Community-acquired Haemophilus influenzae pneumonia in people living with HIV: a 30-year propensity score–matched cohort study

Abstract

Objectives

To compare the clinical characteristics, management, and outcomes of community-acquired pneumonia (CAP) caused by Haemophilus influenzae in people living with HIV (PWH) versus HIV-negative patients, and to evaluate whether HIV infection independently influences disease severity, intensive care unit (ICU) admission, or short- and long-term mortality.

Methods

We conducted a retrospective observational study of prospectively collected data including all consecutive adults hospitalized with CAP at a tertiary teaching hospital in Spain between 1996 and 2025. Microbiological confirmation of H. influenzae was obtained using standard diagnostic methods. Clinical variables, treatments, and outcomes were compared between PWH and HIV-negative patients. Propensity score matching (1:3) was applied to adjust for baseline differences. Multivariable generalized linear and Cox regression models were used to assess the association between HIV status and ICU admission, 30-day, 90-day, and 1-year mortality.

Results

Among 202 patients, 18 (9%) were PWH. Before matching, PWH were younger, more frequently current smokers, and had higher rates of chronic liver disease. After propensity score matching, 18 PWH were compared with 54 HIV-negative controls. No significant differences were observed in disease severity, complications, ICU admission, or mortality. No deaths occurred among PWH during follow-up. HIV status was not independently associated with ICU admission or mortality at any time point. However, confidence intervals were wide and the number of events was limited, reducing statistical power and precision of estimates.

Conclusions

In hospitalized patients with H. influenzae CAP, HIV infection was not associated with worse short- or long-term outcomes. However, given the small sample size and low number of outcome events, clinically meaningful differences cannot be excluded. Preventive strategies, including vaccination and optimized antiretroviral therapy, remain essential to reduce pneumonia burden in PWH.

Supplementary Information

The online version contains supplementary material available at https://doi.org/10.1186/s41479-026-00210-5.

Introduction

Community-acquired pneumonia (CAP) remains an important cause of hospitalization among people living with HIV (PWH), often associated with serious complications, especially in those PWH without antiretroviral therapy (ART) and immunosuppressed [1, 2]. Streptococcus pneumoniae and Haemophilus influenzae remain the most common bacterial pathogens identified in PWH with CAP [3, 4]. While clinical outcomes of pneumococcal CAP have been reported to be similar between virologically suppressed PWH and patients without HIV [5], data comparing outcomes of H. influenzae CAP in PWH versus HIV-negative patients are limited. This gap is important, given the higher prevalence of chronic lung disease in PWH [6], which is a major risk factor for H. Influenzae infection, and the treatment challenges of this microorganism due to the intrinsic resistance to macrolides and ketolides [7]. We aim to compare the clinical characteristics, management, and outcomes of CAP caused by H. influenzae in PWH compared with HIV-negative patients. A secondary objective was to evaluate whether HIV infection independently influenced disease severity, need for intensive care, or short or long–term mortality, after accounting for baseline differences through propensity score matching (PSM). This study was not designed as a non-inferiority or equivalence study.

Methods

This was a retrospective observational study of prospectively collected data from a tertiary-teaching hospital in Spain. We included all consecutive adult patients admitted to the emergency department with a diagnosis of CAP between November 1996 and March 2025. Microbiological diagnosis of H. influenzae pneumonia was established according to standard protocols, primarily based on culture of respiratory and/or blood specimens. Additional diagnostic testing (including urinary antigen detection and RT-PCR for respiratory viruses) was performed according to clinical indication and availability during the study period. RT-PCR was introduced in the institution in the early 2000s and was not available during the earliest years of inclusion. No RT-PCR-based detection of H. influenzae was used. Vaccination status against influenza and pneumococcus was obtained from electronic medical records and/or patient-reported history when documentation was unavailable. Influenza and pneumococcal vaccination status was assessed due to their established role in preventing bacterial pneumonia and reducing severity of respiratory infections in high-risk populations. HIV-associated variables (median years since HIV diagnosis, HIV exposure risk groups (people who inject drugs, men who have sex with men, heterosexual), antiretroviral therapy (ART), current ART, most recent CD4 cell count, plasma HIV-RNA (detectable and undetectable (HIV-RNA of ≤ 50 copies/mL) were collected in all PWH. Virological suppression was defined as HIV-RNA < 50 copies/mL. A secondary exploratory categorization using < 200 copies/mL was used due to assay variability over the long study period. All surviving patients were visited or contacted by telephone at 30 days after discharge, and clinical records together with the Catalunya Health Department database were reviewed at 1 year.

The study was approved by the local Institutional Review Board (HCB 2009/5451). The need for written informed consent was waived due to the non-interventional design.

