Risk Factors Associated with Development and Persistence of Long COVID: A Cross-Sectional Study
This article has been Reviewed by the following groups
Listed in
- Evaluated articles (PREreview)
Abstract
Background Long coronavirus disease (COVID) has been a social concern. Though patient characteristics associated with the development of long COVID are partially known, those associated with its persistence have not been identified. Methods We conducted a cross-sectional questionnaire survey of patients after COVID-19 recovery who visited the National Center for Global Health and Medicine between February 2020 and March 2021. Demographic and clinical data and data regarding the presence and duration of long COVID were obtained. We identified factors associated with the development and persistence of long COVID using multivariate logistic and linear regression analysis, respectively. Results We analyzed 457 of 526 responses (response rate, 86.9%). The median age was 47 years, and 378 patients (84.4%) had mild disease in the acute phase. The number of patients with any symptoms after 6 and 12 months after onset or diagnosis were 120 (26.3%) and 40 (8.8%), respectively. Women were at risk for development of fatigue (odds ratio [OR]: 2.03, 95% confidence interval [CI]: 1.31-3.14), dysosmia (OR: 1.91, 95% CI: 1.24-2.93), dysgeusia (OR: 1.56, 95% CI: 1.02-2.39), and hair loss (OR: 3.00, 95% CI: 1.77-5.09) and for persistence of any symptoms (coefficient: 38.0, 95% CI: 13.3-62.8). Younger age and low body mass index were risk factors for developing dysosmia (OR: 0.96, 95% CI: 0.94-0.98 and OR: 0.94, 95% CI: 0.89-0.99, respectively) and dysgeusia (OR: 0.98, 95% CI: 0.96-1.00 and OR: 0.93, 95% CI: 0.88-0.98, respectively). Conclusion We identified risk factors for the development and persistence of long COVID. Many patients suffer from long-term residual symptoms, even in mild cases.
Article activity feed
-
This Zenodo record is a permanently preserved version of a PREreview. You can view the complete PREreview at https://prereview.org/reviews/6812995.
In this manuscript, the authors describe a study aiming to determine risk factors associated with development and persistence of long COVID-19 in a sample of 457 patients who had been hospitalized evaluated for COVID-19 in a single center. They performed an analytical cross-sectional study involving an online survey sent to patients who had been discharged, with a high response rate. This study is interesting and has strengths like the validation of the questionnaire in a pilot study, high response rate, adequate descriptions of approval by an ethics committee, and adequate statistical analyses. Nonetheless, I have identified several potential biases of this study most of which may …
This Zenodo record is a permanently preserved version of a PREreview. You can view the complete PREreview at https://prereview.org/reviews/6812995.
In this manuscript, the authors describe a study aiming to determine risk factors associated with development and persistence of long COVID-19 in a sample of 457 patients who had been hospitalized evaluated for COVID-19 in a single center. They performed an analytical cross-sectional study involving an online survey sent to patients who had been discharged, with a high response rate. This study is interesting and has strengths like the validation of the questionnaire in a pilot study, high response rate, adequate descriptions of approval by an ethics committee, and adequate statistical analyses. Nonetheless, I have identified several potential biases of this study most of which may compromise the findings and were not adequately discussed by authors (inadequate study design, selection bias, volunteer bias, ascertainment bias, recall bias, type I error). These are my individual comments:
Major comments:
1. The last two sentences of the introduction intend to justify and outline the study objective by mentioning that "it is important to identify risk factors to properly understand and prevent long COVID" and that the authors intended to explore "risk factors for the development and persistence of long COVID in a cohort of patients recovering from COVID-19". There are important inconsistencies since 1) risk factors cannot strictly be identified in a cross-sectional study since this design is not suitable to evaluate directionality of exposure-outcome, 2) the way these sentences are written (i.e. the authors mention the words 'development', 'persistence', 'cohort') would suggest that the most suitable study design to answer the research question would have been a longitudinal study (i.e. cohort study).
2. It would be more adequate to refer to "variables associated with" rather than "risk factors" throughout the manuscript since the authors mentioned having performed a cross-sectional study. Also, instead of 'development' and 'persistence', it could be more accurate to refer to 'self-reported occurrence and self-reported duration of symptoms'.
3. There are several expressions in this manuscript that suggest causality, which is not supported by this study. The authors need to modify the manuscript to avoid all expressions suggesting causality.
4. There are already systematic reviews on this topic which have been published. For instance, https://www.medrxiv.org/content/10.1101/2021.10.17.21265123v1 and https://www.nature.com/articles/s41598-021-95565-8. Please mention in your abstract and introduction how your study could fill any gaps not already covered in systematic reviews.
5. The exposure and outcomes have not been described in the methods section. Please review STROBE recommendations.
6. Line 85-86: How much time passed from recovery to sending the questionnaire to potential participants? If this time was different for all patients, this is another important bias of this study that could very likely alter the results, which would need to be included in the limitations of the study.
7. The study is subject to a very important selection bias and volunteer bias since only those patients who wanted to donate plasma were eligible. These patients are very likely systematically different from those who do not present (or do not want) to donate plasma. Both selection and volunteer bias need to be disclosed in the limitations of the study.
8. There are no other descriptions of inclusion, exclusion, and elimination criteria. Please describe these in your methods and use a flow diagram to report the number of patients meeting inclusion, exclusion, and elimination criteria, as recommended by STROBE.
9. Even though the response rate was an acceptable one for an online survey, non-response bias still needs to be recognized and contextualized to as a potential limitation of the study.
10. Ascertainment bias due to response bias applies to this study as well. This is particularly critical for clinical variables. The authors should have attempted to obtain clinical variables (i.e., diagnosis of pneumonia, ECMO, IMV, drugs received, etc.) from patients' medical records, not by asking these to patients. If possible, please obtain these data from medical records. Otherwise, recognize this bias of your study and do mention in the limitations that clinical variables could have been obtained from medical records to avoid this bias.
