Associations between self-reported SARS-CoV-2 infections and the diagnosis of new health conditions

Abstract

Background: Preexisting health conditions increase the risk of severe outcomes after SARS-CoV-2 infection. Emerging evidence suggests this relationship may be bi-directional with infection potentially increasing the risk of new health condition diagnoses.

Objective: To examine the association between number of SARS-CoV-2 infections and the subsequent diagnosis of new health conditions, excluding longer-term symptoms after SARS-CoV-2 infection, and to examine associations between infection and vaccination characteristics, including longer-term symptoms, and new diagnoses among the infected.

Methods: Population-based self-report cross-sectional survey data with longitudinal follow-up collected from Canadian adults was used to examine new diagnoses of 21 health conditions (and combinations thereof) since the first SARS-CoV-2 infection or, for uninfected adults, since the start of the pandemic. Multivariable binomial logistic regression was used to estimate adjusted odds ratios (aOR) controlling for sex, age, education, income, obesity, smoking status, preexisting health conditions and time at risk.

Results: Relative to adults reporting no SARS-CoV-2 infections, those reporting two or more had significantly greater odds of being diagnosed with a new condition (aOR=1.76, 95% confidence interval [CI]: 1.12–2.76, p=0.0133), respiratory condition (aOR=5.03, 95% CI: 1.72–14.73, p=0.0114), and back problems (aOR=3.79, 95% CI: 1.12–12.89, p=0.0056). Among infected adults, acute symptoms that limited daily life during the first infection (aOR=1.63, 95% CI: 1.08–2.48, p=0.0206) and longer-term symptoms (aOR=3.35, 95% CI: 2.24–5.03, p<0.0001) were also associated with being diagnosed with a new condition.

Conclusion: Repeated SARS-CoV-2 infections may increase the risk of reporting newly diagnosed health conditions. Findings regarding acute infection severity and longer-term symptoms should be interpreted cautiously. Both of these factors may have increased care seeking and opportunities for diagnoses and may be the consequence of a new health condition (reverse causation).

Introduction

Research shows that preexisting chronic conditions increase the risk of more severe outcomes (e.g., hospitalization, death) after a SARS-CoV-2 infection, including the development of longer-term symptoms ^{Footnote 1Footnote 2Footnote 3Footnote 4Footnote 5Footnote 6}. Lesser known, however, is emerging research suggesting this relationship may be bi-directional, as SARS-CoV-2 infection may increase the risk of new health condition diagnoses, with more severe infections associated with greater increases in risk ^{Footnote 7Footnote 8Footnote 9Footnote 10}. In addition, reinfections, pre-Omicron era infections, and not being vaccinated prior to infection have also been identified as potential risk factors for new health conditions ^{Footnote 11Footnote 12Footnote 13}. Hypothesized underlying mechanisms include persistent SARS-CoV-2 reservoirs in tissues, immune dysregulation, autoimmunity, microbiota disruption, clotting and endothelial abnormalities, dysfunctional neurological signalling and organ damage ^{Footnote 14Footnote 15}. The proposed bi-directional relationship is confounded by the indirect effects of the COVID-19 pandemic, which decreased health service use, thereby delaying the diagnosis of health conditions ^{Footnote 16Footnote 17} and potentially increasing the risk of their development due to compromised preventive health care. This highlights the importance of incorporating a control group when estimating the risk of new health conditions due to SARS-CoV-2 infection.

Research examining the association between SARS-CoV-2 infection and the risk of new health condition diagnoses has limitations. Studies may be primarily based on people attending healthcare institutions ^{Footnote 8Footnote 9Footnote 11Footnote 12Footnote 13} or specific subpopulations such as children and adolescents ^{Footnote 11} or veterans ^{Footnote 12Footnote 13}, they may not assess the impact of vaccinations and/or infection characteristics ^{Footnote 7Footnote 8Footnote 9Footnote 11} or they may not examine specific health conditions ^{Footnote 7}.

