Article Text
PDF
Abstract
Introduction Long COVID (LC) is a medical condition first described and documented through anecdotes on social media by patients prior to being recognised by WHO as a disease. Although >50 prolonged symptoms of LC have been described, it remains a diagnostic challenge for military providers and therefore threatens operational readiness.
Methods On 9 September 2021, an online survey was emailed to 2192 Belgian Defence personnel who had previously tested PCR positive for SARS-CoV-2 between 17 August 2020 and 31 May 2021. A total of 718 validated responses were received.
Descriptive analyses determined the prevalence of LC and 10 most common symptoms and their duration following infection. In the explanatory analyses, risk factors related to LC were identified. To establish the health-related impact of LC on quality of life (HRQoL), we used the results from the EuroQol 5 Dimension 5 Level questionnaire.
Results The most frequent symptoms that were reported for >3 months were fatigue, lack of energy and breathing difficulties.
47.35% of the respondents reported at least one persistent symptom, while 21.87% reported more than 3 symptoms lasting for at least 3 months after the initial COVID-19 infection. Most patients with LC suffered from symptoms of a neuropsychiatric nature (71.76%).
LC was significantly associated with obesity; pre-existing respiratory disease and blood or immune disorders. Physical activity of >3 hours per week halved the risk of LC.
The total QoL is reduced in patients with LC. Considering the five dimensions of the questionnaire, only the self-care dimension was not influenced by the presence of LC.
Conclusions Almost half of Belgian Defence personnel developed LC after a confirmed COVID-19 infection, similar to numbers found in the Belgian population. Patients with LC would likely benefit from a multidisciplinary rehabilitation approach that addresses shortness of breath, fatigue and mood disturbance.
- COVID-19
- mental health
- immunology
- epidemiology
- health policy
- public health
Data availability statement
No data are available. Because of the small sample size and the specificity of the information included (ie, patients identified with long COVID) in addition to other information (ie, the presence of specific diseases, gender, age, etc), the anonymity of the participants cannot be guaranteed when sharing the dataset.
This article is made freely available for personal use in accordance with BMJ’s website terms and conditions for the duration of the covid-19 pandemic or until otherwise determined by BMJ. You may use, download and print the article for any lawful, non-commercial purpose (including text and data mining) provided that all copyright notices and trade marks are retained.
https://bmj.com/coronavirus/usageStatistics from Altmetric.com
WHAT IS ALREADY KNOWN ON THIS TOPIC
Long COVID (LC) remains a diagnostic challenge; even today clear and universally accepted definitions are lacking.
In a military population, prolonged symptoms might impact the operational readiness of soldiers; evaluating the prevalence, impact and related factors is therefore important, especially among young adults who could be affected for a greater proportion of their lives.
WHAT THIS STUDY ADDS
We found an overall prevalence proportion of almost half reporting at least one symptom of LC with fatigue or exhaustion, lack of energy, breathing difficulties and loss of smell as the most frequent symptoms lasting at least 3 months.
LC was found to be significantly associated with obesity, pre-existing respiratory disease and blood or immune disease.
Physical activity >3 hours/week reduced the risk of LC.
HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY
LC has an important impact on the quality of life.
Multidisciplinary rehabilitation may be an important first step to protect employees.
