Prevalence of Select New Symptoms and Conditions Among Persons Aged Younger Than 20 Years and 20 Years or Older at 31 to 150 Days After Testing Positive or Negative for SARS-CoV-2

Key Points

Question  Are select new symptoms and conditions more common among persons aged younger than 20 years and 20 years or older who tested positive for SARS-CoV-2 compared with those who tested negative?

Findings  In this cohort study of 338 024 persons younger than 20 years and 1 790 886 persons 20 years or older who were tested for SARS-CoV-2, new diagnoses of shortness of breath, nonspecific heart rate abnormalities, and type 2 diabetes were more common among those hospitalized after positive compared with negative test results; fatigue was more common among those aged 20 years or older.

Meaning  Given these findings, health care professionals should be aware of new symptoms and conditions that may develop after SARS-CoV-2 infection, particularly among those hospitalized.

Importance  New symptoms and conditions can develop following SARS-CoV-2 infection. Whether they occur more frequently among persons with SARS-CoV-2 infection compared with those without is unclear.

Objective  To compare the prevalence of new diagnoses of select symptoms and conditions between 31 and 150 days after testing among persons who tested positive vs negative for SARS-CoV-2.

Design, Setting, and Participants  This cohort study analyzed aggregated electronic health record data from 40 health care systems, including 338 024 persons younger than 20 years and 1 790 886 persons aged 20 years or older who were tested for SARS-CoV-2 during March to December 2020 and who had medical encounters between 31 and 150 days after testing.

Main Outcomes and Measures  International Statistical Classification of Diseases, Tenth Revision, Clinical Modification codes were used to capture new symptoms and conditions that were recorded 31 to 150 days after a SARS-CoV-2 test but absent in the 18 months to 7 days prior to testing. The prevalence of new symptoms and conditions was compared between persons with positive and negative SARS-CoV-2 tests stratified by age (20 years or older and young than 20 years) and care setting (nonhospitalized, hospitalized, or hospitalized and ventilated).

Results  A total of 168 701 persons aged 20 years or older and 26 665 younger than 20 years tested positive for SARS-CoV-2, and 1 622 185 persons aged 20 years or older and 311 359 younger than 20 years tested negative. Shortness of breath was more common among persons with a positive vs negative test result among hospitalized patients (≥20 years: prevalence ratio [PR], 1.89 [99% CI, 1.79-2.01]; <20 years: PR, 1.72 [99% CI, 1.17-2.51]). Shortness of breath was also more common among nonhospitalized patients aged 20 years or older with a positive vs negative test result (PR, 1.09 [99% CI, 1.05-1.13]). Among hospitalized persons aged 20 years or older, the prevalence of new fatigue (PR, 1.35 [99% CI, 1.27-1.44]) and type 2 diabetes (PR, 2.03 [99% CI, 1.87-2.19]) was higher among those with a positive vs a negative test result. Among hospitalized persons younger than 20 years, the prevalence of type 2 diabetes (PR, 2.14 [99% CI, 1.13-4.06]) was higher among those with a positive vs a negative test result; however, the prevalence difference was less than 1%.

Conclusions and Relevance  In this cohort study, among persons hospitalized after a positive SARS-CoV-2 test result, diagnoses of certain symptoms and conditions were higher than among those with a negative test result. Health care professionals should be aware of symptoms and conditions that may develop after SARS-CoV-2 infection, particularly among those hospitalized after diagnosis.

Understanding the frequency with which new symptoms and conditions emerge in the months following SARS-CoV-2 infection is important to inform patients’ expectations for recovery and allow health care professionals and health systems to address patients’ needs. Shortness of breath, fatigue or muscle weakness, and mild subjective cognitive dysfunction (ie, “brain fog”) are among the most commonly reported persistent symptoms in the months following SARS-CoV-2 infection.^1-3 Studies have also documented an increase in diagnoses of new-onset neuropsychiatric conditions and type 2 diabetes among persons after SARS-CoV-2 infection.^4,5 New symptoms and conditions months after a positive SARS-CoV-2 test have been described among small cohorts of nonhospitalized and hospitalized persons.^2,6,7 However, population-based estimates of the occurrence of new symptoms and conditions following a diagnosis of SARS-CoV-2 infection remain undercharacterized in the US. In addition, whether certain new symptoms and conditions occur more frequently among persons with SARS-CoV-2 infection compared with those without has not been well established.