Currently, there is limited research that uses representative, population-based data to examine associations between SARS-CoV-2 infection and the diagnosis of new health conditions, and risk estimates for the general Canadian population are lacking. Considering the average infection-acquired SARS-CoV-2 seroprevalence in Canada's 10 provinces was 81.8% as of the end of October 2023 ^{Footnote 18}, even modest increases in the risk of new health condition diagnoses can translate to substantial additional cases depending on the background risk among the uninfected. The greater the background risk, the greater the additional cases. In this study, population-based survey data collected from Canadian adults was used to conduct exploratory analyses examining associations between the number of self-reported confirmed or suspected SARS-CoV-2 infections and newly diagnosed health conditions, excluding longer-term symptoms after SARS-CoV-2 infection. The secondary objective was to explore whether the time of first SARS-CoV-2 infection, vaccination status prior to first infection, severity of acute symptoms during first infection and longer-term symptoms after any SARS-CoV-2 infection were associated with the diagnosis of new health conditions among adults reporting one or more SARS-CoV-2 infections. The findings from this study will contribute to an evidence base that can inform future public health practice, clinical decision-making, healthcare resource needs and future research.

Methods

Data source and questionnaires

Data was used from the second cycle of the Canadian COVID-19 Antibody and Health Survey (CCAHS-2), a population-based cross-sectional survey that employed probability sampling, and Follow-up Questionnaire (FQ), which collected additional information on a subsample of adults who participated in the earlier CCAHS-2. The analytical sample consists of 8,809 respondents who participated in both the CCAHS-2 and FQ, and agreed to share their data with the Public Health Agency of Canada (Figure 1). Detailed information on the surveys and questionnaires is publicly available ^{Footnote 19}.

Figure 1 - Text description

Figure 1 shows the flow of adults from recruitment to inclusion in analyses. A total of 105,998 adults living in private households across 10 provinces were invited to participate in the Canadian COVID-19 Antibody and Health Survey (CCAHS-2) via invitation letters sent by mail that contained secure access codes for completing the online questionnaire. Three reminder letters, sent at intervals of 9 to 12 days, and follow-up calls by computer assisted telephone interviewers were used to optimize participation. Individuals younger than 18 years of age, persons living in the three territories or on reserves and other Indigenous settlements in the provinces, members of the Canadian Forces living on a base, institutionalized people, and residents of certain remote regions were excluded. Of the 105,998 adults invited to participate, 32,527 (30.7%) completed at least part of the online questionnaire between April and August 2022, and 25,544 (78.5% of 32,527) provided a valid email address which was used to invite them to the Follow-up Questionnaire. Of those providing a valid email address, 10,499 (41.1% of 25,544) completed at least part of the Follow-up Questionnaire online questionnaire between May 23 and June 19, 2023, inclusively, and 8,809 (83.9% of 10,499) agreed to share their data with the Public Health Agency of Canada and are included in the study.

Primary outcome—new diagnosed health conditions

The CCAHS-2 questionnaire asked all respondents whether a healthcare professional had diagnosed them with any of 21 specified health conditions (see Table 1 footnote a for list of conditions), along with the date of first diagnosis of each condition. In the FQ, respondents reporting a confirmed or suspected SARS-CoV-2 infection were asked whether a healthcare provider had diagnosed them with these conditions for the first time since their first infection. Those not reporting an infection were asked about new diagnoses since the start of the pandemic. The FQ did not capture the date of diagnosis, only its occurrence in time relative to the first SARS-CoV-2 infection or start of the pandemic, respectively. In the FQ, respondents were instructed to exclude preexisting conditions or conditions under investigation prior to the first infection or start of the pandemic, respectively, and there was no duration requirement for reported new health condition diagnoses. In order to capture health conditions diagnosed since the CCAHS-2 was conducted, for which a date of diagnosis was not captured, information reported in both the CCAHS-2 and FQ was used to classify respondents into one of three mutually exclusive categories for each health condition: no condition; condition present prior to first infection or the start of the pandemic for those not reporting an infection; or, condition diagnosed for the first time after first infection or since the start of the pandemic for those not reporting an infection.

