To describe current practices and experiences of people with ME/CFS and with long COVID mapped to key NICE guidelines.

1. To describe the time to diagnosis in relation to the year diagnosed.

2. To determine the impact of the NICE guideline on ME/CFS on patients’ lives and the service they received since its publication in October 2021.

This is a cross-sectional survey of the current practices and experiences of people with ME/CFS and long COVID.

This study involved a secondary analysis of anonymised survey data collected to explore the experiences of individuals with ME/CFS and long COVID. The data were analysed in a secure environment in accordance with the University of Exeter’s data protection policies. No new data were collected, and the secondary analysis was conducted within the scope of the original data collection purpose, as outlined to participants at the time of the survey.

An online survey was open for participation between 22 May and 31 July 2023. Eligibility was self-assessed, and the survey was available in English. People with suspected or a diagnosis of ME/CFS or long COVID were invited to participate in answering questions to understand their experiences with diagnosis, symptom management and NHS care, as well as map out coverage and care for people in the UK and determine areas for improvement. A blank copy of the survey measure has been included as a online supplemental file 1. Consent for secondary anonymous data analysis was carried out as follows. The opening page of the survey served as the participant information sheet and consent form for data to be analysed independently. Secondary data analysis was performed on the anonymous data by the University of Exeter following confirmation from the University sponsor (2023).

No formal a priori sample size calculation was performed, and the number of responses during the study period determined the sample size. However, it was estimated that approximately 1000 people would be recruited. Considering an estimated prevalence of 250–265 000 in the UK and a confidence level of 95%, 664 people would give a margin of error of 5%.

A self-selected sample of people on the ME Association https://meassociation.org.uk mailing lists and those that clicked through digital ads on relevant websites. The study aims of the project and how the results would be used were clearly stated to respondents to inform their choice to participate as follows: (a) inform more people about ME/CFS and long COVID to help them obtain an accurate diagnosis if needed and find the right support to improve their life quality; (b) produce an independent report that will raise awareness and help improve the standard of healthcare by working with the NHS and social care services. Participants could only take the survey once but were allowed to complete it in multiple sessions to accommodate fatigue for up to 3 days if they were too fatigued to complete it in one go. Participants had to confirm whether they were completing for themselves or for a close friend/family member had a diagnosis or symptoms of ME/CFS or long COVID. For respondents without a formal diagnosis, the survey asked them to confirm the presence of four key symptoms (fatigue, PEM, cognitive dysfunction and unrefreshing sleep) and their persistence for at least 3 months, as outlined in the NICE guidelines. For those reporting a current diagnosis, no specific question was included to confirm adherence to NICE or other diagnostic criteria.

Data were collected on demographics, diagnosis, symptoms, comorbidities and treatments, experiences around NHS and social care support, and key symptom management strategies.

The current study has not imputed missing data, and the number of responses at variable levels is reported in the results/Appendix. Missing data were handled by incorporating all available responses for each analysis. Participants were not excluded due to incomplete survey submissions, and analyses were conducted using the maximum number of responses available for each variable. The number of respondents included in each analysis is transparently reported in the Results section.

