Clinical outcomes of HIV-infected patients hospitalized with bacterial community-acquired pneumonia

Article Outline

• Summary
• Introduction
• Materials and methods
  • Study design
  • Inclusion criteria
  • Exclusion criteria
  • Study groups
  • Study outcomes
    • Time to clinical stability (TCS)
    • Length of stay (LOS)
    • All-cause mortality
    • CAP-related mortality
  • Confounding variables
  • Statistical analysis
• Results
• Discussion
• Acknowledgement
• References
• Copyright

Summary 

Background

There are limited and conflicting data on clinical outcomes of community-acquired pneumonia (CAP) among HIV-infected patients.

Methods

Secondary analyses of clinical outcomes of CAP were performed for 118 patients with HIV infection and 2790 patients without HIV infection enrolled in the Community-Acquired Pneumonia Organization (CAPO) international study. After adjustment for significant confounders, the effect of HIV infection on length of stay (LOS) and time to clinical stability (TCS) were examined by survival analyses and overall mortality and CAP-related mortality by logistic regression methods.

Results

After adjusting for significant confounders, hospitalized HIV-infected patients with CAP did not have longer times to reach clinical stability (HR 1.126; 95% CI 0.917–1.391; p=0.251) or longer stays in the hospital (HR 1.191, 95% CI 0.979–1.449; p=0.080). In addition, HIV infection did not significantly influence overall mortality rates (OR 1.205, 95% CI 0.686–2.116; p=0.517) or CAP-related mortality rates (OR 1.338; 95% CI 0.623–3.725; p=0.355).

Conclusion

The presence of HIV infection did not influence the clinical outcomes of CAP among patients assessed at CAPO centers. It is not intended that our results be extrapolated to populations receiving limited healthcare for advanced HIV disease, malnourishment and parasitic diseases.

Keywords: HIV infection, Community-acquired pneumonia, Outcome

Back to Article Outline

Introduction 

Pulmonary infections are a major cause of morbidity and mortality in persons infected with HIV.¹ Pneumonia due to unspecified organisms has been reported to be the leading cause of death among HIV-infected individuals in the USA during 1990–1999.² The introduction of highly active antiretroviral therapy (HAART) has led to a notable decline in opportunistic infections (OI), however bacterial pneumonia remains prevalent.³ In addition, bacterial community-acquired pneumonia (CAP) has been reported to be a common cause of hospitalization among HIV-infected patients.⁴, ⁵

The etiology, risk and prognostic factors of CAP among hospitalized HIV-infected patients have been examined. However, studies comparing clinical outcomes of CAP in patients with and without HIV infection are limited and results are conflicting. To date, five studies have evaluated the outcomes of mortality in HIV-infected patients hospitalized with CAP.⁶, ⁷, ⁸, ⁹, ¹⁰ Higher mortality rates among HIV-infected subjects were found in three studies.⁵, ⁶, ⁷ Four of these five studies evaluated length of stay (LOS) and one study found longer LOS among HIV-infected patients. Time to clinical stability (TCS), as a measurable outcome of CAP, was defined in a case–control study of 58 HIV patients done by our group using the Community Acquired Pneumonia Organization (CAPO) database. Similar outcomes were found between HIV-infected patients and those without HIV infection. These five studies did only univariate analyses of clinical outcomes of CAP.

Because of these conflicting results, we performed a study with the objective of evaluating clinical outcomes of bacterial CAP among hospitalized HIV-infected patients using a cohort design and multivariate analyses that made adjustment for confounders.

Back to Article Outline

Materials and methods 

Study design 

A secondary analysis was performed of the Community-Acquired Pneumonia Organization (CAPO) database.¹¹ CAPO is an international, retrospective, observational study of adult patients hospitalized with CAP. Data were analyzed from hospitals in seven countries (USA, Canada, Spain, South Africa, Argentina, Venezuela, Chile), between June 2001 and March 2006. In each participating center, random medical records of hospitalized patients with the diagnosis of CAP were reviewed. The presence of HIV infection was obtained from medical history. Each investigator filled out a case report form that was transferred via the Internet to the CAPO study center at the University of Louisville, KY, USA. A sample of the data collection form is available at the study website (www.caposite.com). Validation of data quality was performed at the study center before the case was entered in the CAPO database. Antibiotic therapy for CAP was examined following each country's guidelines and categorized as: (1) compliance if empirical therapy covered expected CAP bacteria, and (2) non-compliance if antibiotic treatment did not cover expected CAP bacteria.

