Association between SARS-CoV-2 Infection and Select Symptoms and Conditions 31 to 150 Days After Testing among Children and Adults

Background An increasing number of studies have described new and persistent symptoms and conditions as potential post-acute sequelae of SARS-CoV-2 infection (PASC). However, it remains unclear whether certain symptoms or conditions occur more frequently among persons with SARS-CoV-2 infection compared with those never infected with SARS-CoV-2. We compared the occurrence of specific COVID-associated symptoms and conditions as potential PASC 31 to 150 days following a SARS-CoV-2 test among adults (≥20 years) and children (<20 years) with positive and negative test results documented in the electronic health records (EHRs) of institutions participating in PCORnet, the National Patient-Centered Clinical Research Network. Methods and Findings This study included 3,091,580 adults (316,249 SARS-CoV-2 positive; 2,775,331 negative) and 675,643 children (62,131 positive; 613,512 negative) who had a SARS-CoV-2 laboratory test (nucleic acid amplification or rapid antigen) during March 1, 2020–May 31, 2021 documented in their EHR. We identified hospitalization status in the day prior through the 16 days following the SARS-CoV-2 test as a proxy for the severity of COVID-19. We used logistic regression to calculate the odds of receiving a diagnostic code for each symptom outcome and Cox proportional hazard models to calculate the risk of being newly diagnosed with each condition outcome, comparing those with a SARS-CoV-2 positive test to those with a negative test. After adjustment for baseline covariates, hospitalized adults and children with a positive test had increased odds of being diagnosed with ≥1 symptom (adults: adjusted odds ratio[aOR], 1.17[95% CI, 1.11–1.23]; children: aOR, 1.18[95% CI, 1.08–1.28]) and shortness of breath (adults: aOR, 1.50[95% CI, 1.38–1.63]; children: aOR, 1.40[95% CI, 1.15–1.70]) 31–150 days following a SARS-CoV-2 test compared with hospitalized individuals with a negative test. Hospitalized adults with a positive test also had increased odds of being diagnosed with ≥3 symptoms (aOR, 1.16[95% CI, 1.08 – 1.26]) and fatigue (aOR, 1.12[95% CI, 1.05 – 1.18]) compared with those testing negative. The risks of being newly diagnosed with type 1 or type 2 diabetes (aHR, 1.25[95% CI, 1.17–1.33]), hematologic disorders (aHR, 1.19[95% CI, 1.11–1.28]), and respiratory disease (aHR, 1.44[95% CI, 1.30–1.60]) were higher among hospitalized adults with a positive test compared with those with a negative test. Non-hospitalized adults with a positive SARS-CoV-2 test had higher odds of being diagnosed with fatigue (aOR, 1.11[95% CI, 1.05–1.16]) and shortness of breath (aOR, 1.22[95% CI, 1.15–1.29]), and had an increased risk (aHR, 1.12[95% CI, 1.02–1.23]) of being newly diagnosed with hematologic disorders (i.e., venous thromboembolism and pulmonary embolism) 31–150 days following SARS-CoV-2 test compared with those testing negative. The risk of being newly diagnosed with certain conditions, such as mental health conditions and neurological disorders, was lower among patients with a positive viral test relative to those with a negative viral test. Conclusions Patients with SARS-CoV-2 infection were at higher risk of being diagnosed with certain symptoms and conditions, particularly fatigue, respiratory symptoms, and hematological abnormalities, after acute infection. The risk was highest among adults hospitalized after SARS-CoV-2 infection.


222 Exposures and Covariates
223 The exposure of interest was a positive SARS-CoV-2 test, defined as "positive," "presumptive 224 positive," or "detected" ("positive viral test"), versus a negative SARS-CoV-2 test, defined as 225 "negative" or "not detected" ("negative viral test"), on a rapid antigen (1% of patients) or nuclear 226 acid amplification test (NAAT) recorded as polymerase chain reaction (PCR) tests (99% of 227 patients). If patients had any positive SARS-CoV-2 viral test during the study period, they were 228 analyzed as having only a positive test regardless of whether they had prior or subsequent 229 negative tests. Patients categorized as having a negative viral test only had negative viral tests 230 throughout the study period. The index test date from which we examined outcomes was the date 231 of the first positive or negative test.