Categorical variables were summarized as counts and percentages, and continuous variables as means (SD) or medians (IQR), according to distribution. Comparisons were made using χ² or Fisher’s exact test for categorical variables, and Student’s t-test or Mann–Whitney U test for continuous variables. To minimize confounding between people with and without HIV hospitalized with CAP, PSM [8, 9] was applied using logistic regression with HIV status as the dependent variable and relevant clinical covariates as predictors [10]. Optimal 1:3 matching was performed, and covariate balance was assessed using standardized mean differences (SMDs), considering values < 0.1 as indicative of good balance (Supplementary Fig. 1) [11]. Mortality was analyzed with Kaplan–Meier curves (Gehan–Breslow–Wilcoxon test) and Cox regression models [12]. ICU admission was assessed with generalized linear models [13]. Missing data were handled with multiple imputation [14]. All analyses were two-tailed with significance set at p < 0.05 and performed using SPSS 26.0 and R 4.4.2. Additional details of the statistical analysis are provided in the Supplementary Material.

Results

Study population and baseline characteristics

A total of 202 patients hospitalized with CAP were included, of whom 18 (9%) were PWH and 184 (91%) were HIV-negative. After propensity score matching (PSM), 18 PWH were compared with 54 matched controls (see Table 1).

Table 1.

Demographics and clinical characteristics at admission, and complications and outcomes during admission