Introduction
The disease long COVID (LC) was first described and documented on social media by patients prior to being officially recognised by WHO.1 In October 2021, WHO defined the post-COVID-19 condition. According to the definition, LC occurs in individuals with a history of probable or confirmed SARS-CoV-2 infection, usually 3 months from the onset of the infection, with symptoms that last for at least 2 months and cannot be explained by an alternative diagnosis.2
More than 50 prolonged symptoms of LC have been described3 including chest pain, difficulties with breathing, pain when breathing, painful muscles, ageusia or anosmia, tingling extremities, lump in throat, feeling hot and cold alternately, heavy arms or legs and general tiredness.4
LC remains a diagnostic challenge for many reasons including a high proportion of asymptomatic SARS-CoV-2 cases5 and the lack of testing early in the pandemic.6 Positive postinfectious serology may be useful, however, some patients may remain or become seronegative after infection.7 SARS-CoV-2 has a multisystemic effect because the ACE receptor to which it binds is found in several tissues. Broad analysis of relevant scientific publications suggests that LC-related symptoms improve over time.8 In a KCE (Belgian Healthcare Knowledge Centre) literature review, studies including those using outpatient data reported that between 3% and 36% (median value 32%) of patients still present with symptoms even after 3 months.8
Risk factors for LC include increasing age,9–12 experiencing more than five symptoms during the first week of illness and the severity of illness during acute COVID-19 especially hospital admission.10–12 In addition, pre-existing respiratory disease, higher body mass index,10 11 13 the presence of comorbidities, female sex11 12 and dyspnoea at 4–8 weeks.10
Military personnel in general, require physical and mental health usually greater than the civilian population during training or when deployed, as they may have to face harsh physical conditions (hostile climate, carrying heavy loads like ballistic protective gear) and diminished access to medical facilities.
Therefore, the objectives of this study are threefold: first, determine the prevalence of LC and the different clusters of symptoms in the Belgian Defence population while describing the range and severity of symptoms. Second, identify risk factors such as age, gender, blood type, body mass index (BMI), smoking habits (non-smoker, former-smoker, daily or occasional smoker), number of hours of exercise per week before infection and chronic diseases present before COVID-19 for LC. Third, assess the impact of LC on the quality of life (QoL) using the EuroQol 5 Dimension 5 Level (EQ-5D-5L) questionnaire.
Methods
Procedures
On 20 September 2021, an online survey was sent via email to the work addresses of 2192 active Belgian Defence personnel who were known to have been SARS-CoV-2 PCR positive per the National Military Contact and tracing Centre (NMCC) between 17 August 2020 and 31 May 2021. Information including a letter detailing the study goals, procedures and an informed consent document were included (see online supplemental material). The data were collected during 3 weeks via the online survey platform MedSurveys and provided anonymously to the researchers by the data manager of the Directorate General Health & Well-being.
Supplemental material
According to the sample size calculation,14 328 interviewees are needed for a statistically sufficiently large sample to achieve a 95% confidence level.
Variables and instruments
The online survey (see online supplemental material) consisted of three parts: a generic demographics and medical part, a specific COVID-19-related part and the EQ-5D-5L questionnaire as previously described by the EuroQol group,15 valuing health-related QoL (HRQoL).
In the absence of international consensus, for the purposes of this study, LC was defined as the persistence of at least one symptom beyond 12 weeks after the onset of COVID-19 that cannot be explained by another diagnosis.
Statistical analysis
Analyses were performed with IBM SPSS V.25.0 (IBM, Armonk, New York, USA); the significance level for all analyses was set at an alpha of 0.05.
The study sample was first described (frequencies, means and SD) and compared with the target population using data from the NMCC.
Summary statistics then described the prevalence of LC and seven different clusters of LC presentation: chronic fatigue syndrome, cardiorespiratory, neuropsychiatric, dysautonomia, skin, musculo-articular and digestive cluster. Additionally, the most common symptoms during various time frames in the population were also described. The clusters of LC were based on a clinical categorisation of the symptoms (Table 1).
Clusters of long COVID (LC)
LC risk factors
To determine risk factors related to the development of LC, we used binary logistic regression in the explanatory analysis. The independent variables were defined as age, sex, blood type, obesity at the time of infection (BMI ≥30), smoking habits, number of hours of exercise per week prior to infection, chronic disease comorbidity and medical operational category at the time of infection (A indicates medical fitness for all missions abroad, B temporary unfitness and C permanent unfitness. W (waiver) indicates medical fitness under certain conditions. O (overdue) means that the periodic assessment has expired and S (suspended) that the medical file is awaiting complementary information. Chronic diseases were only included in the regression model if at least 10 participants reported the presence of this disease. In a multivariate analysis, all independent variables were added to one model with the dependent variable being LC (yes/no) or a specific cluster of LC.
