ABSTRACT
Background To inform vaccine prioritization guidance in Canada, we reviewed evidence on the magnitude of association between risk factors and severe outcomes of COVID-19.
Methods We updated our existing review by searching online databases and websites for cohort studies providing multivariate adjusted associations. One author screened studies and extracted data. Two authors estimated the magnitude of association between exposures and outcomes as little-to-no (odds, risk, or hazard ratio <2.0, or >0.50 for reduction), large (2.0-3.9, or 0.50-0.26 for reduction), or very large (≥4.0, or ≤0.25 for reduction), and rated the evidence certainty using GRADE.
Results Of 11,734 unique records we included 135 reports. There is probably (moderate certainty) at least a large increase in mortality from COVID-19 among people aged 60-69 vs. <60 years (11 studies, n=517,217), with ≥2 vs. no comorbidities (4 studies, n=189,608), and for people with (vs. without): Down syndrome (1 study, n>8 million), type 1 and 2 diabetes (1 study, n>8 million), end-stage kidney disease (1 study, n>8 million), epilepsy (1 study, n>8 million), motor neuron disease, multiple sclerosis, myasthenia gravis, or Huntington’s disease (as a grouping; 1 study, n>8 million). The magnitude of association with mortality is probably very large for Down syndrome and may (low certainty) be very large for age 60-69 years, and diabetes. There is probably little-to-no increase in severe outcomes with several cardiovascular and respiratory conditions, and for adult males vs. females.
Interpretation Future research should focus on risk factors where evidence is limited in quantity (i.e., no or few small studies) or quality. This may include social factors, some rare conditions, the pediatric population, combinations of comorbidities that increase risk, and long-term outcomes.
Systematic review registration PROSPERO #CRD42021230185.
INTRODUCTION
The novel coronavirus disease 2019 (COVID-19) became a worldwide public health concern in early 2020 [1]. About 20% of affected people will experience severe disease (e.g., requiring hospitalization), and some may have long-lasting medical complications [2]. In late 2020, Health Canada approved two vaccines for the prevention of COVID-19 (Pfizer-BioNTech COVID-19 Vaccine and Moderna COVID-19 Vaccine); this has been followed by the approval of others and continued review of emerging vaccine candidates [3]. As anticipated, vaccine distribution has been constrained by logistical challenges and limited supply [4-6], necessitating prioritization of populations for vaccination [5, 6].
The National Advisory Committee on Immunization (NACI) developed preliminary vaccine prioritization guidance in late 2020 “for the efficient, effective and equitable allocation of safe, efficacious severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) vaccines in the context of staggered arrival of vaccines” [5, 6]. In addition to reviewing evidence on the burden of illness and vaccine characteristics, NACI used its Ethics, Equity, Feasibility and Acceptability (EEFA) framework [7] and evidence-informed Equity Matrix [8] to achieve the goal of their prioritization guidance. The Equity Matrix considers biological and social risk factors (termed P2ROGRESS And Other Factors, informed by the PROGRESS-Plus model [9]) that may result in inequitable health outcomes across a population. The findings of a rapid review of risk factors for severe outcomes of COVID-19 conducted by our research group [10] helped populate the Equity Matrix, and informed NACI’s guidance. This initial review found increasing age (>60 years and particularly >70 years) to be among the strongest risk factors for severe outcomes of COVID-19 [10]. Since this time, numerous relevant primary studies have been published. To inform subsequent [11, 12] and future guidance statements, we performed an updated rapid review to determine the magnitude of association between P2ROGRESS And Other Factors and severe outcomes of COVID-19.
METHODS
Review Approach
Detailed methods are in our pre-defined protocol (PROSPERO #CRD42021230185), and described briefly herein. We followed traditional systematic review methods [13] with modifications to accommodate the expedited timeline needed to inform NACI’s guidance. Single experienced reviewers selected studies and extracted data, with piloting to maintain rigour. We restricted the scope to large, good quality studies with high relevance to Canada.
The eligibility criteria were informed by stakeholders at NACI and the Public Health Agency of Canada; these individuals were not involved in review conduct. This review is reported according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement [14].
Eligibility criteria
Table 1 details our eligibility criteria. We included published (or accepted for publication) cohort studies reporting multivariate adjusted associations between P2ROGRESS And Other Factors and severe outcomes of COVID-19 among: the general population; people with COVID-19; people hospitalized with COVID-19; people with severe COVID-19 (as defined by study authors); children in any of the aforementioned populations. Canadian reports of any design with any analysis type were eligible. For social risk factors, we included only Canadian reports because these were considered as most relevant by NACI.
