Incidence of opportunistic infections and its predictors among HIV/AIDS patients on antiretroviral therapy in Gondar University Comprehensive and Specialized Hospital, Ethiopia

Abstract

Introduction: Opportunistic infections (OIs) are the leading cause of morbidity and mortality among adults living with HIV. Current and accurate information about the occurrence of opportunistic infections in HIV-infected adults is critical for developing more effective treatments and interventions. However, few studies have been conducted in Ethiopia on the prevalence of common opportunistic infections in HIV-infected adults. Thus, the purpose of this study was to determine the prevalence and predictors of opportunistic infections among HIV-infected adults receiving antiretroviral therapy (ART) at the comprehensive specialized hospital affiliated with the University of Gondar.

Methods: Between January 11, 2015, and January 10, 2021, a retrospective cohort study was conducted at the University of Gondar comprehensive specialized hospital. A total of 715 HIV-infected adults on ART were included in the study. Data were extracted from the charts of HIV-infected adults using a data extraction form adapted from the ART entry and follow-up forms. Epi-dataTM Version 4.5 was used to enter data, and StataTM Version 16 was used to analyze the data. The time interval between opportunistic infections was estimated using the Kaplan Meier survival curve. To identify risk predictors of opportunistic infections, bivariate and multivariate semi-parametric and parametric regression models were fitted.

Result: This study included the records of 715 HIV-infected adults-initiated ART between January 11, 2015, to January 10, 2021. During the follow-up period, the overall incidence of opportunistic infections was 4.1 (95 percent CI 3.74 to 4.44) per 10,000 person-year observation, with a median of 57 months (IQR = 40–69 months). Pneumocystis’ pneumonia at 90(16.51%) was the most encountered OI at follow-up. Adults are presenting with baseline CD4 < 200 cells/µl counts (AHR = 1.41, 95% CI 1.18 to 1.69), bedridden baseline functional status (AHR = 1.35, 95% CI 1.01 to 1.82), WHO clinical stage II (AHR = 5.87, 95% CI 3.97 to 8.69) and WHO clinical stage III (AHR = 5.85, 95% CI 3.55 to 9.65) were notably associated with the incidence of opportunistic infections development.

Conclusions: Opportunistic infections are uncommon among HIV-infected adults in this study. In terms of predictors, such as a low CD4 count and an advanced WHO stage (II or III), bedridden functional status was found to be significantly associated with OIs.

Keywords:

• Adults
• antiretroviral therapy
• HIV
• opportunistic infections
• Ethiopia

Background of the study

HIV/AIDS infections continue to be a significant cause of morbidity and mortality worldwide. In a 2018 report, an assessed 37.9 million people globally were living with HIV 54% of PLHIV reside in Eastern and Southern of Africa.¹ Sub-Saharan Africa (SSA) is the most maximum affected area, including Ethiopia, ranking within the top twenty-five countries with the highest new HIV infection rates. In the 2017 report, 36.9 million in the world live with HIV/AIDS, and new infections have seen a reduction by 18% since 2010. Nevertheless, this rate of deterioration is not adequate for the goal of eradication of AIDS by 2030. Only 21.7 million people infected with HIV have access to antiretroviral therapy, with the rest at risk of the potential difficulties of HIV infection.²

Opportunistic infections (OIs) are diseases, which happen more commonly and severely among individuals with declined immune systems of their body, including the most declined immune systems group of people like PLHIV.³ All HIV-infected individuals are susceptible to develop a wide range of opportunistic infections,^4,5 but the prevalence and incidence of HIV-related opportunistic infections vary widely from place to place,

Time to time and individuals to induvial.⁵ Opportunistic infections are the important causes of morbidity and mortality among HIV-infected induvial contributive to 94.1% of HIV-related mortality.^6–8 Without prompt treatment, opportunistic infections adversely affect the treatment outcomes of PLHIV, resulting in decreased quality of life, accelerated disease progression, increased medical costs, increased risk of treatment failure, and impaired patient outcomes’ response to ART drugs.⁹

The World Health Organization (WHO) suggested many varieties of medical treatments to decrease the rate of opportunistic infections among PLHIV. These different treatments include reduction of exposure, chemoprophylaxis (primary/secondary), immunization, and early initiation of ART.¹⁰ The use of highly active antiretroviral therapy (HAART)has effectively reduced opportunistic infections significantly among different age groups of infected individuals living with HIV.¹⁰ In Ethiopia, the Ministry of Health (MOH) has been applying different treatment options and methods to improve the survival of HIV infected individuals, which is reflected, in part, in the increased ART coverage service of the country from 5 percent in 2010 to 9.5 percent in 2013.¹⁰

