Table 1: Comparison of COVID-19 characteristics among PLWH who died or survived
Variable
|
Level
|
Total
|
Discharge
|
Death
|
p
|
|
|
N=255
|
N=195
|
N=60
|
|
^Age
|
Median
|
46 (37-54)
|
45 (36-52)
|
51 (44-59)
|
<0.001
|
Age categories
|
<40
|
77 (30.2%)
|
68 (34.9%)
|
9 (15.0%)
|
0.003
|
|
40-50
|
83 (32.6%)
|
64 (32.8%)
|
19 (31.7%)
|
|
|
>50
|
95 (37.2%)
|
63 (32.3%)
|
32 (53.3%)
|
|
Sex
|
Male
|
74 (29.1%)
|
50 (25.8%)
|
24 (40.0%)
|
0.034
|
|
Female
|
180 (70.9%)
|
144 (74.2%)
|
36 (60.0%)
|
|
Smoker
|
No
|
242 (94.9%)
|
190 (97.4%)
|
52 (86.7%)
|
0.003
|
|
Yes
|
13 (5.1%)
|
5 (2.6%)
|
8 (13.3%)
|
|
Hypertension
|
No
|
145 (56.9%)
|
115 (59.0%)
|
30 (50.0%)
|
0.220
|
|
Yes
|
110 (43.1%)
|
80 (41.0%)
|
30 (50.0%)
|
|
Diabetes Mellitus
|
No
|
180 (70.6%)
|
141 (72.3%)
|
39 (65.0%)
|
0.280
|
|
Yes
|
75 (29.4%)
|
54 (27.7%)
|
21 (35.0%)
|
|
Chronic Kidney Disease
|
No
|
237 (92.9%)
|
183 (93.8%)
|
54 (90.0%)
|
0.390
|
|
Yes
|
18 (7.1%)
|
12 (6.2%)
|
6 (10.0%)
|
|
TB
|
No
|
222 (87.1%)
|
172 (88.2%)
|
50 (83.3%)
|
0.330
|
|
Yes
|
33 (12.9%)
|
23 (11.8%)
|
10 (16.7%)
|
|
Congestive Heart Failure
|
No
|
248 (97.3%)
|
191 (97.9%)
|
57 (95.0%)
|
0.360
|
|
Yes
|
7 (2.7%)
|
4 (2.1%)
|
3 (5.0%)
|
|
Acute kidney injury
|
No
|
212 (83.1%)
|
170 (87.2%)
|
42 (70.0%)
|
0.002
|
|
Yes
|
43 (16.9%)
|
25 (12.8%)
|
18 (30.0%)
|
|
ARDS
|
No
|
244 (95.7%)
|
187 (95.9%)
|
57 (95.0%)
|
0.720
|
|
Yes
|
11 (4.3%)
|
8 (4.1%)
|
3 (5.0%)
|
|
Shock
|
No
|
248 (97.3%)
|
190 (97.4%)
|
58 (96.7%)
|
0.670
|
|
Yes
|
7 (2.7%)
|
5 (2.6%)
|
2 (3.3%)
|
|
Cough
|
No
|
62 (24.3%)
|
46 (23.6%)
|
16 (26.7%)
|
0.630
|
|
Yes
|
193 (75.7%)
|
149 (76.4%)
|
44 (73.3%)
|
|
Fever
|
No
|
153 (61.2%)
|
115 (60.2%)
|
38 (64.4%)
|
0.560
|
|
Yes
|
97 (38.8%)
|
76 (39.8%)
|
21 (35.6%)
|
|
Sore Throat
|
No
|
226 (89.0%)
|
171 (87.7%)
|
55 (93.2%)
|
0.230
|
|
Yes
|
28 (11.0%)
|
24 (12.3%)
|
4 (6.8%)
|
|
Myalgia
|
No
|
198 (77.6%)
|
150 (76.9%)
|
48 (80.0%)
|
0.620
|
|
Yes
|
57 (22.4%)
|
45 (23.1%)
|
12 (20.0%)
|
|
ARV therapy
|
No
|
22 (11.5%)
|
17 (11.6%)
|
5 (11.1%)
|
1.000
|
|
Yes
|
169 (88.5%)
|
129 (88.4%)
|
40 (88.9%)
|
|
*CD4 categories
|
<200
|
73 (36.9%)
|
48 (32.0%)
|
25 (52.1%)
|
0.021
|
|
200-499
|
83 (41.9%)
|
65 (43.3%)
|
18 (37.5%)
|
|
|
500
|
42 (21.2%)
|
37 (24.7%)
|
5 (10.4%)
|
|
*CD4 Count
|
Median (IQR)
|
267 (141-457)
|
296.5 (174-498)
|
188 (72.5-337)
|
0.004
|
Oxygen saturation
|
