3  Results

3.1  Patients
In all, 449 patients admitted to West District of Wuhan Union Hospital between January 25 and March 15, 2020 were included in the study. Subsequently, we excluded the following patients: 186 patients that did not have a decreased ALC upon hospital admission, 35 patients without relevant information in their medical records, 25 patients with duration of hospital stay less than 14 days or death during hospitalization, 25 patients with deteriorated condition requiring ICU care within 14 days after hospital admission, 12 patients aged > 80 years, 11 pregnant or lactating women, two patients with loss of consciousness during hospitalization, seven patients who were uncertain about their sleep quality, seven patients with at least two RCSQ scores between 50 and 70, and four patients with inconsistency between RCSQ and PSQI. Therefore, we included 135 patients (60 in the good-sleep group and 75 in the poor-sleep group) in the final analysis (Fig. 1).

3.2  Demographic and baseline characteristics, as well as laboratory parameters upon hospital admission and treatment
The median age of patients was 63 years (IQR, 55–69 years), and 57.8% of patients were men (Table 1 ). The median interval time from symptom onset to admission was 11.0 days (IQR, 9.0–13.5 days). A total of 83 (61.5%) patients had comorbidities, and 134 (99.3%) patients had bilateral distribution of patchy shadows or ground glass opacity on chest x-ray or CT imaging. There were no significant between-group differences in demographic and baseline characteristics (Table 1), as well as laboratory parameters upon hospital admission and treatment during hospitalization (Table 3).
Table 1 Demographics and baseline characteristics of patients with COVID-19.
Total (n = 135) Good sleep (n = 60) Poor sleep (n = 75) P value
Age, median (IQR), years 63 (55–69) 61 (51–68) 64 (56–70) 0.1126
 < 50 years 22 (16.3) 14 (23.3) 9 (12.0) 0.0818
 50–70 years 83 (61.5) 36 (60.0) 46 (61.3) 0.8747
 ≥ 70 years 30 (22.2) 10 (16.7) 20 (26.7) 0.1649
Sex:
 Male 78 (57.8) 40 (66.7) 38 (50.7) 0.0614
 Female 57 (42.2) 20 (33.3) 37 (49.3)
BMI, median (IQR), kg/m2 23.9 (22.3–25.6) 23.8 (22.5–25.3) 24.2 (22.0–25.7) 0.7886
Marital status:
 Married 129 (95.6) 57 (95.0) 72 (96.0) 0.7794
 Unmarried 3 (2.2) 2 (3.3) 1 (1.3) 0.8447
 Death of a spouse 3 (2.2) 1 (1.7) 2 (2.7) 0.8447
Education level:
 ≤ High school 96 (71.1) 39 (65.0) 57 (76.0) 0.1612
 University or college 32 (23.7) 17 (28.3) 15 (20.0) 0.2579
 Master or doctor 7 (5.2) 4 (6.7) 3 (4.0) 0.4875
Current smoking 4 (3.0) 0 (0) 4 (5.3) 0.1918
Comorbidities: 83 (61.5) 34 (56.7) 49 (65.3) 0.3038
 Cardiovascular and cerebrovascular diseases 42 (31.1) 19 (31.7) 23 (30.7) 0.9008
 Endocrine disease 38 (28.1) 16 (26.7) 22 (29.3) 0.7321
 Malignancy 13 (9.6) 4 (6.7) 9 (12.0) 0.2966
 Chronic respiratory disease 6 (4.4) 1 (1.7) 5 (6.7) 0.3268
 Gastrointestinal disease 4 (3.0) 1 (1.7) 3 (4.0) 0.7766