Even though the incidence of opportunistic infections has decreased significantly since the introduction of HAART, they remain a major cause of morbidity and mortality in a variety of vulnerable populations, including immunocompromised adults.¹¹

Morbidity and mortality associated with HIV disease are caused by underlying immunosuppression, which results in life-threatening opportunistic infections (OIs) during the disease’s natural course.² Since the mid-1990s, widespread use of ART has had the greatest impact on reducing opportunistic infections associated with mortality in HIV-infected individuals in countries where these treatments are accessible and affordable.^12,13 The widespread use of antiretroviral therapy (ART) has had the greatest impact on reducing opportunistic infections associated with HIV-related mortality in countries where these therapies are accessible and affordable.^12,13 This mortality occurs as a result of certain patients’ failure to maintain a sustained response to antiretroviral agents for a variety of reasons, including poor adherence, drug toxicity, drug interactions, or initial acquisition of a drug-resistant strain of HIV-1.¹⁴ However, even with increased ART, opportunistic infections continue to cause morbidity and mortality in HIV/AIDS patients.

Some patients do not have a sustained response to antiretroviral agents for multiple reasons, including poor adherence, drug toxicities, drug interactions, or initial acquisition of a drug-resistant strain of HIV-1. Therefore, opportunistic infections continue to cause substantial morbidity and mortality in patients with HIV-1 infection despite the use of ART. Even though two different studies, such as in Gondar and Areba-Mnchi Hospital, have been conducted on the incidence of opportunistic infections among HIV infected patients on ART in Ethiopia, updated information about the incidence of opportunistic infections and its predictors is scarce in the University of Gondar comprehensive specialized hospital, northwest Ethiopia. Therefore, this study determined the incidence of opportunistic infections and identified its predictors from patients taking ART drugs in the University of Gondar comprehensive specialized referral hospital, northwest Ethiopia.

Methods and materials

Study design, setting, and period

From January 11, 2015 to January 10, 2021, a retrospective study was conducted on HIV-positive patients on HAART who attended chronic HIV care clinics at the University of Gondar comprehensive and specialized hospital in Amhara regional state, northwest Ethiopia. The comprehensive and specialized hospital of the University of Gondar is located in northwest Ethiopia, serving a population of approximately 5 million. The catchment population has been reported to have a variety of diseases, both communicable and non-communicable. Outpatient clinics, maternity clinics, emergency wards, adult in-patients, pediatric in-patients, community clinics, and laboratory services are all examples of health services units.¹⁵ The hospital has 518 beds and approximately 350 to 400 patients per day, of which approximately 100–120 visit the emergency unit. The hospital has four emergency rooms and a triage unit. It employs approximately 270 nurses and 150 physicians.¹⁶

Since 2012, the hospital has provided HIV care and ART follow-up. At the moment, this Hospital’s ART clinic has 5481 ART follower patients, one physician, ten nurses, six data clerks, three porters, two cleaners, six case managers, and seven ART education and adherence counselors. The hospital employs standardized ART monitoring and evaluation tools that were adapted from Ethiopia’s comprehensive HIV care and treatment guidelines. Currently, this site serves a total of 900 HIV-infected adults on a monthly basis, and based on this calculation, it also serves a total of 5400 ART patients annually.

Study participant

The study population included all HIV-infected adults (age 18 years) who had ever been started on antiretroviral therapy (ART) at the University of Gondar’s Comprehensive and Specialized Hospital. Between January 11, 2015, and January 10, 2021, they received at least one follow-up visit. Adults infected with HIV whose ART initiation dates were unknown and who were admitted with incomplete baseline data (CD4 count, hemoglobin level, WHO clinical stage, weight, and height) were excluded from the study.

Sampling procedures

The records of all HIV-infected adults ever started on ART (715) at the University of Gondar comprehensive and specialized hospital were recruited. After excluding incomplete records with substitution, records of 715 HIV-infected adults met the criteria and were included in the study. Data were extracted from the charts of 715 HIV-infected adults on ART.