Impact of LC on QoL
The EQ-5D-5L questionnaire was used to measure the impact of LC on the QoL. All questions were scored for two different time points which included “how I felt before my COVID-19 infection” and “how I feel now” (both questions retrospective).
Five dependent variables were created (mobility, activities of daily living, self-care, anxiety/depression and pain/discomfort) and a distinction was made between an improved, neutral or a deteriorated state for each variable after COVID-19. The improved state was considered ‘missing’ because it occurred very rarely and an improvement of the health status was not expected after COVID-19. Thus, a binary variable was created; neutral/deteriorated.
The impact of LC on QoL was measured by binary logistic regression. The five variables described above were used as the dependent variable after controlling for the aforementioned independent variables.
Finally, a binary variable was created by taking the difference in health estimates out of a possible score of 100 before and after COVID-19. This number was then used to determine whether a higher deviation than the median value in the group occurred. A binary logistic regression was used to analyse which variables could be related to a higher deviation of the difference in health estimate before and after COVID-19. The independent variables were the different clusters of LC corrected for the aforementioned independent variables.
Results
Response rate
A total of 34.7% replied and 718 questionnaires were fully completed, which is statistically large enough for a confidence level of 95%. Thirty-six incomplete questionnaires were not included. Compared with the overall defence target population, the study sample was significantly older (median 43 compared with 36 years) and contained significantly more women (12.53% compared with 9.40%), officers (21.32% compared with 12.64%) and non-commissioned officers (48.81% compared with 40.88%).
Descriptive statistics
Descriptive statistics of the study sample can be found in Table 2. In the fourth quarter of 2020 (second wave), 61.8% (n=444) of the participants tested positive, meaning their infection took place 7–10 months before the study; 3.5% (n=25) of the participants were infected earlier (August or September 2020) and 34.7% (n=249) were infected later (January 2021–May 2021).
Descriptive statistics of the study sample
Of the study participants, 62.12% reported no chronic diseases prior to their infection with COVID-19.
Mild-to-moderate (=non-hospitalised) COVID-19 was reported by 96.8% (n=695) of the participants, while 3.21% (n=23) were hospitalised of which 26% (n=6) needed admission to the intensive care unit.
Prevalence of LC
Among the 718 respondents, 340 (47.35%) people reported at least one persistent symptom lasting >3 months.
Most of the LC cases presented within a neuropsychiatric cluster (71.76%). Fatigue and lack of energy were reported in 57.65% of cases; a cardiorespiratory cluster was found in 45.29%; a dysautonomia cluster in 20%; a musculo-articular cluster in 24.71%; a cutaneous cluster in 8.53% and a digestive cluster in 30.88% of LC cases. Only one in five LC cases presented as a single cluster.
The 10 most frequent symptoms that were reported for >3 months were fatigue or exhaustion (24.23%), lack of energy (19.22%), breathing difficulties (19.08%), loss of smell (12.40%), muscle pain and/or weakness (9.75%), concentration problems (9.33%), loss of taste (8.64%), insomnia (8.64%), joint pain (7.38%) and memory problems (7.24%).
Explanatory analysis
In a multivariate analysis (Table 3), obesity increased the risk for LC (OR 2.225 (95% CI 1.334 to 3.711)) as well as respiratory diseases (OR 2.715 (95% CI 1.108 to 6.652)) and a history of a blood or immune system disorder (OR 12.345 (95% CI 1.068 to 142.677)). Physical activity of 3 hours per week or more decreases the odds for LC (OR 0.446 (95% CI 0.210 to 0.948)).
Significant (p<0.05) p values of the binary logistic regression analysis of long COVID and the clusters of long COVID (without cutaneous cluster n<30) with confounding factors and their ORs
On a subanalysis examining associations between risk factors and clustered symptoms, obesity was significantly associated with the chronic fatigue cluster (OR 1.978 (95% CI 1.180 to 3.315)). With a BMI of 30 or above, the odds for a cardiorespiratory cluster of LC increased (OR 2.68 (95% CI 1.55 to 4.64)). Respiratory disease (OR 3.710 (95% CI 1.086 to 12.671)) and older age (OR 1.038 (95% CI 1.002 to 1.076)) increased the odds of developing a dysautonomia cluster.