Inclusion criteria 

Patients who were included in the study met the following criteria for the diagnosis of CAP: (1) new pulmonary infiltrate on a chest radiograph, and (2) at least one of the following: (a) a new or worse cough, (b) an abnormal body temperature (<35.6°C or >37.8°C), and/or (c) an abnormal serum leukocyte count.

Exclusion criteria 

In an attempt to limit our investigation to bacterial CAP, patients with a presumptive or definitive diagnosis of Pneumocystis pneumonia (PCP), mycobacterial and fungal pneumonia were excluded from this study. Furthermore, patients were excluded if they were treated for a suspected PCP, mycobacterial or fungal etiology.

Study groups 

Two study groups were established, the group of CAP patients with HIV infection based on the presence of positive serology for HIV antibody and a second group categorized as CAP patients without HIV infection based on the presence of negative HIV antibody or unknown HIV status.

Study outcomes 

A patient was defined as clinically stable and ready to be switched to an oral antibiotic the day that the following four criteria were met: (1) improved cough and shortness of breath, (2) lack of fever for at least 8hours, (3) improving leukocytosis (decreased at least 10% from the previous day), and (4) tolerating oral intake with adequate gastrointestinal absorption.¹² Patients were evaluated daily within the first seven days of hospitalization to determine the day when clinical stability was reached.

Defined in days and calculated for each patient as the day of discharge minus the day of admission. Patients hospitalized for more than 14 days were censored at 15 days in an effort to capture LOS data related only to bacterial CAP.

The sum of all patients with a final clinical outcome classified as dead. The final clinical outcome ‘death or alive’ was evaluated at time of hospital discharge. The final clinical outcome for patients who were still hospitalized at day 28 was considered as alive.

Patients in whom the cause of death was considered to be directly related to CAP, due to progression of infection with development of severe sepsis and/or metastatic foci as determined by the local investigator.

Confounding variables 

The following 28 variables were obtained from each study participant. Demographics: age, gender, and nursing home residency. Coexisting conditions: neoplastic disease, chronic obstructive pulmonary disease (COPD), diabetes mellitus, renal disease, liver disease, alcohol and/or illegal drug use, cerebrovascular accident (CVA), and congestive heart failure (CHF). Physical examination: temperature, respiratory rate, heart rate, blood pressure, and altered mental status. Laboratory tests: sodium, arterial partial pressure of O₂, hematocrit, blood urea nitrogen, glucose, and arterial pH. Chest X-ray: multilobar involvement, presence of cavity, and presence of pleural effusion. Severity of CAP: need for ICU admission, risk class according to the pneumonia severity index (PSI), and CRB-65 score.¹³, ¹⁴

Statistical analysis 

Differences in continuous baseline characteristics between patients with HIV infection and without HIV infection were compared by means of unpaired t-tests. Baseline categorical explanatory variables were summarized as frequencies and percentages, and differences between both groups of patients were analyzed using a Chi-square test, the Wilcoxon–Mann–Whitney test, and Fisher's exact test when appropriate and warranted. Since mortality was dichotomized in this study, we examined the association between HIV infection and mortality using a logistic regression model. We then incorporated all standard risk factors into the model by utilizing multiple logistic regression analyses. A stepwise procedure was then employed to obtain the final model. LOS and TCS were analyzed with the Kaplan–Meier method, whereas log-rank tests and Cox proportional-hazards regression analyses were applied to evaluate differences between both groups of patients. In our analyses, LOS was truncated at 14 days and TCS was truncated at 7 days. The Breslow–Day test was performed to evaluate possible heterogeneities of the effect of HIV infection on outcomes according to the country of residence and it demonstrated that the groups were sufficiently homogenous, providing justification for the pooled analyses performed.

Back to Article Outline

Results 

Clinical outcomes of CAP were examined in 118 patients with HIV infection and 2790 patients without HIV infection. Baseline characteristics of both groups of patients are illustrated in Table 1. Patients with HIV infection were younger and predominantly male. Patients with HIV infection had less severe CAP and patients without HIV infection had significantly greater rates of co-morbidities, except for liver illness. Alcohol and drug abuse were present in approximately a third (31%) of HIV-infected patients, around four times more often than in those without HIV infection (8%). Positive bacterial cultures were present in 43 (36%) of the infected patients. Streptococcus pneumoniae was the most common isolated organism and it accounted for 75% of the etiology of pneumonia. A total of 54 (46%) HIV-infected patients had determinations of CD4+ cell count. The CD4+ cell counts ranged from 4 to 1074/mm³ and a count >200/mm³ was present in 57% of patients. Compliance with the local country's guidelines for antibiotic therapy for CAP was 89.8% and 87.6% among individuals with and without HIV infection, respectively.

