Clinical characteristics and prognostic factors in elderly patients with COVID-19

Background: The COVID-19 pandemic posed tremendous threats to the world. Elderly patients were among the high-risk population, and apt to experience worse outcomes. Methods: Elderly patients (age ≥ 60 years old) were enrolled from January 28 to February 29, 2020, in Tongji Hospital, one of designated COVID-19 medical centers in Wuhan, China. A retrospective study was performed to describe clinical characteristics and outcomes of elderly COVID-19 patients. COX regression was used to analyze predictors for 28-day mortality. Linear regression models were constructed to analyze factors associated with length of hospital stay (LOS). Result: A total of 186 elderly patients (aged 70.4 ± 7.1 years, 95 males (51.6%)) were enrolled, 120 patients (64.5%) were severe or critical type, and mortality rate was 16.1%. Patients in non-survival group had a higher smoking rate, more symptoms of dyspnea, lab results indicative of poorer health. Age (HR 1.128, 95% CI 1.066-1.194), lymphocyte count (HR 0.261, 95% CI 0.073-0.930), LDH (HR 1.003, 95% CI 1.002-1.005), procalcitonin (HR 1.061, 95% CI 1.002-1.125), and qSOFA (HR 3.162, 95% CI 1.646-6.072) were independently associated with 28-day mortality. CURB-65 plus LDH on admission were predictors of mortality by ROC analysis (AUROC=0.891). Among surviving patients, consolidation on CTs, elevated ferritin level and neutrophil count were associated with increased LOS. Conclusion: High incidence of comorbidities and mortality were observed in elderly patients. Decreased lymphocyte, older age, higher qSOFA score, procalcitonin and LDH levels were independent factors associated with mortality, CURB-65 plus LDH could be a predictive model of fatal outcome. Consolidation on CTs, elevated ferritin and neutrophil level correlated with increased LOS. Further prospective studies should be performed to test our ndings and

like ARDS, cardiac injury, or acute kidney injury. Until now, few studies explored prognostic factors of elderly patients. Wang [6] showed that comorbidities including cardiovascular disease, chronic obstructive pulmonary disease (COPD) were predictive of fatal outcomes. Chen [7] illustrated that males, comorbidities, and time from disease onset to hospitalization were associated with death in older patients. But we still needed a predictive tool of mortality in clinical practice, especially early in-hospital assessment. The pandemic consumes huge healthcare resources, especially for elderly patients with long-time hospital stay. Prolonged length of hospital stay (LOS) could lead to in-hospital complications like secondary infection, venous thromboembolism, etc., which also worse the prognosis. But studies on in uencing factors of LOS were de cient.
This retrospective study aimed to describe the clinical and treatment courses of elderly COVID-19 patients and explore predictors associated with 28-day mortality and length of hospital stay (LOS).

Study design and participants
Tongji Hospital, one of the largest comprehensive medical centers in Wuhan, was designated as a speci c hospital for COVID-19 patients, especially severe patients, during the outbreak. Isolation wards were managed by supporting medical teams from across China under coordination by the government, due to insu cient medical resources.
In this retrospective study, patients were recruited from six isolation wards of Tongji Hospital, which were managed by medical teams from Peking University Third Hospital, Beijing Hospital and their colleagues.
From January 28, to February 29, 2020, all consecutive patients over age 60 and con rmed to have COVID-19 were enrolled. According to "the Diagnosis and Treatment Protocol for Novel Coronavirus Pneumonia (Trial Version 7)" by the National Health Commission of China [1], con rmed cases must have one of the following virologic or serological evidences: positive SARS-Cov-2 nucleic acid by real-time uorescent RT-PCR or positive serum speci c IgM and IgG antibodies of SARS-Cov-2. The severity of COVID-19 was classi ed according to the guidelines as follows: (1) mild type: clinical symptoms were mild without pneumonia, (2) moderate type: fever and other symptoms presented, with pneumonia on chest computed tomography (CT), (3) severe type: any of the following criteria was met: respiratory rate ≥ 30 per minute, oxygen saturation (S P O 2 ) ≤ 93% on room air at rest, arterial partial pressure of oxygen (PaO 2 )/FiO 2 ratio ≤ 300mmHg, (4) critical type: any of the following conditions presented: respiratory failure that required mechanical ventilation, septic shock or other organ failure that required ICU monitoring.