Use of electronic health records (EHRs) provides an opportunity to characterize the occurrence of symptoms and conditions among large populations. The National Patient-Centered Clinical Research Network, PCORnet, is a national-scale research network of analysis-ready longitudinal EHR data composed of more than 60 health care systems from all census regions of the US.⁸ Starting in April 2020, PCORnet began COVID-19 surveillance among 42 health care systems and 1 health plan with data available for approximately 9 million persons aged 20 years or older and nearly 3 million older than 20 years.

We used PCORnet data to examine the prevalence of select new symptoms and conditions in the 31 to 150 days after SARS-CoV-2 testing among persons aged younger than 20 and 20 years or older with positive and negative SARS-CoV-2 test results. Symptoms (eg, shortness of breath, fatigue, and cognitive dysfunction) and conditions (eg, type 2 diabetes and neurological disorders) were selected based on previous reports of SARS-CoV-2 infection sequelae.^1-5,9,10 We aimed to (1) describe the prevalence of new symptoms and conditions diagnosed at medical encounters between 31 and 150 days after SARS-CoV-2 infection, stratified by age group and care setting, and (2) determine whether new symptoms and conditions were more common among persons with a positive test result for SARS-CoV-2 (hereafter referred to as “positive test”) compared with those with a negative test for SARS-CoV-2 (hereafter referred to as “negative test”). We hypothesized that (1) among persons with a positive test, symptoms and conditions would occur more frequently among those who received a higher level of care after diagnosis, and (2) if new symptoms and conditions were sequelae of SARS-CoV-2 infection, their prevalence would be higher among persons with a positive test.

Descriptions of PCORnet are available elsewhere.^8,11 Briefly, PCORnet uses a Common Data Model to facilitate data interoperability and centralized querying of longitudinal EHR data using modular statistical programs. Starting in April 2020, the network began supporting rapid querying (up to twice monthly), allowing for capture of data on patients tested for SARS-CoV-2 (identified using Logical Observation Identifiers Names and Codes [LOINC]). Queries were performed at each participating health care system using patient-level EHR data; results were transmitted to investigators in aggregated tabular format. This cohort study, included within a general SARS-CoV-2 surveillance project across PCORnet institutions, was deemed exempt from review under the public health surveillance provision of the Common Rule by the Harvard Pilgrim Health Care institutional review board. The design and analysis meets the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline.

In March 2021, 40 sites responded to a query assessing the occurrence of select symptoms and conditions diagnosed in the 31 to 150 days after SARS-CoV-2 testing; 2 sites were excluded because of data-quality issues and response delays (Figure 1; eTable 2 in the Supplement). The query included persons aged 20 years or older and younger than 20 years with SARS-CoV-2 polymerase chain reaction (99%) or antigen (1%) testing between March and December 2020 with follow-up until mid-March 2021 (Figure 1). We stratified age groups by care setting according to the highest level of care (ie, nonhospitalized, hospitalized, and hospitalized and ventilated) recorded between the day prior and 16 days after the index date of their first positive or negative test. Because of small sample sizes, we were unable to characterize a subgroup of ventilated persons younger than 20 years (n = 172). To maximize data completeness on new symptoms and conditions, we restricted the analysis to patients who had at least 1 medical encounter in any care setting recorded within 3 years to 30 days before and 31 to 150 days after SARS-CoV-2 testing (ie, connected to the health care system before and after SARS-CoV-2 testing). Patient-level data were stored behind institutional firewalls; the query returned only aggregate data for analysis. Information on race and ethnicity was abstracted directly from the EHR, which is typically self-reported by patients and captured when patients register for care at an institution.

We captured aggregate information on patient age, sex, race, ethnicity, and underlying symptoms and conditions up to 18 months prior to SARS-CoV-2 testing. Race categories included African American or Black, Asian, White, other (ie, American Indian or Alaska Native, Native Hawaiian or other Pacific Islander, or multiple race), and missing. Ethnicity categories included Hispanic, non-Hispanic, other, and missing. Symptoms and conditions among persons aged 20 years or older (n = 16) and younger than 20 years (n = 12) were defined using grouped International Classification of Diseases, Tenth Revision, Clinical Modification (ICD-10-CM) codes (eTable 1 in the Supplement). We included all ICD-10-CM codes for these symptoms and conditions recorded in encounters 31 to 150 days after SARS-CoV-2 testing. Persons with a single ICD-10-CM code (eg, shortness of breath) recorded in the 31 to 150 days were only counted once; however, patients with multiple symptoms or conditions of interest were counted separately for each symptom or condition of interest.