Study variables

During follow-up, the dependent variable was the occurrence of any type of opportunistic infection. Sociodemographic predictors (i.e. age, sex, marital status, religion, ethnicity, educational attainment, occupation, address, residence, number of households, caregiver, caregiver relation, disclosure, to whom disclosure, and spouse HIV status) were included as independent variables. Clinical and treatment predictors (i.e. baseline CD4 count, viral load, opportunistic infectious disease status, baseline WHO clinical stage, baseline weight, baseline functional status, and prior opportunistic illness); and ART and other medication-related predictors (i.e. eligibility criteria, date of ART initiation, type of baseline ART regimen, original regimen used when the patient first begins ART, and prior opportunistic illness). Is their regimen changing during the follow-up period, and if so, why, the date of switching to a second line regimen, if switched, and Adherence) were selected as independent variables.

Operational definitions

Opportunistic Infections (OIs) are infections that are more prevalent and severe in individuals with weakened immune systems, such as those living with HIV.^3,5,17 Herpes zoster, bacterial pneumonia, pulmonary tuberculosis, extra-pulmonary tuberculosis, oral candidiasis, esophageal candidiasis, mouth ulcer, diarrhea, pneumocystis pneumonia, cryptococcal meningitis, non-Hodgkins lymphoma, Kaposi’s sarcoma, and cervical cancer are all common OI.

In this study, an event occurred when an HIV-infected adult developed any type of OI during the follow-up period after initiating ART.

It was censored when HIV-infected adults were discharged or transferred (dead or alive) to other health institutions or remained on active ART follow-up but did not develop any OIs by the study’s end.

Death: Those patients whose outcome was recorded as died on the follow-up chart.

Lost to follow up: Patients who have missed more than three months of appointment to the same health center since the last planned appointment.^19,20

Transferred out: If PLHIV on HIV care in one health institution shift to another health institution was transferred out of this study.

The Body Mass Index (BMI) is a straight forward weight-for-height index that is frequently used to classify nutritional status as underweight when the BMI is less than 18.50 kg/m², normal when the BMI is between 18.50 and 24.99 kg/m², and overweight when the BMI is greater than 25.00 kg/m².²¹ If adherence was 95%, the percentage of missed doses was two doses of 30 doses or three doses of 60 doses), as documented by ART health personnel.

Fair Adherence: If 85–94 percent adherent, the percentage of missed doses was 3–5 doses of 30 doses or 3–9 doses of 60 doses), as documented by ART health personnel. If adherent at 85%, the percentage of missed doses was three doses of 30 doses or more than nine doses of 60 doses), as documented by ART health personnel.²²

A caregiver is someone who assists others in self-care. This may include patients who are disabled or who have chronic illnesses. These individuals may include physicians, nurses, family members, friends, and social workers.²³

Procedures for data collection and quality assurance

A pretested structured data collection checklist was used to extract routinely recorded data from adult HIV patients who initiated ART treatment in the hospital from march 2021 up to May 2021. All charts containing detailed data, including the baseline CD4 count, viral load, hemoglobin amount, and repeated measures of CD4 cell counts almost every six months was reviewed. Similarly, other characteristics, like socio-demographic (age and sex), clinical (baseline WHO clinical stage, viral load, baseline functional status), residence, baseline regimen, and others, were also collected from ART users’ registration book. Three health professionals, about two clinical nurses, were recruited as data collectors, and one health officer was recruited as a supervisor and trained for two days to retrieve the data. The principal investigator and the supervisor closely monitored the process throughout the data collection period and made corrections. Any laboratory tests performed prior to initiating ART were considered baseline data. If laboratory tests were not performed prior to initiating ART, any tests performed within a month of initiating ART were considered the baseline. The data quality was ensured by pretesting the tool by taking a sample of 25 charts and training data collectors and supervisors, close supervision, and prompt feedback. The training gave instructions on extracting techniques as per data extraction format after announcing the study’s objective. The data were checked for inconsistencies, completeness, accuracy, clarity, missing values, and appropriate corrections were taken by the principal investigator and the supervisor consistently daily.

Data processing and analysis

After ensuring the data’s quality, forms were collected and assigned a sequential number (code) to facilitate data entry. Epi-data version 4.6 software was used to enter data, which was then transferred to Stata version 16 for statistical analysis. Inconsistencies, coding errors, completeness, accuracy, clarity, and missing values are all factors to consider. were checked in the data. To describe the study participants’ characteristics, descriptive statistics were used, including proportions, tables, figures, IQR, and graphs. Survival times were estimated using the Kaplan-Meier and log-rank tests, and survival curves were compared between different exposure groups. The incidence rate of opportunistic infection was determined for each patient characteristic, as well as an overall incidence rate. The Schoenfeld residuals test was used to visually check the proportional hazards assumption and the multicollinearity of each selected predictor. By considering Weibull, exponential, Gompertz, and lognormal baseline distributions, we can create Cox proportional hazard and parametric survival models. Log-likelihood, Akaike Information Criteria (AIC), and Bayesian Information Criteria were used to select the best model (BIC). To determine the model’s goodness of fit, the Cox Snell residual test was used. We estimated adjusted HRs (AHRs) and their respective 95% confidence intervals (CIs), with a p value less than 0.05 indicating a significant association between different predictor variables.

