Age, Sex and Previous Comorbidities as Risk Factors Not Associated with SARS-CoV-2 Infection for Long COVID-19: A Systematic Review and Meta-Analysis

1Long COVID-19 is a term used for defining the persistence of signs and symptoms in people who recovered from an acute Severe Acute Respiratory Syndrome Coronavirus 2 (SARS-CoV-2) infection. Long COVID-19 is defined by the World Health Organization (WHO) as: “post-COVID-19 condition, occurs in individuals with a history of probable or confirmed SARS-CoV-2 infection, usually 3 months from the onset, with symptoms that last for at least 2 months and cannot be explained by an alternative diagnosis.” Several meta-analyses investigating the prevalence of post-COVID-19 symptoms have been published, concluding that around 30–50% of subjects who recover from a SARS-CoV-2 infection develop persistent symptoms lasting up to one year. A recent meta-analysis concluded that two years after the initial spread of coronavirus disease 2019 (COVID-19), up to 42% of infected patients experienced long COVID-19 symptoms.

2Different narrative reviews have mentioned prognostic aspects, but no comprehensive synthesis has been provided so far. Identification of potential risk factors associated with post-COVID-19 syndrome is important since identifying individuals at higher risk can guide management healthcare plans for these patients and reorganize healthcare accordingly. Iqbal et al. tried to pool data, but these authors were only able to pool prevalence data of post-COVID-19 symptomatology, but not risk factors. All these narrative reviews have suggested that female sex, old age, higher number of comorbidities, higher viral load, and greater number of COVID-19 onset symptoms can be potential risk factors for long COVID-19. However, no systematic search or assessment of methodological quality was conducted in these reviews. Two meta-analyses have recently focused on risk factors of long COVID-19. Maglietta et al. identified that female sex was a risk factor for long COVID-19 symptoms, whereas a more severe condition at the acute phase was associated just with long COVID-19 respiratory symptoms. Thompson et al. found that old age, female sex, white ethnicity, poor pre-pandemic health, obesity, and asthma can predict long COVID-19 symptoms. However, this analysis included just studies from the United Kingdom, and used the definition of long COVID-19 proposed by the National Institute for Health Care and Excellence (NICE).

3Accordingly, current evidence on risk factors associated with post-COVID-19 condition is heterogeneous. Risk factors can be classified as pre-infection (e.g., age, sex, previous comorbidities, and previous health status) and infection-associated (e.g., disease severity, symptoms at onset, viral load, hospitalization stay, and intensive care unit admission) factors. The current systematic review and meta-analysis aimed to identify factors not directly associated with acute SARS-CoV-2 infection (i.e., pre-infection factors) such as age, sex, and previous medical comorbidities, which may predict the development of long COVID-19 symptomatology.

4This systematic literature review and meta-analysis aiming to identify the association of age, sex, and comorbidities as predictive factors for development of long COVID-19 was conducted following the Preferred Reporting Items for Systematic reviews and Meta-Analyses (PRISMA) statement of 2020. We also followed specific criteria recommended by Riley et al. to systematic reviews and the meta-analysis of prognostic factor studies. The review study was prospectively registered in the Open Science Framework (OSF) database (https://osf.io/79pdg).

5Two different authors performed an electronic search for articles published up to 15 September 2022 MEDLINE, CINAHL, PubMed, EMBASE, and Web of Science databases, as well as on preprint servers medRxiv and bioRxiv, using the following search terms: “long COVID-19” OR “post-acute COVID” OR “post-COVID-19 condition” OR “long hauler” AND “age” OR “sex” OR “medical comorbidities” OR “transplant” OR “obesity” OR “diabetes” OR “hypertension” OR “pulmonary disease” OR “asthma” OR “chronic obstructive pulmonary disease”. The search was focused on the medical comorbidities likely associated with a more severe COVID-19 condition. Combinations of these search terms using Boolean operators are outlined in Table 1.

6The Population, Intervention, Comparison. and Outcome (PICO) principle was used to describe the inclusion and exclusion criteria:

7Population: Adults (>18 years) infected by SARS-CoV-2 and diagnosed with real-time reverse transcription-polymerase chain reaction (RT-PCR) assay. Subjects could have been hospitalized or not by SARS-CoV-2 acute infection.

8Intervention: Not applicable.

9Comparison: People infected by SARS-CoV-2 who did not develop long COVID-19 symptoms.

10Outcome: Collection of long COVID-19 symptoms developed after an acute SARS-CoV-2 infection by personal, telephone, or electronic interview. We defined post-COVID-19 condition according to Soriano et al., where “post-COVID-19 condition occurs in individuals with positive history of probable or confirmed SARS-CoV-2 infection, usually 3 months from onset of COVID-19, with symptoms that last for at least 2 months and cannot be explained by alternative diagnosis.” We considered any long COVID-19 symptom appearing after the infection, e.g., fatigue, dyspnea, pain, brain fog, memory loss, skin rashes, palpitations, cough, and sleep problems. Results should be reported as odds ratio (OR), hazards ratio (HR), or mean incidence of the symptoms.