To assess prevalence of new symptoms and conditions, we specified 3 time periods relative to the index date of SARS-CoV-2 test: 18 months to 7 days before testing (baseline), 6 days before to 30 days after testing (short-term), and 31 to 150 days after testing (long-term). We set the index date as the date of the first positive or negative test; the cohort of patients with negative tests never had a positive test during the entire query period (ie, up to 150 days after first SARS-CoV-2 test). New symptoms and conditions were defined as those present in the long-term period and absent in the baseline period. Prevalence was calculated by dividing the number of persons having a given symptom or condition in the long-term phase but not at baseline by the total persons in the cohort minus the number of persons with a given symptom or condition in the baseline period (Figure 1).

Persons with a positive and negative test were divided into cohorts by age (20 years or older and younger than 20 years) and care setting. Within cohorts, we compared the distribution of age, sex, race, ethnicity, and underlying conditions using χ² tests. We used a Bonferroni correction to adjust for multiple testing, and statistical significance was considered P < .0008. We were unable to perform multivariable adjustment for differences in demographic characteristics, underlying conditions, and follow-up because of the aggregated format of the data. We compared the prevalence of new symptoms and conditions between persons with a positive and negative test within age and care setting cohorts using unadjusted prevalence ratios (PRs) with 99% CIs. Exclusion of the null (PR of 1.0), rather than P values, was used to determine whether PRs were significantly different between persons with a positive and a negative test.

During March to December 2020, among 1 790 886 persons aged 20 years or older (694 751 men, 1 096 039 women, and 96 other or missing sex) who had medical encounters in the 31- to 150-day period after index (henceforth referred to as medical encounters), we identified 144 768 nonhospitalized and 23 933 hospitalized (of whom 1488 were ventilated) persons with a positive test, and 1 227 510 nonhospitalized and 394 675 hospitalized (13 447 ventilated) persons with a negative test. Among 338 024 persons younger than 20 years (170 206 male, 167 805 female, and 13 other or missing sex) with medical encounters, we identified 25 327 nonhospitalized and 1338 hospitalized persons with a positive test and 260 660 nonhospitalized and 50 699 hospitalized persons with a negative test (Table 1). Among both age cohorts, the proportion of persons with medical encounters was higher among those with a negative test compared with those with a positive test. Persons younger than 20 years with positive tests were older, with more who were aged 13 to younger than 20 years, than those with negative tests (49% vs 35%). More persons aged 20 years or older (18%) and younger than 20 years (26%) with positive tests were Hispanic across all care settings compared with those with negative tests (8% and 14%, respectively) (Table 1).

Type 2 diabetes and severe obesity were more common among persons aged 20 years or older with a positive test relative to those with a negative test across all care settings during baseline (18 months to 7 days prior to SARS-CoV-2 testing) (Table 1). Hospitalized persons younger than 20 years who tested positive had a higher proportion of cancer compared with those who tested negative (10% vs 6%) during baseline.

Among persons aged 20 years or older with a positive test, shortness of breath (4.5% nonhospitalized, 10.5% hospitalized, and 16.6% ventilated), fatigue (4.2% non-hospitalized, 8.0% hospitalized, and 19.8% ventilated), and sleep disorders (3.2% nonhospitalized, 5.6% hospitalized, and 10.4% ventilated) were the most prevalent diagnoses (Table 2). Among persons aged younger than 20 years with a positive test, change in bowel habits (2.3% nonhospitalized and 6.0% hospitalized), fatigue (1.7% nonhospitalized and 3.5% hospitalized), and shortness of breath (1.8% nonhospitalized and 3.9% hospitalized) were the most prevalent diagnoses.

Among nonhospitalized persons aged 20 years or older, the prevalence of shortness of breath in the 31- to 150-day period was higher among those with a positive test compared with those with a negative test (PR, 1.09 [99% CI, 1.05-1.13]), whereas the prevalence of other symptoms was higher among persons with a negative test (Figure 2; eTable 3 in the Supplement). Among nonhospitalized persons younger than 20 years, new diagnoses of shortness of breath were more prevalent among those with positive tests compared with those with negative tests (PR, 1.26 [99% CI, 1.11-1.44]). (Figure 2; eTable 3 in the Supplement). However, for both age cohorts, prevalence differences were smaller than 1%.