Variables	Full cohort (N = 202)			Propensity score matching (N = 72)
	Non-HIV infection (N = 184)	PWH (N = 18)	P-value	Non-HIV infection (N = 54)	PWH (N = 18)	P-value
Age, median (Q1; Q3), years	73.5 (61; 80)	42.5 (34; 53)	< 0.001	60 (50; 75)	42.5 (34; 53)	< 0.001
Male sex, n (%)	118 (64)	10 (56)	0.471	38 (70)	10 (56)	0.248
Smoking habit (current smoke), n (%)	59 (32)	14 (78)	< 0.001	31 (58)	14 (78)	0.142
Alcohol abuse (current alcohol), n (%)	34 (19)	6 (33)	0.210	15 (28)	6 (33)	0.686
Comorbidities, n (%)a	154 (84)	14 (78)	0.513	39 (72)	14 (78)	0.764
Diabetes mellitus	38 (21)	0 (0)	0.027	1 (2)	0 (0)	> 0.999
Chronic lung disease	124 (68)	9 (50)	0.129	34 (63)	9 (50)	0.331
Bronchiectasis	8 (4)	0 (0)	> 0.999	1 (2)	0 (0)	> 0.999
COPD / Chronic bronchitis	86 (47)	4 (22)	0.044	24 (44)	4 (22)	0.094
Asthma	8 (4)	1 (6)	0.578	3 (6)	1 (6)	> 0.999
Other	22 (12)	4 (22)	0.261	6 (11)	4 (22)	0.255
Neurologic disease	22 (12)	1 (6)	0.700	5 (9)	1 (6)	> 0.999
Chronic liver disease	11 (6)	5 (28)	0.008	3 (6)	5 (28)	0.021
Chronic heart disease	22 (12)	1 (6)	0.700	3 (6)	1 (6)	> 0.999
Chronic renal disease	9 (5)	1 (6)	> 0.999	2 (4)	1 (6)	> 0.999
Previous neoplasm	19 (10)	1 (6)	> 0.999	4 (7)	1 (6)	> 0.999
Co-infection HBV	-	1 (6)	-	-	1 (6)	-
Co-infection HCV	-	8 (47)	-	-	8 (47)	-
Nursing-home, n (%)	6 (3)	0 (0)	> 0.999	1 (2)	0 (0)	> 0.999
Previous pneumonia, n (%)	26 (16)	8 (44)	0.007	10 (21)	8 (44)	0.069
Days since initial symptoms to admission, median (Q1; Q3)	5 (3; 7)	4 (3; 7)	0.598	4 (3; 7)	4 (3; 7)	0.747
Cough, n (%)	154 (86)	17 (94)	0.477	43 (83)	17 (94)	0.435
Dyspnea, n (%)	147 (83)	12 (67)	0.116	38 (75)	12 (67)	0.534
Fever, n (%)	121 (67)	16 (89)	0.054	37 (71)	65 (70)	0.203
Decreased level of consciousness, n (%)	33 (18)	2 (11)	0.745	6 (11)	2 (11)	> 0.999
Treatment before admission, n (%)
Systemic corticosteroids	9 (7)	0 (0)	0.603	3 (7)	0 (0)	0.548
Inhaled Corticosteroids	51 (28)	1 (6)	0.078	15 (28)	1 (6)	0.094
Previous Antibiotic	35 (20)	1 (6)	0.204	6 (12)	1 (6)	0.685
Influenza vaccine	50 (44)	1 (7)	0.008	11 (28)	1 (7)	0.148
Pneumococcal vaccine	32 (28)	1 (7)	0.113	4 (10)	1 (7)	> 0.999
Characteristics at admission
Respiratory rate ≥ 30 rpm, n (%)	77 (44)	5 (31)	0.324	19 (37)	5 (31)	0.699
PSI IV-V, n (%)	93 (65)	5 (42)	0.126	19 (49)	5 (42)	0.669
SOFA score, median (Q1; Q3)	2 (1; 3)	2 (2; 5)	0.242	2 (2; 3)	2 (2; 5)	0.346
Laboratory findings, median (Q1; Q3)
Leucocyte count, 109/L	13.8 (9.6; 17.7)	9.3 (6.4; 11.7)	0.008	11.6 (7.8; 17.2)	9.3 (6.4; 11.7)	0.205
Lymphocyte count, 109/L	0.81 (0.55; 1.5)	1.15 (0.45; 1.97)	0.442	0.71 (0.16; 1.51)	1.15 (0.45; 1.97)	0.284
Neutrophil/Lymphocyte ratio	10.2 (6; 17.6)	7.1 (4.2; 8.3)	0.061	9.2 (4.8; 17.2)	7.1 (4.2; 8.3)	0.318
Empiric antibiotic therapy, n (%)
Monotherapy	28 (16)	7 (39)	0.023	7 (13)	7 (39)	0.037
Fluoroquinolones	15 (8)	3 (17)	0.220	2 (4)	3 (17)	0.103
β-lactams	13 (7)	3 (17)	0.170	5 (10)	3 (17)	0.415
Macrolide	0 (0)	1 (6)	0.092	0 (0)	1 (6)	0.257
Combination therapies	150 (84)	11 (61)	0.023	45 (87)	11 (61)	0.037
β-lactams plus fluoroquinolones	25 (14)	3 (17)	0.727	12 (23)	3 (17)	0.744
β-lactams plus macrolides	103 (58)	7 (39)	0.122	29 (56)	7 (39)	0.217
Other combination therapies	22 (12)	1 (6)	0.701	4 (8)	1 (6)	> 0.999
Severe CAP, n (%)b	46 (33)	5 (29)	0.775	15 (37)	5 (29)	0.601
Respiratory support, n (%)c			0.321			0.292
Non-invasive mechanical ventilation	6 (4)	0 (0)	-	4 (8)	0 (0)	-
Invasive mechanical ventilation	20 (12)	4 (24)	-	6 (12)	4 (24)	-
Complications during admission, n (%)
Pleural effusion	23 (13)	3 (17)	0.712	6 (11)	3 (17)	> 0.999
Sepsisd	124 (74)	11 (85)	0.521	36 (77)	11 (85)	0.713
Multilobar infiltration	46 (25)	6 (33)	0.412	18 (33)	6 (33)	> 0.999
ARDS	11 (6)	2 (13)	0.291	6 (12)	2 (13)	> 0.999
Septic shock	13 (7)	2 (11)	0.630	5 (9)	2 (11)	> 0.999
Bacteremia	18 (13)	2 (13)	> 0.999	4 (10)	2 (13)	0.654
Acute renal failure	33 (18)	4 (22)	0.750	8 (15)	4 (22)	0.490
Outcomes
Length of hospital stay, median (Q1; Q3), days	8 (6; 12)	8 (6; 17)	0.915	10 (6; 14)	8 (6; 17)	0.661
ICU admission, n (%)	39 (21)	4 (22)	> 0.999	13 (24)	4 (22)	> 0.999
ICU mortality, n (%)e	5 (13)	0 (0)	> 0.999	0 (0)	0 (0)	-
Length of ICU stay, median (Q1; Q3), dayse	5 (4; 7)	2 (2; 2)	0.115	7 (5; 27)	2 (2; 2)	0.180
In-hospital mortality, n (%)	6 (3)	0 (0)	> 0.999	1 (2)	0 (0)	> 0.999
30-day mortality, n (%)f	6 (3)	0 (0)	> 0.999	0 (0)	0 (0)	-
90-day mortality, n (%)g	9 (5)	0 (0)	> 0.999	3 (6)	0 (0)	0.568
1-year mortality, n (%)h	13 (7)	0 (0)	0.609	3 (6)	0 (0)	> 0.999