University of Gondar’s College of Medicine and Health Sciences, Institute of Public Health ethical review committee granted clearance and approval to conduct the research. Due to the fact that this study analyzed secondary data from patient charts, we were granted an informed consent waiver. To maintain confidentiality, the data collection tool did not include names or other personally identifiable information such as unique identification numbers.

Result

Socio-demographic characteristics of HIV/AIDS patients on ART treatment

A total of 715 HIV-infected adults on ART were included in the final analysis. More than half, 414 (57.9%), of the study participants were females. By the time of enrollment into ART care, 292 (40.84%) of the participants were between 31 and 40 years of age. The majority of the study participants were married, accounting for 521 (72.87%) of the total sample. About educational status, 293 (40.98%) of patients had secondary education. About 270 (37.76%) occupational status was a merchant. Most of 542 (75.08%) the study participants spouse HIV status were positive (Table 1).

Table 1. Socio-demographic variables of HIV/AIDS patients on ART treatment at University of Gondar Compressive Specialized Hospital, January 11, 2015, to January 10, 2021.

Variables	Category	Frequency(N)	Percentage (%)
Sex	Female	414	57.9
	Male	301	42.1
Age	18–25	47	6.57
	26–30	92	12.87
	31–40	292	40.84
	41–50	190	26.57
	51–60	75	10.49
	61–76	19	2.66
Marital	Single	112	15.66
	Married	521	72.87
	Divorced	82	11.47
Religion	Orthodox	377	52.72
	Muslim	251	35.1
	Protestant	87	12.17
Education	No education	120	16.78
	Elementary	137	19.16
	Secondary	293	40.98
	Tertiary	165	23.08
Occupation	Student	36	5.03
	Daily labor	84	11.75
	Housewife	93	13.01
	Merchant	270	37.76
	Government	232	32.45
Residence	Urban	712	99.58
	Rural	3	0.42
House hold	1–2	12	1.68
	3–5	417	58.32
	6–8	286	40
Care giver	Yes	710	99.3
	No	5	0.7
Disclosure	Disclosed	709	99.16
	Not disclosed	6	0.84
Spouse HIV status	Positive	542	75.08
	Unknown	173	24.2

Clinical and treatment characteristics of HIV/AIDS patients on ART treatment

Among the study subjects, 464 (64.89%) had baseline CD4 count >200 cell/μl. Based on baseline WHO clinical stage 516 (72.17%) of patients were stage II followed by stage I patients, 156 (21.81%). Most, 399 (55.80%) of the participants had a baseline working functional status, and 501 (70.06%) of study participants were taken 1^st line regimen. The majority of the study participants were normal BMI accounting for 634 (88.67%). About 27 (3.78%) were had a good drug adherence (Table 2).

Table 2. Clinical variables of HIV/AIDS patients on ART treatment at University of Gondar Compressive Specialized Hospital, January 11, 2015, to January 10, 2021.

Variables	Category		Frequency(N)	Percentage (%)
OIs	Yes		545	76.22
	No		170	23.78
Prophylaxis		CPT	272	38.04
		INH	297	41.53
		Fluconazole	146	20.42
CD4 count		≤200 cell/µl	251	35.1
		>200 cell/µl	464	64.89
Hgb		≤12	589	82.38
		13–17.5	126	17.62
BMI		Normal	634	88.67
		Mild	66	9.23
		Moderate	15	2.09
Functional Status		Bed ridden	69	9.65
		Ambulatory	247	34.54
		Working	399	55.80
WHO clinical stage		Stage I	156	21.81
		Stage II	516	72.17
		Stage III	43	6.01
Side effect		Yes	353	49.37
		No	362	50.63
Appointment		Yes	561	78.46
		No	154	21.54
ART Drug		1^st line regimen	501	70.06
		2^nd line regimen	3	0.42
		3^rd line regimen	211	29.51
Drug adherence		Good	27	3.78
		Fair	2	0.28
		Not recorded	686	95.84

Opportunistic infections incidence from HIV/AIDS patients on ART treatment

Study subjects were followed for a specified period with a median time of 57 months (IQR = 40–69 months). The patient’s minimum time was followed for one month, and the maximum time was 72 months after the start of ART treatment. Based on this, the total person-time observation was found to be 1283449 person-year. During the follow-up period, 76.22% (CI, 72.95% to 79.21%) of participants developed different opportunities for infections.

