Socio-Demographic Characteristics
The median age of the patients was 58((IQR = 41 – 66) years. The minimum and maximum ages of patients were 19 and 86 years respectively. Four hundred seventy two (69.01%) of the patients were males and six hundred twenty six (91.52 %) of the patients were from Addis Ababa (Table 1).
Table 1: Socio–demographic characteristics of adult severe COVID-19 patients who received supplemental oxygen therapy admitted at EKGH, Addis Ababa Ethiopia, March April 2020 to 2022 (n=684).
Variables
|
Status at last follow up of SO therapy
|
Total (%)
|
No. of
Event (%)
|
No. of
Censored (%)
|
Age Median(IQR)
|
56.5(40 – 66)
|
60(49 – 70)
|
58( 41 – 66)
|
Age
|
≤60
|
341(82.17)
|
74 (17.83)
|
415 (60.67)
|
> 60
|
215(79.93)
|
54 (20.07)
|
269 (39.33)
|
Sex
|
Male
|
404(85.59)
|
68(14.41)
|
472(69.01)
|
Female
|
152(71.70)
|
60(28.30)
|
212(30.99)
|
Clinical Characteristics
All adult severe COVID-19 patients were symptomatic at admission. At admission, the median Spo2 those who were on supplemental oxygen therapy was 94 %( IQR=92 – 95%) and at the last date of follow up median SpO2 those who were censored and on oxygen therapy was 93% (IQR = 88 – 96%) and those who were experienced an event and on room air was 92 %( IQR = 91 – 94%).
The most common symptom at admission was cough (93.86%) and only few had diarrhea (8.92%) (Fig1). Among overall COVID-19 patients, four hundred forty (64.33%) had a history of at least one preexisting medical condition. The majority of patients had diabetes mellitus, followed by hypertension, HIV/AIDS, cardiac illness, tuberculosis, asthma and COPD (Fig2) (Table 2).
HTN: Hypertension; DM: Diabetes Mellitus; CI: Cardiac illness; COPD: Chronic Obstructive Pulmonary Disease; TB: Tuberculosis; CKD: Chronic Kidney Disease; CLD: Chronic Liver Disease; HIV: Human Immunodeficiency Virus
Table 2: Clinical features and preexisting comorbidities of adult severe COVID-19 patients who received supplemental oxygen therapy admitted at EKGH, Addis Ababa Ethiopia, March 2020 to April 2022 (n=684).
Clinical characteristics and co- morbidities
|
Status at last follow up of
SO therapy
|
Total (%)
|
No. of
Event (%)
|
No. of
Censored (%)
|
Duration of symptoms onset prior to admission (Median(IQR) )
|
7(5 – 10)
|
10(7 – 14)
|
7(5 – 11)
|
Easily fatigue
|
yes
|
487(80.36)
|
119(19.64)
|
606(88.60)
|
no
|
69(88.46)
|
9(11.15)
|
78(11.40)
|
Febrile sensation(fever)
|
yes
|
369(86.01)
|
60(13.99)
|
429(26.72)
|
no
|
187(73.33)
|
68(26.67)
|
255(37.28)
|
Headache
|
yes
|
278(91.45)
|
26(8.55)
|
304(44.44)
|
no
|
278(73.16)
|
102(26.84)
|
380(55.56)
|
Arthralgia
|
yes
|
224(89.60)
|
26(10.40)
|
250(36.55)
|
no
|
332(76.50)
|
102(23.50)
|
434(63.45)
|
Myalgia
|
yes
|
196(91.59)
|
18(8.41)
|
214(31.29)
|
no
|
360(76.60)
|
110(23.40)
|
470(68.71)
|
Sore throat
|
yes
|
57(78.08)
|
16(21.92)
|
73(10.67)
|
no
|
499(81.67)
|
112(18.33)
|
611(89.33)
|
Diarrhea
|
yes
|
53(86.89)
|
8(13.11)
|
61(8.92)
|
no
|
503(80.74)
|
120(19.26)
|
623(91.08)
|
Temperature (℃) (Median(IQR))
|
36.4(36.1 – 37.1)
|
36(35 – 37 )
