In-hospital mortality of acute HIV-associated pulmonary morbidity among COVID-19 negative medical admissions to Dr. George Mukhari Academic Hospital

INTRODUCTION

Human immunodeficiency virus (HIV)/acquired immunodeficiency syndrome (AIDS) remains a cause of significant ill health and death world-wide, with 1.5 million people becoming newly infected with HIV and 650,000 AIDS-related deaths in 2021.^[1] South Africa is estimated to have an overall HIV prevalence rate of 13.7%, with approximately 8.2 million individuals living with the disease.^[2] HIV-related admissions remain very common in the country.^[3,4] Almost 86,000 (85,769) South Africans succumbed to AIDS-related deaths in 2022.^[5]

Anti-retroviral therapy (ART) has decreased the incidence of opportunistic infections and allowed for an increase in the life expectancy of HIV-infected persons.^[6-8] ART has been freely available to South Africans for more than a decade and a half. Adherence to ART is an important determinant of virologic suppression and has been shown to reduce morbidity and mortality.^[9-11]

Pulmonary complications are among the most frequent causes of morbidity and mortality in HIV-infected individuals.^[12-14] The spectrum includes infectious and non-infectious complications.^[15-17]

Pulmonary infections remain the most frequent cause of hospital admissions for HIV-infected people.^[18-20] South African studies suggest pulmonary tuberculosis (PTB), bacterial community-acquired pneumonia (CAP), and Pneumocystis jirovecii pneumonia (PJP) be the more common infectious complications.^[21,22]

Comorbid conditions have been suggested to significantly increase the odds of poorer health outcomes in HIV-infected persons.^[23]

Our primary aim was to determine the in-hospital mortality of COVID-19 negative HIV-positive patients admitted with an acute pulmonary illness. Secondary objectives were to describe the diagnoses, diagnostic work-up, and treatment of the cohort and to assess the contribution of comorbidities and HIV-associated factors to mortality. The focus of our study was on COVID-19 negative patients as we felt concomitant COVID infection would be a significant confounder to our results and, in general, less attention was being paid to this group during the pandemic.

The study was approved by the Institutional Research and Ethics Committee (Sefako-Makgatho University and Dr. George Mukhari academic hospital]) (SMUREC/ M32/2020:PG). Due to the observational nature of the study, informed consent was waived.

MATERIALS AND METHODS

This was a cohort study of unselected COVID-19 negative HIV-infected patients admitted to the medical wards of Dr. George Mukhari Academic Hospital with an acute respiratory illness (defined as presentation to hospital within 10 days of symptom onset). The hospital is located in the Tshwane district of the Gauteng Province, South Africa and is surrounded by low- and middle-income communities. The study was conducted between April 1, 2020, and March 31, 2021.

All consecutive admissions of medical patients with a known diagnosis of HIV or who were highly suspected of having and subsequently confirmed with HIV and an acute respiratory illness were included in the study. The researchers were made aware of these cases by the admitting medical teams. The teams had been made aware of the project after approval had been granted by the Institutional Review Board. The investigations and treatment of the patients were at the discretion of the medical teams. We observed the clinical progress of the patients and prospectively documented the results of the diagnostic investigations, the management of the patients and outcomes. The variables of interest were as follows: The degree of immunosuppression as evidenced by the CD4 count and viral load (VL), comorbidities, pathogens isolated and sources of isolation (in cases of suspected infections), % on ARTs, treatment administered, and outcome (death or discharge).

RESULTS

The study took place over a period of 12-month (April 1, 2020–March 31, 2021). Our study cohort consisted of 272 patients. The mean age of the group was 42.0 ± 10.8. The majority (62.4%) were male.

One hundred and forty-four patients (52.6%) were on ART. There were significantly more female patients on ART than males (64.1% vs. 45.6%; P = 0.0436).

The median CD4 count of the cohort was 76 cells/mm³ (IQR 39–310). The median CD4 count of the male patients was 63 (IQR 20–172) cells/mm³). Female patients had a median CD4 count of 98 cells/mm³. This difference was not statistically significant (P = 0.1630).

Two hundred and seven patients (76.1%) had a CD4 count below 200 cells/mm³; 93 (45%) of these patients were on ART.

Seventy-one patients (26%) had a concomitant comorbid illness with 17 (24%) of these having two or more comorbidities [Table 1]. The most frequent clinical diagnosis was community-acquired pneumonia. Other respiratory reasons for admission are summed up in [Table 2].