Median (IQR)
|
93 (86-97)
|
94 (90-98)
|
86.5 (74.5-93)
|
<0.001
|
PaO2
|
Median (IQR)
|
8 (6.8-9.9)
|
8.1 (6.9-9.9)
|
7.3 (6.1-11.2)
|
0.160
|
FiO2
|
Median (IQR)
|
21 (21-21)
|
21 (21-21)
|
21 (21-40)
|
0.016
|
HGT
|
Median (IQR)
|
7.4 (6-11.9)
|
7.3 (5.7-12.7)
|
7.9 (6.8-11.8)
|
0.100
|
Creatinine
|
Median (IQR)
|
74 (58-116)
|
71 (55-100)
|
101 (71-262)
|
<0.001
|
CRP
|
Median (IQR)
|
145 (76-223)
|
139 (70-207)
|
204 (124-285)
|
<0.001
|
White Cell Count
|
Median (IQR)
|
8.64 (6.56-11.16)
|
7.93 (6.35-10.59)
|
9.42 (7.8-13.32)
|
0.002
|
Lymphocytes
|
Median (IQR)
|
1.65 (1.22-2.20)
|
1.68 (1.27-2.22)
|
1.37 (0.9-1.855)
|
0.031
|
Neutrophils
|
Median (IQR)
|
5.62 (4.12-8.22)
|
5.34 (3.97-7.70)
|
7.3 (5.60-9.35)
|
0.002
|
Platelets
|
Median (IQR)
|
280 (221-374)
|
287 (223-374)
|
273 (207-369)
|
0.480
|
HbA1c
|
Median (IQR)
|
12.8 (11.2-14)
|
12.9 (11.3-14.1)
|
12.0 (10.4-13.4)
|
0.038
|
^ Median (IQR), *CD4 counts: At least six months prior to SARS-CoV-2 infection
Abbreviations: ARDS: acute respiratory distress syndrome, ARV: antiretroviral therapy, CD4: cluster of differentiation 4, CRP: C-reactive protein, IQR: interquartile range, FiO2: fraction of inspired oxygen, HGT: Hemo Glucose test, PaO2: partial pressure of oxygen, TB: Tuberculosis
Table 2: Univariate and multivariate level analysis of factors associated with COVID-19 mortality among PLWH
Characteristic
|
RR (95% CI)
|
p
|
ARR (95% CI)
|
p
|
Age categories
<40
|
1
|
.
|
|
|
40-49
|
1.96 (0.94-4.07)
|
0.072
|
2.24 (0.76 – 6.64)
|
0.143
|
≥50
|
2.88 (1.46-5.67)
|
0.002
|
2.32 (0.67 – 8.14)
|
0.186
|
Sex: Female
|
1.62 (1.04-2.52)
|
0.032
|
2.00 (0.85- 4.67)
|
0.111
|
Smoker
|
2.86 (1.75-4.69)
|
<0.001
|
4.17 (1.50-11.61)
|
0.006
|
Hypertension
|
1.32 (0.85-2.05)
|
0.22
|
0.93 (0.40-2.19)
|
0.869
|
Diabetes Mellitus
|
1.29 (0.82-2.04)
|
0.271
|
1.04 (0.51-2.16)
|
0.904
|
Chronic Kidney Disease
|
1.46 (0.73-2.93)
|
0.283
|
|
|
TB
|
1.35 (0.76-2.38)
|
0.309
|
|
|
Congestive Heart Failure
|
1.87 (0.77-4.52)
|
0.168
|
|
|
Acute kidney injury
|
2.11 (1.36-3.30)
|
0.001
|
0.71 (0.33-1.57)
|
0.401
|
ARDS
|
1.17 (0.43-3.15)
|
0.76
|
|
|
Shock
|
1.22 (0.37-4.03)
|
0.742
|
|
|
Cough
|
0.88 (0.54-1.45)
|
0.624
|
|
|
Fever
|
0.87 (0.55-1.39)
|
0.565
|
|
|
Sore Throat
|
0.59 (0.23-1.50)
|
0.265
|
|
|
Myalgia
|
0.868 (0.50-1.52)
|
0.621
|
|
|
ARV therapy
|
1.04 (0.46-2.36)
|
0.922
|
|
|
*CD4: 200-499
|
1
|
.