 Urologic diseases 2 (1.5) 0 (0) 2 (2.7) 0.5025
Onset of symptom to hospital admission, median (IQR), days 11.0 (9.0–13.5) 11.0 (9.0–15.0) 11.0 (8.5–13.0) 0.5954
Signs and symptoms at disease onset:
 Fever 123 (91.1) 53 (88.3) 70 (93.3) 0.3104
 Cough 92 (68.1) 38 (63.3) 54 (72.0) 0.2828
 Shortness of breath 75 (55.6) 31 (51.7) 44 (58.7) 0.4160
 Fatigue 68 (50.4) 32 (53.3) 36 (48.0) 0.5380
 Expectoration 37 (27.4) 14 (23.3) 23 (30.7) 0.3425
 Myalgia or arthralgia 34 (25.2) 15 (25.0) 19 (25.3) 0.9646
 Diarrhoea 18 (13.3) 7 (11.7) 11 (14.7) 0.6104
 Chills 18 (13.3) 8 (13.3) 10 (13.3) > 0.9999
 Nausea or Vomiting 15 (11.1) 9 (15.0) 6 (8.0) 0.1984
 Anorexia 12 (8.9) 7 (11.7) 5 (6.7) 0.3104
 Headache 10 (7.4) 5 (8.3) 5 (6.7) 0.9707
 Dizziness 5 (3.7) 2 (3.3) 3 (4.0) 0.7989
 Chest pain 5 (3.7) 2 (3.3) 3 (4.0) 0.7989
 Pharyngalgia 3 (2.2) 3 (5.0) 0 (0) 0.1704
 Rhinorrhoea 2 (1.5) 1 (1.7) 1 (1.3) > 0.9999
 Haemoptysis 2 (1.5) 2 (3.3) 0 (0) 0.1957
 Abdominal pain 2 (1.5) 2 (3.3) 0 (0) 0.1957
Vital signs on admission:
 Body tempreture, median (IQR), ℃ 36.9 (36.5–38.0) 36.9 (36.6–38.0) 36.8 (36.5–37.9) 0.4967
 Systolic pressure, median (IQR), mmHg 131.0 (120.0–143.0) 131.5 (117.5–145.0) 131.0 (120.0–142.5) 0.5040
 Heart rate, median (IQR), beat per minute 85.0 (75.5–100.0) 80.5 (74.5–99.0) 87.0 (77.5–100.5) 0.2444
 Respiratory rate, median (IQR), breaths per minute 20.0 (20.0–24.5) 20.0 (20.0–24.0) 21.0 (20.0–25.0) 0.2136
 Percutaneous oxygen saturation, median (IQR), %† 95 (92–96) 95 (91–96) 94 (92–96) 0.9816
Bilateral distribution of patchy shadows or ground glass opacity on chest x-ray or CT imaging: 134 (99.3) 59 (98.3) 75 (100%) 0.2618
Values are numbers (percentages) unless stated otherwise. Abbreviations: IQR, interquartile range; n, number. P values indicate differences between good-sleep and poor-sleep patients. P < 0.05 was considered statistically significant.
†Data were missing for the measurement of percutaneous oxygen saturation in eight patients (13.3%) of the good-sleep group, and nine patients (12.0%) of the poor-sleep group.

3.3  Self-reported sleep quality
Patients in the poor-sleep group had lower RCSQ scores in the first week (median, 40.0 [IQR, 30.0–50.0] vs 75.0 [IQR, 70.0–82.0], P < 0.0001), second week (median, 45.0 [IQR, 40.0–50.0] vs 80.0 [IQR, 75.5–86.0], P < 0.0001) and third week (median, 45.0 [IQR, 40.0–50.5] vs 80.0 [IQR, 75.0–88.0], P < 0.0001) than those in the good-sleep group (Table 2 ). The PSQI scores were higher in patients in the poor-sleep group than those in the good-sleep group (median, 12.0 [IQR, 11.0–14.0] vs 5.0 [IQR, 4.0–6.0], P < 0.0001). Among the etiological causes of poor-sleep during hospitalization, these included environmental factors, psychosocial factors, discomfort caused by the illness, and chronic insomnia for 85.3%, 60.0%, 56.0%, and 32.0% of patients, respectively (Table 2).
Table 2 Sleep quality assessment in patients with COVID-19.