Descriptive analyses were conducted to characterise gender, age, diagnoses, symptoms, comorbidities, treatments, experiences of support from NHS and social care, and key symptom management strategies. These were reported as frequencies and percentages for both ME/CFS and long COVID. Comparisons, when performed, were reported using χ² tests, independent samples median tests or t-tests according to the variable type and were accompanied by descriptive statistics.11–13 For the 10 most frequently reported symptoms, counts of how many times each pair of symptoms were reported by a participant were calculated to obtain symptom co-occurrence matrices for ME/CFS and long COVID. Hierarchical clustering analysis (HCA) was then performed using the Euclidean distance method to obtain distance matrices, with the complete agglomeration method used for clustering (default settings in the heatmap.2 function in R V.4.3.2).14 15 HCA was conducted using symptom co-occurrence matrices as input. Symptom co-occurrence matrices were constructed by calculating the frequency with which pairs of symptoms were reported together. Clustering was performed using the Euclidean distance metric to compute dissimilarities and the complete agglomeration method for grouping. Clusters were visualised using dendrograms to highlight relationships among symptoms. Graph theory was applied to provide a further comprehensive analysis of symptom interactions. For graph theory analysis, symptom co-occurrence matrices were converted into weighted, undirected graphs where nodes represented symptoms and edges indicated co-occurrences. The strength of each edge was determined by the frequency of co-occurrence. Metrics such as degree centrality (to identify the most connected symptoms), betweenness centrality (to determine symptoms acting as bridges) and clustering coefficients (to measure the tendency of symptoms to form clusters) were calculated using Python libraries. Network graphs were visualised to interpret the relationships between symptoms. The strength of each edge was determined by the frequency of co-occurrence, forming weighted networks for both ME/CFS and long COVID. Key graph metrics were calculated: degree centrality to identify the most connected symptoms, betweenness centrality to determine symptoms that act as bridges within the network and clustering coefficient to measure the tendency of symptoms to cluster together. Data analysis was completed using SPSS V.29.0 (SPSS, Chicago, Illinois, USA), Python V.3.12 with the PyCharm IDE 2024.4.1 (JetBrains) and R. Statistical significance was indicated at p<0.05.16 17

Patients and the public were actively involved in the development of this study. The survey was designed in consultation with individuals living with ME/CFS and long COVID, as well as with representatives from the ME Association, to ensure that the questions addressed the key concerns and experiences of those directly affected by these conditions. Feedback was gathered during the survey design phase through a pilot test, where individuals with ME/CFS trialled the survey and provided suggestions to improve its accessibility and relevance.

Symptom prevalence with ME/CFS: a total of 450 participants within the subset of those diagnosed with ME/CFS responded with information about symptom presentation (Q5=diagnosis, Q8=symptom) (figure 4A). The 10 most commonly reported symptoms with ME/CFS (starting with the most commonly reported) were unrefreshing sleep or sleep disturbance (n=398, 88.22%), debilitating fatigue/exhaustion (n=397, 71.11%), brain fog or cognitive difficulties (n=394, 88.44%), pain in muscles, joints or nerves (n=352, 87.65%), PE M (n=320, 78.22%), increased sensitivity to light and noise (n=269, 58.67%), headache (n=264, 59.78%), irritable-bowel-type symptoms (IBS) (n=256, 54.00%), increased sensitivity to hot and cold temperatures (n=253, 56.22%) and dizziness (n=243, 56.89%).

Symptom prevalence with long COVID: there was a total of 267 participants within the subset of those diagnosed with long COVID that responded with information about symptom presentation (Q5=diagnosis, Q8=symptom) (figure 4A). The 11 (there were two in joint 10th position) most commonly reported symptoms with long COVID (starting with the most commonly reported) were debilitating fatigue/exhaustion (n=189, 42.00%), brain fog or cognitive difficulties (n=186, 41.33%), pain in muscles, joints or nerves (n=171, 38.00%), unrefreshing sleep or sleep disturbance (n=165, 36.66%), breathlessness (n=134, 29.78%), headache (n=117, 26.00%), PEM (n=109, 24.22%), IBS (n=103, 22.89%), dizziness (n=98, 21.78%), ringing in the ears (n=84, 18.67%) and excessive sweating when asleep (n=84, 18.67%).

Most reported symptoms with ME/CFS or long COVID taken together: of the 10 most prevalent symptoms for ME/CFS and long COVID, 8 were common to both: brain fog, dizziness, fatigue, headache, IBS, pain in muscles, joints or nerves, PEM and sleep disturbances. Light/noise sensitivity and temperature sensitivity were only within the 10 most common symptoms with ME/CFS, and breathlessness, ears ringing and sleep sweats were only within the 10 most common symptoms with long COVID.