Out of a total of OIs development, the largest proportion, 90 (16.51%) was found to be Pneumocystis’ pneumonia followed by Chronic diarrhea 89 (16.33%), Bacterial pneumonia 59 (10.82%), Pulmonary tuberculosis 57(10.46%), and the smallest proportion was found to be wasting 1 (0.18%) (Figure 1).

Figure 1. Frequency distribution of the type of OIs disease among HIV/AIDS patients on ART treatment at University of Gondar compressive specialized Hospital, January 11, 2015, to January 10, 2021.

The overall incidence of OIs was found to be 4.1 (95% CI 3.74 to 4.44) per 10,000 person year observation and from this, the incidence of Zoster was 13.5 (95% CI: 10.29,17.7), pneumocystis pneumonia 11.86 (95% CI: 0.96, 14.65), CNS toxoplasmosis 11.87 (95%CI: 0.87, 16.19), Cryptococci meningitis 10.88 (95%CI: 7.46,15.87) bacterial pneumonia 12.64 (95%CI:0.97, 16.43), pulmonary tuberculosis 13.94 (95%CI: 10.73, 18.12), thrush oral 10.58 (95%CI: 0.70, 15.92), esophageal candidiasis 15.97 (95%CI: 0.96, 26.50), ulcer mouth 17.34 (95%CI: 12.62, 23.83), chronic diarrhea12.59 (95%CI: 10.17, 15.60) and acute diarrhea12.07 (95%CI: 0.80,18.17) were the most different incident type of opportunistic infections

The cumulative probability of developing OIs among HIV/AIDS patients who were free from any development at the start of the follow-up time was 0.9986, and there was 0.8154 at month 25, 0.5986 at month 50, 0.0000 at month 72 during the follow-up time. The median survival time was found to be 57 months (Figure 2).

Figure 2. The Kaplan-Meier curve showing the cumulative probability of OIs disease among HIV/AIDS patients on ART treatment at University of Gondar compressive specialized Hospital, January 11, 2015, to January 10, 2021.

Predictors of opportunistic infections among HIV/AIDS patients on ART treatment

Separate graphs representing the Kaplan-Meier function estimates for some of the categorical variables were estimated (Figure 3 and 4). The upper curve in each figure indicates that the particular group experiences more survival time than the one below. To investigate if there was a significant difference between the opportunistic infections across different base line CD4 count groups, this figure shows that the > 200 cell/μl base line CD4 count group had more survival time than that of ≤200 cell/μl baselines CD4 count group one.

Figure 3. The Kaplan Meier Survival Curves for covariates curve showing CD4 count, ART side effect of HIV/AIDS patients on ART treatment at University of Gondar compressive specialized Hospital, January 11, 2015, to January 10, 2021.

Figure 4. The Kaplan Meier Survival Curves for covariates curve showing Base line WHO clinical stage, baseline functional status of HIV/AIDS patients on ART treatment at University of Gondar compressive specialized Hospital, January 11, 2015, to January 10, 2021.

The log-rank statistical method was used to check whether there is a significant difference in the survival functions among categories shown using Kaplan Meier estimates of survival functions. Based on the result of the log-rank test, there was a significant difference in survival among categories of educational status, occupation, prophylaxis, baseline CD4 count, baseline functional status, ART side effect, and baseline WHO clinical stage. However, there is no significant difference among categories of age and sex. HIV/AIDS patients who have >200 cell/μl baselines CD4 count at the start of ART treatment had longer survival experience than HIV/AIDS patients with ≤ 200 cell/μl baselines CD4 count, which is supported by log-rank test (log-rank Chi²(1) = 34.73, p-value = 0.00).

HIV/AIDS patients who have baseline WHO Clinical Stage I at the start of ART treatment had longer survival experience than HIV/AIDS patients with baseline WHO Clinical Stage III, which is supported by log-rank test (log-rank Chi²(1)= 108.77, p-value = 0.00 (Table 3).

Table 3. The log-rank test results for categorical variables in OIs HIV/AIDS patients on ART treatment in University of Gondar Compressive Specialized Hospital, January 11, 2015, to January 10, 2021.