Table 1. Baseline patient characteristics stratified by HIV status

Variables	Patients with HIV infection (n=118)	Patients without HIV infection (n=2790)	p-Valuea
Mean age	40	67	<0.0001
Males (%)	87 (74%)	1710 (61%)	0.0065
ICU admission	14 (12%)	343 (12%)	0.890
Mean PSI	50	93	<0.0001
Risk class
I	38 (32%)	226 (8%)	<0.0001
II	36 (30%)	422 (15%)	0.0175
III	23 (19%)	606 (22%)	0.5621
IV	15 (13%)	1067 (38%)	<0.0001
V	6 (5%)	469 (17%)	0.0072
CRB-65	0.35	0.95	<0.0001
0	81 (69%)	907 (32%)	<0.0001
1	32 (27%)	1249 (45%)	0.0002
2	5 (4%)	524 (19%)	<0.0001
3	0 (0%)	93 (3%)	0.0438
4	0 (0%)	17 (1%)	0.3951
Chronic morbidities
Cancer	3 (2%)	279 (10%)	0.007
Renal disease	9 (8%)	318 (11%)	0.327
Liver disease	17 (14%)	101 (4%)	<0.001
Cerebrovascular accident	2 (2%)	453 (16%)	<0.001
Congestive heart failure	4 (3%)	588 (21%)	<0.001
COPD	13 (11%)	836 (30%)	<0.001
Diabetes mellitus	8 (7%)	548 (20%)	<0.001
Alcohol/drug use	37 (31%)	231 (8%)	<0.001
Chest X-ray
Multilobar infiltrate	40 (34%)	820 (29%)	0.014
Cavitary lesion	0 (0%)	32 (1%)	0.205
Pleural effusion	11 (9%)	566 (20%)	0.001
Time to first antibiotic, hours	7.3	6.6	0.240

ICU, intensive care unit; PSI, pneumonia severity index; CRB-65, confusion, respiratory rate >30 per minute, systolic blood pressure <90 mmHg or diastolic blood pressure ≤60 mmHg and ≥65 years of age; COPD, chronic obstructive pulmonary disease.

The rates of all-cause mortality and CAP-related mortality were not influenced by the HIV status before adjusting for confounders (Table 2). Similarly, the logistic regression model indicated that HIV status does not have a significant effect on all-cause mortality (OR 1.205, 95% CI 0.686–2.116; p=0.517) when adjusted for other confounders (Table 3). In addition, HIV infection was eliminated in the stepwise selection and excluded from the final model. The saturated logistic regression analysis for CAP-related mortality is shown in Table 4. These results also indicate that HIV infection does not have a significant effect on CAP-related mortality (OR 1.338; 95% CI 0.623–3.725; p=0.355).

Table 2. Comparison of mortality rates by HIV status before adjustment for confounders

CAP, community-acquired pneumonia.

Table 3. Logistic regression model investigating the influence each variable has on overall mortality when adjusting for all other confounders

OR, odds ratio; CI, confidence interval; PSI, pneumonia severity index; CRB-65, confusion, respiratory rate >30 per minute, systolic blood pressure <90 mmHg or diastolic blood pressure ≤60 mmHg and ≥65 years of age.

Table 4. Logistic regression model investigating the influence each confounder has on CAP-related mortality when adjusting for all other confounders

CAP, community-acquired pneumonia; OR, odds ratio; CI, confidence interval; PSI, pneumonia severity index; CRB-65, confusion, respiratory rate >30 per minute, systolic blood pressure <90 mmHg or diastolic blood pressure ≤60mmHg and ≥65 years of age.

The Kaplan–Meier survival analysis for TCS stratified by HIV status suggests that HIV-infected patients have a shorter TCS (Chi-square 12.26; p<0.001). However, HIV infection did not have significant influence on TCS (HR 1.126, 95% CI 0.917–1.391; p=0.251) after the adjustment for confounders in the Cox regression analysis (Table 5). Similarly, the Kaplan–Meier survival analysis for LOS stratified by HIV status showed that HIV-infected individuals had a shorter LOS (Chi-square 15.85; p<0.001), while the Cox regression analysis showed that HIV infection does not have significant influence on LOS (HR 1.191, 95% CI 0.979–1.449; p=0.080) (Table 6).