Symptom co-occurrence network for individuals with ME (figure 4B): this network uses graph theory to visualise how symptoms reported by the survey respondents inter-relate, based on their co-occurrence: nodes represent different symptoms experienced by individuals. Edges indicate that these symptoms frequently co-occur. The thickness of an edge correlates with the frequency of co-occurrence; thicker edges mean that the symptoms appear together more often among respondents. The network layout is structured so that symptoms with more connections are generally positioned more centrally, suggesting they may be common or pivotal symptoms in the condition.

The network graph visually represents how different symptoms interconnect based on their co-occurrence in patient reports. Nodes in the network represent various symptoms, and edges between them indicate that these symptoms are frequently experienced together by individuals with ME. Thicker edges denote a higher frequency of co-occurrence, highlighting symptom pairs or clusters that are more commonly reported.

This network analysis shows that certain symptoms such as fatigue, pain and brain fog not only show strong connections to many other symptoms but also feature centrally within the network. This suggests that these symptoms are particularly prevalent and might be central to the experiences of many individuals with ME. Additionally, severity analysis indicates that these symptoms are often associated with higher impact severity scores, suggesting they significantly affect patients’ quality of life.

Symptom co-occurrence network for individuals with long COVID (figure 4B): a symptom co-occurrence network was constructed using graph theory to understand the relationships between different symptoms of long COVID and their co-occurrence patterns.

The graph was created by treating each symptom as a node. Edges between nodes were drawn if the corresponding symptoms co-occurred in at least one individual, and the weight of each edge was proportional to the number of individuals in which both symptoms appeared together. Visualising this network provided a graphical representation of how symptoms cluster together.

Symptoms such as fatigue, brain fog, and breathlessness tend to be linked with higher severity ratings. These symptoms, represented by nodes connected with thicker edges in the network, signify common co-occurrence and a stronger impact on the individuals’ health.

Associations of symptoms: for both ME/CFS and long COVID, we explored co-occurrence (how often a participant reported both of a pair of symptoms) within the 10 most frequently reported symptoms, then used HCA to identify symptom clusters.

ME/CFS symptom clusters: there were 448 participants with a diagnosis who responded to relevant questions (Q5=diagnosis, Q8=symptoms) and who reported at least one of the 10 most common symptoms. There were three clear clusters of co-occurrence for ME/CFS (figure 4C).

Cluster 1: debilitating fatigue/exhaustion, unrefreshing sleep or sleep disturbance, and ‘brain fog’ or cognitive difficulties.

Cluster 2: PEM, and pain in muscles, joints or nerves.

Cluster 3: headache, increased sensitivity to light and noise, dizziness, increased sensitivity to hot and cold temperatures, and IBS.

Long COVID symptom clusters: there were 261 participants with a diagnosis who responded to relevant questions (Q5=diagnosis, Q8=symptoms) and who reported at least one of the 10 most common symptoms. There were three clear co-occurrence clusters with long COVID (figure 4C).

Cluster 1: debilitating fatigue/exhaustion, ‘brain fog’ or cognitive difficulties, pain in muscles, joints or nerves, and unrefreshing sleep or sleep disturbance.

Cluster 2: ringing in the ears and excessive sweating when asleep.

Cluster 3: breathlessness, headache, PEM, IBS and dizziness.

Symptoms and impact of condition: we explored the relationship between the impact of the condition (ME/CFS or long COVID) and the prevalence of symptoms within the subset of participants with a diagnosis who responded to relevant questions (Q5=diagnosis, Q8=symptoms, Q15b=impact severity). With both ME/CFS and long COVID, the more prevalent a symptom was, the lower the proportion (%) of respondents with that symptom that reported ‘severe’ or ‘very severe impact’ of the condition. For example, rarer symptoms are associated with more severe impact on life (figure 4D).