Variables	Df	Chi- square	P-value
Sex	1	0.01	0.76
Age	5	4.90	0.43
Education	3	26.02	0.00*
Occupation	4	9.66	0.04*
Prophylaxis	2	34.73	0.00*
CD4 count	2	34.73	0.00*
Functional status	2	18.19	0.00*
WHO Clinical Stage	3	108.77	0.00*
ART Side effect	1	5.78	0.02*

(*) The difference is significant at α = 0.05.

Assessing the proportional hazard assumption

To fit a model, we have to assess some requirements of the model that means the model should be assessed whether it describes our data well or not. In this instance, the primary objective was to validate. the proportional hazard assumption and to determine the model’s overall goodness of fit. The proportional hazard assumption states that the study subjects’ risk of failure must remain constant regardless of how long they are followed.

The log-log plot (survival probability) versus the log of survival time was done for different categories of predictor variables (Figure 4). As observed from the plot of two categories of baseline CD4 count and baseline Hgb amount, the two lines were nearly parallel, which means that the proportional hazard assumption was valid (Figure 5). On the other hand, the graphical display shows plots of the scaled Schonfield residuals against the survival time for each covariate: baseline CD4 count and baseline functional status of the patients was estimated (Figure 6). In plots of scaled Schoenfeld residuals show randomness. Moreover, the smoothed curve is an approximately horizontal line, so the assumption of proportional hazards was also satisfied repeatedly using this method.

Figure 5. Plot of log (−log(survival probability)) vs. log (survival time) by Hgb and baseline CD4 count for type HIV/AIDS patients on ART treatment at University of Gondar compressive specialized Hospital, January 11, 2015, to January 10, 2021.

Figure 6. Baseline functional status and CD4 count Shenofelide test for type HIV/AIDS patients on ART treatment at University of Gondar compressive specialized Hospital, January 11, 2015, to January 10, 2021.

The global test of proportional-hazards assumption based on the Schoenfeld residuals was also done, and it was found that all covariates and full model satisfy the proportional hazard assumption (Chi square= 17.32, p-value= 0.2999) in the process of model development.

Comparative analysis of models

After confirming the proportional hazard assumption, semi-parametric and parametric proportional hazard models were fitted to estimate the survival incidence of opportunistic infections and to identify predictors in HIV/AIDS patients. Using information criterion (AIC, BIC) and log-likelihood results, the most sparsity model was chosen. Based on all the three comparison techniques used, the Gompertz regression model (AIC = 1207.696, BIC = 133.147, log likelihood= −576.8479) was more efficient than Cox-PH and other parametric models. Interpretations and conclusions were thus be based on the Gompertz model (Table 4).

Table 4. Summary of model comparison between semi-Cox proportional hazard models and parametric Cox- Regression models using AIC, BIC, and log-likelihood.

Methods	Comparison models type
	Cox PH	Exponential	Gompertz	Loglogistic	Weibull	Lognormal
Log likelihood	−2954.238	−744.16944	−576.8479	−671.793	−598.20	−694.05
AIC	5958	1540.339	1207.696	1391.58	1250.4	1443.87
BIC	607.784	1659.218	133.147	1501.32	1373.8	1567.32

After fitting a univariate Gompertz proportional hazard model, all the predictor variables were found to have p-value <0.2; after this, a multivariable model was fitted, and covariates like baseline CD4 count, baseline bedridden functional status, and baseline WHO clinical stage II and III were found to be significant predictors for incidence of opportunistic infections among HIV/AIDS patients at 5% level of significance.

The hazard of developing opportunistic infections is increased by 41% among HIV/AIDS patients who have baseline CD4 count ≤200 cell/μl than those patients who have baseline CD4 count >200 cell/μl, (AHR= 1.41 (1.18, 1.69).

The hazard of developing opportunistic infections among HIV/AIDS patients with bedridden baseline functional status increases by 35% compared to patients with working baseline functional status (AHR = 1.35 (1.01, 1.82).

The hazard of developing opportunistic infections among HIV/AIDS patients at baseline WHO clinical stage II is 5.87 times higher than patients at baseline WHO clinical stage I (AHR = 5.87 (3.97, 8.69)

The hazard of developing opportunistic infections among HIV/AIDS patients at baseline WHO clinical stage III is 5.85 times higher than patients at baseline WHO clinical stage I status HIV/AIDS patients (AHR = 5.85 (3.55,9.65) (Table 5).