233
We controlled for several a priori confounders in our regression analyses. For both children and 234 adults, we controlled for age as a continuous variable, age squared to account for nonlinear effect 235 of age, sex (female, male, and missing sex), race (Asian, Black, White, other race, missing), 236 ethnicity (Hispanic, non-Hispanic, missing), weight class (children: BMI less than the 95 th 237 percentile, BMI greater or equal to 95th percentile, missing BMI; adults: BMI < 30 kg/m 2 , ≥ 30 238 kg/m 2 , missing BMI), and number of visits or encounters with a health system in the 150-to 31-239 day period prior to the index date. For adults, we additionally controlled for combined 240 comorbidity score [34] assessed based on conditions that occurred in the 540 to 7 days prior to 241 the index date and current smoking status (current smoker; never or missing smoking), assessed 242 based on the record closest to the index date in that same period. For hospitalized adults and 243 children, we additionally controlled for length of stay, dexamethasone use, and mechanical for use under a CC0 license. This article is a US Government work. It is not subject to copyright under 17 USC 105 and is also made available preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this this version posted December 19, 2022. ; https://doi.org/10.1101/2022.12.18.22283646 doi: medRxiv preprint 244 ventilation during the hospitalization to account for variation in disease severity among 245 hospitalized patients. Mechanical ventilation was identified from the index date through 16 days 246 following the index date; this time period was chosen to account for the possibility that it may 247 take more than two weeks for respiratory failure to develop.

249 Analyses
250 All analyses were conducted using distributed regression modeling, in which each site separately 251 executed identical regression models, returning summary output including parameter estimates, 252 standard errors, covariance matrices, convergence status, and number of observations. Based on 253 the convergence of each regression at each site, results were either discarded or included in the 254 meta-analysis. Results from a specific site could be discarded for some outcomes and included 255 for others. Once the convergence was assessed, results from the selected sites were combined 256 using meta-analytic techniques (eTable 2). The random-effects model based on the DerSimonian 257 and Laird method was used to obtain pooled estimates [35].

259
We used different methodological approaches to examine condition and symptom outcomes.
260 Among adults, we examined each of the seven conditions in separate models. For each model, 261 we excluded all patients who had a diagnostic code for the relevant condition that was the 262 outcome for the model during the 540 to 31 days prior to the index date (e.g., patients with 263 hematologic conditions in the baseline were excluded from the model examining the outcome of 264 hematologic conditions). We used Cox proportional hazard regression models, accounting for 265 time from the beginning of the post-acute period (31 days post) to the earliest documentation of for use under a CC0 license. This article is a US Government work. It is not subject to copyright under 17 USC 105 and is also made available preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this this version posted December 19, 2022. ; https://doi.org/10.1101/2022.12.18.22283646 doi: medRxiv preprint 266 the first diagnostic code for each condition (event) and the end of the outcome period (150 days 267 post-censoring). We controlled for all covariates described above in these models.

268
269 For the symptom outcomes, we did not exclude patients who had diagnostic codes for these 270 symptoms during the baseline period. We took this approach because of how common these 271 symptoms are in routine clinical care. Instead, we controlled for the presence of these symptoms 272 in the 150 to 31 days prior to index date. We used logistic regression models to assess the odds 273 of having any of these four symptom outcomes associated with SARS-CoV-2 infection in the 274 entire 31 to 150 days post index period. We controlled for the same covariates as we did in the 275 condition outcome models, with the addition of a covariate indicating the presence of relevant 276 symptoms during the baseline period (e.g., for the "any symptom" outcome, we controlled for 277 the presence of any of the symptoms during the baseline period as one of the covariates; for the 278 fatigue outcome, we controlled for the presence of fatigue during the baseline period as one of 279 the covariates). As a secondary analysis, we examined two symptom outcomes, at least one   This article is a US Government work. It is not subject to copyright under 17 USC 105 and is also made available preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. 298 hospitalized children (Table 1).