In both age cohorts, among hospitalized persons with a positive test, the prevalence of shortness of breath (≥20 years: PR, 1.89 [99% CI, 1.79-2.01]; <20 years: PR, 1.72 [99% CI, 1.17-2.51]) and nonspecific heart rate abnormalities (ie, tachycardia, bradycardia, or palpitations) (≥20 years: PR 1.25 [1.16-1.36]; <20 years: PR 1.47 [1.02-2.12]) was significantly higher compared with those with a negative test (Figure 2; eTable 3 in the Supplement). Fatigue (PR, 1.35 [99% CI 1.27-1.44]), sleep disorders (PR, 1.12 [99% CI 1.04-1.22]), and cognitive dysfunction (PR, 1.17 [99% CI 1.08-1.26]) were higher among hospitalized persons aged 20 years or older with a positive compared with a negative test; however, absolute differences in prevalence were smaller than 1% for sleep disorders and cognitive dysfunction. Differences in symptom prevalence between persons aged 20 years or older with a positive vs negative test were greatest among those who were ventilated; prevalence of fatigue and shortness of breath were 10% and 7% higher among those with a positive test, respectively (Table 2).

The prevalence of new conditions among nonhospitalized persons aged 20 years or older with a positive test was less than 2% (Table 2). Among hospitalized persons aged 20 years or older with a positive test, type 2 diabetes (7.2%), anxiety and depression (4.9%), and ataxia or trouble walking (2.2%) were the most prevalent new diagnoses. Among ventilated persons aged 20 years or older with a positive test, prevalence was highest for type 2 diabetes (16.7%) and anxiety and depression (9.5%). New-onset peripheral nerve disorders (7.2%), ataxia and trouble walking (7.3%), and myoneural disorders (5.7%) were common among ventilated persons aged 20 years or older with a positive test, although these conditions were rare among their nonventilated hospitalized and nonhospitalized counterparts (≤2.5% prevalence). Among persons younger than 20 years with a positive test, anxiety and depression (2.6% nonhospitalized and 4.6% hospitalized) were the most prevalent new diagnoses.

Among nonhospitalized persons aged 20 years or older, several new conditions were more prevalent among those with a negative test compared with those with a positive test; however, prevalence differences were less than 1%. Among nonhospitalized persons younger than 20 years, the prevalence of anxiety and depression (PR, 1.17 [99% CI, 1.05-1.31]) was higher among those with positive vs a negative test (Figure 3; eTable 3 in the Supplement).

Among hospitalized persons aged 20 years or older, the prevalence of new type 2 diabetes (PR, 2.03 [99% CI, 1.87-2.19]) was significantly higher among those with a positive test compared with those with a negative test (Figure 3; eTable 3 in the Supplement). Among ventilated persons aged 20 years or older with a positive test, the prevalence of new type 2 diabetes (PR, 2.25 [99% CI, 1.82-2.77]), peripheral nerve disorders (PR, 2.05 [99% CI, 1.55-2.70]), ataxia and trouble walking (PR, 2.06 [99% CI, 1.57-2.70]), and myoneural disorders (PR, 5.79 [99% CI, 4.06-8.25]) were significantly higher compared with those with a negative test. Among hospitalized persons younger than 20 years, the prevalence of new type 2 diabetes (PR, 2.14 [99% CI, 1.13-4.06]) was higher among those with a positive test compared with those with a negative test; however, the prevalence difference was smaller than 1%.

In this cohort study, using EHR data from 168 701 persons aged 20 years or older and 26 665 aged younger than 20 years who tested positive for SARS-CoV-2, we observed that symptoms and conditions reported as possible sequelae of SARS-CoV-2 infection occurred infrequently (≤11% prevalence) in nonventilated persons with medical encounters at 31 to 150 days after SARS-CoV-2 testing. New diagnoses of shortness of breath were significantly higher among persons with a positive test compared with those with a negative test in both age cohorts. Diagnoses of fatigue, cognitive dysfunction, sleep disorders, heart rate abnormalities, myoneural disorders, and type 2 diabetes among hospitalized persons aged 20 years or older and heart rate abnormalities and type 2 diabetes among hospitalized persons younger than 20 years were more common for those with a positive vs negative test. Although new symptoms and conditions occurred infrequently, applying the proportions of these rare events to the millions of persons infected with SARS-CoV-2 means that a substantial number might experience new symptoms and conditions after their acute illness. In addition, new symptoms can be long-lasting¹² and involve chronic conditions, such as type 2 diabetes. Increasing awareness of new symptoms and conditions among health care professionals and health systems is paramount to meet the needs of patients with prolonged or chronic sequelae of SARS-CoV-2 infection.