Primary explanatory variable—number of self-reported confirmed or suspected SARS-CoV-2 infections

In the FQ, all respondents were asked "Since the start of the COVID-19 pandemic, how many different COVID-19 infections did you think or know you had?". Respondents were instructed to include infections with a positive test result and infections suspected to be COVID-19 because of the symptoms experienced or recent contact with a COVID-19 case. The date of the first SARS-CoV-2 infection was also captured. To investigate the potential cumulative effect of repeated SARS-CoV-2 infections, the number of self-reported infections was categorized as 0, 1, or 2 or more.

Additional explanatory variables

Time at risk for diagnosis of new conditions varied by self-reported SARS-CoV-2 infection status. Respondents reporting a confirmed or suspected infection were at risk as of the date of their first infection, while all others were considered at risk beginning January 1, 2020. This date was chosen to align with earliest infections reported by respondents. All respondents continued to be at risk for new diagnosed health conditions until they completed the FQ.

Time period of first SARS-CoV-2 infection was dichotomized as occurring either before or during the less virulent Omicron wave, defined here as beginning December 2021 ^{Footnote 20}.

Vaccination status prior to first SARS-CoV-2 infection was determined using the number and dates of receipt of COVID-19 vaccine doses, captured in both the CCAHS-2 and FQ, along with the date of first SARS-CoV-2 infection. Respondents were categorized as having received either one or fewer, or two or more doses (i.e., fully vaccinated) prior to the month of first infection. Respondents receiving one dose prior to infection were too few to examine separately.

Severity of acute symptoms during first SARS-CoV-2 infection was determined using information from both the CCAHS-2 and FQ and categorized as either impacting daily activities or not. Longer-term symptoms related to SARS-CoV-2 infection were defined as symptoms three or more months after a SARS-CoV-2 infection that could not be explained by anything else. Finally, because of their potential association with the diagnosis of new health conditions and/or number of SARS-CoV-2 infections, socioeconomic and health characteristics captured in the CCAHS-2 were also considered (Table 1).

A supplementary file, including details on imputations applied to COVID-19 vaccination dates collected in CCAHS-2 and to the self-reported date of first SARS-CoV-2 infection, as well as key questionnaire content, is available from the corresponding author.

Statistical analyses

Statistical procedures that acknowledge the complex survey design were used in combination with final and bootstrap weights supplied by Statistics Canada. Descriptive statistics (mean, percentage) and corresponding 95% confidence intervals (CIs) were used to characterize the Canadian adult population overall, as well as by diagnosis of a new health condition. The CIs for weighted means use the t-distribution, and the t-test was used to test for differences between group means (alpha=0.05, two-tailed). Confidence intervals for weighted proportions were calculated using the Clopper-Pearson exact method, and the design-based first-order Rao-Scott chi-square test of association was used to test for group differences.

Binomial logistic regression was used to assess whether the number of SARS-CoV-2 infections was independently associated with the diagnosis of new conditions after adjusting for sex at birth, age at start of time at risk, highest educational attainment among household members, national neighbourhood household income quintile, obesity, smoking status, number of preexisting health conditions and time at risk. Given the relatively small number of new diagnoses during time at risk, a parsimonious modelling approach was essential to ensure convergence and stable estimates. Older age ^{Footnote 21} and preexisting chronic conditions have been shown to increase the risk of diagnosis of a new condition ^{Footnote 22Footnote 23Footnote 24}. Additionally, longer time at risk increases the opportunity for new diagnoses, regardless of infection status. To determine which socioeconomic and health characteristics to include in all models and how to model them, binomial logistic regression was employed to identify those characteristics significantly (p<0.05, two-tailed) associated with a new diagnosis of at least one of the specific health conditions, examined separately, after adjusting for number of SARS-CoV-2 infections, sex, age, number of preexisting health conditions and time at risk.