11This review included observational cohort, cross-sectional, and case-control studies whether presence of risk factors for development of symptoms appearing after an acute SARS-CoV-2 infection were investigated in COVID-19 survivors, either hospitalized or non-hospitalized. The current review was limited to human studies and English language papers. Editorials, opinion, and correspondence articles were excluded.

12Two authors screened title and abstract of publications obtained from database search and removed duplicates. Full text of eligible articles was retrieved and analyzed. The following data were extracted from each study: authors, country, design, sample size, age range, assessment of symptoms, long COVID-19 symptoms, and effect (measure) of risk factor studied. Discrepancies between reviewers in any part of the screening and data extraction process were resolved by a third author.

13The Quality in Prognosis Studies (QUIPSs) tool was used to determine the risk of bias (RoB) of the studies. The QUIPS consists of six domains such as study participation, study attrition, prognostic factor measurement, outcome measurement, adjustment for other prognostic factors, and statistical analysis. RoB was initially evaluated by two authors. If there is disagreement, a third researcher arbitrated a consensus decision. Risk of bias was scored as low, moderate, or high as follows: 1 if all domains are classified as having low RoB, or just one as moderate RoB, the paper was classified as low RoB (green); 2 if one or more domains are classified as having high RoB; or ≥3 if all domains have moderate RoB, the paper was classified as high RoB (red). All papers in between were classified as having moderate RoB (yellow).

14We conducted a qualitative synthesis of data for those risk factors where the heterogeneity of results did not permit to perform a meta-analysis. We only included articles in the meta-analysis that followed the Soriano et al. definition of post-COVID-19 condition, hence meta-analysis was possible for age and sex.

15To synthesize the association between age and sex with post-COVID-19 condition, random-effects meta-analyses were performed using MetaXL software ( https://www.epigear.com/index_files/metaxl.html) to estimate weighted mean differences (for age) and pooled odds ratios (ORs) with 95% confidence intervals (CIs) for sex and age above 60 years (old adults). A p-value < 0.05 was considered statistically significant. Given the heterogeneity expected, a random-effects model was employed. Measures of heterogeneity such as the I square statistics and Cochran’s Q test statistic and p-value are also reported. When each age group was reported using median and interquartile range values, the method described by Wan was used for transformation in mean and standard deviation.

16The electronic search allowed to initially identify 1978 titles for screening. After removing duplicates (n = 154) and papers not directly related to risk factors (n = 1352), 472 studies remained for abstract examination. Four hundred and twenty-five (n = 425) were excluded after reading the abstract, thus leading to a total of 47 text articles for eligibility (Figure 1). Nine articles were excluded because there were no comparisons between subgroups (n = 2), inappropriate methodology (n = 2), data not extractable (n = 1), unrelated to association of risk factors (n = 1), and type of literature commentary, case reports, and case series (n = 3). A total number of 37 peer-reviewed studies and one pre-print study were finally included. All papers could be included in qualitative analysis, whereas seven of these could also be pooled in the meta-analysis.

17A total of 18 articles, including 819,884 COVID-19 survivors analyzed age as a risk factor for developing long COVID-19 symptoms (Table 2). Four articles used percentages, five used means, seven OR, one adjusted OR (aOR), and one adjusted hazard ratio (aHR). Eight articles included population samples aged ≥50 years old, eight individuals aged between 40 and 49 years, and one a population between 18 and 64 years. Two studies included children aged 10–12 years, but data from these age groups were not considered in the main analyses.

18Overall, most articles observed that old age was associated with long COVID-19 symptoms. Contrastingly, Peghin et al. did not find an association between age and long COVID-19 symptoms. Subramanian et al. stated that adults aged >70 years displayed lower risk of developing long COVID-19 symptoms than those aged 30–39 years.

19Three articles (n = 30,371 patients) were included in the meta-analysis based on their similar study design, study outcomes, and long COVID-19 definition. We grouped individuals aged over 60 years old, since this age group is considered to be at higher risk of severe COVID-19. The meta-analysis did not reveal a significant association between old age and long COVID-19 symptomatology (OR 0.86, 95% CI 0.73 to 1.03, Q = 3.54, p = 0.17, I2: 44%, Figure 2). Another three articles reporting data as mean (with their standard deviation) or median (interquartile range) were also pooled. We pooled these data through a random effects model, resulting in a non-significant weighted mean difference (WMD) of −0.25 (95% CI −3.78 to 3.27, Q = 3.27, p = 0.19, I2: 39%, Figure 3).