The prevalence of new symptoms and condition diagnoses following a positive test was higher among hospitalized persons compared with their nonhospitalized counterparts; new symptoms and conditions also were most prevalent among ventilated persons aged 20 years or older. Although the primary reason for hospitalization following SARS-CoV-2 testing was not available for this study, this study suggests that long-term symptoms and conditions may be more common with increased SARS-CoV-2 infection severity. Previous reports have highlighted a higher incidence of neuropsychiatric conditions and diabetes among hospitalized and ventilated adults with SARS-CoV-2 infection compared with those without.^4,5,13 The higher prevalence of anxiety and depression, myopathies, peripheral nerve disorders, and weight loss among ventilated persons highlights the role of post–intensive care syndrome in compounding the effects of SARS-CoV-2 infection.¹⁴ Symptom prevalence among nonhospitalized persons aged 20 years or older in our study was similar to other studies that have found a prevalence of selected symptoms of less than 10% in the 6 months following SARS-CoV-2 infection.⁶ Similar to our findings, no increased risk for new symptoms and conditions in the postacute phase, other than shortness of breath and venous thromboembolism, was observed in a large population-based study of nonhospitalized adults with SARS-CoV-2 infection.¹⁵ Whether these differences were due to underlying demographic differences in patients infected with SARS-CoV-2 (compared with those who were not), differences in the types of care received (eg, aggressive medical treatment when hospitalized), or direct effects of the SARS-CoV-2 virus is unknown. Longitudinal studies extending beyond 6 months after acute infection will be important to quantify the true impact of COVID-19 sequelae.

The prevalence of new-onset diabetes in our population may be overestimated, as hyperglycemia or receipt of glucocorticoids during the acute event might have unmasked diabetes that was only coded during a follow-up visit when persistent hyperglycemia was recognized. Nevertheless, there is evidence of direct cytopathic effects of SARS-CoV-2 infection in pancreatic tissue and increases in insulin resistance owing to inflammation.^16,17 Future studies should evaluate whether evidence for type 2 diabetes persists beyond the time period that could be affected by treatments.

Data regarding postacute symptoms and conditions in children are scarce. Studies have highlighted prolonged fatigue, headaches, sleep disturbances, cognitive dysfunction, and sensory disturbances in this population.^9,10,18 In our study, shortness of breath and nonspecific heart rate abnormalities were significantly higher among hospitalized persons younger than 20 years who tested positive for SARS-CoV-2 compared with those who tested negative, whereas minimal differences were observed between these groups when not hospitalized. Our data suggest that, similar to adults, the risk of selected new symptoms and conditions among nonhospitalized persons younger than 20 years may be low.

There are several limitations to our study. First, use of diagnostic codes may have led to an underestimation of prevalent symptoms and conditions, as we only observed patients when they presented for care, and coding for illnesses and symptoms may not be comprehensive during visits. In addition, we defined ICD-10-CM code groups broadly based on reviews of the medical literature and clinical knowledge, which may have resulted in including symptoms and conditions unrelated to SARS-CoV-2 infection. Second, we defined new symptoms or conditions as the occurrence of 1 diagnostic code during the 31- to 150-day period to enhance sensitivity for detection of possible SARS-CoV-2 sequelae. Using 2 or more occurrences of diagnostic codes may have enhanced specificity; however, given that patients may have not had the opportunity for additional visits during the 120-day follow-up period, we selected enhanced sensitivity. Third, we were unable to determine whether new symptoms and conditions were due to effects of SARS-CoV-2 infection or the result of recognition of previously undiagnosed health problems. Fourth, we were unable to control for differences in demographic characteristics, underlying conditions, and follow-up between persons with positive and negative tests. Fifth, we used care setting following SARS-CoV-2 testing as a proxy for COVID-19 severity, which may have resulted in misclassification if persons with a positive test were hospitalized for reasons other than acute SARS-CoV-2 illness. Sixth, we were unable to ascertain whether SARS-CoV-2 testing was conducted because of symptoms, as a part of routine surveillance, or for travel purposes. Hospitalized persons who tested negative for SARS-CoV-2 included those hospitalized for nonviral illness (eg, pregnancy, trauma, chronic conditions) and may have biased our estimates, as there may be multiple etiologies for symptoms and conditions examined in this report. Finally, not all individuals in the study had 150 days of follow-up, which may have led to an underestimation of new symptoms and conditions among persons who tested positive in the months of November and December 2020.