Separate models assessed the new diagnosis of any of the 21 conditions, any condition within defined categories (cardiovascular, respiratory, neurological and musculoskeletal) and each specific condition. Respondents with a preexisting condition were excluded from the corresponding condition-specific model. For condition categories, respondents reporting all conditions within the category as preexisting were excluded from the category's model. To examine associations between infection and vaccination characteristics and new diagnoses, the analytical sample was limited to adults reporting at least one SARS-CoV-2 infection and the same models were rerun with the addition of the characteristic; separate models were run for each characteristic. When examining the effect of longer-term symptoms after SARS-CoV-2 infection, respondents reporting their first infection within 82 days of completing their FQ were excluded from the model due to insufficient follow-up to ascertain the presence of longer-term symptoms three or more months after infection. An 82-day threshold was used rather than 89 days, to account for the imprecision in self-reported infection dates recorded as occurring in the early (assigned 5^th day), middle (15^th) or late (25^th) part of the month. Due to potential multicollinearity between time at risk and both "period of infection" and "number of vaccine doses received prior to first infection", sensitivity analyses were conducted excluding time at risk when assessing the effects of these two characteristics. For all modelling, analyses were limited to adults with complete data for all the variables included in the model. The complexity of hypothesized relationships between the variables examined reinforces the exploratory nature of the study (Figure 2).

Figure 2 - Text description

Directed acyclic graph displaying hypothesized relationships between new health condition diagnoses, main explanatory variables of interest, and additional covariates. Main explanatory variables of interest included: number of self-reported confirmed or suspected SARS-CoV-2 infections; severity of acute SARS-CoV-2 infection symptoms; continuing symptoms three or more months after SARS-CoV-2 infection; time period of first SARS-CoV-2 infection; and vaccination status prior to first SARS-CoV-2 infection. Covariates included: sex at birth; age at start of time at risk; number of preexisting health conditions; obesity; smoking status; highest education achieved by household members; and national neighbourhood household income quintile. Arrows show the direction of the hypothesized relationships. The directed arrows are as follows: number of self-reported confirmed or suspected SARS-CoV-2 infections to continuing symptoms three or more months after SARS-CoV-2 infection, and new health condition diagnosis; severity of acute SARS-CoV-2 infection symptoms to continuing symptoms three or more months after SARS-CoV-2 infection, and new health condition diagnosis; continuing symptoms three or more months after SARS-CoV-2 infection to new health condition diagnosis; time period of first SARS-CoV-2 infection to severity of acute SARS-CoV-2 infection symptoms, continuing symptoms three or more months after SARS-CoV-2 infection, vaccination status prior to first SARS-CoV-2 infection, and new health condition diagnosis; vaccination status prior to first SARS-CoV-2 infection to number of self-reported confirmed or suspected SARS-CoV-2 infections, severity of acute SARS-CoV-2 infection symptoms, continuing symptoms three or more months after SARS-CoV-2 infection, and new health condition diagnosis; sex at birth to severity of acute SARS-CoV-2 infection symptoms, continuing symptoms three or more months after SARS-CoV-2 infection, and new health condition diagnosis; age at start of time at risk to vaccination status prior to first SARS-CoV-2 infection, and new health condition diagnosis; number of preexisting health conditions to number of self-reported confirmed or suspected SARS-CoV-2 infections, severity of acute SARS-CoV-2 infection symptoms, continuing symptoms three or more months after SARS-CoV-2 infection, vaccination status prior to first SARS-CoV-2 infection, and new health condition diagnosis; obesity to severity of acute SARS-CoV-2 infection symptoms, continuing symptoms three or more months after SARS-CoV-2 infection, and new health condition diagnosis; smoking status to new health condition diagnosis; highest education achieved by household members to number of self-reported confirmed or suspected SARS-CoV-2 infections, vaccination status prior to first SARS-CoV-2 infection, and new health condition diagnosis; and national neighbourhood household income quintile to number of self-reported confirmed or suspected SARS-CoV-2 infections, vaccination status prior to first SARS-CoV-2 infection, and new health condition diagnosis.

Since a minimum of five events (i.e., new diagnoses) per model parameter is desirable for modelling purposes ^{Footnote 25}, outcomes (e.g., respiratory condition, chronic lung condition, weakened immune system, etc.) with fewer than 60 new diagnoses were not modelled separately.