Table 5. Multivariable analysis using the Gompertz cox-regression model for predictor’s HIV/AIDS patients on ART treatment at University of Gondar Compressive Specialized Hospital, January 11, 2015, to January 10, 2021.

Covariates	OIs		CHR (95% CI)	AHR (95% CI)	P-value
	Yes	No
Education
No education	105	15	1.56(1.19, 2.03)	1.28(0.98, 1.69)	0.073
Elementary	112	25	1.39 (1.06, 1.81)	1.21(0.92, 1.58)	0.167
Secondary	210	83	0.99(0.79, 1.25)	0.89 (0.71,1.13)	0.344
(Tertiary)	118	47	1	1
Spouse HIV
Positive	430	112	1	1
Unknown	115	58	0.76 (0.61, 1.81)	0.96 (0.77,1.20)	0.721
CD4 count
≤200	231	20	1.98 (1.66, 2.36)	1.41 (1.18,1.69)	0.00*
(>200)	314	150	1	1
Base line Hgb
≤12	482	107	2.26 (1.71, 2.97)	1.28(0.97,1. 70)	0.084
(13–17.5)	63	63	1	1
Functions
Bedridden	59	10	1.50 (1.13,2.00)	1.35 (1.01, 1.82)	0.040*
Ambulatory	203	44	1.31 (1.09,1.57)	1.16 (0.96,1.41)	0.120
(Working)	283	116	1	1
WHO Stage
(Stage I)	30	126	1	1
Stage II	473	43	9.13 (6.20,13.44)	5.87 (3.97,8.69)	0.000*
Stage III	42	1	9.82 (6.02,16.01)	5.85 (3.55,9.65)	0.000*

LR test chi²(25) = 286.34 Prob > chi2 = 0.0000.

The shape parameter gamma was 0.02 (95% CI: 0.023, 0.029), which is positive. This indicates that the hazard of opportunistic infections increases exponentially with time.

The goodness of fit test

The Cox- Snell residuals (together with their Nelson-Aalen cumulative hazard function) had been obtained from fitting using the exponential, Weibull, Gompertz, lognormal log-logistic models to our data. It can be seen that the plot of the Nelson Aalen cumulative hazard function against Cox-Snell residuals was closest to 45° relatively straight lines through the origin for the Gompertz model when compared to Weibull and others. This suggests that the Gompertz model provided the best fit for our data set (Figure 7).

Figure 7. Cox-Snell residual obtained by fitting Gompertz model for HIV/AIDS patients under ART treatment in University of Gondar compressive specialized Hospital, January 11, 2015, to January 10, 2021.

Discussion

This study mainly investigated the incidence and predictors of opportunistic infections among HIV/AIDS patients in the University of Gondar compressive and specialized hospital, Ethiopia. Many other studies reported different risk predictors for incidence of opportunistic infections; our study also assessed the patients’ socio-demographic, clinical, and treatment characteristics based on the records from their medical follow-up chart. As a result, predictors like baseline CD4 count, baseline functional status, and baseline WHO clinical stages II and III were significant predictors of the incidence of opportunistic infections. This finding is similar to study done in Gondar,²⁴ Arba-Minch,²⁵ Mekelle,²⁶ and Debre Markos,²⁷ which reported the WHO clinical stage and CD4 count variables as risk predictors for the development of opportunistic infections. Out of 715 patients in the cohort study, 76% (CI, 72.95 to 79.21) developed different opportunistic infections. This study is comparable to one conducted in a Tertiary Care Hospital., China, which reports 810 had OIs (72.8%), but this result was a little bit higher than the study done in Mekele²⁶ which reports the proportions of OIs to be 57.1% after a median follow-up time of three months. This might be due the difference of patients’ awareness, follow up time and ART service difference in this two-study area. Among the developed OIs the highest most common one was pneumocystis pneumonia which proportion and its incidence rate was 90 (16.51%) and 16.51% respectively. This finding is similar to with the study done in Shanghai and Uganda^28–30 with proportions Pneumocystis pneumonia and bacterial coinfection (42.1%) was found to be the most common.

During the study period, the overall incidence rate of opportunistic infections after a median follow-up time of 57 months was 4.1 (95% CI 3.74 to 4.44) per 10,000person-year observation. This result was a little bit lower than the study done in Africa,³¹ which showed the cumulative incidence to be 53.6 per 100 people year, in Arba Minch, which showed 23.9/100 PY (95% CI; 18.3, 29.5),^25,32 in Gondar²⁴ reported 7 cases per 100 PY of observation, Asella³³ reported 7.9 cases per 100 PY of observation in Tanzania³⁴ and other Sub-Saharan countries³⁵ which reported incidence ranging from 7.6 to 8.2 per 100 PY of observation.