Median (IQR) P value
Good sleep (n = 60) Poor sleep (n = 75)
RCSQ scores in the first week, median (IQR)
All ages 75.0 (70.0–82.0) 40.0 (30.0–50.0) < 0.0001
 < 50 years 76.0 (71.0–84.0) 44.5 (40.0–47.5) < 0.0001
 50–70 years 75.0 (70.0–80.5) 40.0 (30.0–50.0) < 0.0001
 ≥ 70 years 78.5 (72.0–80.0) 45.0 (32.5–50.0) < 0.0001
RCSQ scores in the second week, median (IQR)
All ages 80.0 (75.5–86.0) 45.0 (40.0–50.0) < 0.0001
 < 50 years 86.0 (76.0–87.0) 46.0 (42.5–50.0) < 0.0001
 50–70 years 80.0 (75.0–85.0) 45.0 (40.0–50.0) < 0.0001
 ≥ 70 years 78.5 (76.0–82.0) 44.0 (35.0–50.0) < 0.0001
RCSQ scores in the third week, median (IQR)
 All ages 80.0 (75.0–88.0) 45.0 (40.0–50.5) < 0.0001
 < 50 years 80.0 (75.0–89.0) 51.5 (45.0–60.0) 0.0002
 50–70 years 81.0 (75.0–89.0) 45.0 (40.0–55.0) < 0.0001
 ≥ 70 years 80.0 (75.0–82.0) 45.0 (40.0–50.0) < 0.0001
PSQI scores, median (IQR)
All ages 5.0 (4.0–6.0) 12.0 (11.0–14.0) < 0.0001
 < 50 years 6.0 (4.0–6.0) 12.0 (10.5–15.0) < 0.0001
 50–70 years 5.0 (4.5–6.0) 12.0 (11.0–13.0) < 0.0001
 ≥ 70 years 5.0 (4.0–7.0) 13.0 (12.0–15.0) < 0.0001
Factors affecting sleep quality
 Environmental factor, n (%) – 61 (85.3) –
 Psychosocial factor, n (%)a – 45 (60.0) –
 Discomfort caused by the illness, n (%) – 42 (56.0) –
 Chronic insomnia, n (%) – 24 (32.0) –
Values are median (IQR [range]) unless stated otherwise. aPsychological factors include fear, anxiety, helplessness and depression. Abbreviations: IQR, interquartile range; n, number; PSQI, Pittsburgh Sleep Quality Index; RCSQ, Richards Campbell sleep questionnaire. P values indicate differences between the good-sleep and the poor-sleep patients. P < 0.05 was considered statistically significant.

3.4  Complications, HAI, and duration of hospital stay
During hospitalization, there were no important between-group differences in organ function damage, including acute liver injury (47 [62.7%] vs 37 [61.7%], P = 0.9052), hyperglycemia (33 [44.0%] vs 31 [51.7%], P = 0.3754), acute kidney injury (three [4.0%] vs 0 [0%], P = 0.3275), and acute cardiac injury (two [2.7%] vs one [1.7%], P = 0.8447) in the good-sleep group and poor-sleep group (Table 3 ). Patients in the poor-sleep group had an increased incidence of HAI (seven [9.3%] vs one [1.7%]) compared to those in the good-sleep group; however, this difference was not significant (P = 0.1316) (Table 3). One patient with HAI in the good-sleep group had pulmonary infection of extended-spectrum β-lactamase (EBSL)-positive klebsiella pneumoniae. Meanwhile, HAI in the poor-sleep group included the following: two patients had pulmonary infection of carbapenem-resistant Acinetobacter Baumanii and Candida tropicalis, two patients had urinary tract infection, one patient had blood stream infection of EBSL-positive klebsiella pneumoniae, one patient had both pulmonary infection of Candida tropicalis and blood stream infection of bacillus, and one patient had pulmonary infection of Stenotrophomonas maltophilia and carbapenem-resistant Acinetobacter baumannii. The health condition of nine [12.0%] patients in the poor-sleep group deteriorated and required ICU care (P = 0.0151); meanwhile, none of the patients in the good-sleep group needed ICU care. In addition, patients in the poor-sleep group had an increased hospital stay (median, 33.0 [IQR, 23.0–47.0] days vs 25.0 [IQR, 20.5–36.5] days, P = 0.0116) compared to those in the good-sleep group.
Table 3 Laboratory parameters on hospital admission as well as complications, treatments and outcomes of patients with COVID-19.