11. In the survey, the option "I don't know" is given for COVID-19 severity variables. The authors did not describe how many participants reported these options. Participants were not given the option to respond "I am unsure" or "I don't remember" for all symptoms recorded in the survey and were asked to report a specific number of days. This is another important limitation of the questionnaire since it would be almost impossible to recall the specific time elapsed from infection to the last day the symptom was present, especially when the median time from acute COVID-19 to responding the questionnaire was 248.5 days.
12. Recall bias applies to this study, although it has already been recognized by the authors.
13. The classification of disease severity is not the internationally recognized classification. See page 24 of this WHO guideline https://apps.who.int/iris/rest/bitstreams/1394399/retrieve. I would suggest classifying patients by obtaining clinical data from their medical records. Otherwise, clarify in your manuscript methods and limitations that you have used non-standard international definitions of disease severity.
14. A sample size was seemingly not calculated. Please mention this as a limitation of the study.
15. Azithromycin, ivermectin, hydroxychloroquine, and azithromycin were identified as antivirals by the authors. However, these drugs do not fall in the antivirals category.
16. There is a high risk of type I error since several regression analyses were performed on different outcomes which are not mutually exclusive. This needs to be recognized as a limitation of the study.
17. There are several and important reporting deficiencies of this paper which may not allow it to be reproducible or properly evaluated by readers. Thus, it is of utmost importance that the authors review the full STROBE explanation and elaboration document (https://doi.org/10.1371/journal.pmed.0040297) to report the manuscript accordingly. Please provide the STROBE checklist for a cross-sectional study for the re-submission of your revised manuscript as supplementary material for peer-review only. This checklist can be easily built in https://www.goodreports.org/
Minor Comments:
• Please change the title to avoid mentioning "risk factors". Also, since this was not a longitudinal study, "development" and "persistence" may need to be changed for more suitable words. Please change throughout the manuscript, too.
• The correct term to refer to "long COVID" is "post-acute COVID-19 syndrome". https://www.ncbi.nlm.nih.gov/mesh/?term=C000711409 Please correct throughout the manuscript.
• See other official abbreviations for the virus and the disease by the WHO and correct your manuscript accordingly: https://www.who.int/emergencies/diseases/novel-coronavirus-2019/technical-guidance/naming-the-coronavirus-disease-(covid-2019)-and-the-virus-that-causes-it. For example, line 67.
• Line 43: I suggest changing this for "We conducted an analytical cross-sectional study, through a web-based questionnaire, on patients who recovered from acute COVID-19 after receiving care in the National Center …"
• Line 45: "data" is mentioned twice.
• Line 46: "development" and "persistence" may not be adequate for the previously outlined reasons. Same for lines 60-61 and 63-64 and please avoid "risk factors".
• Line 67-68: It is not useful to mention the number of deaths up to a single point in time since this will rapidly be outdated. Please start your introduction in a different way to avoid these statements that may make your paper seem outdated when it is published.
• Lines 68-70: Please give more context for these numbers. Out of all those patients, how many had mild, moderate, severe, or critical COVID-19 during the acute phase?
• Line 83: Please change "cross-sectional survey" for "analytical cross-sectional study". Also, add the word "web-based" or "online" before "questionnaire".
• Line 84-85: Does this mean that two different surveys were sent to participants or only one survey with different reminders to complete the same survey? Please be more specific to avoid any confusions.
• Line 84: Was any online platform used, or how was the questionnaire delivered (i.e., what file format)?
• Line 98: A systematic literature review would have involved conducting a systematic review per se. Please be more specific on how this systematic review was conducted. If this is not what you intended to say, please change wording, perhaps only mention "literature review". If you indeed performed a systematic search of studies, please describe your search strategy, records retrieved, eligibility criteria, etc. in the supplementary materials section.
• Line 101-102: Were these 13 patients who participated in the pilot study included for analyses of the main study? Please do mention this in the manuscript.
• How was the diagnosis of COVID-19 confirmed in all patients? Please describe this in your methods.
• Depression is a medical diagnosis, not a symptom. Was the questionnaire applied in Japanese? Maybe there are no distinctions between the different symptoms of depression and the diagnosis in Japanese?
• Sputum could be a better translation than phlegm if you are referring to expectoration of mucus.
• Dysosmia refers to any alterations of smell, whereas anosmia refers to loss of smell. Please correct as necessary since the questionnaire in the appendix mentions "Dysosmia (loss of smell)". Similarly, please review if you have used an adequate translation for dysgeusia or not.
• Statistical analysis: What is the justification for using median and IQR instead of mean and standard deviation? It would seem as if data could have been presented as mean with standard deviation.
• Lines 126-130 should be in a different section of your methods, not the statistical analysis section.
• Paragraph starting in line 132: There is important information missing for the models.
i. What statistical assumptions were verified to create the models?
ii. How ere variables selected (i.e., stepwise, Enter method?).
iii. Describe that different models were created for all individual symptoms.
iv. Why were only some symptoms included?
• Line 135: Please fully describe all the individual confounding variables that you selected. It is desirable to include a reference or explanation of why those confounding variables were selected.
• Lines 138-143 are unclear. This paragraph needs to be improved since it is difficult to understand.
• Mention if this was a binary logistic regression.
• The definition of statistical significance in your study could be placed at the end of this paragraph to avoid interrupting the flow of ideas.
• Were age and BMI managed as continuous quantitative variables? Please clarify this in statistical methods.
• Please change figures in appendix 4a, 4b to tables since they are difficult to interpret. Include the values for OR, 95% CI, and the regression coefficient.
-