20A total of 16 articles including 504,044 COVID-19 survivors were used in the analysis of sex as a risk factor for developing long COVID-19 symptomatology (Table 3). Data were presented as OR, aOR, HR, aHR, and percentage. Seven articles used OR, two used aOR, another two used both OR and aOR, while three articles used percentage, one article used both percentage and OR, and one article used both HR and aHR.

21Fourteen articles observed that female sex (n = 276,953) was associated with higher risk of long COVID-19, whilst two articles (n = 475) reported that female sex was not associated with higher risk of long COVID-19.

22Seven articles (n = 386,234 COVID-19 patients who recovered from acute SARS-CoV-2 infection) were included in the meta-analysis due to their similarities in study design, definition of long COVID-19, as well as similarities in data presentation. The meta-analysis revealed that female sex was significantly associated with nearly 50% higher risk (OR 1.48, 95% CI 1.17 to 1.86, Q = 17.2, p = 0.01, I2: 65%, Figure 4) of long COVID-19 symptomatology.

23A total of 12 articles with 677,045 COVID-19 survivors were analyzed for association between long COVID-19 and comorbidities (Table 4). Four comorbidities were included: pulmonary disease (n = 4), diabetes (n = 1), obesity (n = 6), and organ transplantation (n = 1). Data were presented as means, medians, percentages, odds ratio (OR), and incident rate ratio (IRR). One study used mean, one used both median and percentage, three used percentage only, five used OR, and two used both OR and IRR.

24Three articles on pulmonary disease revealed an association between asthma and longer symptom duration among patients recovering from COVID-19. However, both asthma and chronic pulmonary disease were not associated with long COVID-19 in one study. For diabetes, no difference was found in the number of long COVID-19 symptoms among diabetic and non-diabetic patients. For obesity, all six articles noted that this metabolic disease was associated with worse health due to increased number of long COVID-19 symptoms, longer persistence of symptoms, more presence of pathological pulmonary limitations, and metabolic abnormalities. Meanwhile, one study on kidney transplant patients revealed that patients have higher susceptibility to developing long COVID-19 symptoms, although this did not affect mortality rate.

25From out of 18 papers evaluating age as risk factor, three were classified as low risk of bias (green), five as moderate risk of bias (yellow), and the remaining ten as high risk of bias (red). Figure 5 visually graphs that the most frequent risk of bias was adjustment for other prognostic factor (i.e., if important potential confounding factors were appropriately accounted for), which was properly performed in just one study.

26On the other hand, from 16 papers evaluating sex as a risk factor, four studies were classified as low risk of bias (green), five as moderate risk of bias (yellow), and the remaining seven as high risk of bias (red). Figure 6 visually graphs that the most frequent risk of bias in this group of studies were adjustment for other prognostic factors and study attrition (i.e., the representativeness of the participants with follow-up data with respect to those originally enrolled in the study, selection bias).

27Lastly, from 12 papers evaluating previous medical comorbidities as a risk factor, four were classified as low risk of bias (green), two as moderate risk of bias (yellow), and the remaining six as high risk of bias (red). Figure 7 visually graphs that the most frequent risk of bias in this group of studies was concerned prognostic factor measurement (i.e., if the prognostic factors were measured in a similar way for all the participants).

28This systematic review and meta-analysis explored the association of long COVID-19 with risk factors not directly related to an acute SARS-CoV-2 infection (i.e., pre-infection factors), including age, sex, or previous comorbidities. The results support that female sex may be a predictor of long COVID-19 while old age was reported to be associated with long COVID-19 in single studies; however, the pooled evidence was not significant. Finally, prior medical comorbidities can also be potential predictors of long COVID-19 symptoms. These results should be considered with caution because most studies exhibited moderate to high risk of bias.

29Old age is an important risk factor of poor outcomes in COVID-19 hospitalization; however, the impact of age on long COVID-19 is controversial. Old age is associated with higher risk of long COVID-19 symptomatology in single studies and in two previous reviews, but not in the meta-analysis by Maglietta et al.. Results from our qualitative analysis suggest that older adults can develop more long COVID-19 symptoms than younger adults; however, this assumption was not supported when pooling data into a meta-analysis. We conducted two meta-analyses, the first one categorizing those adults older than 60 years (Figure 2), and a second one considering age as a continuous variable (Figure 3); neither analysis revealed an association between old age and risk of developing long COVID-19. Nevertheless, the number of studies pooled in our analyses of age was notably limited (n = 3). Our data are consistent with the meta-analysis of Maglietta et al. but disagree with Thompson et al.. Several differences can explain the discrepancy with Thompson et al.. It is possible that the use of a different definition of long COVID-19 by these authors can lead to inconclusive comparisons of results. In addition, Thompson et al. did not pool data of age and long COVID-19 into a meta-analysis, but only calculated regression of proportions of subjects at each age group developing long COVID-19 symptoms. The significance of old age as a risk factor for long COVID-19 development requires further investigation. In fact, just three out of eighteen papers (16%) analyzing age as prognostic factor showed low risk of bias. The most significant bias of these studies was the proper control of other cofounding factors observed in older people, i.e., higher presence of medical comorbidities, or longer hospitalization stay, which can also be associated with long COVID-19.

