DOI: https://doi.org/10.21203/rs.3.rs-2260357/v1
The CD4 + count is used to evaluate the clinical status of HIV-infected patients when deciding whether to initiate ART. To study the progression of HIV-infected patients on ART, CD4 + counts in each individual could be measured repeatedly to monitor the patient's AIDS progression and monitor treatment success. Therefore, this study aimed to identify predictors of CD4 + progression in HIV-positive patients receiving ART at the Debre Berhan Referral Hospital.
Retrospective data were collected from 322 HIV-infected patients who started ART in the hospital from September 2013 to February 2019. Exploratory analyses were applied to assess subject-specific and mean differences in terms of patients’ CD4 + progression. A linear mixed model was used as data analysis to account random effects.
Of the 322 HIV-infected patients considered in the study, 225 (69.88%) were females. The baseline mean CD4 + counts was 335.7 and changed to 408.61 over 7 follow-up years. Moreover, predictors such as patients’ gender (male) (β =-0.7512, p-value = 0.019), age at initiation of ART (β = -0.02705, p-value = 0.047), bedridden functional status of the patients at initiation of ART (β = -0.03365, p-value = 0.021), TDF-3TC-NVP regimen class (β = -0.1474, p-value = 0.031), unmarried patients (β = 0.610, p-value = 0.011), patients’ WHO clinical stage-II (β = -0.402, p-value = 0.047), baseline CD4 count (β = 0.020, p-value = 0.0001) and follow-up time (β = 0.613, p-value = 0.0001) were positively as well as negatively associated and had significant impact on CD4 count progression.
Patients’ gender, age at initiation of ART, bedridden functional status at ART initiation, TDF-3TC-NVP treatment class, unmarried marital status, WHO clinical stage II, baseline CD4 count and follow-up time was found to be a significant predictor of the progression of a patient's CD4 count. Therefore, HIV-positive patients can be advised to start ART treatment as early as possible. Special guidance and attention is also required, especially in elderly patients, males with bedridden functional status, and late WHO clinical stage patients.
Human immunodeficiency virus (HIV) destroys CD4 cell numbers, raises HIV RNA plasma levels, and causes long-term acquired immunodeficiency syndrome (AIDS) [1], [2]. Worldwide, 37.7 million (30.2–45.1 million) people lived with HIV in 2020, of which 28.2 million were receiving antiretroviral therapy as of June 30, 2021). Of the total of 37.7 million people living with HIV, the majority come from low- and middle-income countries, and sub-Saharan Africa accounts for two-thirds (67%) of people living with HIV [3], but only about 12% of the world's population [2].
The CD4 + count is used as assessing the clinical status of HIV-infected patients in making decisions regarding the initiation of antiretroviral therapy (ART) [4]. Measuring CD4 + count is a strong predictor of progression to Human Immunodeficiency Syndrome (AIDS) as well as a means of monitoring the success of such antiretroviral therapy [5]. The possible increase or decrease in CD4 + counts are directly related to HIV replication. Low CD4 + counts are associated with a greater risk of patients living with HIV developing opportunistic infections, which may then progress to advanced disease and death [4], [5]. In a healthy adult, a normal CD4 + count can vary enormously (by population, age group, etc.), but is typically around 500 to 1500 cells per cubic milliliter of blood (mm3) [6]. When it falls below 200, however, then the disease is technically classified as AIDS.
In ART-stable patients, CD4 cell counts are not required to monitor treatment response when HIV viral load tests are available [7]. However, CD4 remains the best measure of a patient's immune and clinical status, risk of opportunistic infections, and supports diagnostic decision-making, especially in patients with advanced HIV disease[7]. Moreover, the use of combinations of antiretroviral drugs (ART) greatly results in the supervision of virus replication and hence increased levels of CD4 + count. The CD4 + count is also used to decide when to start antiretroviral therapy. Initiation of Highly Active Antiretroviral Therapy (HAART) at higher CD4 + counts has been demonstrated to “the risk of death, opportunistic infections, and non-HIV related comorbidities”[8]. Combined with at least two to three types of antiretroviral therapy, HAART effectively lowers the levels of the virus in the body by increasing the immune system's known CD4 + count.
More previously, limited studies were focused on modeling the CD4 + counts trend over time for patients on ART [2]. However, [9] suggested that the progression of CD4 + T-cells follows a pattern of a long periods of slow decline followed by a rapid decline just before the onset of AIDS. Here, it indicates the need further investigation for the question what happens about the progression after the start of HAART.