A number of diagnostic investigations were carried out on the patients as deemed appropriate by the treating clinicians. All patients underwent chest radiography. Four patients (3.3%) underwent CT pulmonary angiography; three were positive for pulmonary emboli. Six patients underwent non-contrast chest CT scans. The remainder of the investigations and the frequency with which they were done is summed up in [Table 3]. Sixty patients (28.3%) with CAP had microbiological documentation of infection. [Table 4] is a summary of the isolates, frequency of detection, and sample source.

The proportion of patients with Gram-negative isolates was higher in those with comorbidities compared to patients without (18.3% vs. 6.0%; P = 0.004).

Two hundred and seventy patients (99.3%) received empiric antibiotic therapy (combination of a beta-lactam/beta-lactamase penicillin and a macrolide) as initial treatment. Half of these patients (49.63%) also received concomitant anti-tuberculosis (TB) therapy and just over a quarter (26.47%) received doses of cotrimoxazole appropriate for PJP.

The median length of hospital stay of the patients was 11 (IQR 7.0–17.5 days). The hospital length of stay was not significantly different between survivors and non-survivors (13.8 [±9.86] vs. 14.8 [±14.44]; P = 0.568).

Overall, in-hospital mortality was 25.7% (n = 70). Survival rates were not statistically significantly different between male and female patients (70.8% vs. 80.4%; P = 0.1679). The presence of a comorbid illness did not adversely affect patient survival [Table 5]. HIV-related variables had no impact on mortality rates [Table 6].

The proportion of patients with comorbid illness and on ART who demised was not statistically significantly different from those with comorbidities, not on ART (2.96% vs. 2.59%; P = 1.00).

DISCUSSION

The study confirms previous findings that CAP is a common respiratory infectious complication of and a frequent reason for hospitalization (including intensive care unit [ICU] admissions) among HIV-infected individuals despite widespread availability of ART.^{[12,21,24-26]}

Studies in South Africa that have reported on the causes of respiratory infections among admitted HIV-infected patients have suggested TB and Gram-positive bacteria to be the more common isolates.^[4,21,22] The proportion of patients diagnosed with PTB in this study was much lower than would be expected from the low to middle income communities that the hospital serves. The explanation for this is several-fold. It should be noted that in the three South African studies cited, there was a concerted/deliberate effort to systematically screen every patient for PTB by obtaining multiple samples for TB diagnosis. Our study was, in effect, a review of everyday “standard of care” of such patients by our physician colleagues. In addition, only 35% of the patients had a sputum sample submitted and not all were for TB diagnosis.

Gram-negative pathogens were the more frequent isolates in our study. While this may be a true reflection of the status of isolates in the community we serve, we would recommend a cautious interpretation of this finding as the bulk of the bacterial isolates which were on sputum. The quality of the sputum samples was not assessed. The usefulness of sputum culture in making a microbiological diagnosis of pulmonary bacterial infections has been controversial.^[27,28]

About 16% of the infections were poly-microbial. Respiratory infections with more than one pathogen are not unusual in HIV-infected individuals, especially with advanced immunosuppression.^[20,21,29] While it is usually encouraged to work with a single diagnosis and treat accordingly, it is worth bearing in mind that respiratory infections in HIV-infected patients may be the result of multiple pathogens that may require their own unique treatments.^[11] In a study undertaken by Barbier et al. of the causes and outcomes of acute respiratory failure in HIV-associated admissions to an ICU, two or more respiratory processes were identified in 22% of the cases.^[20]

Community-acquired Pseudomonas aeruginosa and S taphylococcus aureus infections were isolated in 18% and 10% of the CAP cases in this study, respectively. Both pathogens have been reported to occur as community-acquired pathogens with increased frequency in HIV-infected persons.^[14,25] The number of cases with PTB in our cohort was unusually low compared to studies previously undertaken in the country.^[4,22] This is most likely an underestimate as not only does South Africa have a high TB burden, but HIV-positive individuals are much more likely to develop TB. It is worth noting that only 35% (n = 95) of the patients had a sputum sample submitted. We did not pay attention to the quality of the sputum samples collected.

Two hundred and seven patients (76%) had CD4 counts below 200 during the current admission. Almost half of them (47.4%) were ART naïve. Admission of patients with advanced HIV disease and ART naivety in South African hospitals is not uncommon.^[4,21,26] ART is freely available in South Africa and, globally, South Africa is reported to support the largest ART program.^[9] This finding likely reflects delays in diagnosis and/or patient reluctance to be initiated on ART.