To ensure confidentiality, all descriptive estimates based on fewer than 30 respondents or having fewer than five respondents in the numerator of a proportion (unweighted) are suppressed. To ensure adequate reliability, descriptive estimates with coefficients of variation greater than 33.3% are also suppressed. In addition, multivariable models are not presented when the overall likelihood ratio test for the model is non-significant, any categorical variable includes fewer than 30 respondents or five new diagnoses in a specific category (unweighted), or any bootstrap replicate samples are excluded from variance estimation due to absent response levels or non-convergence.

SAS Enterprise Guide 7.1 (SAS 9.4TM) was used for all analyses, and R version 4.3.2 was used to produce the forest plots (forestploter package) and directed acyclic graph (dagitty and ggdag packages).

The study was exempt from research ethics board review under article 2.2 of the Tri-Council Policy Statement: Ethical Conduct for Research Involving Humans–TCPS 2 (2022) ^{Footnote 26}.

Results

Adults who reported being diagnosed with a new condition during their time at risk were more likely to be older, current smokers and obese, to identify as having a disability, to have greater time at risk for diagnosis of a new condition and to report at least one preexisting health condition (Table 1 and Table 2). They were also less likely to report a SARS-CoV-2 infection or to live in a household where someone had completed post-secondary education.

Among adults reporting at least one infection, those diagnosed with a new condition were more likely to have been first infected prior to the Omicron wave; to report that acute symptoms during their first infection affected daily life; and to report longer-term symptoms three or more months after infection. They were also less likely to have been fully vaccinated prior to their first infection (Table 3).

Compared to adults reporting no infections, those reporting two or more had significantly (p<0.05) greater odds of being diagnosed with a new condition (adjusted odds ratio [aOR]=1.76, 95% CI: 1.12–2.76, p=0.0133), a respiratory condition (aOR=5.03, 95% CI: 1.72–14.73, p=0.0114), and back problems (aOR=3.79, 95% CI: 1.12–12.89, p=0.0056) (Figure 3). Although not statistically significant, strong associations (i.e., aOR>2) were also observed between one infection and respiratory conditions, as well as between both one and two or more infections and sleep apnea.

Figure 3 - Text description

New diagnosed condition	Number of infections	aOR	95% CI	p-value
Any of select conditions	1	1.12	0.67–1.85	0.0133
	2+	1.76	1.12–2.76	-
Respiratory	1	3.59	0.81–15.96	0.0114
	2+	5.03	1.72–14.73	-
Cardiovascular	1	0.95	0.41–2.17	0.3944
	2+	1.44	0.69–2.99	-
High blood pressure	1	0.61	0.25–1.48	0.2477
	2+	1.14	0.52–2.48	-
Musculoskeletal	1	1.09	0.42–2.83	0.3762
	2+	1.66	0.77–3.6	-
Arthritis	1	1.76	0.62–4.96	0.4057
	2+	0.83	0.28–2.49	-
Back problems	1	0.73	0.13–4.07	0.0056
	2+	3.79	1.12–12.89	-
Sleep apnea	1	2.95	0.58–14.92	0.4178
	2+	3.09	0.48–20.03	-
Mental health condition	1	0.29	0.08–1.03	0.0487
	2+	1.11	0.39–3.16	-

Among adults reporting a SARS-CoV-2 infection, being first infected prior to the Omicron wave (vs. during) or having received fewer than two COVID-19 vaccine doses prior to the first infection (vs. two or more doses) was not significantly associated with new diagnoses (Figure 4). However, sensitivity analyses showed that after removing time at risk from the model, being infected prior to the Omicron wave was significantly associated with greater odds of being diagnosed with a new condition (aOR=2.17, 95% CI: 1.46–3.23, p=0.0001) and cardiovascular condition (aOR=2.64, 95% CI: 1.34–5.19, p=0.0051), with a borderline significant association for high blood pressure (aOR=1.88, 95% CI: 0.89–4.00, p=0.0987). Similarly, after removing time at risk from the model, not being fully vaccinated prior to the first infection was significantly associated with greater odds of being diagnosed with a new condition (aOR=1.92, 95% CI: 1.30–2.84, p=0.0010), cardiovascular condition (aOR=2.67, 95% CI: 1.50–4.77, p=0.0009), and high blood pressure (aOR=2.38, 95% CI: 1.22–4.61, p=0.0105).

