SR, et al.: Deep phenotyping of post-infectious Myalgic Encephalomyelitis/Chronic Fatigue Syndrome. Nat Commun. 2024; 15(1): 907. PubMed Abstract | Publisher Full Text | Free Full Text

[13] 13. Devereux-Cooke A, Leary S, McGrath SJ, et al.: DecodeME: community recruitment for a large genetics study of Myalgic Encephalomyelitis / Chronic Fatigue Syndrome. BMC Neurol. 2022; 22(1): 269. PubMed Abstract | Publisher Full Text | Free Full Text

[14] 14. Gallagher CS, Ginsburg GS, Musick A: Biobanking with genetics shapes precision medicine and global health. Nat Rev Genet. 2024: 26(3); 191–202. PubMed Abstract | Publisher Full Text

[15] 15. Conroy M, Sellors J, Effingham M, et al.: The advantages of UK Biobank’s open-access strategy for health research. J Intern Med. 2019; 286(4): 389–397. PubMed Abstract | Publisher Full Text | Free Full Text

[16] 16. Bycroft C, Freeman C, Petkova D, et al.: The UK Biobank resource with deep phenotyping and genomic data. Nature. 2018; 562(7726): 203–209. PubMed Abstract | Publisher Full Text | Free Full Text

[17] 17. UK Biobank: About our data. 2024. Reference Source

[18] 18. Fry A, Littlejohns TJ, Sudlow C, et al.: Comparison of sociodemographic and health-related characteristics of UK Biobank participants with those of the general population. Am J Epidemiol. 2017; 186(9): 1026–1034. PubMed Abstract | Publisher Full Text | Free Full Text

[19] 19. Wang Q, Dhindsa RS, Carss K, et al.: Rare variant contribution to human disease in 281,104 UK Biobank exomes. Nature. 2021; 597(7877): 527–532. PubMed Abstract | Publisher Full Text | Free Full Text

[20] 20. Hajdarevic R, Lande A, Mehlsen J, et al.: Genetic association study in Myalgic Encephalomyelitis/Chronic Fatigue Syndrome (ME/CFS) identifies several potential risk loci. Brain Behav Immun. 2022; 102: 362–369. PubMed Abstract | Publisher Full Text

[21] 21. Canela-Xandri O, Rawlik K, Tenesa A: An atlas of genetic associations in UK Biobank. Nat Genet. 2018; 50(11): 1593–1599. PubMed Abstract | Publisher Full Text | Free Full Text

[22] 22. Neale Lab: UK Biobank. 2018. Reference Source

[23] 23. UK Biobank: The verbal interview within ACE centres. Reference Source

[24] 24. UK Biobank: UK Biobank verbal interview stage. 2012; 1–17. Reference Source

[25] 25. UK Biobank: Experience of Pain Questionnaire. Reference Source

[26] 26. UK Biobank: Resource 138483 hospital episode statistics. Reference Source

[27] 27. UK Biobank: Resource 591 - primary care linked data. Reference Source

[28] 28. Cho HJ, Menezes PR, Hotopf M, et al.: Comparative epidemiology of Chronic Fatigue Syndrome in Brazilian and British primary care: prevalence and recognition. Br J Psychiatry. 2009; 194(2): 117–122. PubMed Abstract | Publisher Full Text

[29] 29. Wessely S, Chalder T, Hirsch S, et al.: The prevalence and morbidity of chronic fatigue and Chronic Fatigue Syndrome: a prospective primary care study. Am J Public Health. 1997; 87(9): 1449–1455. PubMed Abstract | Publisher Full Text | Free Full Text

[30] 30. Nacul LC, Lacerda EM, Pheby D, et al.: Prevalence of Myalgic Encephalomyelitis/Chronic Fatigue Syndrome (ME/CFS) in three regions of England: a repeated cross-sectional study in primary care. BMC Med. 2011; 9: 91. PubMed Abstract | Publisher Full Text | Free Full Text

[31] 31. Samms GL, Ponting CP: Unequal access to diagnosis of myalgic encephalomyelitis in England. BMC Public Health. in press.