Of particular concern was the large number of patients (45%) on ART that still had evidence of inadequate immune reconstitution (CD4 counts <200 cells/mm³). Challenges with adherence to treatment and/or drug resistance may be the reasons behind this. Reports of short-term ART adherence among South African patients from multiple urban HIV clinics indicated adherence rates of between 63% and 88%.^[30,31] Challenges of retaining people in HIV care through treatment programs in South Africa have been highlighted.^[1]

A study of long-term adherence and successful treatment outcomes by Moosa et al. concluded that long-term adherence was possible within a structured ART program with close adherence monitoring.^[9]

There are significantly more women than men on ART. The explanation for this finding remains obscure. The previous studies have suggested that fewer men go for voluntary counseling and testing and were less likely to initiate ART if they tested positive.^[32,33] Other studies have disputed this, suggesting that men are as willing to be tested and treated as non-pregnant women.^[34-37]

The investigations carried out for an infectious etiology in this study appear standard. There is no consensus among clinicians on an algorithm for diagnosis of pulmonary infections in HIV patients. The recommendation has been for the diagnostic approach to be guided by the local epidemiologic findings.^[14] Flexible bronchoscopy was conspicuously absent as an investigational tool. Communication with the local respiratory team indicated that the study took place just as the COVID-19 pandemic was starting, with concerns about the risk of transmission of the infection to the staff posed by the technique. In addition, some patients were deemed too unwell for the procedure to be undertaken safely. Ordinarily bronchoscopy is a procedure with a high yield for pulmonary disease in HIV-infected patients.^[14]

The majority of the patients with a suspected infective etiology were initiated on empiric standard antibiotic therapy. About a quarter of the patients received also empiric cotrimoxazole for PJP. Both the World Health Organization (WHO) and South African thoracic society (SATS) advise addition of empiric therapy for PJP when patients fulfill the WHO case definition.

Empiric anti-TB therapy was started in almost half of the patients. Guidelines recommend that empiric anti-TB treatment should not be routinely initiated unless there is a miliary pattern on a chest X-ray or the patient is severely ill and the suspicion for TB is high.^[38,39]

A mortality rate of 26% is unacceptably high. We could not find comparative data (mortality data related to pulmonary morbidity in HIV-infected admitted patients) from our review of the literature. Most published studies referring to mortality as an outcome has focused on HIV-associated complications in general and not specifically as they relate to pulmonary morbidity. Barbier et al. found in-hospital mortality of 19.7% among their cohort of HIV-infected patients admitted to an ICU in acute respiratory failure.^[20] Spanish researchers found an overall 30-day mortality of 7% among HIV-infected admissions for community-acquired pulmonary sepsis.^[28] Results of admissions for CAP in a teaching hospital in Kwazulu-Natal Nyamande found a 15.9% in-hospital mortality among patients infected with HIV.^[22] These comparisons are to be interpreted cautiously given the differences in patient populations.

Being male was not associated with an increased risk of death in this study. For reasons not clearly understood, increased mortality has been reported among HIV-infected men on ART over women.^[40-42] Some of the reasons offered by the researchers were late initiation of treatment and challenges with follow-up.

We could not establish a relationship between CD4 count and in-hospital mortality. Similar results were described by Barbier et al. in France as well as by Meintjes’s and Mkoko’s groups in South Africa.^[4,20,26] Shortcomings of using the CD4 as a sole determinant of the degree of immune-depression have been well described.^[43] Data on VL were not available. VL is a better predictor of mortality and HIV/AIDS progression than CD4 cell count.^[44] Suppression of VL to undetectable levels has been associated with reduced HIV-related mortality.^[45] Compliance with ART in our cohort was not examined. Adherence to ART has been associated with better chances of survival for HIV-infected patients.^[10]

Survival rates were not significantly different between patients with comorbidities and those without. This finding should be interpreted as being equivocal, because the numbers of patients computed for comorbid illnesses were too small (analysis was under-powered). HIV-infected patients are living longer and develop comorbidities related to aging and the HIV infection itself.^[12,13,17] We believe that it would be reasonable to evaluate the impact of comorbidities on survival in HIV-infected patients presenting with an acute respiratory illness.

CONCLUSION

Patients admitted with HIV-associated acute pulmonary morbidity have significant mortality. Approximately half of the patients were ART naïve, with advanced disease despite ready availability of ART. Programs that encourage voluntary testing and treatment are likely to reduce the high number of late presentations and reduce the poor outcomes. More studies looking at the pathogens in CAP in our community are necessary to clarify the dominant pathogens. Future studies looking at other HIV-associated variables such as the VL and treatment adherence in relation to mortality are indicated. The SATS guidelines for the management of CAP in adults recommends among others, that in HIV-infected patients presenting with CAP, clinicians should have a low-threshold for carrying out testing for PTB.^[38] TBLAM antigen testing is advised in those with CD4 counts below 100 cells/mm³ or stage 3 or 4 disease if sputum is unavailable.^[38]