In this cohort study, we determined the prevalence of select new symptoms and conditions in a large health care–seeking population, highlighting symptoms and conditions that occurred more frequently among persons with a positive compared to those with a negative SARS-CoV-2 test, and found an increase in symptoms and conditions in those with higher severity of acute SARS-CoV-2 infection. These estimates highlight the need for health care professionals and patients to monitor for development of new symptoms and conditions beyond the first month after SARS-CoV-2 infection, particularly for individuals who required hospitalization for acute COVID-19.

Accepted for Publication: December 14, 2021.

Published: February 4, 2022. doi:10.1001/jamanetworkopen.2021.47053

Open Access: This is an open access article distributed under the terms of the CC-BY License. © 2022 Hernandez-Romieu AC et al. JAMA Network Open.

Corresponding Authors: Jason Block, MD, MPH, Department of Population Medicine, Harvard Pilgrim Health Care Institute/Harvard Medical School, 401 Park Dr, Ste 401, Boston, MA 02215 (jblock1@partners.org); Alfonso C. Hernandez-Romieu, MD, MPH, Epidemic Intelligence Service, Centers for Disease Control and Prevention, 1600 Clifton Rd NE, Building 16, Atlanta, GA 30333 (vvb7@cdc.gov; ahernandez4@cdc.gov).

Author Contributions: Drs Hernandez-Romieu and Block had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

Concept and design: Hernandez-Romieu, Carton, Azziz-Baumgartner, Garrett, Bailey, Draper, Mayer, Trick, Wanga, Chevinsky, Jackson, Goodman, Cope, Gundlapalli, Block.

Acquisition, analysis, or interpretation of data: Hernandez-Romieu, Carton, Saydah, Azziz-Baumgartner, Boehmer, Bailey, Cowell, Draper, Mayer, Nagavedu, Puro, Rasmussen, Wanga, Chevinsky, Jackson, Cope, Gundlapalli, Block.

Drafting of the manuscript: Hernandez-Romieu, Azziz-Baumgartner, Mayer, Gundlapalli, Block.

Critical revision of the manuscript for important intellectual content: Hernandez-Romieu, Carton, Saydah, Azziz-Baumgartner, Boehmer, Garrett, Bailey, Cowell, Draper, Mayer, Nagavedu, Puro, Rasmussen, Trick, Wanga, Chevinsky, Jackson, Goodman, Cope, Gundlapalli.

Statistical analysis: Hernandez-Romieu, Mayer, Wanga, Block.

Obtained funding: Carton, Boehmer, Garrett, Gundlapalli, Block.

Administrative, technical, or material support: Hernandez-Romieu, Carton, Azziz-Baumgartner, Boehmer, Garrett, Bailey, Cowell, Draper, Mayer, Nagavedu, Puro, Chevinsky, Gundlapalli, Block.

Supervision: Carton, Saydah, Azziz-Baumgartner, Mayer, Jackson, Goodman, Cope, Gundlapalli, Block.

Conflict of Interest Disclosures: Dr Trick reported receiving grants from the Centers for Disease Control and Prevention (CDC) during the conduct of the study. No other disclosures were reported.

Funding/Support: This publication was supported by Cooperative Agreement number 6-NU38OT000316, funded by the CDC.

Role of the Funder/Sponsor: Scientists from the CDC, the funder for this project, were involved in the design and conduct of the study; analysis and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication. Several of the authors of this study are affiliated with the CDC.

Disclaimer: The findings and conclusions in this report are those of the authors and do not necessarily represent the official position of the CDC. Its contents are solely the responsibility of the authors and do not necessarily represent the official views of the CDC or the Department of Health and Human Services.