Unexpected low frequency of respiratory symptoms in an HIV-positive urban sub-Saharan population compared to an HIV-negative control group

Background Chronic respiratory illnesses and respiratory infections are common in HIV-positive populations. It seems reasonable that HIV-positive people experience more respiratory symptoms, such as coughing and breathlessness, than those who are HIV-negative. Objectives This study aims to determine the frequency of respiratory symptoms in an urban African HIV-positive population. Method A cross-sectional study was conducted in Johannesburg, South Africa, in 2016–2017. Four groups of participants were included: HIV-positive participants (1) not yet on antiretroviral therapy (ART), (2) on first-line ART, (3) on second-line ART and (4) age- and sex-matched HIV-negative controls. Data were collected on socio-demographics, respiratory risk factors and respiratory symptoms. A logistic regression analysis was performed to determine if respiratory symptoms differed between groups and to identify determinants associated with symptoms. Results Overall, 547 participants were included, of whom 62% were women, with a median age of 37 years. Of these patients, 63% (347) were HIV-positive, 26% were ART-naïve, 24% were on first-line ART and 50% were on second-line ART. Cough and/or productive cough was reported by 27 (5%), wheezing by 9 (2%) and breathlessness by 118 (22%) of the participants. The frequency of these symptoms did not differ by HIV status after adjustment for age and sex. Breathlessness was associated with age, female sex, obesity, a history of respiratory infection and a history of airway hyper-reactivity. Conclusion The frequency of respiratory symptoms was low in our study population except for breathlessness. HIV-positive participants, whether or not on ART, did not experience more symptoms than HIV-negative participants.


Introduction
In the last two decades, the introduction of antiretroviral therapy (ART) has substantially improved life expectancy of HIV-positive patients. Patients who are adequately treated with ART have a near-normal lifespan compared to the general population and, as a consequence, HIV has become a chronic disease. 1 One of the reasons for the increase in life expectancy in HIV-positive patients on ART is the lower frequency of severe and life-threatening opportunistic lung infections, such as tuberculosis and bacterial pneumonia. 2 However, non-communicable chronic respiratory illnesses, such as chronic obstructive pulmonary disease (COPD), are more frequently seen in people living with HIV (PLHIV) compared to people without HIV. 3 Studies have shown that PLHIV report more respiratory symptoms, such as coughing, productive cough and shortness of breath. 4 This is of interest when considering that PLHIV are ageing and respiratory symptoms are more frequent in an older population. 5 However, most of the literature related to respiratory symptoms in HIV infection emanate from North America. There are insufficient data on the extent of this problem in lowand middle-income countries (LMIC) where the majority of PLHIV reside and where the burden of tuberculosis and bacterial pneumonia is higher than in high-income countries (HIC) like North America.
We studied the frequency of respiratory symptoms in PLHIV whether or not on ART in an urban area in South Africa in comparison to an HIV-negative control group, as well as the determinants of respiratory symptoms.

Methods
We conducted a cross-sectional study in Johannesburg, South Africa, from July 2016 to November 2017. We recruited four groups of participants from the Johannesburg area: HIV-positive participants not yet on ART, HIV-positive participants on first-line ART, HIV-positive participants on second-line treatment and HIV-negative control participants. The HIV-positive participants were recruited from past or ongoing randomised controlled trials (RCTs) comparing different ART regimens in a governmental HIV care facility in central Johannesburg. 6,7 The control group was recruited by HIV-positive participants who invited their family or friends with a negative or unknown HIV status with the same age range (+/−5 years) and sex to participate in the study. All control participants underwent HIV counselling and testing according to the South African Department of Health guidelines. 8 If a control participant tested HIV-positive and there was no history of ART use, they were counted in the first group (HIV-positive, ART-naïve) and referred to a local clinic to initiate ART. If they were on ART already, they were included in the group on first-or secondline ART, depending on their current ART regimen.