We applied criteria at the selection stage to include only the most informative, good quality studies. Studies needed to take place in Organization for Economic Cooperation and Development (OECD) countries and include ≥1,000 participants (i.e., sufficiently powered for multivariate adjustment). Adjustment by age and sex was required in all studies, except Canadian reports. Findings for age and/or sex required adjustment for at least one pre-existing condition, and findings for most pre-existing conditions required adjustment for body mass index (BMI). After screening, we loosened our selection criteria for studies among children, requiring a sample size of ≥300 participants and adjustment by age and sex (but not BMI), as limited evidence was available for this population.
When no primary studies were located on a risk factor of high relevance to NACI, we included systematic reviews with broader selection criteria.
Literature search and study selection
Full details of the literature search are in Appendix 1. A research librarian searched online databases (Ovid Medline® ALL 1946-, Epistemonikos COVID-19 in L-OVE Platform for individual predictors of outcomes; 2-3 December 2020) and websites suggested by NACI (6 January - 3 February, 2021). We updated the Medline and website searches in April 2021 for the pediatric population, immune compromise, and autoimmune conditions, to further inform NACI’s guidance for risk factors. We exported records to Endnote X9 (Clarivate Analytics, Philadelphia, PA), removed duplicates and pre-prints, then uploaded the records to Covidence systematic review platform (https://www.covidence.org/). After piloting, a single experienced reviewer screened each record by title and abstract, then by full text. A second reviewer was consulted as needed.
Following screening, gaps in evidence (i.e., no study meeting our criteria) were apparent for asplenia/splenic dysfunction, cystic fibrosis, sickle cell disease, β-thalassemia, and learning disability (potentially important based on inclusion in priority lists from other jurisdictions [15-18]). To locate evidence for these, we scanned all studies excluded based on sample size (<1,000 participants), preprints, and systematic reviews that we had located during our scoping exercises (September, 2020) using Smart Searches in Endnote. On 18 February 2021, we conducted targeted searches in Ovid Medline® ALL 1946- for studies and reviews on these risk factors.
Data extraction
After piloting, a single reviewer extracted data from each study into Excel (v. 2016; Microsoft Corporation, Redmond, WA): study and population characteristics, exposures, covariates, outcomes and their definitions, findings for multivariate associations. We held regular team meetings to troubleshoot and ensure consistency. Because we attempted to include only good quality studies, we did not formally assess their quality, but recorded potential risk of bias concerns (Appendix 2) for consideration during assessment of the certainty of evidence.
Synthesis and drawing conclusions
Appendix 2 details our approach to synthesis and drawing conclusions. Results from the Canadian studies, none of which met the criteria of having sufficiently adjusted analysis, were not included in the main synthesis and are reported separately. We did not pool findings statistically for any outcome, due to large heterogeneity in comparisons and measures of association, and at least some overlap in populations across studies. Instead, two reviewers reached consensus on a best estimate of the magnitude of association for each outcome (based on adjusted odds [aOR], risk, or hazard ratios): little-to-no (<2.0, or >0.5 for reduction), large (2.0-3.9, or 0.5-0.26 for reduction), or very large (≥4.0, or ≤0.25 for reduction). Two reviewers assessed the certainty of evidence for each exposure-outcome association, informed by elements of the Grading of Recommendations, Assessment, Development and Evaluations approach [19].
Ethics approval
Ethics approval was not required.
RESULTS
Included studies
Of 11,734 unique records identified, we included 135 reports (Figure 1) [20-154]. Full text exclusions are listed at [https://doi.org/10.7939/DVN/VDELOG]. We included one systematic review of case series for β-thalassemia [119]. No studies or reviews were located with relevant findings for asplenia/splenic dysfunction, cystic fibrosis, sickle cell disease, or learning disability. No study reported on long-term outcomes. Table 2 shows a summary of the characteristics of the studies presenting multivariate adjusted findings that were included in the main synthesis (n=124) [20-121, 131-152]; full details by study, including 11 Canadian reports [122-130, 153, 154], are in Appendix 3. The studies contributing multivariate associations originated primarily from the United States (n=55), the United Kingdom (n=15), and Mexico (n=13), and included more than 76 million participants (median 5,327; range 34 to 61,414,470).