In a previous study, HAART brought a significance improvement on CD4 + count and provided further quantitative evidence about aspects of the therapy effect such as the changes in the slope of CD4 + cell count [10]. Moreover, patients who had already experienced an AIDS defining event at the point of initiating HAART were also at higher risk of developing a new event, irrespective of their CD4 + count evolution during treatment [11].
In the previous studies, predictors such as baseline CD4 cell count [12]–[21], follow-up time [12], [14], [16], [18], [21]–[23], regimen d4t-3TC-NVP and AZT-3TC-NVP,[12], initial regimen of AZT + 3TC + EFV[20], but all ART regimens without significance difference that increased mean CD4 count at the study period [13], functional status [17], [22], working functional status [13], [18], [19], baseline age [14], [16], [17], [22], [23], younger age [13], [18], baseline WHO stage [17], [19], [21], [22], baseline WHO clinical stage II [20], [24], marital status [14], [16], [17], sex of patients [16], [21], [23], patients’ level of education [16], tertiary level of education [20] were identified as significant factors that significantly contribute to the changes of patients CD4 cell count.
Even when HIV-infected patients are advised to initiate ART, assessing the improvement in CD4 + counts over time after the patient initiates HAART, and whether the pattern of therapy varies depending on the patient’s characteristics such as gender, age, functional status, and dietary intake, etc. is very crucial. Moreover, since the data is measured on the same subject, repeatedly, taking this correlated data as independent leads to inappropriately estimated standard errors and inefficient estimators. Hence, using linear mixed models is also important to incorporate the random effects of between-subjects variability. This study therefore aimed to examine the progression of CD4 + counts over time and to evaluate the determinants factors from September 11, 2013 to February 8, 2019 in Debre Berhan Referral Hospital.
A retrospective cohort study of HIV/AIDS patients who started ART at Debre Berhan Specialty Hospital (DBRH) from September 11, 2013, to February, 2019 was performed. DBRH is found in Amhara Regional state, 130 km Northeast of Addis Ababa, which is the capital of Ethiopia. Study subjects include HIV-positive adults aged 16 years and older who have started ART in a hospital. All patients who had started ART and had their CD4 + counts measured at least twice, including baseline, and who started first-line ART were included in the study population. Patients under 16 years of age and those who started ART before September 11, 2013, or after March 10, 2016 were excluded from the study and CD4 + counts per mm3 of blood were collected approximately every six months. Finally, of the enrolled subjects, each of the 322 patients was followed 6 times at 6-month intervals, and finally 1,932 observations were recruited.
The response variable in this study was each person's CD4 + count, measured approximately every 6 months. The CD4 + number is a continuous variable measured in an individual patient and is expressed as the number of cells per cubic millimeter (cells/mm3) of blood. Whereas, the independent variables were baseline patient age, baseline CD4 + count, time to follow-up (in months), patient’s sex, patient marital status at baseline, WHO clinical stage, patient grade of treatment, body mass index (BMI) at baseline, patient’s functional status and patient’s level of education.
Exploratory data analysis was used to gain insight into the raw data and individual curves were also performed before the model was formally fitted. Therefore, individual profile plots were used to observe the evolution of a particular subject over time and to determine which random effects should be included in the model. Mean profiles are tested to select a fixed effect for the model, and subgroup mean profiles are also performed separately to check for possible differences between groups. A variance structure was also plotted to explore changes in CD4 + counts over time compared to baseline. Finally, a linear mixed-effects model was performed to determine predictors of CD4 progression.
A key characteristic of linear mixed-effects models is that the mean response is modeled as a combination of a population characteristic β, assumed common to all people, and thematic effects unique to a particular individual. The former is called the fixed effect and the latter is called the random effect by Laird and Ware (2004).This linear mixed effects model can be expressed as:
Where β is the \(p\times 1\) vector of fixed effects, \({b}_{i }\)is the \(q\times 1\) vector of random effects, \({X}_{i}\) is the \({n}_{i}\times p\) matrix of covariates for fixed effects \({Z}_{i}\) is a matrix of covariates for random effects with \(q\le p\) and \({e}_{i}{\prime }s\) considers measurement errors. The random effects\({ b}_{i}\), and the error term\({ϵ}_{i}\), assumed to have a multivariate normal distribution with \(E\left[\begin{array}{c}{b}_{i}\\ {ϵ}_{i}\end{array}\right]\)=\(\left[\begin{array}{c}0\\ 0\end{array}\right]\) and\(Cov\left[\begin{array}{c}{b}_{i}\\ {ϵ}_{i}\end{array}\right]=\left[\begin{array}{c}D\\ R\end{array}\right]\), where, D is between-subject covariance matrix and R is within-subject covariance matrix.