|
36.4 (36 – 37)
|
Temperature (℃)
|
<37.5
|
454(80.35)
|
111(19.65)
|
565(82.60)
|
≥37.5
|
102(85.71)
|
17(14.29)
|
119(17.40)
|
RR(breath/min) (Median(IQR))
|
28 (26 – 36)
|
32 (28 – 40)
|
30 (26 – 36)
|
HR( beats/min) (Median(IQR))
|
89(82 – 102)
|
96 (96 – 109)
|
94 (83 – 103)
|
HR( beats/min)
|
< 100
|
376(83.00)
|
77(17.00)
|
453(66.23)
|
≥ 100
|
180(77.92)
|
51(22.08)
|
231(33.77)
|
SBP(mmHg) (Median(IQR) )
|
125(110 – 135)
|
121(110 – 141)
|
125 (110 – 135)
|
HTN
|
Yes
|
169(83.23)
|
34(16.75)
|
203(29.68)
|
no
|
387(80.46)
|
94(19.54)
|
481(70.32)
|
DM
|
Yes
|
210(83.33)
|
42(16.67)
|
252(36.84)
|
no
|
346(80.09)
|
86(19.91)
|
432(63.16)
|
CI
|
Yes
|
50(100)
|
0(0.00)
|
50(7.31)
|
no
|
506(79.81)
|
128(20.19)
|
634(92.69)
|
COPD
|
Yes
|
19(67.86)
|
9(32.14)
|
28(4.09)
|
no
|
537(81.86)
|
119(18.14)
|
656(95.91)
|
Asthma
|
Yes
|
33(78.57)
|
9(21.43)
|
42(6.14)
|
no
|
523(81.46)
|
119(18.54)
|
642(93.86)
|
TB
|
Yes
|
32(65.31)
|
17(34.69)
|
49(7.16)
|
no
|
524(82.52)
|
111(17.48)
|
635(92.84)
|
CKD
|
Yes
|
1(10.00)
|
9(90.00)
|
10(1.46)
|
no
|
555(82.34)
|
119(17.66)
|
674(98.54)
|
CLD
|
Yes
|
8(100)
|
0(0.00)
|
8(1.17)
|
no
|
548(81.07)
|
128(18.93)
|
676 (98.83)
|
HIV
|
Yes
|
56(76.71)
|
17(23.29)
|
73(10.67)
|
no
|
500(81.83)
|
111(18.17)
|
611(89.33)
|
Chronic neurological disease
|
Yes
|
2(20.00)
|
8(80.00)
|
10(1.46)
|
no
|
554(82.20)
|
120(17.80)
|
674(98.54)
|
Treatments and Medications Histories
Most of the patients (79.68%) were received anti-pains. But small numbers of patients were treated with redmesivir (6.58%), ACEI/ARBS (12.57%). Only 3.80% of adult severe COVID-19 patients had a history of chronic corticosteroid use prior to admission (Table 3).
Table 3: Treatment and medication histories of adult severe COVID-19 patients who received supplemental oxygen therapy admitted at EKGH, Addis Ababa Ethiopia, March 2020 to April 2022 (n=684).
Variables
|
Status at last follow up of SO therapy
|
Total (%)
|
No. of
Event (%)
|
No. of
Censored (%)
|
Redmesivir
|
yes
|
28(62.22)
|
17(37.78)
|
45(6.58)
|
no
|
528(82.63)
|
111(17.67)
|
639(93.42)
|
ACEI/ARBS
|
yes
|
77(89.53)
|
9(10.47)
|
86(12.57)
|
no
|
479(80.10)
|
119(19.90)
|
598(87.43)
|
Anti – pains
|
yes
|
442(81.10)
|
103(18.90)
|
545(79.68)
|
no
|
114(82.01)
|
25(17.99)
|
139(20.32)
|
Type of Ant – pains
|
NSAIDS
|
40(59.70)
|
27(40.30)
|
67(12.29)
|
Other types
|
376(84.68)
|
68(15.32)
|
444(81.47)
|
Mixed
|
26(76.47)
|
8(23.53)
|
34(6.24)
|
Chronic corticosteroids use prior to admission
|
yes
|
17(65.38)
|
9(34.62)
|
26(3.80)
|
no
|
539(81.91)
|
119(18.09)
|
658(96.20)
|
Baseline Laboratory Biomarkers of the Patients
Nearly one third of adult severe COVID-19 patients had lymphocytopenia (72.66%). More than half of the patients had RBS above 140mg/dl (53.36%) and high neutrophil count (66.37%). Leukocytosis was found in 21.19%, thrombocytopenia in 31.58% and anemia in 4.18% of the patients. Elevated level of AST (45.18%) and BUN (42.25%) were more common than elevated levels of ALP (19.74%), ALT (21.93%) and creatinine (17.69%) (Table 4).