Data collection
Data were collected from electronic medical records by a trained team of physicians. The parameters consisted of demographic, clinical characteristics, comorbidities, laboratory tests, radiological ndings, treatment, and outcome. Comorbidities were mainly determined by past medical history, but diabetes was also con rmed by hemoglobin A1c (HbA1c) ≥ 6.5%. Chronic respiratory diseases (CRD) were con rmed not only by history of COPD, chronic bronchitis, bronchiectasis, and asthma, but also by existence of bilateral emphysema and bullae on chest CTs. Laboratory tests consisted of blood routine, biochemistry, coagulation, cardiac markers, infection-related indexes, and cytokines. Chest CTs were evaluated by two pulmonary specialists separately. The results were described as the number of lobes in ltrated, the distribution pattern (peripheral, random, or diffuse) and characteristics of pulmonary lesions (groundglass opacities, consolidation, patchy shadow, and pleural effusion). CURB-65 (short for assessment of Consciousness, serum Urea nitrogen, Respiratory rate, Blood pressure, and age ≥ 65 years) and qSOFA (quick Sequential Organ Failure Assessment) were assessed on admission to quickly evaluate severity of disease. Activities of daily living (ADL) were assessed by Barthel Index score on admission for describing functional abilities [8]. Medications, such as antivirals, antibiotics, and corticosteroids, and advanced supportive procedures, such as mechanical ventilation, were recorded.

Outcome
Follow-up continued for the hospitalized patients until April 4, 2020. The primary measured outcome was survival or death after 28 days. The secondary outcome was LOS until discharge or endpoint of the study (a study ow diagram is shown in Supplement 1). According to the Chinese guidelines [1], patients could be discharged when symptoms improved signi cantly, with obvious absorption of in ltrates on chest imaging and negative results for at least two consecutive tests of SARS-CoV-2 nucleic acids.

Statistical analysis
Continuous variables were expressed as mean ± SD or median (IQR) with or without normal distribution. Categorical variables were reported as frequencies, and percentages. The independent sample T test or the Mann-Whitney Test were used for comparison between continuous data, and the chi-square test was used for comparison between categorical data. Univariate and multivariate Cox regression analysis, and ROC analysis were used to analyze risk factors of 28-day mortality. Spearman correlation test and multivariate linear regression were used to analyze factors associated with LOS. A two-sided p < 0.05 was considered statistically signi cant, and IBM SPSS Statistics (version 22.0) was used for all data analyses.

Results
Demographic data and clinical characteristics A total of 186 patients with con rmed COVID-19 were enrolled in the study, 64.5% were classi ed as severe or critical type on admission (shown in Table 1). The mean age was 70.4 ± 7.1 years old, and 96 patients (51.6%) were male. The mortality rate of patients over age 75 was much higher than that of patients aged 60-74 (28.9% (13/45) vs 12.1% (17/141), p = 0.008). Lower respiratory rates and higher S P O 2 levels were found among survivors, but peak temperatures were similar between the two groups.

Laboratory ndings and chest imaging
On admission, lymphocytopenia (count <1.0 × 10 9 /L) was found in 58.9% of patients. As shown in Table  2, signi cantly higher neutrophil count or lower lymphocyte levels were found among non-survivors. sensitivity Troponin I (hsTnI, 16.1 vs 5.3 pg/ml) and N-terminal pro-brain natriuretic peptide (NT-proBNP, 616 vs 224 pg/ml), were higher in non-survival group, but only slightly exceeded the normal range. All p values above were < 0.05.