The form of induced random effect covariance structure in the linear mixed model, first distinguish the conditional mean of \({Y}_{i}\) given random effects\({b}_{i}\),
From the marginal population averaged mean of\({Y}_{i}\), when averaged over the distribution of random effects\(,{b}_{i}\) is \(E\left({Y}_{i}\right)={\mu }_{i}\) \({=X}_{i}^{{\prime }}\beta\).
Maximum likelihood (ML) was used to estimate the parameters. The best fit model was selected using Akaike Information Criterion (AIC), Bayesian Information Criterion (BIC), and Likelihood Ratio (LR) Criterion. Residual plots, on the other hand, are used to test the normality of these effects and to identify categories of external effects. Statistical analysis system, SAS-Version 9.4 was used for statistical analysis. For statistical tests, a 5% significance level is considered.
The requirement for Informed consent was waived by Institutional Review Board (IRB) of Debre Berhan University (DBU) for the current study. A statement to confirm that all experimental protocols were also approved by Institutional Review Board (IRB) of Debre Berhan University (DBU) and the study was conducted according to the principles of the Declaration of Helsinki.
A total of 322 HIV-positive adults receiving ART at the DBRH were included in the study. And, these patients measured 1932 CD4 levels during 7 years of follow-up. Of the total of 322 patients, 225 (69.88%) were females. Regarding the patient’s marital status, 155 (48.14%) were married and only 45 (13.98%) was divorced. Of all patients under study, 90 (27.95%) did not educated, 132 (40.99%) have a certificate of primary school, and the remaining 31.06% had secondary and higher education (Table 1).
Of the patients in the study, 296 (91.93%) had working functional status, whereas the remaining 8.07% less than one-tenth were in ambulatory and bedridden status. Moreover, 141 (43.79%) were in the first WHO clinical stage, 72(22.36%) were found in the second stage, 98(30.43%) were in the third stage and only 3.42% of patients were in the fourth stage. When looking at baseline ART regimen classes, 22 (6.83%) of the patients were starting with the AZT-3TC-EFV class of ART regimen, 36(11.18%) were the AZT-3TC-NVP ART type, 27(8.39%) were the TDF-3TC-NVP, 215(66.77%) were the TDF-3TC-EFV class of ART regimen and the remaining 22(6.83%) were initiated on other different ART regimen classes (Table 1).
|
Variables |
Categories |
Frequency |
Percentage, % |
|---|---|---|---|
|
Sex |
Males |
97 |
30.12 |
|
Females |
225 |
69.88 |
|
|
Marital Status |
Married |
155 |
48.14 |
|
Never married |
66 |
20.50 |
|
|
Divorce |
45 |
13.98 |
|
|
Separated and widowed |
56 |
17.39 |
|
|
Educational Level |
No education |
90 |
27.95 |
|
Primary |
132 |
40.99 |
|
|
Secondary |
70 |
21.74 |
|
|
Tertiary |
30 |
9.32 |
|
|
Functional status |
Working |
296 |
91.93 |
|
Ambulatory |
23 |
7.14 |
|
|
Bedridden |
3 |
0.93 |
|
|
WHO Clinical stage |
Stage I |
141 |
43.79 |
|
Stage II |
72 |
22.36 |
|
|
Stage III |
98 |
30.43 |
|
|
Stage IV |
11 |
3.42 |
|
|
Regimen class |
AZT-3TC-EFV |
22 |
6.83 |
|
AZT-3TC-NVP |
36 |
11.18 |
|
|
TDF-3TC-NVP |
27 |
8.39 |
|
|
TDF-3TC-EFV |
215 |
66.77 |
|
|
Others |
22 |
6.83 |
The mean ages of study participants with standard deviation were 36.8 and 10.12, respectively. Patients' minimum and maximum CD4 cell counts at baseline were 7 and 1344, respectively, with an average of 335.7. However, the average CD4 cell count at baseline increased from 335.7 to 408.61 within 7 years. Patients' body mass index (BMI) mean and standard deviation at baseline were 21.34 and 2.78, respectively (Table 2).
|
Variables |
N |
Min. |
Max. |
Mean |
Std. Dev |
|---|---|---|---|---|---|
|
Age |
322 |
16 |
66 |
36.8 |
10.12 |
|
BMI |
322 |
11.26 |
27.92 |
21.34 |
2.78 |
|
Baseline CD4 count |
322 |
7 |
1344 |
335.7 |
222.82 |
|
CD4 count after follow-up |
1932 |
7 |
1894 |
408.61 |
230.90 |
Since the actual CD4 + count data was not normally distributed at initial and should undergo possible transformations before a complete analysis. As a result, we performed logarithmic and square root transformations and applied various normality checking methods such as QQ-plot, histograms, and Shapiro-Wilk tests. The CD4 + count data was then become normalized using the square root conversion method.