Data collection
Data were collected during a single visit. Information on demographics and smoking was assessed with a modified version of the WHO STEPs instrument 9 . Information on medical history, respiratory symptoms, working and living circumstances and occupational exposure to potentially harmful agents was obtained using following questionnaires: The British Medical Research Council (MRC) Respiratory Questionnaire, 10,11 the MRC dyspnoea scale, 12 the World Health Survey, 13 the ATS-DLD-78-A 14 and questions used in other publications 15,16 . Respiratory symptoms evaluated in this study were 'cough', 'bringing up phlegm', 'breathlessness' and 'wheezing or whistling' according to the MRC Respiratory Questionnaire and the ATS-DLD-78. A questionnaire also included 'cough' and 'bringing up phlegm' as symptoms which were evaluated separately (Table 1).
A physical examination was performed, which included measurements of height and weight. Blood sample was collected for measurement of HIV viral load and CD4-cell count (HIV-positive participants only). For participants who were recruited from one of the RCTs, laboratory data were retrieved from the RCT visit closest to our study visit.

Data analysis
Outcomes were described as median with interquartile range for continuous variables (all non-normally distributed) and count with percentage for categorical variables. Differences in continuous variables across the four groups were tested using a Mann-Whitney U test and categorical variables using a Fisher's exact test. The frequency of respiratory symptoms across the four groups was displayed in bar charts.
We combined the four respiratory symptoms in a composite outcome 'any respiratory symptom' that was defined as the occurrence of at least one of the respiratory complaints, namely coughing, bringing up phlegm, shortness of breath and/or wheezing or whistling. We first analysed if the frequency of respiratory symptoms differed according to HIV or ART status in three logistic regression models using 'any respiratory symptom' as outcome. In the first model, we assessed the unadjusted association between HIV and ART status and the occurrence of any respiratory symptom using the HIV-negative group as the reference group. The second model was adjusted for sex and age, and the third model additionally adjusted for body mass index (BMI), ever smoking, passive smoking, respiratory infections in the past (pneumonia and/or tuberculosis) and bronchial hyperreactivity. To investigate the influence of HIV-and ARTrelated characteristics, we repeated the models described above including the HIV-positive participants only and using the ART-naïve group as the reference group. In the third model, HIV viral load and CD4+ cell counts were added.
Finally, we analysed which determinants were associated with any respiratory symptom. The following factors were considered in univariable analysis: HIV status, age, sex, BMI, ever smoking and passive smoking, respiratory illnesses in the past such as tuberculosis and pneumonia, history of bronchial hyper-reactivity and environmental factors (worked in the mining industry or worked in a dusty job or exposure to gas, chemical fumes or pesticides in work). All factors with a p-value of < 0.2 in univariate analysis as well as age and sex were then included in multivariable analysis Which of the following statements best describes your situation? 1. Not troubled by breathlessness except on strenuous exercise 2. Short of breath when hurrying on the level or walking up a slight hill 3. Walks slower than most people on the level, stops after 1.5 km or so, or stops after minutes of walking at own pace 4. Stops for breath after walking about 100 m or after walking a few minutes in level ground 5. Too breathless to leave house, or breathless when undressing Wheezing or whistling: Have you had attacks of wheezing or whistling in your chest at any time in the last 12 months? Yes or no http://www.sajhivmed.org.za Open Access using forced entry. A p-value < 0.05 was considered to be statistically significant.
For the statistical analyses, we used the statistic programme 'IBM SPSS statistics' version 24.0 (IBM SPSS Statistics for Windows, Version 24.0. IBM Corp., Armonk, NY).

Ethical considerations
Ethical permission was obtained from the Human Research Ethics Committee of the University of the Witwatersrand (HREC number M160131). All participants provided written informed consent prior to participation.