Figure 4 - Text description

New diagnosed condition	Infection characteristics	aOR	95% CI	p-value
Any of select conditions	Infected prior to Dec 2021 (ref=Dec 2021 onward)	0.71	0.32–1.55	0.3885
	Less than two COVID-19 vaccine doses received pre-infection (ref=2 or more)	0.64	0.32–1.28	0.2079
	Acute symptoms of first SARS-CoV-2 infection affected daily life (ref=no effect)	1.63	1.08–2.48	0.0206
	Longer-term symptoms three or more months after infection (ref=no longer-term symptoms)	3.35	2.24–5.03	<0.0001
Cardiovascular	Infected prior to Dec 2021 (ref=Dec 2021 onward)	2.28	0.60–8.74	0.2279
	Less than two COVID-19 vaccine doses received pre-infection (ref=two or more)	2.39	0.75–7.62	0.1416
	Acute symptoms of first SARS-CoV-2 infection affected daily life (ref=no effect)	1.63	0.82–3.24	0.1615
	Longer-term symptoms three or more months after infection (ref=no longer-term symptoms)	2.54	1.35–4.77	0.0039
High blood pressure	Infected prior to Dec 2021 (ref=Dec 2021 onward)	1.69	0.42–6.77	0.4563
	Less than two COVID-19 vaccine doses received pre-infection (ref=two or more)	3.06	0.91–10.27	0.0696
	Acute symptoms of first SARS-CoV-2 infection affected daily life (ref=no effect)	2.13	0.99–4.56	0.0527
	Longer-term symptoms three or more months after infection (ref=no longer-term symptoms)	2.22	1.07–4.57	0.0313

Adults reporting that their daily life was impacted by acute symptoms during their first SARS-CoV-2 infection (vs. not impacted) had significantly greater odds of reporting a newly diagnosed condition (aOR=1.63, 95% CI: 1.08–2.48, p=0.0206), with a borderline significant association for high blood pressure (aOR=2.13, 95% CI: 0.99–4.56, p=0.0527) (Figure 4).

Adults reporting longer-term symptoms three or more months after any SARS-CoV-2 infection (vs. no longer-term symptoms) had significantly greater odds of reporting a newly diagnosed condition (aOR=3.35, 95% CI: 2.24–5.03, p<0.0001), cardiovascular condition (aOR=2.54, 95% CI: 1.35–4.77, p=0.0039), and high blood pressure (aOR=2.22, 95% CI: 1.07–4.57, p=0.0313) (Figure 4). Additional details regarding all modelling results are available from the corresponding author.

Discussion

Our exploratory analyses using population-based self-report survey data indicated that reporting two or more SARS-CoV-2 infections (vs. no infection) was significantly associated with greater odds of being diagnosed with a new health condition, respiratory condition and back problems. Among adults reporting at least one SARS-CoV-2 infection, having daily life negatively impacted by acute symptoms during the first infection (vs. not impacted) was significantly associated with greater odds of being diagnosed with a new health condition, while longer-term symptoms (vs. no longer-term symptoms) after any SARS-CoV-2 infection was significantly associated with greater odds of being diagnosed with a new health condition, cardiovascular condition and high blood pressure.

The lack of association between being diagnosed with a new condition and both period of first SARS-CoV-2 infection and vaccine doses received prior to first infection could have been the result of correlations with time at risk, as evidenced by the sensitivity analyses. The findings with respect to more severe acute infections or prolonged symptoms after infection should be interpreted cautiously. Both situations may have increased care seeking and opportunities for diagnoses and may be the consequence of a new health condition rather than a risk factor (i.e., reverse causation).