Non-smokers were less likely to develop a cardiorespiratory cluster than daily smokers (OR 0.364 (95% CI 0.173 to 0.764)). Physical activity, including practising sport between 3 and 7 hours, or >7 hours per week reduced the odds of developing a neuropsychiatric cluster (OR 0.405 (95% CI 0.194 to 0.843) and OR 0.344 (95% CI 0.147 to 0.804), respectively).
Obesity (OR 3.216 (95% CI 1.626 to 6.247)), the presence of metabolic disease (OR 3.824 (95% CI 1.054 to 13.870)) and a haematological or immune system disease (OR 8.359 (95% CI 1.299 to 53.781)) were associated with the musculo-articular cluster of LC, while blood type O was associated with lower odds (OR 0.504 (95% CI 0.278 to 0.914)).
The odds of developing a digestive cluster of LC were increased among those with a history of digestive disease (OR 7.160 (95% CI 1.903 to 26.935)), respiratory disease (OR 4.302 (95% CI 1.219 to 15.185)) or obesity (OR 2.534 (95% CI 1.361 to 4.716)). The odds of developing a digestive cluster of LC decreased among those reporting a comorbid condition (ie, at least one existing chronic disease) (OR 0.229 (95% CI 0.065 to 0.812)).
Impact on QoL
The impact of LC on the QoL was measured using binary logistic regression (Table 4).
Significant (p<0.05) p values of the binary logistic regression analysis of four domains of QoL and the clusters of long COVID and their ORs
Mobility was negatively associated among those with LC reporting a cardiorespiratory cluster (OR 2.62 (95% CI 1.53 to 4.52)) of symptoms.
Deterioration in the activities of daily living was observed among those with LC reporting three different clusters of symptoms: chronic fatigue cluster (OR 4.11 (95% CI 2.18 to 7.76)), cardiorespiratory cluster (OR 2.66 (95% CI 1.48 to 4.79)) and autonomic nervous system cluster (OR 2.33 (95% CI 1.11 to 4.92)). For former or non-smokers, the impact seems to be smaller compared with smokers (OR 0.23 (95% CI 0.08 to 0.64) and OR 0.37 (95% CI 0.16 to 0.88), respectively).
Anxiety or depression was associated with four clusters of LC: chronic fatigue cluster (OR 4.15 (95% CI 2.10 to 8.17)), neuropsychiatric cluster (OR 7.13 (95% CI 2.95 to 17.23)), cutaneous cluster (OR 5.56 (95% CI 1.46 to 21.12)) and digestive cluster (OR 3.05 (95% CI 1.61 to 5.79)).
Increased levels within the pain spectrum were associated with LC in general (OR 3.33 (95% CI 1.50 to 7.38)) and more specifically within three clusters: chronic fatigue cluster (OR 2.48 (95% CI 1.44 to 4.27)), dysautonomia cluster (OR 2.72 (95% CI 1.32 to 5.63)) and musculo-articular cluster (OR 3.18 (95% CI 1.66 to 6.07)). For people reporting previous digestive or immune/blood disorders, the effect of LC on pain was less when compared with people who did not report any chronic conditions (OR 0.18 (95% CI 0.03 to 0.96) and OR 0.09 (95% CI 0.01 to 0.89), respectively).
No variables were associated with a deterioration in the self-care dimension.