Table 4: Baseline laboratory biomarkers results at admission of adult severe COVID-19 patients who received supplemental oxygen therapy admitted at EKGH, Addis Ababa Ethiopia, March 2020 to April 2022 (n=684).
Variables
|
Status at last follow up of
SO therapy
|
Total (%)
|
No. of
Event (%)
|
No. of
Censored (%)
|
Total WBC cells×10^3/µl) (Median(IQR))
|
6.72(5.31 – 9.04)
|
12.35( 6.42 – 15.71)
|
7.52(5.32 –10.90)
|
Lymphocytes count (%) (Median(IQR))
|
11.8(7.1 – 24.7)
|
8.2(3.5 – 10.4 )
|
10.30(6.1 0–24.05)
|
Lymphocytes count (%)
|
≥21
|
187(100.00)
|
0(0.00)
|
187(27.34)
|
<21
|
369(74.25)
|
128(25.75)
|
497(72.66)
|
Neutrophils (%) (Median(IQR))
|
80.9(66.9 – 88)
|
86.3(80 – 93.9)
|
81.85(68.30 – 90.60)
|
Platelets (cells×10^3/µl) (Median(IQR)
|
188(146 – 262 )
|
197(126 – 221 )
|
188(145 – 228)
|
Platelets
(cells×10^3/µl)
|
≥150
|
384(82.05)
|
84(17.95)
|
468(68.42)
|
<150
|
172(79.63)
|
44(20.37)
|
216(31.58)
|
Hemoglobin (g/dl) (Median(IQR))
|
14.6(13.9 – 16)
|
14.2(12.7 – 15.8 )
|
14.5(13.80 – 15.80)
|
Anemia
|
yes
|
73(75.26)
|
24(24.74)
|
97(14.18)
|
no
|
483(82.28)
|
104(17.72)
|
587(85.82)
|
RBS (mg/dl) (Median(IQR))
|
143(122 – 199)
|
222(126 – 272)
|
144(122 – 206)
|
RBS (mg/dl)
|
≤140
|
267(83.70)
|
52(16.30)
|
319(46.64)
|
>140
|
289(79.18)
|
76(20.82)
|
65(53.36)
|
AST(IU/L) (Median(IQR))
|
36(26 – 63)
|
42(31 – 85)
|
36(27.50 – 63)
|
AST (IU/L)
|
≤ 37
|
323(86.13)
|
52(13.87)
|
375(54.82)
|
> 37
|
233(75.40)
|
76(24.60)
|
309(45.18)
|
ALT(IU/L) (Median(IQR))
|
39(30 – 61)
|
39(24 – 80)
|
39(29 – 61)
|
ALT(IU/L)
|
≤ 63
|
441(82.58)
|
93(17.42)
|
534(78.07)
|
> 63
|
115(76.67)
|
35(23.33)
|
150(21.93)
|
ALP (IU/L) (Median(IQR))
|
68(56.5 – 101)
|
73(50 – 127)
|
68(56 – 104)
|
ALP (IU/L)
|
≤116
|
464(84.52)
|
85(15.48)
|
549(80.26)
|
> 116
|
92(68.15)
|
43(31.85)
|
135(19.74)
|
Creatinine(mg/dl)(Median(IQR)
|
0.94(0.78 – 1.1)
|
1.26(1.06 – 1.75)
|
1.01(1 – 13)
|
Creatinine(mg/dl)
|
≤1.3
|
495(87.92)
|
68(12.08)
|
563(82.31)
|
> 1.3
|
61(50.41)
|
60(49.59)
|
121(17.69)
|
BUN(mg/dl) (Median(IQR))
|
15(11 – 22)
|
23(19 – 41)
|
7(0.82 – 1.16)
|
BUN(mg/dl)
|
≤18
|
369(93.42)
|
26(6.58)
|
395(57.75)
|
> 18
|
187(64.71)
|
102(35.29)
|
289(42.25)
|
RBS: Random Blood Sugar; AST: Aspartate Transaminase; ALT: Alanine Transaminase; ALP: Alkaline Phosphatase; BUN: Blood Urea Nitrogen
Censoring Status at the Last Follow-up
Among the 684 patients, 556 (81%) of the patients achieved the event (supplemental oxygen therapy discontinued) while 128 (19%) were censored. Among the 128 censored observation, 9 (7.03%) died and 119(92.97 %) were transferred to ICU for further care (Figure 3).