The findings align, in part, with previous studies and contribute to available research on the general population. Regarding the diagnosis of a new condition, while the aOR for two or more infections relative to no infection (aOR=1.76, 95% CI: 1.12–2.76) is consistent with De Ridder et al.'s ^{Footnote 7} aOR for any confirmed infection (aOR=2.15, 95% CI: 1.43–3.23), no significant association for one infection was found. Bowe et al. ^{Footnote 12} found that, relative to adults with no confirmed SARS-CoV-2 infections, the risk of at least one sequelae across multiple organ systems increased with the number of confirmed SARS-CoV-2 infections from 37% (hazard ratio [HR]=1.37, 95% CI: 1.36–1.38) for one infection, to 107% (HR=2.07, 95% CI: 2.03–2.11) for two infections, to 135% (HR=2.35, 95% CI: 2.12–2.62) for three or more infections.

Based on a meta-analysis, Gaudet et al. ^{Footnote 9} reported moderate increases in the risk of diagnosis of respiratory disorders (HRs: 1.83–1.89) and sleep apnea (HRs: 1.5–1.7) in adult outpatient/mixed outpatient and inpatient samples. Gaudet et al. also found that the risk of diagnosis of new conditions was often higher among inpatients than outpatients or mixed samples, highlighting the importance of infection severity. Although research has demonstrated an association between infection and musculoskeletal disorders ^{Footnote 12}, no studies were found that specifically examined back problems.

Consistent with the findings during sensitivity analyses, Xie et al. ^{Footnote 13} found that the risk of new post-acute SARS-CoV-2 infection sequela across multiple organ systems during the first year after infection was about 34% and 23% greater in the pre-Delta and Delta era relative to the Omicron era among unvaccinated adults, respectively; and, that the risk of sequela among unvaccinated adults was 78% and 122% greater than in vaccinated adults in the Delta and Omicron era, respectively, although Xie et al. did not acknowledge the sequencing of vaccination and infection.

The CCDSS, which includes data to fiscal year 2023–2024 for select chronic conditions, did not indicate notable sustained increases in the overall age-standardized incidence rate of asthma, chronic obstructive pulmonary disease, ischemic heart disease, stroke, osteoarthritis and rheumatoid arthritis relative to the pre-pandemic period ^{Footnote 21}. Although there was no significant association between number of SARS-CoV-2 infections and new high blood pressure diagnoses, the CCDSS documented an 8.1% increase in the overall age-standardized incidence rate for high blood pressure between fiscal year 2019–2020 (2,093 per 100,000, 95% CI: 2,085–2,101) and fiscal year 2022–2023 (2,262 per 100,000, 95% CI: 2,254–2,270); however, population-based surveillance data should be interpreted cautiously. True changes in incidence may be confounded by the indirect effects of the pandemic, including changes in the availability and likelihood of accessing health services during the pandemic, or changes in coding practices in administrative health data. The CCDSS indicated declines in the incidence of many conditions during the pandemic, and more recent increases could be the result of diagnostic "catch-up".