Table 5. Cox regression model for time to clinical stability (TCS)

HR, hazards ratio; CI, confidence interval; PSI, pneumonia severity index; CRB-65, confusion, respiratory rate >30 per minute, systolic blood pressure <90 mmHg or diastolic blood pressure ≤60 mmHg and ≥65 years of age.

Table 6. Cox regression model for length of stay (LOS)

HR, hazards ratio; CI, confidence interval; PSI, pneumonia severity index; CRB-65, confusion, respiratory rate >30 per minute, systolic blood pressure <90 mmHg or diastolic blood pressure ≤60mmHg and ≥65 years of age.

Back to Article Outline

Discussion 

Our study indicates that clinical outcomes of CAP among hospitalized HIV-infected patients are similar to those without HIV infection among CAPO centers in the USA, Europe, South Africa and South America. That is, patients with HIV infection and patients without HIV infection had similar time to clinical stability, length of stay, all-cause mortality, and CAP-related mortality rates in the multivariable Cox regression and the multiple logistic regression analyses. In comparison with previous studies, our results were adjusted for confounders.

Upon initial assessment, patients with HIV infection had lower PSI scores than patients without HIV infection. This implies that physicians had a low threshold for admission of HIV-infected patients with CAP, likely because of the presumption that HIV-infected patients have worse outcomes than patients without HIV infection. National organizations from several countries have developed guidelines for the management of patients with CAP. Most guidelines, however, have excluded HIV-infected patients due to the paucity of data regarding management decisions such as need for hospitalization, timing to switch to oral therapy and hospital discharge.¹² One important implication of our study results is that several of the management recommendations offered by national guidelines can be safely applied to HIV-infected patients with CAP. The presence of HIV infection by itself should not be considered a risk factor for poor clinical outcomes and should not alter important management decisions such as site of care, timing for oral switch, or criteria for hospital discharge. Our study showed that the increasing severity of pneumonia measured by CRB-65 and PSI determines the outcomes of CAP patients regardless of the status of HIV infection.

Studies that have examined CAP among hospitalized HIV-infected patients have reported conflicting clinical outcomes. HIV-infected patients were found to have higher mortality rates in four studies.⁶, ⁷, ⁸, ¹⁵ Park et al. reported high rates of homelessness, tobacco smoking, and intravenous drug use among HIV-infected patients.⁶ Thus, these results may not be generalized to the entire population of HIV-infected patients. In addition, none of these patients was taking HAART. The International Pneumococcal Study Group has recently examined the outcome of bacteremic pneumococcal pneumonia in a cohort of 200 HIV-infected patients.¹⁵ HIV-infected patients had greater mortality than those without HIV infection. The HIV-infected patients were predominantly from South Africa and had median CD4+ cell counts of 119/mm³. Touchie and Marrie enrolled 32 HIV-infected patients with a mean CD4+ cell count of 70±98.5/mm³ and 50% of CAP etiology was PCP.⁷ The low CD4+ cell count and high incidence of PCP indicate advanced immunosuppression, which could explain the higher mortality rate among HIV-infected patients than among HIV-uninfected patients. Johnson et al. included mainly subjects with CD4+ cell counts <200/mm³ (74%) and did not describe the etiology of pneumonia.⁸ HIV-infected patients are expected to have poor CAP outcomes in settings with endemic malnourishment and recurrent parasitic diseases receiving limited healthcare. In comparison to these studies, our study examined a cohort of HIV-infected patients with less advanced immunosuppression with proven and presumed bacterial CAP after the exclusion of PCP, tuberculosis and fungal pneumonia.

Similar LOS seen in both our groups of patients has also been reported by others.⁷, ⁹, ¹⁰ Johnson et al. found longer LOS among HIV-infected patients. This could be explained by the same reasons cited for higher mortality rates in the group of HIV-infected patients.⁸ TCS directly correlates with length of hospitalization. It is defined by the time a patient is considered stable enough to be switched from intravenous to oral antibiotic therapy as defined by Ramirez et al.¹⁶ Although it has been reported in several studies, TCS has not been well defined among HIV-infected patients with CAP. Our group is the first to compare the TCS in relation to CAP between hospitalized HIV-infected patients and patients without HIV infection. After adjusting for confounders, the results of our study proved that the TCS between the two groups of patients was equivalent.