Abbreviations. HIV indicates human immunodeficiency virus; Q1, first quartile; Q3, third quartile; COPD, chronic obstructive pulmonary disease, PSI, pneumonia severity index; SOFA, sequential organ failure assessment; PaO₂, partial pressure of arterial oxygen; FiO₂, fraction of inspired oxygen; CAP, community acquired pneumonia; ICU, intensive care unit; ARDS, acute respiratory distress syndrome

Note: Percentages calculated on non-missing data. P-values marked in bold indicate numbers that are statistically significant at the 95% confidence limit. ^a May have > 1 comorbid condition. ^b Severe CAP was defined according to the 2019 ATS/IDSA severe community-acquired pneumonia criteria. ^c Patients who initially received non-invasive ventilation but subsequently needed intubation were included in the invasive mechanical ventilation group. ^d Sepsis was defined according to the Sepsis-3 consensus definition. ^e Calculated only for patients admitted to intensive care unit in the full cohort (39 in the non-HIV infection group and 4 in the HIV infection group) and in the propensity score matching (13 in the non-HIV infection group and 4 in the HIV infection group). ^f Calculated only for patients with 30-day follow-up in the full cohort (180 in the non-HIV infection group and 18 in the HIV infection group) and in the propensity score matching (54 in the non-HIV infection group and 18 in the HIV infection group). ^g Calculated only for patients with 90-day follow-up in the full cohort (178 in the non-HIV infection group and 18 in the HIV infection group) and in the propensity score matching (54 in the non-HIV infection group and 18 in the HIV infection group). ^h Calculated only for patients with 1-year follow-up in the full cohort (179 in the non-HIV infection group and 17 in the HIV infection group) and in the propensity score matching (54 in the non-HIV infection group and 17 in the HIV infection group)

Before matching, PWH were significantly younger, more frequently current smokers, had a lower prevalence of COPD/chronic bronchitis, and had a higher prevalence of chronic liver disease and previous pneumonia, while none had diabetes compared with more than one-fifth of HIV-negative patients. Hepatitis C virus (HCV) co-infection was frequent among PWH (47%). Ten patients were male (56%). The most common HIV-risk factor was heterosexual contact (77.7%), followed by injecting drug use (17%); none were men who have sex with men (MSM). Six patients (33%) had previously been diagnosed with AIDS. The median CD4 + T-cell count was 228 cells/µL (IQR 110–500). A total of 13 patients (81%) were receiving antiretroviral therapy (ART), of whom 7 (53%) were virologically suppressed (plasma HIV viral load < 200 copies/mL). Vaccination coverage against influenza and pneumococcus was very low, at 7% for each.

At admission, laboratory findings showed that PWH presented with lower leukocyte counts and lower CRP levels. Empiric antibiotic monotherapy was prescribed more often to PWH compared with HIV-negative patients. After matching, differences in smoking status and some laboratory findings persisted, and the use of antibiotic monotherapy remained significantly more frequent among PWH. These characteristics are detailed in Table 1 and Supplementary Tables 1–2.

Clinical outcomes before and after matching

Before and after matching, there were no significant differences between PWH and HIV-negative patients in disease severity, complications, need for respiratory support, ICU admission, ICU length of stay, or mortality at any time point. Notably, no deaths occurred among PWH throughout follow-up. However, confidence intervals were wide and estimates imprecise due to the small sample size and low number of events.

Multivariable analysis of ICU admission and mortality

In multivariable generalized linear and Cox regression analyses, HIV status was not significantly associated with ICU admission or mortality at 30 days, 90 days, or 1 year (Table 2 and Supplementary Figs. 2–4). Although the point estimates for 90-day and 1-year mortality were low (HR 0.03–0.04), these estimates were driven by the absence of deaths among PWH and were accompanied by very wide confidence intervals, reflecting the limited sample size and lack of statistical precision. In the matched analysis, 30-day mortality could not be estimated because no events occurred in either group. Full results are shown in Table 2.

Table 2.

Association between HIV status and clinical outcomes: crude and propensity score matched estimates from generalized linear and Cox proportional hazards models

Variable	OR or HR	95% CI	P-value
ICU admission
Crude (full cohort)	1.06	0.33 to 3.41	0.919
PSM	0.90	0.25 to 3.22	0.873
30-day mortality
Crude (full cohort)	0.04	0.00 to 9,510.06	0.617
PSM	-a	-a	-a
90-day mortality
Crude (full cohort)	0.04	0.00 to 954.61	0.538
PSM	0.03	0.00 to 2,003.15	0.540
1-year mortality
Crude (full cohort)	0.04	0.00 to 191.67	0.463
PSM	0.03	0.00 to 2,003.15	0.540

Abbreviations. OR indicates odds ratio; HR, hazard ratio; CI, confidence interval; ICI, intensive care unit; PSM, propensity score matched

Note: P-values marked in bold indicate numbers that are statistically significant at the 95% confidence limit. ^a In the PSM analysis, 30-day mortality could not be estimated because no events occurred in either group, resulting in non-identifiable models

Discussion

This is the first study comparing outcomes of CAP caused by H. influenzae between PWH and HIV-negative controls. We found that HIV infection was not associated with increased disease severity, complications, ICU admission, or mortality. However, these findings should be interpreted cautiously, as the study is underpowered and confidence intervals are wide, meaning that clinically important differences cannot be excluded.