Overall Survival Estimates for Time to Discontinuation of Supplemental Oxygen Therapy and Comparison of Survival Status among Groups
The median follow up time was 9 days (IQR= 5-13) and the lowest and the highest length of follow-up time were 2 and 61 days respectively, and the total person-time risk was 6829. The overall supplemental oxygen therapy discontinuation rate was 8.14 (95% CI: 7.49 – 8.85) per 100 person day observation.
The estimated median survival time to discontinuation of supplemental oxygen therapy was 10 days (IQR= 6-15).
The survival status of adults with severe COVID-19 who were received supplemental oxygen therapy was estimated by the Kaplan-Meier survival curve. The curve tends to decrease rapidly within the first ten days indicating that most the cases discontinued from supplemental oxygen therapy within this time (Fig 4). The survival estimates of adult severe COVID-19 patients who were received supplemental oxygen therapy were varied in relation to presence of comorbidity, SOB, chest pain, SBP ≥140 mmHg, high total WBC, neutrophil count and elevated level of ALP and serum creatinine. The KM survival function graph showed that those severe COVID-19 patients presented with conditions (variables) mentioned above at admission had a longer survival experience (time to discontinuation of supplemental oxygen therapy) as compared to those without such conditions. For instance as shown in figure 5 patients come with SOB at admission had a longer survival experience than those without SOB. This means patients with SOB required longer duration of supplemental oxygen therapy than those without the complaint of SOB at admission.
Log-Rank test was also used test survival function differences between categories of predictors. According to the log rank test result, found that there is a significant difference in survival experience among the following categories of predictors at p-value < 0.05. The median duration of oxygen therapy was longer among patients with age above 60 years(11days) compared to those ≤ 60 years(9days) ,female sex(12 days) than male patients(10days) and a complaint of shortness of breath (11 days) compared to those with no such complaint (7 days), Chest pain (19days) compared to those with no chest pain (10 days), sore throat (11days) compared to those with sore throat (10 days), diarrhea (16 days) to those with no such complaint (10days). Patients with SBP≥140mmHg, RR > 30 breath/min also needed a longer duration of oxygen therapy. On the other hand patients with a complaint of febrile sensation, headache or myalgia were needed shorter duration of oxygen therapy. But the log rank test didn’t show any significant difference in the survival function among the other sign and symptom groups (all p-values >0.05). The survival time were significantly longer among patients with a history of one or more pre-existing co-morbid illness, hypertension, COPD, tuberculosis and chronic liver disease
It was also found that there was a statistically significant longer median survival time to discontinuation of oxygen therapy among those who were treated with anti-pains and had leukocytosis ,high neutrophil count, lymphocytopenia , anemia ,elevated AST,ALT,ALP, serum creatinine and BUN whereas on contrary treatment with ACEI/ARBS reduced the survival time
Predictors of the Survival Time (Time to Discontinuation of Supplemental Oxygen Therapy)
Figure 6 showed that the log minus log survival curve seems to be parallel among the groups classified by presence of SOB and high WBC count but clearly crossed each other (i.e. non- proportional) among the groups of anemia and HIV/AIDS. Global fitness test of multivariable Cox PH model were found to be p-value < 0.001 and the time- varying covariates (TVC) were also found that anemia, myalgia, respiratory rate and HIV/AIDS had time varying effect. Therefore, there were significant evidences that suggested the assumptions of proportional-hazard were violated. Even if the proportional hazard assumptions were found to be violated, the model adequacy and fitness of final multivariable Cox PH model was checked by Cox-Snell residual plot showed it was not a well fitted model.