Normal range Total (n = 135) Good-sleep (n = 60) Poor-sleep (n = 75) P value
Blood routine:
 White blood count, ×109/L 3.50–9.50 5.36 (4.16–7.18) 6.06 (4.77–7.57) 5.16 (4.09–6.78) 0.1260
 Neutrophil count, ×109/L 1.80–6.30 4.23 (2.98–5.99) 4.60 (3.30–6.61) 3.92 (2.92–5.54) 0.1700
 Neutrophil percentage, % 40.0–75.0 78.9 (72.1–86.7) 79.7 (73.6–86.9) 76.5 (71.9–85.9) 0.1515
 Lymphocyte count, ×109/L 1.10–3.20 0.75 (0.55–0.95) 0.72 (0.55–0.93) 0.81 (0.55–0.96) 0.5126
 Lymphocyte percentage, % 20.0–50.0 14.5 (8.5–18.1) 11.8 (8.1–17.8) 14.8 (8.9–19.2) 0.2271
 Platelet count, ×109/L 125–350 207.0 (153.0–258.0) 207.0 (145.0–264.5) 207.0 (154.5–247.0) 0.9358
Coagulation parameters†:
 Activated partial thromboplastin time, S 27.0–45.0 38.0 (34.8–42.2) 38.6 (35.4–42.9) 37.9 (34.7–41.6) 0.7342
 Prothrombin time, S 11.0–16.0 13.5 (12.6–14.2) 13.4 (12.7–14.2) 13.5 (12.6–14.2)
 D-dimer ≥ 0.5, μg/mL, n (%) 0.0–0.5 72 (53.3) 29 (48.3) 43 (57.3) 0.2976
Blood biochemistry and electrolyte‡:
 Albumin, g/L 33.0–55.0 29.7 (26.8–32.6) 30.4 (27.5–34.2) 28.7 (26.1–32.0) 0.0386
 Alanine transaminase, U/L 5–40 38.0 (24.5–58.5) 42.5 (29.0–72.5) 33.0 (23.0–54.0) 0.0424
 Total bilirubin, μmol/L 3.0–20.0 10.7 (7.3–14.0) 11.7 (8.6–14.0) 10.1 (6.7–14.0) 0.1305
 Blood urea nitrogen, mmol/L 2.90–8.20 4.77 (3.63–6.47) 5.52 (3.52–7.42) 4.65 (3.64–5.83) 0.1422
 Serum creatinine, μmol/L 57.0–111.0 69.3 (58.5–85.0) 69.7 (58.7–86.3) 68.0 (57.3–82.6) 0.9991
 Creatine kinase, U/L‡ 24–194 69.0 (47.0–130.0) 81.0 (52.0–165.0) 64.5 (40.0–105.5) 0.0288
 Lactate dehydrogenase, U/L 109–245 305.0 (224.5–380.5) 326.0 (225.0–402.0) 277.0 (224.5–367.0) 0.1233
 Glucose, mmol/L 3.90–6.10 6.40 (5.53–8.63) 6.37 (5.77–9.78) 6.55 (5.47–8.20) 0.3565
 Total carbon dioxide, mmol/L 20.0–29.0 24.1 (21.6–28.0) 23.6 (21.7–27.6) 24.6 (21.6–28.4) 0.3503
 Sodium, mmol/L 137.0–147.0 138.2 (135.9–140.3) 137.8 (135.3–140.1) 138.5 (136.5–140.4) 0.2621
 Potassium, mmol/L 3.50–5.30 3.75 (3.41–4.11) 3.82 (3.50–4.17) 3.65 (3.39–4.07) 0.1872
Cardiac biomarkers§:
 Creatine kinase-MB ≥ 6.6, ng/mL < 6.6 0 (0) 0 (0) 0 (0) –
 Hypersensitive cardiac troponin I ≥ 26.2, ng/L < 26.2 7 (5.2) 5 (8.3) 2 (2.7) 0.2779
Complications
 Acute liver dysfunction – 84 (62.2) 37 (61.7) 47 (62.7) 0.9052
 Hyperglycaemia – 64 (47.4) 31 (51.7) 33 (44.0) 0.3754
 Hospital-acquired infection – 8 (5.9) 1 (1.7) 7 (9.3) 0.1316
 Acute kidney injury – 3 (2.2) 0 (0) 3 (4.0) 0.3275
 Acute cardiac injury – 3 (2.2) 1 (1.7) 2 (2.7) 0.8447
Treatment:
 Oxygen therapy
 Nasal cannula – 102 (75.6) 48 (80.0) 64 (85.3) 0.4128
 Oxygen mask – 32 (23.7) 16 (26.7) 16 (21.3) 0.4690
 Without oxygen – 3 (2.2) 1 (1.7) 2 (2.7) 0.8447
 Antiviral therapy – 132 (97.8) 59 (98.3) 73 (97.3) 0.6953
 Antibiotics – 105 (79.5) 44 (73.3) 61 (81.3) 0.2666
 Immunoregulatory therapy – 79 (58.5) 37 (61.7) 42 (56.0) 0.5067
 Corticosteroids – 74 (54.8) 32 (53.3) 42 (56.0) 0.7570
 Immunoglobulin – 39 (28.9) 21 (35.0) 18 (24.0) 0.1612
 Vasoconstrictive agents – 13 (9.6) 6 (10.0) 7 (9.3) 0.8962
Requring ICU care – 9 (6.7) 0 (0) 9 (12.0) 0.0151
Total length of hospital stay, median (IQR), d – 29.0 (21.0–45.0) 25.0 (20.5–36.5) 33.0 (23.0–47.0) 0.0116
Values are median (IQR [range]) or numbers (percentages). Abbreviations: IQR, interquartile range; n, number. aP values indicate differences between good-sleep and poor-sleep patients. P < 0.05 was considered statistically significant.