Associations between risk factors and outcomes
Based on stakeholder input, we considered people with confirmed COVID-19 to be the population of highest relevance, thus have focused on these in our presentation of findings. When no data were available from the COVID-19 population for a risk factor-outcome comparison, we relied on findings from the general population. Table 3 summarizes the risk factors showing large or very large multivariate-adjusted associations with severe outcomes. Appendix 2 includes findings for all risk factors and populations investigated in the review. Data supporting the conclusions are available at [https://doi.org/10.7939/DVN/VDELOG].
The certainty of evidence was moderate for at least a large increase in mortality from COVID-19 for people aged 60-69 years versus <60 years (11 studies, n=517,217) [24, 48, 63, 68, 71, 84, 92, 93, 98, 101, 106, 113], people with two or more versus no comorbidities (4 studies, n=189,608) [48, 59, 101, 105], and for people with (versus without): Down syndrome (1 study, n>8 million) [35], type 1 and 2 diabetes (1 study, n>8 million) [35], end-stage kidney disease (1 study, n>8 million) [35]; epilepsy (1 study, n>8 million) [35]; motor neuron disease, multiple sclerosis, myasthenia gravis, or Huntington’s disease (as a grouping; 1 study, n>8 million) [35]. The magnitude of association with mortality is probably (moderate certainty) very large for Down syndrome, and may (low certainty) be very large for age 60-69 years, and diabetes. We located no evidence on the combination of conditions that would place people with two or more comorbidities at increased risk.
The certainty of evidence was low for a large increase in mortality or hospitalization among: people with cerebral palsy (1 study, n>8 million); major psychiatric disorder (schizophrenia, schizoaffective disorder, or bipolar disorder, with drug use for the condition in the past 6 months; 1 study, n=11,122) [105]; obesity class III (BMI ≥40 kg/m2) versus normal weight (1 study, n=1,612) [113]; Parkinson’s disease (1 study, n>8 million) [35]; sickle cell disease or severe immunodeficiency (1 study, n>8 million) [35]; solid organ transplant (2 studies, n=18,038 for mortality [105, 113]); transplant of any type (3 studies, n=24,227 for hospitalization [49, 64, 105]); kidney transplant (1 study, n>8 million) [35]; recent bone marrow or stem cell transplant (1 study, n>8 million) [35]; metastatic cancer among those <70 years (2 studies, n=38,377) [58, 113]; current or recent (past year) chemotherapy or radiotherapy (past 6 months; 1 study, n>8 million); pregnancy (any stage; 1 study, n=409,462) [34]; frailty (2 studies, n=404,079) [77, 95]. The magnitude of association with mortality may (low certainty) be very large for recent chemotherapy. There was moderate certainty for a large increase in hospitalization with previous cerebrovascular accident (2 studies, n=12,376) [49, 115], and vasculitis (1 study, n=4,543,249) [135].
Among children, there was evidence of moderate certainty for a large increase in hospitalization among those having two or more versus no chronic conditions (1 study, n=804) [137], and evidence of low certainty for a large increase in mortality and hospitalization for children of Black versus White non-Hispanic race/ethnicity (1 study, n=12,198) [146]. There was low certainty for a large reduction in hospitalization for children 0-19 years versus adults aged 50-59 years (1 study, n=2,199) [105], and a large increase in hospitalization for children with one versus no chronic conditions (1 study, n=790) [137], and specifically for children with versus without immunodeficiency or immunosuppression (2 studies, n=21,875) [82, 138], and hypertension (1 study, n=21,116) [82]. Additionally, there was evidence of low certainty for a large increase in severe outcomes (ICU admission, mechanical ventilation, or ‘severe disease’) among children with versus without: one or more chronic conditions (1 study, n=759 for mechanical ventilation and severe disease [138]; 2 studies, n=1,341 for ICU admission), asthma (1 study, n=21,116) [82], endocrine conditions (1 study, n=5,374) [133], metabolic conditions (1 study, n=5,374) [133], and malignancy (1 study, n=5,374) [133].
There were several risk factors with moderate certainty evidence for little-to-no association with increased severity of COVID-19, for example many pre-existing conditions (notably several cardiovascular and respiratory conditions), and for adult males versus females (Appendix 2).
Table 4 shows the risk factors identified in Canadian reports and potential associations with severe outcomes of COVID-19. All data collected from these studies are in Appendix 2. This evidence was used primarily to draw conclusions about social factors, and was often descriptive and lacking adjustment for important covariates. These studies showed the potential for a large increase in mortality among people living in long-term care (6 reports, n≈2 million) [122, 123, 125, 127, 128], visible minority groups (mainly South Asian, Chinese, Black, Filipino, Latin American, Arab, Southeast Asian, West Asian, Korean, Japanese; 1 report, n=8,796) [129], and people living in the general population (off-reserve) versus First Nations people living on-reserve (1 report, n=9,715) [124].
