Characteristics and Risk Factors of Hospitalized and Nonhospitalized COVID-19 Patients, Atlanta, Georgia, USA, March–April 2020

We compared the characteristics of hospitalized and nonhospitalized patients who had coronavirus disease in Atlanta, Georgia, USA. We found that risk for hospitalization increased with a patient’s age and number of concurrent conditions. We also found a potential association between hospitalization and high hemoglobin A1c levels in persons with diabetes.

I nformation about care-seeking behavior, symptom duration, and risk factors for progression to severe illness in nonhospitalized patients with coronavirus disease (COVID-19) aids in resource planning, disease identifi cation, risk stratifi cation, and clinical management of nonhospitalized patients (1)(2)(3)(4)(5)(6). We built on a previous analysis comparing hospitalized and nonhospitalized CO-VID-19 patients, which found that hospitalized patients were more likely to be >65 years of age, men, Black, diabetic, or obese (7). We describe symptom patterns, duration of illness, and care-seeking behavior among nonhospitalized patients and explore the relationships between hospitalization and the number, control, and interaction of concurrent medical conditions and age. We defi ned control as how well a disease is managed in the patient, as measured by hemoglobin A1c levels in diabetics, number of classes of hypertension medication being taken by patients with hypertension, and BMI among patients with obesity.

The Study
We enrolled hospitalized and nonhospitalized patients >18 years of age with laboratory-confi rmed COVID-19 (defi ned as a positive real-time reverse transcription PCR result for severe acute respiratory syndrome coronavirus 2) treated at 6 acute care hospitals and outpatient clinics affi liated with a single academic hospital system in the Atlanta, Georgia, USA, metropolitan area during March 1-April 7, 2020, as previously described (7). In this investigation, we compared characteristics and symptoms of hospitalized and nonhospitalized persons using χ 2 , Fisher exact, or t-test as appropriate.
We conducted univariable and multivariable logistic regression analyses to explore the associations between age group and number of underlying conditions on risk for hospitalization. We conducted separate analyses to model the associations and interactions of diabetes, hypertension, and obesity with hospitalization for COVID-19. We then used univariable and multivariable logistic regression to investigate whether the severity or control of concurrent conditions was associated with increased risk for hospitalization. We modeled the association between body mass index (BMI) and hospitalization. Further, we investigated whether use of multiple classes of hypertension medication was associated with hospitalization among patients with hypertension and the association of elevated levels (>7%) of hemoglobin A1c (HbA1c) and hospitalization among patients with diabetes. This level was chosen because a value <7% is considered an indicator of adequate blood glucose control in patients with diabetes (8). Multivariable models were adjusted for characteristics previously associated with hospitalization in these populations (7) (Appendix, https://wwwnc.cdc.gov/EID/ article/27/4/20-4709-App1.pdf).
Of 311 nonhospitalized patients, 135 (43%) had their first contact with the healthcare system for their COVID-19 illness on a telephone triage line, 23 (7%) at the emergency department, and 141 (45%) at an ambulatory care clinic (Table 1). Of nonhospitalized patients, 85% sought in-person care (i.e., ambulatory care, emergency department, or urgent care) a single time for their COVID-19 illness. A subset of 188 non-

Figure.
Symptoms of coronavirus disease among hospitalized and nonhospitalized patients, Atlanta, Georgia, USA, 2020. Gastrointestinal symptoms include vomiting, nausea, diarrhea, and abdominal pain. Upper respiratory symptoms include sore throat, rhinorrhea, and nasal congestion. *p>0.01. Among patients with hypertension, the odds of hospitalization demonstrated a possible dose-dependent increase among patients taking multiple classes of hypertension medications; however, precision of estimates was limited by small sample size (Table 2). Among patients with diabetes, those with a recent HbA1c score >7% had an increased risk for hospitalization (aOR 4.1, 95% CI 0.9-19.1); however, precision of estimates was limited by small sample size. Among obese patients (BMI >30), BMI was not associated with increasing odds of hospitalization (Table 2). In the multivariable analyses, we did not detect significant additive or multiplicative interaction between diabetes and obesity, hypertension and obesity, or hypertension and diabetes (Appendix Table 4).

Conclusions
Symptoms lasting >21 days were common among nonhospitalized patients in this investigation; however, <20% of these patients had >1 in-person healthcare visit for COVID-19 during acute illness. These extended symptom durations, in conjunction with limited care-seeking behavior, suggest that many mildly ill COVID-19 patients can self-manage their symptoms. Because telemedicine was the second most common healthcare delivery method in our investigation, we hypothesize that it might have provided ongoing patient support and decreased the need for in-person healthcare visits (9). These findings can assist healthcare providers with anticipatory guidance for patients and caregivers and can inform decisions about allocation of resources for healthcare delivery. We found that age and number of underlying conditions were associated with a dose-dependent increase in likelihood of hospitalization. Elderly COVID-19 patients frequently have multiple conditions that increase risk for hospitalization and serious infection (10). However, we did not find a significant additive or multiplicative interaction between the 3 most common underlying conditions among study participants: hypertension, diabetes, and obesity.
We hypothesized that degree of control of underlying conditions would affect risk for hospitalization. We found that the aORs for hospitalization were higher among patients with diabetes who had elevated mean levels of HbA1c and among patients with hypertension taking an increasing number of hypertension medications. Although not statistically significant, these findings may suggest an association between the management of concurrent conditions and COVID-19 disease severity. Despite obesity's association with increased risk for severe illness and death from COVID-19 (11,12), we did not find an increasing risk for hospitalization with increasing BMI among persons with obesity.
A limitation of our study is that, because of small sample sizes, our analyses might have lacked power to detect a significant association between degree of control of underlying conditions and hospitalization. In addition, our sample comprised patients at a single hospital system during a limited timeframe, and thus our results might not be generalizable to other populations. Because this hospital system prioritized certain persons (e.g., older patients, patients with underlying conditions, and healthcare personnel) for outpatient SARS-COV-2 testing, these persons might be overrepresented among the nonhospitalized patients in our sample. We were also not able to assess symptom resolution among all patients during the timeframe of this investigation and therefore might not have accounted for all follow-up healthcare visits for COVID-19.
In conclusion, although many nonhospitalized patients in this study reported symptoms lasting >21 days, most cases of COVID-19 among nonhospitalized patients were managed with a single ambulatory care visit and telehealth follow-up appointments. Patients of increasing age, with a greater number of underlying conditions, and with poor management of those conditions might be at higher risk for hospitalization and severe disease from COVID-19.