†Data were missing for the measurement of activated partial thromboplastin time, prothrombin time, or D-dimer in five patients (8.3%), five patients (8.3%), and 10 patients (16.6%) in the good-sleep group, and six patients (8.0%), six patients (8.0%), and 10 patients (13.3%) in the poor-sleep group, respectively.
‡Data regarding creatine kinase were missing for seven patients (11.7%) in the good-sleep group, and 11 patients (14.7%) in the poor-sleep group.

3.5  Absolute lymphocyte count and neutrophil-to-lymphocyte ratio
Compared to patients in the good-sleep group, patients in the poor-sleep group had significantly decreased ALC on day 14 (median, 1.10 [IQR, 0.88–1.35] vs 1.32 [IQR, 1.04–1.59], P = 0.0055) and day 21 after hospital admission (median, 1.18 [IQR, 0.92–1.45] vs 1.48 [IQR, 1.10–1.98], P = 0.0034). Patients in the poor-sleep group also had an increased NLR on day 14 (median, 3.17 [IQR, 2.27–4.66] vs 2.44 [IQR, 2.05–3.75], P = 0.0284) and day 21 (median, 2.73 [IQR, 2.10–5.90] vs 2.23 [IQR, 1.63–2.92], P = 0.0092) than those in the good-sleep group (Fig. 2 ). The recovery rate based on ALC was higher in patients in the good-sleep group on day 14 (median, 56.91 [IQR, 8.67–94.54] vs 69.40 [IQR, 43.67–132.40], P = 0.0255) and day 21 (median, 61.40 [IQR, 26.81–102.82] vs 111.47 [IQR, 65.38–183.05], P = 0.0003) than those in the poor-sleep group; similar results were found regarding the deterioration based on an increased NLR on day 7 (median, −15.01 [IQR, −43.02–0.82] vs −39.24 [IQR, −60.34–-0.24], P = 0.0314), day 14 (median, −39.65 [IQR, −62.06–-12.94] vs −61.09 [IQR, −70.87––23.50], P = 0.0155), and day 21 (median, −51.40% [IQR, −61.08––22.36] vs −75.43 [IQR, −82.19–-47.77], P = 0.0003) (Fig. 2).
Fig. 2 Effects of self-reported sleep quality on recovery from lymphopenia. Dynamic changes in absolute lymphocyte count (ALC; A) and neutrophil-to-lymphocyte ratio (NLR; B) within 24 h (baseline value), on day 7, 14, and 21 after hospital admission in patients with COVID-19 in the good-sleep and poor-sleep groups. Dynamic changes in the ALC recovery rate (C) and NLR deterioration rate (D) on day 7, 14, and 21 after hospital admission in patients with COVID-19 in the good-sleep and poor-sleep groups. Data were missing in five patients (8.3%) on day 7, and three patients (5.0%) on day 21 in the good-sleep group. Data were missing in five patients (6.6%) on day 7, three patients (4.0%) on day 14, and four patients (5.3%) on day 21 in the poor-sleep group. ***P < 0.001 and ****P < 0.0001 when compared to the baseline value of the good-sleep group; §§P < 0.01 and §§§§P < 0.0001 when compared to baseline values of the poor-sleep group; ##P < 0.01 and ###P < 0.001 when compared to values on day 7 in the good-sleep group; &&P < 0.01 and &&&P < 0.001 when compared to values on day 7 in the poor-sleep group; N.S., not significant. P < 0.05 was considered statistically significant.