After COVID-19, a significantly greater decrease in the median QoL score was observed in the LC group compared with the group without LC (11 vs 4 points decrease, respectively). A change in the health status estimate (before and after COVID-19) is influenced by the following four clusters of LC: fatigue (OR 3.08 (95% CI 1.79 to 5.30)), cardiorespiratory (OR 2.91 (95% CI 1.61 to 5.24)), digestive (OR 3.39 (95% CI 1.68 to 6.82)) and musculo-articular cluster (OR 5.39 (95% CI 2.14 to 13.53)). We also noted a greater decrease in the QoL for women compared with men (OR 2.29 (95% CI 1.19 to 4.39)). People who reported >7 hours of sports per week before their infection would report a higher deterioration of their estimated health compared with people who did not exercise (OR 2.54 (95% CI 1.05 to 6.18)). Patients with an existing cardiovascular disease and locomotor disease reported a smaller deterioration after a COVID-19 infection when compared with those without any cardiac or locomotor disease (OR 0.31 (95% CI 0.11 to 0.86) and OR 0.23 (95% CI 0.06 to 0.90), respectively).
Discussion
Prevalence of LC
In the survey, 47.35% of participants reported at least one symptom lasting over 3 months after initial SARS-CoV-2 infection, which is similar to the prevalence of 47% found in the COVIMPACT study by Sciensano, within the Belgian population.16 Fatigue or exhaustion was the most prevalent LC symptom in both studies.
In the absence of a broad scientific consensus regarding classification and underlying pathological mechanisms of LC, we decided to reduce the myriad COVID-19-related symptoms into seven clinical clusters. This classification is primarily based on symptomatology and body systems, not on the presence of shared underlying pathological mechanisms. We have deliberately grouped neurological and psychiatric symptoms into one cluster since the mechanism(s) explaining these symptoms is unknown. Grouping LC symptoms in this manner is less synoptic than the King’s College classification which only uses three predominant profiles which are not validated and therefore not used to guide clinical practice.17 On the other hand, it created the opportunity to bring clinical nuance and further clarify clinical presentation that could be used to target treatment approaches.
The most prevalent LC cluster was the neuropsychiatric cluster (71.76%). This could be partially explained by pathogen factors related to the Alpha and Delta variants. SARS-CoV-2 variants appeared in Belgium beginning in December 2020. Likely many of the participants responding to the survey were infected by the Alpha or Delta variant in addition to the original strain. Some evidence suggests that these variants induce more neuropsychiatric symptoms than the original strain.18 Furthermore, neurological symptoms are more frequently reported in case of infections with the Alpha and Delta variants, compared with the original strain.17 Concurrent stressors related to the pandemic such as social isolation, stress and uncertainty can also contribute to the symptomatology and drive a higher prevalence of neuropsychiatric LC19 cluster-related symptoms.
LC explanatory factors
We identified obesity, immune and blood disorders and respiratory disease as important risk factors associated with LC and were associated with multiple symptom clusters. Obesity was identified as a risk factor for all LC clusters except the neuropsychiatric and dysautonomia cluster.
Increased BMI or obesity, and airway disease such as asthma are risk factors for LC in other studies as well.13 20 Older age was found to be a risk factor in other studies including a systematic review of the literature that found an age between 40 and 49 years is associated with developing LC.21 Additionally, in another systematic review the 35–69 years age group seemed to be the most affected by LC.8 However, in our study age was only found to be associated with the dysautonomia cluster.
The association between obesity and LC might be expected because of the multisystemic changes associated with obesity.22 LC in general and the neuropsychiatric LC cluster were positively impacted by fitness, specifically, the amount of athletic practice before the COVID-19 infection. This effect could be related to a healthier weight and the presence of good physical and mental health.20 Female gender and mental disease have been found to be risk factors in other studies, but this was not observed in our study.23
The association of the musculo-articular cluster to existing blood or immune system disorders could be related to the production of autoantibodies. In a study by Richter et al24 of patients with COVID-19 receiving intensive care treatment, SARS-CoV-2 infection is associated with the production of a limited profile of autoantibodies specific to skin, muscle, skeletal and cardiac tissues.
We did not identify an effect of medical operational categories on the development of LC. This is in contrast to recent unpublished work in the UK military that identified 20%–30% of those who developed LC were not fully medically deployable. This could be due to the fact that the medical operational category is not necessarily a reflection of the general health of the personnel but rather an administrative state demonstrating the execution of medical screenings, dental evaluation and vaccination status. This may be different in the UK, where predeployment screenings can be more sensitive to LC symptoms.