Individual trajectories were plotted against a patient's CD4 count (Fig. 1). As can be seen from the graph, in most patients, square root CD4 counts were in the range of 22 cells/mm3 at the intersection of each trajectory, with significant differences. Similarly, some trajectories are steeper and others are almost horizontal, indicating possible differences in CD4 count slopes. Therefore, the use of the mixed model fits the data very well due to the slope of the intercept and the slope of the trajectory.
Individual profile plots, as shown in Fig. 1, show that most of the square root CD4 + cell counts are concentrated around 22, the CD4 + cell count varies significantly at baseline than at the end, and it has to appear increasing. Despite individual profile graphs showing thematic changes in the number of CD4 + patients over time, the graph did not essentially determine what the evolution of each subject looked like as the number of observations increased over time (Fig. 1).
In this regard, a more meaningful graph is created, which is an overlay of individual profiles and a graph showing the average trend (Fig. 2). Therefore, the single red curve in Fig. 2 shows that the mean square CD4 + count plot is linearly related to time. This is because the square root of the number of CD4 + cells increases until the 12th month (2nd follow-up time), but the rate of increase slows down after the 12th measurement point and becomes almost constant near the 24th month (4th follow-up time).
The changes in CD4 + cell counts were more pronounced in women than in men, and the effect was dependent on the patient's baseline CD4 + count. It implies that patients with a high CD4 + count of at baseline had a high rate of change during follow-up periods (Fig. 3).
Autoregressive (1) or type-one covariance structure was taken based on its smaller ‘-2res LL, AIC, and BIC’ values after comparing exchangeable, autoregressive (1) and unstructured covariance (Table 3).
|
Test |
Exchangeable |
Autoregressive(1) |
Unstructured |
|---|---|---|---|
|
-2 Res Log Likelihood |
10319.8 |
10307.6 |
10318.7 |
|
AIC (Smaller is Better) |
10326.7 |
10323.8 |
10323.8 |
|
BIC (Smaller is Better) |
10332.6 |
10331.4 |
10341.8 |
The intra-class correlation coefficient (ICC), which represents the variation between bins with respect to the total variation, is calculated to be 47.1% in the null model. This value is above the minimum recommended value (25%) [25], so higher ICC values further enhance the suitability of using a mixed model for the data. Or it can be suggested that the data structure is best captured by using the random effects model. The total variability associated with intra-subject variability was 0.852, indicating that 85.2% of the data variability was due to intra-subject variability. Variations between subjects accounted for the remaining 14.8% of the total variability (Table 4).
|
Covariance Parameter |
Estimate |
|---|---|
|
Intercepts, b1i |
0.791 |
|
Covariance, Cov(b1i, b2i) |
0.701 |
|
Random slope-AR(1), b2i |
0.887 |
|
Residual, €ij |
9.704 |
AR (1): Fist order autoregressive
The variance covariance estimates in Table 4 are generally summarized in the matrix below.
In the "G" matrix, a value 0.791 represents the variance of the random intercept. The value 0.887 is the variance of the slope with respect to the time effect, and 0.701 is the covariance of the random intercept and slope over time. A value of 0.701 indicates a strong positive correlation between the slope and the intercept, implying that people with higher baseline CD4 + counts tend to have a higher rate of change over time and those with lower baseline CD4 + counts tend to have a lower rate.