[32] 32. Tanigawa Y, Li J, Justesen JM, et al.: Components of genetic associations across 2,138 phenotypes in the UK Biobank highlight adipocyte biology. Nat Commun. 2019; 10(1): 4064. PubMed Abstract | Publisher Full Text | Free Full Text

[33] 33. Neale Lab UK Biobank GWAS round 2 results. 2018. Reference Source

[34] 34. Zhou W, Nielsen JB, Fritsche LG, et al.: Efficiently controlling for case-control imbalance and sample relatedness in large-scale genetic association studies. Nat Genet. 2018; 50(9): 1335–1341. PubMed Abstract | Publisher Full Text | Free Full Text

[35] 35. Pan UK Biobank Team. Reference Source

[36] 36. Canela-Xandri O, Rawlik K, Tenesa A: An atlas of genetic associations in UK Biobank. Nat Genet. 2018; 50(11): 1593–1599. PubMed Abstract | Publisher Full Text | Free Full Text

[37] 37. Beentjes SV, Kaczmarczyk J, Cassar A, et al.: Replicated Blood-based biomarkers for Myalgic Encephalomyelitis not explicable by inactivity. medRxiv. 2024. Publisher Full Text

[38] 38. UK Biobank: Category 1008 physical activity measurement. Reference Source

[39] 39. Sun BB, Chiou J, Traylor M, et al.: Plasma proteomic associations with genetics and health in the UK Biobank. Nature. 2023; 622(7982): 329–338. PubMed Abstract | Publisher Full Text | Free Full Text

[40] 40. UK Biobank: Imaging study updates. 2024. Reference Source

[41] 41. Office for National Statistics: Population estimates for the UK, England, Wales, Scotland, and Northern Ireland: mid-2022. Reference Source

[42] 42. Leitsalu L, Haller T, Esko T, et al.: Cohort profile: Estonian Biobank of the Estonian Genome Center, University of Tartu. Int J Epidemiol. 2015; 44(4): 1137–47. PubMed Abstract | Publisher Full Text

[43] 43. Kurki MI, Karjalainen J, Palta P, et al.: FinnGen provides genetic insights from a well-phenotyped isolated population. Nature. 2023; 613(7944): 508–518. PubMed Abstract | Publisher Full Text | Free Full Text

[44] 44. All of us data browser - Chronic Fatigue Syndrome. Reference Source

[45] 45. Pendergrast T, Brown A, Sunnquist M, et al.: Housebound versus nonhousebound patients with Myalgic Encephalomyelitis and Chronic Fatigue Syndrome. Chronic Illn. 2016; 12(4): 292–307. PubMed Abstract | Publisher Full Text | Free Full Text

[46] 46. UK Biobank: Apply for access website. https://www.ukbiobank.ac.uk/enable-your-research/apply-for-access

[47] 47. Extended Data Table 1. UK Biobank Field IDs used in analyses. Extended Data Table 2. Definition of ME/CFS codes in Primary care records. https://osf.io/rp5ca

Defining a High-Quality Myalgic Encephalomyelitis/Chronic Fatigue Syndrome cohort in UK Biobank

Abstract

Background

Methods

Results

Conclusions

Plain Language Summary

Plain English summary

Keywords

Introduction

Methods

Public and Patient Involvement

Data acquisition

Baseline measurements

Pain Questionnaire (PQ)

Results

Cohort 1: Self-reported CFS diagnosis

Table 1. Self-reported overall health ratings in C1.

Cohort 2: ME/CFS defined in the Pain Questionnaire

Figure 1. ‘Yes’ responses to ME/CFS-relevant symptom and comorbidity questions as asked in the PQ.

Figure 2. ‘Yes’ responses as a percentage of each cohort for post-exertional malaise-related questions.

Cohort 3: Hospital Inpatient Data: ICD-10 G93.3

Cohort 4: Primary care records

ME/CFS status in 4 cohorts

Table 2. Prevalence and female-to-male (F/M) ratio of ME/CFS across different sources of ME/CFS data in UKB.

Table 3. Numbers of UK participants with 1, 2, 3 or 4 pieces of evidence for ME/CFS.

Discussion

Figure 3. Concordance of ME/CFS evidence across data sources for each of the 4 UKB cohorts.

Ethical approval

Consent

Data availability

Author contributions

Acknowledgements

References

Comments on this article Comments (0)

Open Peer Review

Comments on this article Comments (0)

Open Peer Review

Reviewer Status

Reviewer Reports

Comments on this article

Competing Interests Policy

Stay Updated

Are you an NIHR-funded researcher?

Thank you!

Table 1. Self-reported overall health ratings in C₁.