Treatments and outcomes
As shown in Table 1 Table 3. According to ROC analysis, the cut-off value of procalcitonin and qSOFA were 0.09 ug/ml and 0.5 scores, which having little clinical signi cance. The optimal cut-off of LDH, CURB-65 and lymphocyte count were 360.5U/L, 1 score and 0.665×10 9 /L, with the AUROC being 0.838, 0.775, 0.720, respectively. Combined indexes of CURB-65, LDH or lymphocyte count were used to obtain a more accurate prognostic value.

Predictive factors for the LOS in elderly patients
The correlations of demographic and clinical characteristics with the LOS were analyzed among the surviving 156 patients, which is shown in were associated with an increased LOS.

Discussion
This was an observational study of elderly COVID-19 patients, in which we successively assessed factors associated with 28-day mortality and length of hospital stay. The patients were enrolled consecutively from January 28 to February 29, which was an intense period of the epidemic in Wuhan, and was relatively representative of the similar situations many countries now faced with. They were carefully treated and followed up until nearly all patients had reached to endpoint of discharge or death. A few studies analyzed characteristics and fatal outcomes of elderly COVID-19 patients [6,7], but little attention was received to LOS, which was also important for prognosis. To our knowledge, this was the rst study clarify factors associated with LOS in elderly patients.
According to a nationwide epidemiological data of whole population, 18.5% of con rmed COVID-19 patients in China were classi ed as severe/critical type, with a crude fatality rate of 2.3% [5]. In this study, among elderly COVID-19 patients, 64.5% were classi ed as severe or critical type, and the mortality rate was 16.1%, which is higher than previous studies. This was not only owing to the characteristics of elderly patients, but also to the fact that Tongji hospital was designated as the center for severe COVID-19 patients in Wuhan. 75.3% patients had at least one comorbidity, which was a higher rate than that of COVID-19 patients overall [9][10][11][12]. 32.8% of patients were diagnosed with diabetes in this study, which was higher than that in previous data [6,13]. It should be noted that most studies acquired diabetes information only from the past medical history, which might underestimate of its actual prevalence in COVID-19 patients. In this study, diabetes was also con rmed by HbA1c ≥ 6.5% according to the American Diabetes Association 2010 criteria [14,15]. Hence, 22 more patients were newly found to combined with diabetes. The ADL scores on admission were higher among survivors, which indicated better functional abilities in this group. But standard deviation was relatively large in both groups, limiting the reliability of the result. Mean ADL score among non-survivors was less than 60, indicating that these patients could not ful ll daily activities and relied highly on caregivers. In this study, the time before admission was longer than that in other large studies [12], which might be attributed to the severe epidemic situation and relatively limited medical resources during this period in Wuhan. It has been reported that time from disease onset to hospitalization was an independent risk factor for mortality [7]. However, in our study, the time before admission in non-survival group was shorter than that of the survival group, possible reason being priority treatment of the more severe patients.
Previous studies had found that patients with COVID-19 were often associated with decreased lymphocyte count and increased in ammatory indicators [9,16]. In COVID-19 patients with severe and fatal diseases, biomarkers of in ammation, cardiac and muscle injury, liver function, kidney function and coagulation function were signi cantly elevated, and IL-6, IL-10, and ferritin were strong predictors for severe disease [17]. Li et al. reported [19] that incidences of consolidation, crazy-paving patterns, and bronchial wall thickening in severe/critical patients were signi cantly higher than those in the moderate patients. Another study showed that the median CT score of the mortality group was higher compared with that of the survival group, with higher frequencies of consolidation and air bronchogram [20]. In this study, manifestations on CTs could not predict mortality, but might affect the LOS. More affected lobes, diffuse distribution and existence of consolidation indicated severe injury of lungs, which needed longer duration of treatment and LOS.
At present, a few studies had discussed the prognosis of elderly COVID-19 patients [6,7,21]. This study found that increased age, LDH, procalcitonin levels, and qSOFA scores and reduced lymphocyte were predictors of death. Even among the older adults, age was still a risk factor of death. Mortality rate was higher in older patients (≥ 75 years) than those aged 60 to 74. LDH existed in the heart, liver, skeletal muscle, and red blood cells, and increased LDH level indicates damage of these tissue or cells. On other hand, LDH was also an in ammatory biomarker and, combined with other factors like ferritin, could be strong discriminators for severe COVID-19 or ARDS [17,18]. Procalcitonin had emerged as an indicator of bacterial infection and helped initiate or discontinue antibiotic therapy [22,23]. In elderly patients with community acquired pneumonia, procalcitonin allowed early detection of severe courses and initiation of suitable treatment [24]. The qSOFA score was rst used to identify risk of patients with sepsis for a poor outcome, it only required a few parameters and vital signs, which made it easy to operate in isolation wards. Recent study showed that the prognostic performance of qSOFA for community-acquired pneumonia was similar with that of CURB-65 [25]. Studies had shown that T lymphocyte reduced in COVID-19 [26] and speculated that coronavirus acted directly on ACE2 receptors of lymphocyte to cause damage [27], or through virus-mediated lymphocyte apoptosis by activating in ammatory responses, similar with SARS [28]. Immune system was weakened in patients with severe lymphocytopenia, they were apt to suffer secondary infection and poor prognosis [29]. The optimal cut-off value of procalcitonin or qSOFA had little clinical signi cance by ROC analysis, making neither of these two indexes ideal indicators of death. The areas under the ROC for LDH, CURB-65 and lymphocyte count were all higher than 0.7, and LDH was the highest of the single factor index in predicting death. CURB-65 was another useful assessment in predicting 30-day mortality of CAP, it contained ve parameters and was easy to operate. But its predictive value in viral pneumonia were relatively limited [30]. CURB-65 plus LDH might be a stronger predictive index of 28-day mortality in elderly COVID-19 patients, and can be easily obtained on admission, with high sensitivity and speci city.
Based on result from a Fangcang shelter hospital, which mainly offered medical care for mild, moderate or asymptomatic COVID-19 patients during the outbreak, patients with fever before admission and bilateral pneumonia had signi cantly longer LOS [31]. In this study, consolidation of chest CTs, increased ferritin level and neutrophil count were independent predictors of longer LOS. As mentioned above, consolidation on CTs, higher ferritin, and neutrophil were all associated with severity of diseases [17][18][19].
Severe patients usually needed more advanced treatment and longer recovery times, which were the main reasons for prolonged hospital stay. On the other hand, consolidation indicated lling of airspace on CTs, which was related to hypoxemia and sometimes required mechanical ventilation. Patients with elevated ferritin or neutrophil might have uncontrolled underlying in ammation and need anti-bacterial treatment, which would have prolonged LOS.