Strengths and limitations

The primary strengths of this study are that it is population-based, considers a wide range of health conditions and allows for the examination of infection and vaccination characteristics. Its primary limitations are its reliance on participation and self-reporting, which make it susceptible to survivor, selection, misclassification, recall and surveillance biases. The most severely affected adults, who did not survive their SARS-CoV-2 infection(s), were excluded from the study. Of those adults invited to participate in the CCAHS-2, 8.3% (n=8,809 of 105,998) were included in the analytical sample. Although weights were adjusted for non-response and calibrated using auxiliary information to reflect the target population, biased estimates may still occur if participants differed systematically from the target population in ways not corrected through weighting. Some suspected infections may have been due to conditions or infections unrelated to SARS-CoV-2, while others who did not report an infection may have actually been infected, potentially biasing the true underlying association toward the null ^{Footnote 27}. There was no serology data on FQ participants as of June 2023 to incorporate it into the analyses. The FQ data collected between May 23 and June 19, 2023 indicated that about 67% of adults reported at least one SARS-CoV-2 infection (Table 1), which is comparable to infection-acquired SARS-CoV-2 seroprevalence in Canada's 10 provinces as of June 6, 2023 (78%), considering the latter estimate included children and youth: age groups with the highest seroprevalence ^{Footnote 18Footnote 28}. Although self-report is imperfect, other measures such as seroprevalence cannot provide information about the timing or number of infections and is affected by issues like non-seroconversion and seroreversion ^{Footnote 14}. Adults reporting an infection may have been more likely to recall a new diagnosis or visit their healthcare provider, increasing the chances of receiving a diagnosis. Similarly, seeking care for a health condition may have increased the likelihood of having a SARS-CoV-2 infection diagnosed. There was no information to adjust for pre-pandemic health seeking behaviour, which could bias the findings if health-seeking behaviour differed by self-reported SARS-CoV-2 infection status. Although adjustment was made for time at risk, differences in the start of follow-up between adults reporting and not reporting a SARS-CoV-2 infection may have biased the results; adults reporting an infection had less time at risk for new health condition diagnoses and had a greater proportion of their time at risk distributed later in the calendar-time period of follow-up. Adjustment was not made for all potential confounders, such as province of residence and healthcare service availability, which may be associated with both number of SARS-CoV-2 infections and diagnosis of new health conditions. Considering the FQ was conducted more than three years ago, the findings may not apply today given the evolving viral, vaccine, and population immunity landscape. The relatively small number of newly diagnosed conditions decreased the power of the analyses and limited the number of specific conditions modelled. Finally, since p-values were not adjusted for the number of analyses performed, some of the findings may be due to chance (i.e., type I error).

Conclusion

The exploratory analyses found several significant associations between new condition diagnoses and SARS-CoV-2 infection and vaccination characteristics. Compared to adults reporting no SARS-CoV-2 infections, those reporting two or more infections had significantly greater odds of being diagnosed with a new health condition, reinforcing the continued importance of public health efforts to prevent infection. Although more population-based research is needed to disentangle the effects of infection characteristics, vaccination status and surveillance bias, the exploratory analyses indicated that being first infected early in the pandemic, not being fully vaccinated prior to first infection, experiencing acute symptoms that limit daily life during the first infection, and reporting prolonged symptoms after any infection may be associated with an increased risk of diagnosis of new health conditions. Continued investigation into the long-term sequelae of SARS-CoV-2 infections, particularly reinfections, and the biological, behavioural and systemic factors that may mediate the effects of these infections is needed. Such investigations could assist healthcare practitioners in tailoring COVID-19 vaccine recommendations and post COVID-19 monitoring for new health conditions and guide public health practice and messaging when promoting COVID-19 vaccine uptake and personal protective measures (e.g., masking) to prevent and minimize the severity of infections, as well as the potential development of new health conditions.

Authors' statement

DZ — Conceptualization, methodology, writing–original draft, writing–review & editing, formal analysis, visualization
XH — Conceptualization, methodology, writing–original draft, writing–review & editing, formal analysis
SA — Conceptualization, methodology, writing–original draft, writing–review & editing
LO — Conceptualization, methodology, writing–original draft, writing–review & editing
JC — Conceptualization, methodology, writing–original draft, writing–review & editing

Competing interests

None.

ORCID numbers

Dianne Zakaria — 0000-0001-8610-1248
Xue feng Hu — 0000-0001-6794-5699

Acknowledgements

The data used in this article were made possible through a collaboration between the Public Health Agency of Canada, Statistics Canada, and the COVID-19 Immunity Task Force. The participation of the citizens of Canada in Cycle 2 of the Canadian COVID-19 Antibody and Health Survey and Follow-up Questionnaire is greatly appreciated. The generous and inciteful reviewer feedback improved the quality of this article. The analyses, conclusions, opinions and statements expressed herein are solely those of the authors. No endorsement by the Public Health Agency of Canada or Statistics Canada is intended or inferred.

Funding

This work was supported by the Public Health Agency of Canada and Statistics Canada.

Appendix

Supplemental material is available upon request to the corresponding author: dianne.zakaria@phac-aspc.gc.ca

• Previous Page
• Table of contents
• Next page