Results
Of the 548 participants, 547 were included in the study. One participant was excluded as there was no information on HIV status. Almost all participants were black Africans (99.6%), 341 (62%) were women, 394 (72%) were HIV-positive, and the median age was 37 years (Table 2). Of the HIV-positive group, 103 (26%) were ART-naïve, 94 (24%) were on first-line treatment and 197 (50%) were on second-line treatment ( Table 2).
The HIV-positive group on second-line treatment had a higher percentage of women (73%, p < 0.001) and was older  The numbers were especially high in the group on secondline treatment (51%).
Occupational exposure was uncommon, with only 26 participants (5%) reporting any exposure. Almost all participants used gas or electricity for cooking (99.5%) and heating (99%).
In the unadjusted comparison across the four groups, the HIV-positive group on second-line ART had significantly more respiratory symptoms compared to the HIV-negative group (odds ratio [OR] 2.5, 95% confidence interval [CI] 1.5-4.1) (Table 3). However, after adjustment for sex and age, there was no difference in the frequency of respiratory symptoms related to HIV or ART status. Further adjustment for respiratory risk factors did not change this finding.
When we restricted the analysis to the HIV-positive participants, the same crude trend was observed with participants on second-line ART having more respiratory symptoms than the other participants in the unadjusted analysis. This result disappeared after adjustment for sex and age. Additional adjustment for CD4-cell count and viral load did not change the findings.
The occurrence of any respiratory complaint was associated with age, female sex, BMI and a history of bronchial hyper-reactivity (Table 4). When limiting the analysis to breathlessness, age, female sex, BMI, bronchial hyper-reactivity and a history of pulmonary infection were the associating factors (Table 5).

Discussion
The frequency of respiratory complaints in our study was surprisingly low when compared to what has been reported in the literature in studies from both HIC and LMIC. For coughing, a frequency of 17% 17 -40% 18 has been reported in HIC and 7% -48% 19,20,21,22,23 in LMIC. In contrast, only 3.3% of the participants reported cough in our study. Bringing up phlegm was reported by 30% of HIV-negative and 42% of PLHIV in studies conducted in HIC, 18,21,24 whereas this was reported by only 2% of the participants in our study. Both coughing and bringing up phlegm were reported to happen more in PLHIV than in HIV-negative participants. 18,21,24 Wheezing and whistling have only been evaluated in two studies from HIC. No difference was found between PLHIV and HIV-negative participants. 24,25 In contrast, breathlessness was a frequently expressed complaint with 22% of the participants in our study, indicating that they experienced breathlessness. This is in line with the literature, mostly from HIC, reporting that between 1.4% and 42% of the population experience breathlessness. People living with HIV were found to experience breathlessness more often than HIV-negative individuals, 18,25,26 a finding that we could not confirm in our study.
The following reasons should be considered to understand the low number of respiratory complaints, except for breathlessness, noted among the participants in our study compared to previous studies. Firstly, our study population differed from study populations in HIC. People living with HIV from HIC represent a particular group with specific health risk behaviour, such as men having sex with men (MSM) or intravenous drug users with a higher percentage of smokers than the HIV-negative population. 27 In our study cohort, PLHIV smoked significantly less than   Despite these possible explanations for our findings, it is still particularly surprising that HIV-positive, ART-naïve participants in our study hardly reported respiratory symptoms. A possible explanation could be the fact that the ART-naïve group was still young and had not had the chance to develop symptoms yet. Additionally, they had a median CD4 count of 281 cells/mm 3 and therefore were still at low risk for opportunistic infections. 31 This is supported by the low prevalence of a history of tuberculosis in the ART-naïve group compared to the group on first-and second-line ART, which might indicate that ART-naïve participants had not had uncontrolled viremia for a long period of time.
Another unexpected finding of our study was that despite the high burden of respiratory infections in the past in PLHIV, this did not result in a higher frequency of respiratory symptoms except for breathlessness. It may be the case that, given the relatively young age of the population, there was still enough pulmonary reserve capacity to limit complaints in daily life. Clearly, the next step to clarify the relationship between respiratory infections in the past, particularly tuberculosis, and current pulmonary function would be to include lung function testing in the analysis.
Limitations of this study include the recruitment process of the study population which may influence the generalisability of our results. Participants were not randomly sampled from the general HIV-positive population but were recruited from RCTs. However, we think that the participants were representative of people with the same HIV and ART status as they were all recruited from routine local HIV diagnostic services. Furthermore, it is possible that the control group experienced more breathlessness as a result of the high percentage of smokers in this group. This could have reduced the contrast between the HIV-positive and HIV-negative participants.
Lastly, although we tried to define respiratory complaints clearly, there may still be uncertainty around the definition. Future studies should seek to standardise the definition of respiratory symptoms more strictly and it would be helpful if all studies used the same definitions.
In conclusion, the results of this study among a large group of HIV-positive and -negative patients do not support the view that respiratory symptoms are more common among patients with HIV.