Post-acute Sequelae of SARS-CoV-2 Infection Among Adults Aged ≥18 Years

Long Beach, California, April 1-December 10, 2020

Kyle Yomogida; Sophie Zhu; Francesca Rubino, MSc; Wilma Figueroa, MPH; Nora Balanji, MPH; Emily Holman, MSc


Morbidity and Mortality Weekly Report. 2021;70(37):1274-1277. 

In This Article

Abstract and Introduction


Post-acute sequelae of COVID-19, also known as "long COVID," is used to describe the long-term symptoms that might be experienced weeks to months after primary infection with SARS-CoV-2, the virus that causes COVID-19. Among persons with a previous COVID-19 diagnosis, estimates of the prevalence of sequelae range from 5% among nonhospitalized persons to 80% among hospitalized persons.[1,2] Studies have analyzed the aftereffects of COVID-19, but few have assessed the demographic characteristics associated with long COVID.[3,4] Health disparities resulting from pervasive structural and socioeconomic barriers in the U.S. health care system might contribute to differences in these effects and might continue to exacerbate existing inequities.[5] To identify trends in post-acute sequelae, the Long Beach Department of Health and Human Services (LBDHHS) interviewed a random sample of 366 persons aged ≥18 years who received a positive SARS-CoV-2 test result during April 1–December 10, 2020. One third of the persons interviewed reported having at least one symptom 2 months after their positive test result, with higher odds of sequelae among persons aged 40–54 years, females, and those with preexisting conditions. Black or African American (Black) participants had higher odds of reporting dyspnea and myalgia/arthralgia compared with other racial/ethnic groups. Persons who were aged ≥40 years, female, Black, or who reported known preexisting conditions also reported higher numbers of distinct sequelae. As the number of recovered COVID-19 patients increases, monitoring the prevalence of post-acute sequelae among larger cohorts in diverse populations will be necessary to understand and manage this condition. Identification of groups disproportionately affected by post-acute COVID-19 sequelae can help develop efforts to prioritize preventions and treatment strategies, including vaccination of groups at higher risk for these long-term sequelae, and access to testing and care for post-acute sequelae.

Data were collected by LBDHHS (California Code of Regulations, Title 17) under the authority of the Long Beach City Health Officer during case investigation and follow-up. Among 28,594 Long Beach residents aged ≥18 years who received a positive SARS-CoV-2 reverse transcription–polymerase chain reaction (RT-PCR) test result during April 1–December 10, 2020, approximately 3% (791) were randomly selected for follow-up interviews. Persons with an intellectual or developmental disability or who had died were excluded from the study. During the first round of sampling, 400 persons with positive test results during April 1–August 26, 2020 were randomly selected. Because of the winter surge in cases, a subsequent round of sampling was conducted during August 27–December 10, 2020, in which 391 persons were selected. The second round of sampling maintained the same proportion of selected persons to all confirmed cases as the first round. Overall, 366 (46.3%) of the 791 selected persons agreed to be interviewed. Participants were interviewed during October 1, 2020–March 3, 2021 by telephone at least 2 months after the positive test result (median = 202 days; range = 78–368 days) using a standardized survey instrument. Interviews were conducted in English, Spanish, or Khmer. Questions were adapted from the 30-item California Reportable Disease Information Exchange COVID-19 case investigation questionnaire. Interviewers recorded questionnaire responses in Veoci, a secured virtual emergency operation center software application.

Multivariable logistic regression was used to assess associations between symptoms experienced 2 months after receiving a positive test result and participant characteristics (demographics and preexisting conditions) by calculating adjusted odds ratios (aORs) for having any sequalae and each of the most common sequelae. Multivariable Poisson regression was used to assess adjusted incidence rate ratios (aIRRs) for the number of reported symptoms. Racial and ethnic groups were combined, and persons identifying as both White and a racial minority group were categorized by their respective racial minority group. Because of limited availability of diagnostic and screening testing during the sampling phase, persons who experienced symptoms within 14 days before and 10 days after testing were classified as being symptomatic at the time of diagnosis.§ Disease severity at diagnosis was classified according to the National Institutes of Health groupings as mild, moderate, or severe.[6] Because small sample sizes precluded analysis of specific preexisting conditions, a yes/no variable was used to indicate any preexisting condition. Persons were given the opportunity to report symptoms not covered in the survey. Responses with related symptoms were combined for analysis.** All analyses were conducted in R (version 4.0.5; R Foundation); p-values <0.05 were considered statistically significant. Collection and analysis of human surveillance data fall under routine public health activities in the State of California and were exempt from Institutional Review Board review.