Predictors that had association at a p-value of < 0.2 in bivariable Cox PH regression were included in multivariable Cox regression. On the bivariable Cox PH model Age, sex ,cough ,SOB ,febrile sensation , chest pain ,headache ,myalgia ,sore throat , nausea/vomiting ,diarrhea ,duration of symptoms, body temperature ,RR, SBP ,presence of comorbidities , HTN ,DM ,COPD ,TB ,CLD ,HIV ,remdesivir ,ACEI ,Anti-pains , total WBC , Lymphocyte, Neutrophil count , Anemia , AST, ALT,ALP, Creatinine and BUN were found to be possible predictors of time to discontinuation of supplemental oxygen therapy among adult severe COVID-19patients. The forward stepwise approach was used for variable selection to fit the final multivariable Cox PH model at a p-value of <0.05. Finally multivariable stratified Cox regression model stratified by HIV/AIDS, Myalgia, RR and anemia, and including ten predictors in the model that were not violated PH assumption (Preexisting comorbidities, cough, SOB, chest pain, nausea/vomiting, SBP, and total WBC, Neutrophil count, ALP and serum creatinine level) . All were statistically significant predictors of time to discontinuation of supplemental oxygen therapy (Table 5).
Before interpretation of the result of the fitted multivariable stratified Cox proportional hazard regression model, overall model adequacy and model fitness were checked by Cox-Snell residual plot and global fit test and found that it was appropriate. After stratified Cox proportional hazard model fitted there was no a significant evidence suggested the violation of assumptions of proportional hazard (global test p-value= 0.1308). This supported the Cox-Snell residual plot finding of model adequacy (Fig 7).
The result of multivariable analysis after adjusting for other covariates revealed that the rate of discontinuation of supplemental oxygen therapy of adult severe COVID-19 patients with preexisting comorbidities decreased by 55% as compared to those without comorbidities (AHR; 0.45, 95% CI (0.36 – 0.57), p-value<0.001). Having a complaint of shortness of breath at admission was associated with a 51% lower rate of discontinuation of supplemental oxygen therapy compared to those patients with no such complaint (AHR= 0.49, 95%CI (0.36 – 0.66), p-value<0.001).Supplemental oxygen therapy discontinuation rate among adult severe COVID-19 patients admitted with chest pain reduced by 57% as compared to those who had no (AHR; 0.43, 95%CI (0.26 – 0.71), p-value=0.001). The rate of discontinuation of supplemental oxygen therapy among patients with nausea/vomiting was 49% lower than patients without such complaint at admission (AHR= 0.51, 95%CI (0.38 – 0.68), p-value<0.001). Discontinuation of supplemental oxygen therapy among patients admitted with systolic blood pressure ≥140mmHg the rate was decreased by 55% as compared to patients with systolic blood pressure <140mmHg at admission (AHR= 0.45, 95%CI (0.33 – 0.61), p-value<0.001). On contrary patients presented with cough at admission the rate of discontinuation of supplemental oxygen therapy was 91% higher than patients had no cough (AHR; 1.91, 95%CI (1.19 –3.04), p-value=0.007) (Table 5).
Laboratory biomarkers that found to be an independent predictor of time to discontinuation of supplemental oxygen therapy were high total WBC count (AHR; 0.65, 95%CI (0.49 – 0.85), p-value=0.002),high Neutrophil count (AHR; 0.56, 95%CI(0.43 – 0.73), p-value<0.001), elevated Alkaline phosphatase (ALP) (AHR; 0.20, 95%CI (0.15 – 0.29), p-value<0.001), elevated serum creatinine (AHR; 0.32, 95%CI (0.22 – 0.47), p-value<0.001) (Table 5).
The Cox- Snell plot of Figure 14 indicates that at beginning the plot makes Nelson-Aalen cumulative hazard function straight lines through the origin and follows the Cox-Snell estimate of residuals 45 degree line very closely except for some large values of time suggesting that the fitted model is appropriate. It is very common for models with censored data to have some wiggling at large values of time and it is not something which should cause much concern.