A total of 2892 cases of diagnosed ME/CFS were reported. The distribution of these cases over time is as follows: 8.89% of the cases were diagnosed in the last month. 12.07% were diagnosed 2–6 months ago. 5.08% were diagnosed 7–11 months ago. 11.10% were diagnosed 1–2 years ago. 16.32% were diagnosed 3–4 years ago. A significant proportion (43.40%) were diagnosed 5 years ago or longer. 2.70% of the respondents reported that they have not been seen. 0.45% of the respondents reported not knowing when they were diagnosed.

We found that the number of comorbidities was significantly related to time to diagnosis in people with ME/CFS, whereas this was not the case in people with long COVID. Other factors such as age, ethnicity and gender were not significantly related to time to diagnosis in participants with either ME/CFS or long COVID, R<0.1 and p>0.05.

We observed similar time to diagnosis in the first 12 months and in general for those diagnosed 3 years before (n=979) and after (n=743) the 2021 NICE guideline (figure 5, left).

Figure 5 (right and left) shows a potential trend to faster diagnosis in the 3-, 3 to 6- and 7 to 12-month periods after 2022, although trends were fluctuating over the reporting periods.

To determine the impact of the NICE guideline on people with ME/CFS’s life and service experience since its publication in October 2021. According to this study result, 80 (10.1%) respondents from 4761 thought that the NICE guideline’s recommendations had made a positive difference to the healthcare they received, with 70 (1.5%) people noting a positive significant impact.

480 (10.1%) people from 4761 respondents thought the NICE guideline’s recommendations had made a positive difference to the healthcare they received, with 70 (1.5%) people noting a positive significant impact.

This study has several limitations. First, the use of a convenience sample relying on self-reported eligibility and data introduces potential biases, including selection and reporting biases. Our sample may overrepresent individuals who are more engaged with patient advocacy groups or have access to online platforms, potentially excluding those with limited internet access or differing healthcare experiences. Our sample comprised 6053 females, 1077 males, 129 non-binary individuals and 3155 participants who preferred not to disclose their gender, spanning all four nations of the UK. While this demographic distribution reflects the typical population engaged with ME/CFS and long COVID services, it may not represent the broader patient population, limiting our findings’ generalisability. Additionally, the reliance on self-reported diagnoses without verification through clinical records or standardised diagnostic tools may affect the reliability of the data, as self-reported symptoms can lead to both overestimation and underestimation of disease prevalence. However, the substantial sample size of 8804 participants enhances the study’s statistical power, providing a margin of error between 1% and 1.5% at a 99% CI. Including caregivers reporting on behalf of individuals with ME/CFS and long COVID may have introduced variability in responses; nonetheless, sensitivity analyses indicated no significant differences, supporting the inclusion of perspectives from those with severe disease, as recognised in the NICE guidelines. The study offers a comprehensive analysis of the survey findings by using all available data without excluding incomplete responses. However, this approach may introduce variability, as participants who did not complete all sections of the survey may systematically differ from those who did. The distinction between ME/CFS (ICD-10 G93.31) and long COVID (ICD-10 U09.9) is complex and may not always be clearly understood by patients or clinicians. This limitation is particularly relevant in cases of post-COVID-19 fatigue where diagnostic ambiguity may arise. While our survey allowed participants to select multiple diagnoses, self-reporting may not capture nuances in clinical criteria, and future studies could benefit from incorporating clinician-verified diagnoses or more specific diagnostic questions to address this overlap. Finally, data collection occurred in the UK during a period of notable dissatisfaction with NHS service provision, as highlighted in the 2023 British Social Attitudes survey, which may have influenced participants’ responses and should be considered when interpreting the conclusions.

Consent obtained directly from patient(s).

This study involves human participants. The project used secondary data previously collected by the ME Association. As this data was gathered in accordance with the University of Exeter’s ethical policy and participants consented to its use in publications, the university granted an ethical exemption for this project. Informed consent was obtained from all participants at the time of data collection, as outlined in the survey information sheet. Participants were informed about the use of their anonymised responses for research purposes, and consent was implicitly provided uponon survey completion.