INTERPRETATION
This review provides a methodologically rigorous synthesis of a large volume of emerging evidence on factors associated with severe outcomes of COVID-19. There is now strong (moderate certainty) evidence of at least a large increase in mortality from COVID-19 among people aged 60 to 69 versus <60 years (and over 70 years from our previous review [117]), people having two or more versus no comorbidities, and for people affected by (versus unaffected): Down syndrome; type 1 and type 2 diabetes; end-stage kidney disease; epilepsy; motor neuron disease, multiple sclerosis, myasthenia gravis, or Huntington’s disease.
Our earlier review suggested a potentially large increase in severe outcomes among males versus females, and for people with some pre-existing conditions (heart failure, dementia, liver disease) [10]. The current evidence no longer supports this increased risk, underscoring the importance of interpreting low certainty evidence cautiously, and of continually reviewing emerging data to support vaccine prioritization guidance. There is now relatively strong evidence for little-to-no increase in severe outcomes for several pre-existing diseases, notably many cardiovascular and respiratory conditions. In contrast, many jurisdictions have identified people with these conditions as priorities for vaccination [15-18]. The reason for this is not fully clear, because the evidence on which prioritization decisions are made is often not publically available, and such decisions are informed by multiple considerations (e.g., local epidemiology, ethics, risk-benefit analyses, vaccine characteristics and availability [5, 6, 155]) and varying levels of evidence.
A large reliance on unadjusted data may provide some explanation [156]. The data synthesized within this review have the advantage of accounting for the impact of multiple co-existing conditions, social and demographic factors. For example, in a large study (n=418,794) of the general population in the United Kingdom [94], a large increase in hospitalization among people with heart failure (OR 2.24) and chronic obstructive pulmonary disease (OR 2.67) was observed. Following adjustment for age, sex, social factors, BMI, and other chronic conditions, these associations were substantially attenuated to the range of little-to-no association (aOR 1.09 and 1.51, respectively) [94]. We observed similar relationships in other studies. The thresholds of magnitude used to define ‘large’ and ‘very large’ increases in severe outcomes may also impact the conditions prioritized. Finally, there are some rare conditions (e.g., thalassemia, splenic dysfunction) for which to our knowledge, no large and well-adjusted studies have been published. In these cases, governing bodies may need to rely on evidence available for other infectious diseases, or from smaller cohorts suggesting the potential for increased risk [18, 157-160].
The findings of this review, among other considerations, have informed NACI’s vaccine prioritization guidance. Emerging data should continue to be synthesized to improve on the certainty of evidence for some populations and outcomes. Though children seem to be at reduced risk of severe outcomes of COVID-19 compared to adults [161], there are few large, well-adjusted analyses available for children. Therefore, the current evidence about risk factors within this population is of low certainty. However, data is beginning to accumulate; we located six new eligible studies in our April 2021 update search to help inform NACI’s guidance [133, 137, 138, 146, 150, 154]. As we did not locate data to inform the combination of comorbidities that may place a person at increased risk of severe outcomes, a focus on this in future research would be valuable. Information from the development of risk prediction models may also provide insight [162]. Finally, while we looked for studies examining a range of factors based on the “P2ROGRESS And Other Factors” framework, the majority of studies focused on medical comorbidities. High quality research on P2ROGRESS And Other Factors would allow for a more nuanced approach to optimize vaccine rollout [155].
Limitations
Because we used a rapid approach, there is a small possibility of undetected errors in study selection or data extraction. We mitigated this through piloting and engaging experienced reviewers. Though the findings are most applicable to OECD countries, we often relied on data largely from the United States and other countries without universal healthcare systems. We located no evidence relevant to long-term outcomes.
Conclusion
There is strong evidence to support at least a large increase in mortality from COVID-19 among older adults aged 60 to 69 years versus <60 years; people having two or more versus no comorbidities; and for people affected by Down syndrome; type 1 and 2 diabetes; end-stage kidney disease; epilepsy; and motor neuron disease, multiple sclerosis, myasthenia gravis, or Huntington’s disease (as a grouping). Future research should focus on risk factors where evidence remains limited (no studies) or of low quality, the pediatric population, combinations of comorbidities that may increase risk, and long-term outcomes.