Using the fixed effect estimates of the linear mixed model, the association of several factors with CD4 count change was investigated. Hence, predictors such as male patients (β =-0.7512, p-value = 0.019), age at the initiation of ART (β = -0.02705, p-value = 0.047), bedridden functional status of the patient at the initiation of ART (β = -0.03365, p-value = 0.021), TDF-3TC-NVP regimen class (β = -0.1474, p-value = 0.031), never married (β = 0.610, p-value = 0.011), patients’ WHO clinical stage-II (β = -0.402, p-value = 0.047), baseline CD4 count (β = 0.020, p-value = 0.0001) and follow-up time (β = 0.613, p-value = 0.0001) were positively as well as negatively associated and had significant impact on CD4 count progression. Whereas, patients educational level and body mass index (BMI) were not found to be significantly associated with CD4 count change (Table 5).
|
Effect |
Categories |
Estimate, βi |
Sd. Error |
t-value |
Pr > |t| |
|---|---|---|---|---|---|
|
Intercept |
Continuous |
11.9299 |
1.391 |
8.58 |
.0001*** |
|
Sex |
Male |
-0.7512 |
0.320 |
-2.35 |
0.019** |
|
Female |
Reference |
. |
. |
. |
|
|
Age |
Continuous |
-0.02705 |
0.014 |
-1.90 |
0.047** |
|
BMI |
Continuous |
0.01115 |
0.051 |
0.22 |
0.126 |
|
Functional Status |
Ambulatory |
-0.3617 |
0.533 |
-0.68 |
0.138 |
|
Bedridden |
-0.03365 |
1.416 |
-0.02 |
0.021** |
|
|
Working |
reference |
. |
. |
. |
|
|
Regimen Class |
AZT-3TC-NVP |
0.9382 |
0.645 |
1.45 |
0.216 |
|
TDF-3TC-NVP |
-0.1474 |
0.692 |
-0.21 |
0.031** |
|
|
TDF-3TC-EFV |
0.1073 |
0.533 |
0.20 |
0.841 |
|
|
Others |
0.9957 |
0.745 |
1.34 |
0.181 |
|
|
AZT-3TC-EFV |
reference |
. |
. |
. |
|
|
Educational Level |
Primary |
0.024 |
0.335 |
0.07 |
0.943 |
|
Secondary |
-0.046 |
0.388 |
-0.12 |
0.906 |
|
|
Tertiary |
-0.300 |
0.520 |
-0.58 |
0.565 |
|
|
No-educated |
reference |
. |
. |
. |
|
|
Marital Status |
Never Married |
0.610 |
0.382 |
1.59 |
0.011** |
|
Divorced |
0.509 |
0.419 |
1.21 |
0.225 |
|
|
Separated &widowed |
-0.280 |
0.388 |
-0.72 |
0.470 |
|
|
Married |
reference |
. |
. |
. |
|
|
WHO Clinical Status |
II |
0.203 |
0.353 |
0.57 |
0.565 |
|
III |
-0.096 |
0.342 |
-0.28 |
0.779 |
|
|
IV |
-0.402 |
0.781 |
-0.51 |
0.047** |
|
|
I |
reference |
. |
. |
. |
|
|
Baseline CD4 |
Continuous |
0.020 |
0.001 |
30.60 |
.0001*** |
|
Time |
Continuous |
0.613 |
0.065 |
9.37 |
.0001*** |
**P-value <0.05, ***P-value <0.001 significant
Insignificant Shapiro-Wilk test was estimated (0.9891, p-value < 0.8431), and it suggested that there is insufficient evidence to conclude that the model residuals are not normally distributed at the 5% level of significance. Furthermore, the histogram and QQ-plots displayed in (Figs. 4 (a) and 4(b)) for residual, respectively, also indicated that the residuals for square root CD4 count of patients were normally distributed.
This study aimed to identify predictors associated with changes in CD4 cell count of patients those who started ART at Debre Berhan Referral Hospital from September 11, 2013, to February 2019. After the data was collected from the hospital, various exploratory and model-based analyzes were performed. The normality of CD4 + cell count was verified using the QQ plot, histogram, and Shapiro-Wilk test prior to detailed analysis. Logarithmic and square root transformations were performed after the plot led to a test showing that the CD4 + count data needed to be converted. However, the square root transformation of CD4 + count proved to be more normal than the log transformation.
In exploratory analyses, the mean profile graph shows that mean CD4 + counts increase until the twelfth month and then show a fairly consistent rate of increase over time. This is supported by results from [26] who determined that after patients started ART, their CD4 + counts increased by up to a third point from baseline. It was also observed that HIV-positive female patients receiving ART had a higher mean CD4 + count change than males in the mean profile plot. The stochastic intercept and stochastic slope models were chosen as the final random effects models based on the local correlation to which the total variability was attributed to each random model.