299
300 Individuals with a positive viral test were older than those with a negative viral test in both age 301 cohorts across most care settings, although non-hospitalized adults who tested positive were 302 younger than those who tested negative (mean age: 49 vs 53 years, P < 0.001). Among both age 303 cohorts, compared to those with a negative viral test, more patients with a positive viral test were 304 Black (26% vs 18% among adults, P < 0.001 and 25% vs 18% among children, P < 0.001) 305 among hospitalized patients and Hispanic (17% vs 10% among adults, P < 0.001 and 23% vs 306 16% among children, P < 0.001) in both care settings. Adults with a positive viral test were more 307 likely to have obesity in both care settings (5-percent-point difference in both care settings, P < 308 0.001). Hospitalized children with a positive viral test were more likely to have obesity than 309 those with a negative viral test (21% vs 17%, P = 0.01).
for use under a CC0 license.
This article is a US Government work. It is not subject to copyright under 17 USC 105 and is also made available preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.   This article is a US Government work. It is not subject to copyright under 17 USC 105 and is also made available preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

397 Incidence of New Conditions Among Adults
398 Hospitalized adults with a positive viral test had higher incidence of almost all conditions, except 399 for mental health conditions, compared with those with negative viral test (Table 2). Hospitalized 400 adults with a positive viral test were most likely to be newly diagnosed with respiratory for use under a CC0 license.
This article is a US Government work. It is not subject to copyright under 17 USC 105 and is also made available preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.   This article is a US Government work. It is not subject to copyright under 17 USC 105 and is also made available preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

536
537 This study is subject to several limitations. First, use of EHR data to ascertain symptoms and 538 conditions may have led to an underestimation of real prevalence and incidence as we only 539 observe diagnosis codes for these symptoms and conditions when patients have a clinical 540 encounter with health systems of participating sites. Symptoms may be much less likely to be 541 entered into the EHR as diagnostic codes; clinicians often describe symptoms only in 542 unstructured notes. This is particularly an issue among patients with limited healthcare access.
543 Similarly, patients who always tested negative might have had a positive test that was not 544 captured in EHR (e.g., self-test at home). Thus, it is possible that some patients in the control 545 group may have tested positive at some point, perhaps within the follow-up period of 30-150 546 days after their negative test. This differential misclassification would bias results toward the 547 null. Second, we defined symptoms or conditions as the occurrence of one ICD-10-CM 548 diagnostic code 31 to 150 days following SARS-CoV-2 infection. This approach was used to 549 enhance sensitivity for detection of possible SARS-CoV-2 sequelae in the short interval of 31 to 550 150 days after the test but may have lower specificity. Although using 2 or more occurrences of for use under a CC0 license. This article is a US Government work. It is not subject to copyright under 17 USC 105 and is also made available preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this this version posted December 19, 2022. ; https://doi.org/10.1101/2022.12.18.22283646 doi: medRxiv preprint 551 diagnostic codes may have enhanced specificity, this would have restricted the study to patients 552 with multiple encounters in the same site, such as sicker patients or patients with better access to 553 healthcare. Third, although we have adjusted for a comprehensive set of confounders in 554 regression models, certain important covariates, such as vaccination status, were not included 555 due to data limitations. Vaccination data is often missing in EHR systems for most health 556 systems because of incomplete capture of data from state immunization registries. Fourth, we 557 used hospitalization within 16 days of a positive test for SARS-CoV-2 infection as a proxy for 558 COVID-19 severity, which may have resulted in misclassification if patients with a positive test 559 were hospitalized for reasons other than acute COVID-19 illness. Fifth, we were unable to 560 ascertain whether SARS-CoV-2 testing was conducted because of symptoms, as a part of routine 561 surveillance, or for travel purposes. Hospitalized persons who tested negative for SARS-CoV-2 562 included those hospitalized for nonviral illness (e.g., pregnancy, trauma, chronic conditions, 563 elective procedures) and may have biased our estimates if these illnesses were associated with 564 conditions or symptoms assessed in this study. There may be multiple etiologies for symptoms 565 and conditions examined in this report; the same is true for patients tested in the non-hospital 566 setting prior to certain tests or procedures that were for illnesses that also might have been 567 associated with the conditions or symptoms assessed in this study. Future studies may compare 568 patients hospitalized for SARS-CoV-2 infection only with patients hospitalized for influenza and 569 other lower respiratory tract illnesses. Finally, for covariates with missing values (e.g., sex and 570 race), we adjusted for missing values as a separate category in the analyses. Imputing missing 571 values may be a more robust approach.

572
for use under a CC0 license. This article is a US Government work. It is not subject to copyright under 17 USC 105 and is also made available preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.