Before matching, PWH presented different demographic and clinical characteristics compared with HIV-negative patients. They were younger, more often current smokers, and had higher rates of chronic liver disease and prior episodes of pneumonia, while none had diabetes. HCV co-infection was frequent, reflecting overlapping risk profiles. Despite 69% of PWH being on ART, the median CD4 count at admission was low, at 228 cells/µL (IQR 110–550) and HIV-RNA remained detectable in half of the cases (50%), indicating incomplete immune restoration in a subset of individuals. Vaccination coverage against influenza and pneumococcus was low, highlighting gaps in preventive care in this population. Co-infection with Streptococcus pneumoniae was uncommon and occurred at similar frequencies in both groups (Supplementary Table 3), suggesting that it was unlikely to have substantially influenced the observed clinical outcomes.

Together with previous evidence on pneumococcal [5] and Legionella CAP [15], our findings suggest that HIV infection may not confer a worse prognosis when PWH receive appropriate HIV care.

Although our multivariable analyses did not show an independent association between HIV status and outcomes, the estimates were imprecise and should not be interpreted as evidence of equivalence.

These results are consistent with prior studies suggesting improved outcomes in PWH in the modern ART era; however, our study spans nearly three decades (1996–2025), during which major changes in ART and pneumonia management occurred, and residual temporal confounding cannot be excluded [4, 16, 17].

Exploratory subgroup analyses according to viral suppression and CD4 cell count did not reveal clear differences in outcomes, although these analyses were limited by small sample size.

The absence of men who have sex with men in our cohort and the high prevalence of HCV co-infection likely reflect historical HIV transmission patterns in our region during earlier years of the study period, when injecting drug use and heterosexual transmission were more prevalent.

This study has limitations, including the small number of PWH and the low frequency of outcome events, which may have reduced statistical power and precision of effect estimates. Therefore, clinically relevant differences cannot be excluded.

Additionally, the long study period (1996–2025) introduces potential temporal confounding due to major changes in ART, microbiological diagnostic methods, antimicrobial therapy, critical care, and pneumonia management. Although propensity score matching was used, residual confounding cannot be excluded.

However, our study adds novel evidence by directly evaluating H. influenzae CAP outcomes in PWH using prospectively collected data and robust statistical methods.

In conclusion, our study demonstrates that, although PWH hospitalized with H. influenzae pneumonia differ in demographic and clinical characteristics, no statistically significant differences in outcomes were observed between groups. However, the study is limited by the small sample size and low event rates, and clinically meaningful differences cannot be excluded. Therefore, these findings should be interpreted with caution and should not be considered evidence of equivalent outcomes between groups.

Our efforts should focus on improving preventive measures including vaccination, smoking cessation, and optimized ART adherence to reduce the burden of pneumonia and improve lung health in PWH.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

Conceptualization (CC, AC, RG, AG, MAM, JMM, AT), data curation (CC, AG), formal analysis (AG), investigation (all), methodology (CC, AG, JMM, AT), supervision (CC, AT, JMM), writing – original draft (CC, JMM, AT), and writing – review & editing (all). All authors provided final approval of the version submitted for publication.

Funding

This study was supported by Ciber de Enfermedades Respiratorias (CibeRes CB06/06/0028), and 2009 Support to Research Groups of Catalonia 911; IDIBAPS. Dr Cilloniz is the recipient of a SEPAR fellowship 2024. JM Miro received a personal 80:20 research grant from Institut d’Investigacions Biomèdiques August Pi i Sunyer (IDIBAPS), Barcelona, Spain, during 2017–26.

Data availability

Data from this study is available from the corresponding author, AT, upon reasonable request.

Declarations

Ethical approval

The study was approved by the Hospital Clinic Institutional Ethics Committee (HCB 2009/5451). The requirement for written informed consent was waived because of the study's non-interventional design.

Artificial intelligence

No artificial intelligence (AI) tools were used in the preparation or writing of this article.

Competing interests

The authors declare no competing interests.