Estimates of the fixed effects were also confirmed using an exploratory plot in which the mean squared rate of change in CD4 + cell count was 0.75 times significantly lower in males than in females (p-value = 0.019). It was consistent with [22] that women responded more to ART than men. Others [16], [21], [23] also reported that patient gender had a significant effect on changes in mean CD4 cell count. However, they did not indicate which gender was superior in changing the patient's CD4 cell count. In contrast, a study conducted in northwestern Ethiopia [18] found that gender was not a significant predictor of CD4 + cell number progression. The general rationale is that women can participate in voluntary counseling and testing as part of routine health care during pregnancy, but male patients are less likely to be tested for HIV and therefore do not seek treatment.
Baseline age was negatively correlated with the CD4 count response and was supported by other studies [14], [16], [17], [22], [23]. When patients at the start of ART were 1 year older, the mean change in CD4 cell count was 0.0271 times lower than in patients 1 year younger (p-value = 0.047). Elderly people have also been shown to be late for the clinic, which may slow their response to ART treatment [13], [18].
The findings of this study showed that at the onset of ART, the mean change in CD4 counts was 0.034 square-fold lower in patients with the bedridden functional state than in patients in the working state (p-value = 0.021). This was confirmed by previous studies [13], [18], [19], where patients with working functional status showed better improvement in CD4 counts after initiating ART. However, other studies such as [17], [22]did not specify which patients were affected and simply assumed that the patient's functional status was an important predictor of mean changes in the patient's CD4 number.
Patients who started TDF-3TC-NVP ART therapy had fewer changes in CD4 count compared to patients who started AZT-3TC-EFV therapy (β = -0.1474, p-value = 0.031). Other studies have shown that d4t-3TC-NVP and AZT-3TC-NVP [12] and AZT + 3TC + EFV [20] are types of therapy that significantly affect changes in mean CD4 counts in HIV-infected patients. In contrast, a study in northeastern Ethiopia [13], suggested all ART regimens with no significant differences in increasing mean CD4 counts over the study period.
In this study, the average CD4 + counts of single patients was 0.610 squared higher than that of married patients (p-value = 0.011). However, widowed and separated patients had lower mean changes in CD4 counts over time, although no statistically significant differences were observed. In other studies [14], [16], [17], marital status was also found to be an important predictor of changes in a patient's CD4 count, but patient specific status was not reported.
The current study has shown that baseline WHO stage IV was a factor with negative significance for the progression of CD4 + cell counts. HIV-infected patients had WHO stage IV at baseline ART and had a 0.402-fold lower change in mean CD4 cell count than patients with WHO stage I (β = − 0.402, p-value = 0.047). In contrast, studies [20], [24] revealed that the baseline WHO Phase II clinical stage is an important predictor of CD4 cell progression in patients.
Baseline CD4 cell count was found to be one of the key predictors of changes in patient CD4 cell count (p-value = 0.0001). Patients with high baseline CD4 tended to show a rate of change 0.020 square times higher than patients with low baseline CD4. And, it was supported by many other studies [13], [16], [17], [21], [22], [27]. Another study [28]also found that patients with high pre-ART CD4 counts had a less long-term increase in CD4 counts during ART. Follow-up time also proved to be positively significant in relation to the mean change in the patient's CD4 cell count, consistent with other studies [12], [14], [16], [18], [21]–[23].
This study showed that at baseline, the mean CD4 count increased from 335.7 to 408.61 over 7 years. Moreover, patients’ gender, age at initiation of ART, bedridden functional status at ART initiation, TDF-3TC-NVP treatment class, unmarried marital status, WHO clinical stage II, baseline CD4 count and follow-up time was found to be a significant predictor of the progression of a patient's CD4 count. On the other hand, education level and obesity index were not significantly associated with the change in CD4 count. Therefore, HIV-positive patients can be advised to start ART treatment as early as possible. Special guidance and attention is also required, especially in elderly patients, males with bedridden functional status, and late WHO clinical stage patients.
AIC: Akaike Information Criterion
AIDS: acquired immunodeficiency syndrome
ART: Antiretroviral therapy
BIC: Bayesian Information Criterion
BMI: Body mass index
HAART: Highly active antiretroviral therapy
HIV: Human immunodeficiency virus
LR: Likelihood Ratio
RNA: Ribonucleic acid
WHO: world health organization
Data Availability
The datasets used and analyzed during the current study are available from the corresponding author upon reasonable request.
Conflict of interest
The author declared that there is no conflict of interest.
Funding Statements
Not applicable
Acknowledgments
Authors would like to acknowledge Debre Berhan Referral hospital for providing the study data
Author’s contributions
A.W Kassie done all the study design, data management and analysis, interpretation, report writing, reviewing and editing.