Among the 366 participants, the largest percentages were aged 25–39 years (144; 39%), female (207; 57%), and Hispanic/Latino (240; 66%) (Table 1). These were elevated relative to the general population because persons aged 25–39 years account for approximately 25% of the population, females for 50%, and Hispanic/Latino persons for 40%. Approximately one half (46%) of participants reported having a chronic preexisting condition before their COVID-19 diagnosis. Nineteen (5%) participants were hospitalized because of COVID-19. Participants reported an average of 5.26 symptoms (standard deviation SD = 3.82), and most (92.3%) experienced at least one symptom related to COVID-19 around the time of testing. Ageusia, parosmia/anosmia, myalgia/arthralgia, fatigue, and headache, were reported by 54.1%, 50.3%, 51.4%, 48.4%, and 46.4% of participants, respectively (Table 2). Two months after a positive SARS-CoV-2 test result, 128 (35.0%) participants reported an average of 1.30 (SD = 2.40) symptoms. Participants reported fatigue (16.9%), ageusia (12.8%), parosmia/anosmia (12.6%), dyspnea (12.8%), and myalgia/arthralgia (10.9%). The frequency of symptoms reported by persons 2 months after receiving a positive SARS-CoV-2 test result varied with the severity of illness at diagnosis; 55.5% reported severe/critical symptoms, 52.6% reported moderate symptoms, 29% reported mild symptoms, and 3.7% reported no symptoms (Supplementary Table 1, Nearly one third of participants (115; 31.4%) had symptoms at the time of interview; fatigue (50; 13.7%), dyspnea (38; 10.4%), and parosmia (35; 9.6%) were most frequently reported.

In the multivariable regression model, the odds of experiencing symptoms 2 months after a positive SARS-CoV-2 test result were significantly higher among females (aOR = 2.83), persons with at least one preexisting condition (aOR = 2.17), and those aged 40–54 years (versus 25–39 years) (aOR = 1.86) (Table 3). Analyses of the four most common symptoms experienced 2 months after a positive test result (fatigue, dyspnea, parosmia/ageusia, and myalgia/arthralgia) revealed similar findings in persons with at least one preexisting condition, females, and aged ≥40 years. (Supplementary Table 2, Females had higher adjusted odds of ageusia/parosmia/anosmia and fatigue than did males; whereas persons aged ≥40 years had higher adjusted odds of both, as well as myalgia/arthralgia, compared with persons aged 18–39 years. Persons with at least one preexisting condition had higher adjusted odds of all four of the most common symptoms compared with persons without preexisting conditions. Among Black persons compared with other racial/ethnic groups, the aORs of experiencing dyspnea and myalgia/arthralgia 2 months after testing were 2.52 and 3.67 times higher, respectively.

More symptoms were reported by females (aIRR = 2.13; 95% CI = 1.40–3.25), persons with preexisting conditions (aIRR = 1.96; 95% CI = 1.32–2.91), persons aged ≥40 years (aIRR = 1.73; 95% CI = 1.14–2.63), and Black persons (aIRR = 1.95; CI = 1.02–3.73) than by males, persons without preexisting conditions, persons aged 25–39 years, and non-Hispanic White persons (Table 3).

*These authors contributed equally to this report.
Multivariable Poisson regression was used to assess adjusted incidence rate ratios (aIRRs) of the number of reported symptoms experienced 2 months after receipt of a positive SARS-CoV-2 test result across participant characteristics. Model assumptions for Poisson regression were assessed. Because of overdispersion, a quasi-Poisson link function was applied; all other assumptions were met. Forward stepwise selection was used to identify confounding variables using significance level of α = 0.05. Model selection was based upon minimizing the Akaike information criterion (an estimator of prediction error) as well as considering the public health significance of predictors with strong effect sizes.
§These cutoffs were determined based upon previous knowledge that persons infected with SARS-CoV-2 might yield a positive RT-PCR result 2 days before symptom onset and most persons receive negative test results 10–12 days after symptom onset.
Preexisting conditions measured in the questionnaire included diabetes, cardiovascular disease, hypertension, asthma, chronic lung disease, chronic kidney disease, chronic liver disease, stroke, neurologic or neurodevelopmental conditions, cancer, immunocompromising conditions, obesity, or history of smoking. Participants were allowed to report other preexisting conditions not directly assessed in the questionnaire. These included anxiety, depression, arthritis, allergies, hypothyroidism, chronic migraine headaches, fibromyalgia, and history of heart surgery.
**Responses related to joint pain (e.g., knee pain, joint pain, or bone aches) were categorized as arthralgia and grouped with myalgia for analysis. Responses related to alteration of sense of taste (ageusia) and changes in sense of smell (parosmia and anosmia) were grouped for analysis. Only symptoms reported by ≥40 participants were analyzed at the individual level.