In our study, HIV/AIDS patients who had baseline CD4 count ≤ 200 cell/μl is increased by 41% than among HIV/AIDS patients who have baseline CD4 count >200 cell/μl, (AHR= 1.41 (1.18, 1.69). This finding is similar to other research which is done in Debre-Berhan Referral Hospital,^36,37 in Cape Town, South Africa,³⁸ and in Bejing Youan Hospital of China.³⁹ The finding proves that CD4 cells play a central role in activating both humoral and cellular immune responses of patients’ bodies to fight against infection. Hence, a low CD4 count increases susceptibility to OIs than the higher one.⁴⁰

In our study, HIV/AIDS patients with bedridden baseline functional status are increased by 35% compared to patients with working baseline functional status (AHR = 1.35 (1.01, 1.82). This could be due to restrictions from physical activities and the inability to perform daily tasks, which, indirectly, compromise the immune system of the patients.⁴⁵ This finding is also similar to other research which is done in Debre Markos referral hospital,⁴¹ in Eastern Zone of Tigray,⁴² in Addis Ababa,⁴³ and study done in sub-Saharan Africa.⁴⁴

In this study, the advanced stage of baseline WHO clinical stage II, III was found to increase the risk of opportunistic infections. The hazard of developing opportunistic infections among HIV/AIDS patients on baseline WHO clinical stage II is 5.87 times higher than patients at baseline WHO clinical stage I, (AHR = 5.87 (3.97,8.69) and the hazard of developing opportunistic infections among HIV/AIDS patients at baseline WHO clinical stage III is 5.85 times higher than the hazard of patients at baseline WHO clinical stage I status of HIV/AIDS patients, (AHR = 5.85 (3.55,9.65). This finding is similar to other studies in Ethiopia^24,45 and China.⁴⁶ Often advanced baseline WHO clinical stages exhibit severe immune deficiency when the stage becomes more advanced, and the occurrence and recurrence of OIs also increase. The most serious and life-threatening OIs are more common among HIV-infected people with stage II and IV.⁵

The clinical importance of this study was to provide information for health professionals, researchers, policymakers, and patients about predictors that are associated with the risk of opportunistic infections development during ART follow uptime and to act on them to minimize the risk and maximize their effort on prevention of having the problem and also its public health importance of this study is to prevent economic loss associated with the infections and its complications as a result of cost investment to prevention, treatment and control mechanisms.

Conclusion

The incidence rate of OIs among HIV-positive people treated at the University of Gondar compressive and specialized hospital, ART clinic was relatively less than other study reports. More importantly, patient baseline CD4 ≤ 200 cell/μl, state-of-the-art baseline whose clinical stage, and baseline bed ridden functional status were associated with a high risk of the incidence rate of OIs development.

Limitation of the study

Since this study is a retrospective follow-up study, using the patients’ baseline socio-demographic, clinical, and treatment characteristics, there may be a change of these predictors (change of exposure predictors) after a time. Additionally, this study used secondary data, as robust data on some potentially significant predictors, such as behavioral, distance to the hospital, and monthly income, were not available in patient charts.

Authors’ contributions

The research concept, design, data collection, analysis, and interpretation, as well as the initial manuscript write-up. Collecting data, analyzing and interpreting it, and editing the manuscript. The final manuscript draft has been read and approved by all authors participations.

Consent to participate and approval of the ethics

The Public Health Institute, the College of Medicine, and Health Sciences, University of Gondar’s ethical review committee granted clearance and approval to conduct the research under the reference letter Ref No/IPH 22/07/2021. Due to the fact that this study analyzed secondary data from patient charts, we were granted an informed consent waiver. To maintain confidentiality, the data collection tool did not include names or other personally identifiable information such as unique identification numbers.

Consent for publication

Not applicable.

Acknowledgment

The authors would like give expressions their gratitude to the health care professionals at the University of Gondar compressive specialized Hospital for their generosity and invaluable assistance in collecting data and retrieving charts. Moreover, the authors wish to express their gratitude to the data collectors and supervisors.

Data and materials accessibility

The corresponding author will have the right access to the data upon request.

Disclosure statement

No potential conflict of interest was reported by the author(s).