Data Availability
The data pertaining to the conclusions are available on reasonable request using the following link: https://doi.org/10.7939/DVN/VDELOG.
CONTRIBUTOR’S STATEMENT
MG contributed to the conception and design of the work, selection of studies, data extraction, analysis and interpretation of data, and drafted the initial version of the manuscript. JP contributed to the conception and design of the work, analysis and interpretation of data, and revised versions of the manuscript. AW, SG, SR, BZ, and AG contributed to the selection of studies, data extraction, organizing data for analysis, and revised versions of the manuscript. LH contributed to the conception and design of the work, methodological oversight, interpretation of findings, and revised versions of the manuscript. All authors approved of the final version of the manuscript submitted and agree to be accountable for all aspects of the work.
ROLE OF FUNDERS
The funders had no role in the conduct of the study nor decision to submit for publication.
DATA AVAILABILITY
The data pertaining to the conclusions are available on reasonable request using the following link: https://doi.org/10.7939/DVN/VDELOG.
ACKNOWLEDGMENTS
We thank Kelsey Young, Kelly Farrah, Shainoor Ismail, and Matthew Tunis from the NACI Secretariat for expert input into the review’s scope and eligibility criteria, and for reviewing the initial draft of the manuscript; Sharon Bartholomew, Louise McRae, Anne-Marie Robert, Samina Aziz, Alain Demers, Catherine Pelletier, Cynthia Robitaille, Larry Shaver, Helene Gardiner, and Dianne Zakaria from the Public Health Agency of Canada for input into the necessary adjustment variables for analyses of chronic diseases; Elizabeth Dennett and Erica Wright for developing and running literature searches; Andrew Beck for assistance with screening; and Mackinna Hauff for article retrieval and document formatting.
Footnotes
Funding statement: This study was completed under contract to the Public Health Agency of Canada, contract #4600001536. LH is supported by a Tier 1 Canada Research Chair in knowledge synthesis and translation.
Declaration of competing interests: The authors have no competing interests to declare.
The following risk factors were updated to 9-13 April 2021: -All risk factors within children -Immune compromise -Autoimmune -Pregnancy
REFERENCES
- 1.↵
- 2.↵
- 3.↵
- 4.↵
- 5.↵
- 6.↵
- 7.↵
- 8.↵
- 9.↵
- 10.↵
- 11.↵
- 12.↵
- 13.↵
- 14.↵
- 15.↵
- 16.
- 17.
- 18.↵
- 19.↵
- 20.↵
- 21.
- 22.
- 23.
- 24.↵
- 25.
- 26.
- 27.
- 28.
- 29.
- 30.
- 31.
- 32.
- 33.
- 34.↵
- 35.↵
- 36.
- 37.
- 38.
- 39.
- 40.
- 41.
- 42.
- 43.
- 44.
- 45.
- 46.
- 47.
- 48.↵
- 49.↵
- 50.
- 51.
- 52.
- 53.
- 54.
- 55.
- 56.
- 57.
- 58.↵
- 59.↵
- 60.
- 61.
- 62.
- 63.↵
- 64.↵
- 65.
- 66.
- 67.
- 68.↵
- 69.
- 70.
- 71.↵
- 72.
- 73.
- 74.
- 75.
- 76.
- 77.↵
- 78.
- 79.
- 80.
- 81.
- 82.↵
- 83.
- 84.↵
- 85.
- 86.
- 87.
- 88.
- 89.
- 90.
- 91.
- 92.↵
- 93.↵
- 94.↵
- 95.↵
- 96.
- 97.
- 98.↵
- 99.
- 100.
- 101.↵
- 102.
- 103.
- 104.
- 105.↵
- 106.↵
- 107.
- 108.
- 109.
- 110.
- 111.
- 112.
- 113.↵
- 114.
- 115.↵
- 116.
- 117.↵
- 118.
- 119.↵
- 120.
- 121.↵
- 122.↵
- 123.↵
- 124.↵
- 125.↵
- 126.
- 127.↵
- 128.↵
- 129.↵
- 130.↵
- 131.↵
- 132.
- 133.↵
- 134.
- 135.↵
- 136.
- 137.↵
- 138.↵
- 139.
- 140.
- 141.
- 142.
- 143.
- 144.
- 145.
- 146.↵
- 147.
- 148.
- 149.
- 150.↵
- 151.
- 152.↵
- 153.↵
- 154.↵
- 155.↵
- 156.↵
- 157.↵
- 158.
- 159.
- 160.↵
- 161.↵
- 162.↵