|
|
|
<200
|
1.58 (0.94-2.65)
|
0.085
|
1.27 (0.62-2.61)
|
0.509
|
≥500
|
0.55 (0.22-1.38)
|
0.202
|
0.35 (0.09 – 1.29)
|
0.114
|
pa02
|
1.012 (0.96-1.07)
|
0.665
|
|
|
HGT
|
0.999 (0.973-1.03)
|
0.944
|
|
|
CRP
|
10.03 (10.01-10.05)
|
<0.001
|
10.02 (10.002-10.1)
|
0.032
|
Creatinine
|
1.001 (1.001-1.002)
|
<0.001
|
1.00 (0.99-1.00)
|
0.187
|
WCC
|
1.05 (1.01-1.08)
|
0.005
|
|
|
Neutrophils
|
1.024 (1.01-1.03)
|
<0.001
|
1.03 (1.02-1.06)
|
<0.001
|
HbA1c
|
1.01 (1.007-1.01)
|
<0.001
|
1.01 (1.001-1.02)
|
0.011
|
Platelets
|
0.999 (0.997-1.00)
|
0.477
|
|
|
Lymphocytes
|
0.882 (0.58-1.34)
|
0.557
|
|
|
*CD4 counts: At least six months prior to SARS-CoV-2 infection
Abbreviations: ARDS: Acute respiratory distress syndrome, ARV: antiretroviral, CD4: cluster of differentiation 4, CRP: C-reactive protein, HbA1c: Glycated haemoglobin A1, HGT: haemo glucose test, PaO2: partial pressure of oxygen, TB: tuberculosis, WCC: white cell count
3.1. Characteristics of PLWH with SARS-COV-2
A total of 255 PLWH were admitted during the first and the second waves. The cohort was comprised mainly of females (70.9%) and mostly above 50 years old (38.8%), with 55.2% (n=32) of all patients who died being 50 years and older (Table 1). Among PLWH, the main pre-existing co-morbidities were hypertension (43.1%), diabetes mellitus (29.4%), chronic kidney disease (7.1%), TB (12.9%), and acute kidney injury (16.9%) (Table 1). The presenting clinical signs and symptoms suggestive of COVID-19 included fever (38.8%), cough (75.7%), sore throat (11.0%), and myalgia (22.4%) (Table 1). Most of this PLWH cohort (88.5%) were on ARV therapy. Among them, 36.9% of this cohort had severe immunosuppression with CD4< 200mm3 with only 21% having a CD4≥500mm³ (Table 1). The median oxygen saturation and PaO2 were 93% (86-97) and 8kPa (6.8-9.9), respectively (Table 1). Higher median CRP was observed among those who died 20.4 (12.4-28.5) as compared to those discharged 13.9 (7.0-20.7) characterised this PLWH cohort and, high median (IQR) HbA1c was a predictor of mortality with 12% (10.35%-13.35%) (Table 1).
3.2. Association of demographic, haematological, and biochemical parameters with mortality among PLWH with SARS-CoV-2 infection
Table 2 shows the association between demographic, haematological, and biochemical parameters with survival among PLWH. creatinine (1.001, 95% CI: 1.001–1.002, p < 0.001), Neutrophils (1.02, 95%CI: 1.01–1.03; p < 0.001), HbA1c (1.01, 95%CI: 1.007–1.010; p < 0.001), and CRP (10.03, 95%CI: 10.01–10.05, p < 0.001) were significantly associated with the risk of mortality. In multivariate analysis smoking, neutrophils, HbA1c and CRP were all significantly associated with increased risk of mortality (aRR 4.17: 1.50-10.01); (aRR: 1.03, 95%CI: 1.02–1.05; p < 0.001); (aRR: 1.01, 95%CI: 1.001–1.02; p = 0.021); (aRR; 10.02, 95%CI: 10.00–10.01; p0.032).
3.3. ROC curves and cut-offs
As the adjusted RR was significant for neutrophils and CRP at the borderline, we determined the optimal cut-offs to predict non-survival and test performance of these two parameters using ROC curves. ROC curves were drawn with sensitivity as the horizontal coordinate and the 1‑specificity as the vertical coordinate to predict COVID-19 severity and mortality among PLWH admitted to the hospital. The proposed optimum cut-off points for neutrophils derived from ROC analysis was ≥ 5.6 × 109/L with sensitivity = 76% and specificity = 56% (Table 3 and Figure 2). An optimal cut-off of 202 mg/L rendered 52 % sensitivity and 74% specificity for the CRP (Table 3 and Figure 3). The area under the ROC curves (AUC) for the neutrophils and CRP were 0.66 and 0.65, respectively (Table 3). However, the performance of both was suboptimal to use as a predictive marker on their own. The combination of the two variables and their predictive value on the roc curve changed slightly to AUC=0.67.
Table 3: Optimal cut-off, sensitivity, specificity, and AUC for CRP, and Neutrophils.
Analyte
|
direction
|
Optimal cut point
|
sensitivity
|
specificity
|
AUC
|
CRP (mg/L)
|
≥
|
202
|
0.52
|
0.74
|
0.65
|
Neutrophils (×109/L)
|
≥
|
5.6
|
0.76
|
0.56
|
0.66
|
3.4. Kaplan-Meier survival estimates between males and females
The rate of death seemed to be higher among male patients during the whole duration of the hospitalization. However, Poisson regression analysis was used to compare male and female mortality, no significant difference was found (p=0.160) (Figure 4). The median stay was 7 (IQR: 3-12) days for females compared to 5 (IQR: 3-9.5) days for males. The overall median stay was 6 (IQR: 3-10) days. (Figure 4). The plots of the scaled Schoenfeld residuals of each covariate against log-time were used to determine whether the proportional hazards assumption was violated (Figure 5). The Schoenfeld residuals test revealed that a proportional hazard assumption had the same effect on male and female survival rates (p = 0.3905) (Figure 5).