Limitations
There were several limitations of the study. First, data might be incomplete due to its retrospective design, and we could not testify effectiveness of medication or supportive treatment. Second, it was a singlecenter study with limited number of patients, and Tongji hospital was a designated center mainly for severe COVID-19 patients in Wuhan. Selected bias would be inevitable. Third, due to rare cases nowadays in China, we could not perform an external validation to testify our predictive model.

Conclusion
High incidence of comorbidities and high mortality rates were seen in elderly COVID-19 patients.

Declarations
Ethics approval and consent to participate The study was approved by the Ethics Committee of Peking University Third Hospital (IRB00006761-M2020060), and written informed consent was waived because of its retrospective design.

Consent for publication
Not applicable.

Availability of data and materials
The datasets used for the current study are available from the corresponding authors upon reasonable request.

Competing interests
The authors declare that they have no competing interests Funding Ning Shen is currently receiving a grant (#2018YFC1311900) from National Key R&D Program of China.
The funding body had no role in the design of the study and collection, analysis, interpretation of data or in writing the manuscript.
Authors' contributions NS and YmL contributed to the conception and design of the study. QC contributed to the design of the study, data acquisition, analysis, and interpretation of the results. YxL contributed to the design of the study, analysis, and interpretation. YjL contributed to the design of the study, data acquisition, and interpretation. LT contributed to statistical analysis of the study. HW and ML contributed to data acquisition and analysis. QG contributed to the design of the study. All authors read and approved the nal manuscript.