Quality of life
The EQ-5D-5L questionnaire data identified a greater negative impact in at least one of the LC clusters compared with the group without LC. Only the self-care dimension was not affected by any of the LC clusters.
Our study detected a significant decrease of 11 points on the EQ Visual Analogue Scale in the LC group. In comparison, a decrease of 10.4 points in the QoL score was observed in the COVIMPACT study by Sciensano16 and of 24 points in the KCE LC patient needs study.8
Treatment
The current study did not explore the effect of possible treatments on LC. However, based on literature, military personnel exposed to high volume/intensity physical exercise, or hospitalised during acute illness, may benefit from cardiopulmonary monitoring and appropriate rehabilitation beyond 5 months25 and should be medically assessed before returning to work.26 Chronic fatigue patients could consider experimental hyperbaric oxygen treatment (HBOT), as suggested by a small-scale study in the UK, showing positive results after 10 sessions.27 A HBOT study was started in the Belgian Defence targeting all LC clusters.
Study limitations
Our study does not consider the effect of vaccination against SARS-CoV-2 on LC development, despite its proven effectiveness in other studies28 and the fact vaccination is mandatory in deployable personnel and healthcare workers in Belgium. However, 40% of our sample was infected in October 2020 (at the peak of the second wave in Belgium) and 62% before January 2021, which means the vaccination level in our sample was likely to be low. Recent studies show positive effects of vaccination on the occurrence of prolonged symptoms (beyond 4 weeks) after the second dose.28 In another study with at least two doses, 7 of the 10 most common LC symptoms were reduced by 54%–82%.29
Approximately two-thirds of the participants in our study responded to the online questionnaire 7–8 months after their initial infection. This relatively long interval could create a memory or recall bias regarding the duration of symptoms. In addition, selection bias may have appeared in which those who continued to have symptoms are more likely to reply to the survey than those who do not. Due to the self-reporting of the symptoms and their duration, the prevalence of LC in our population may be an overestimation.
Because of the small numbers of some pre-existing conditions (n<10), we could not analyse their relation with the development of LC.
Conclusion
The prevalence of LC in the Belgian active defence population (civil and military) is similar to the prevalence found in the Belgian population16 and the QoL impact is substantial. Regardless of initial disease severity or SARS-CoV-2 laboratory confirmation, patients with LC would likely benefit from multidisciplinary rehabilitation assessment, especially when addressing the most prevalent symptoms of shortness of breath, fatigue and mood disturbance.30
Data availability statement
No data are available. Because of the small sample size and the specificity of the information included (ie, patients identified with long COVID) in addition to other information (ie, the presence of specific diseases, gender, age, etc), the anonymity of the participants cannot be guaranteed when sharing the dataset.
Ethics statements
Patient consent for publication
Ethics approval
This study was approved by the ethics committee of the Université Catholique de Louvain on 17 August 2021 under the reference 2021/25MAI/242. Participants gave informed consent to participate in the study before taking part.
Acknowledgments
The authors would like to thank Audrey Collée and Ive Van Cauwenbergh for their contribution in the development of the research process.We would like to acknowledge the support of Col (Ret.) Dr.G MM Kerr and CDR Andrew G. Letizia for thoroughly reviewing our manuscript.
References
Supplementary materials
Supplementary Data
This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
Footnotes
Correction notice This article has been corrected since it first published. Acknowledgements have been added.
Contributors HM as the head researcher was responsible for the planning and conduct of the research and for the analysis of the data. SL as a contributor was responsible for the interpretation of the data and for reporting the work. ED as a contributor was responsible for the internal contacts and for the coordination of the internal processes (planning and conduct). NS as a contributor was responsible for the dissemination of the results of the study and for the contextualisation of the final results (reporting). KG as a coordinator and guarantor was responsible for the conception of the work, for recruiting the respondents and the interpretation of the results (planning and conduct). HM, NS, ED, SL and KG have read and approved the final manuscript.
Funding The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors.
Competing interests None declared.
Provenance and peer review Not commissioned; externally peer reviewed.
Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.