30Sex is another important risk factor which has been studied in relation to COVID-19 and long COVID-19. Evidence supports that men and women exhibit the same probability of being infected by SARS-CoV-2; however, males are at a higher risk of worse outcomes and death than females during the acute phase of infection. Results from our systematic review and meta-analysis support that female sex may be associated with higher risk of developing long COVID-19 (OR 1.48, 95% CI 1.17 to 1.86). Our results are similar to those previously observed by Maglietta et al., who also reported that female sex was associated with long COVID-19 symptoms (OR1.52, 95% CI 1.27–1.82), and with results (OR1.60, 95% CI 1.23–2.07) previously reported by Thompson et al.. Based on available data, females are more vulnerable to develop long COVID-19 than males. Hence, considering sex differences in diagnosis, prevention and treatment are necessary, and fundamental steps towards precision medicine in COVID-19. Biological (i.e., hormones and immune responses), and sociocultural (i.e., sanitary-related behaviors, psychological stress, and inactivity) aspects play a significant role in creating sex-differences in long COVID-19 symptoms, although mechanisms behind increased risk of long COVID-19 in females remain unknown and warrant investigation.

31Such as with old age, the presence of prior medical comorbidities (e.g., hypertension, obesity, diabetes, chronic kidney disease, cerebrovascular disease, chronic obstructive pulmonary disease, or cardiovascular disease) is known to induce a more severe COVID-19 disease progression. A potential reason is that such comorbidities can contribute to degradation of angiotensin-converting enzyme 2 (ACE2). Since the SARS-CoV-2 virus uses this receptor as entry pathway in host cells, higher degradation of ACE2 could lead to a long-lasting inflammatory cytokine storm, oxidative stress, and hemostasis activation, which are all hallmarks of severe/critical COVID-19 illness. Nevertheless, this hypothesis is not yet supported by the literature.

32The current qualitative analysis suggests that prior comorbidities may contribute the risk of developing long COVID-19. Among different comorbidities, obesity seems to be associated; however, this assumption should be considered with caution at this stage, since potential cofounding factors, particularly those related to hospitalization (obese patients have more severe COVID-19 disease and higher hospitalization rates than non-obese patients), were not properly controlled in these studies. Moreover, the association of long COVID-19 with other medical comorbidities such as diabetes or transplants was only investigated in one prior study.

33The results of this systematic review and meta-analysis should be considered according to potential strengths and limitations. Among the strengths, we conducted a systematic search of all the currently available evidence on factor not related to an acute SARS-CoV-2 infection but associated with higher risk of developing long COVID-19. This led to identification of thirty-eight studies. Second, this is the first time that several medical comorbidities have been systematically investigated as risk factors of long COVID-19.

34One of the limitations is the lack of a consistent definition of long COVID-19 in available literatures. We included all identified studies within the qualitative analysis, but only those using the definition by Soriano et al. of long COVID-19 were included in the meta-analyses. This assumption led to a small number of studies in the meta-analyses. Future studies using a more consistent definition of long COVID-19 are needed for improved quantification of the results. Another limitation is the lack of differentiation of risk factors between hospitalized and non-hospitalized patients. Similarly, no study investigating risk factors considered the SARS-CoV-2 variants of concern. Therefore, studies identifying long COVID-19 risk factors not directly associated with SARS-CoV-2 infection differentiating between hospitalized and non-hospitalized patients, and among different SARS-CoV-2 variants of concern are now needed. Finally, it should be considered that this systematic review and meta-analysis only investigated risk factors not associated with an acute SARS-CoV-2 infection. Other potential SARS-CoV-2-associated factors, such as severity of disease during the acute phase of infection or the number of COVID-19-associated onset symptoms have also been preliminarily identified as risk factors associated with long COVID-19 symptoms, particularly with respiratory symptoms. Similarly, it is possible that some long COVID-19 symptoms can also be related to hospitalization factors which were also not investigated in this review.

35The current review demonstrates that female sex and previous medical comorbidities may be predisposing factors for the development of long COVID-19 symptomatology. The current literature does not conclusively confirm that old age would significantly influence long COVID-19 risk. These results should be considered with caution due to moderate to high risk of bias in most published studies. These findings highlight the need for further research with improved control of confounding factors and use of a consistent and validated definition of long COVID-19.