It is recognized that patients with HIV infection are at increased risk of developing bacterial CAP. An HIV patient with two episodes of S. pneumoniae CAP meets criteria for AIDS.¹⁷ The increased risk of S. pneumoniae CAP is primarily related to abnormalities of T-cells and immunoglobulins. The pulmonary host defenses are initiated and coordinated by alveolar macrophages in bacterial CAP. Gordon et al. found a similar ability of alveolar macrophages to attach and phagocytize opsonized pneumococci in HIV-positive and negative patients.¹⁸ These data suggest that alveolar macrophages of HIV-infected patients maintain the ability to kill CAP-related bacteria. Even though HIV-infected patients are at increased risk of developing bacterial CAP, laboratory data and clinical data of this study suggest that the host response to bacteria is similar regardless of the HIV status. Similarly to previous studies, incomplete data for CD4+ cell counts, HIV-RNA levels and HAART are limitations of our study.

We acknowledge that our study has important limitations. It is limited by the nature of a retrospective design, incomplete CD4+ cell count data, and lack of information regarding prophylaxis for opportunistic infections, HAART, and prior vaccination status. Confirmation of etiology of pneumonia was unknown in 66% of our patients, a rate higher than the rate of 26–35% reported by other studies.⁶, ⁷, ⁹ Lastly, even though our patients are from countries with low prevalence of HIV infection, the numbers of HIV-infected patients might also be underestimated since HIV testing is not done in all hospitalized patients with CAP enrolled in the CAPO study. Conversely, a major strength of our study is the utilization of multivariate techniques to adjust potential confounders for more precisely measuring the impact of HIV infection on outcomes. In view of limitations of the current literature, future studies of HIV-infected individuals with bacterial CAP should include complete data of CD4+ cell counts, HIV-RNA levels, HAART intake and HAART adherence.

In summary, our study indicates that the presence of HIV infection does not influence the clinical outcomes of patients with CAP after adjustment for confounders. It is not intended that these results be extrapolated to populations receiving limited healthcare for advanced HIV disease, malnourishment and parasitic diseases. The decision regarding initial site of care for an HIV-infected patient with CAP should be based on the patient's severity of disease and not influenced by the presence of HIV infection. Our study suggests that several strategies recommended by national organizations for the management of non-immunocompromised patients with CAP may be safely used to manage HIV-infected patients with bacterial CAP.

Back to Article Outline

Acknowledgement 

Authors of this paper greatly appreciate the editorial support of Mrs RoseMarie Leone.

Conflict of interest: No conflict of interest to declare.

Back to Article Outline

References 

1. Murray JF, Felton CP, Garay SM, Gottlieb MS, Hopewell PC, Stover DE, et al. Pulmonary complications of the acquired immunodeficiency syndrome: report of a National Heart, Lung, and Blood Institute workshop. N Engl J Med. 1984;310:1682–1688
   • View In Article
   • MEDLINE
   • CrossRef
2. Selik RM, Byers RH, Dworkin MS. Trends in diseases reported on US death certificates that mentioned HIV infection, 1987–99. J Acquir Immune Defic Syndr. 2002;29:378–387
   • View In Article
   • MEDLINE
   • CrossRef
3. Wolff AJ, O’Donnell AE. Pulmonary manifestations of HIV infection in the era of highly active antiretroviral therapy. Chest. 2001;120:1888–1893
   • View In Article
   • MEDLINE
   • CrossRef
4. Polsky B, Gold JW, Whimbey E, Dryjanski J, Brown AE, Schiffman G, et al. Bacterial pneumonia in patients with the acquired immunodeficiency syndrome. Ann Intern Med. 1986;104:38–41
   • View In Article
   • MEDLINE
5. Floris-Moore M, Lo Y, Klein RS. Gender and hospitalization patterns among drug users before and after the availability of highly active antiretroviral therapy. J Acquir Immune Defic Syndr. 2003;34:331–337
   • View In Article
   • MEDLINE
   • CrossRef
6. Park DR, Sherbin VL, Goodman MS, Pacifico AD, Rubenfeld GD, Polissar NL, et al. Harborview CAP Study Group The etiology of community acquired pneumonia at an urban public hospital: influence of HIV infection and initial severity of illness. J Infect Dis. 2001;184:268–277
   • View In Article
   • MEDLINE
   • CrossRef
7. Touchie C, Marrie TJ. Comparison of community-acquired pneumonia requiring admission to hospital in HIV- and non-HIV-infected patients. Can J Infect Dis Med Microbiol. 1996;7:253–258
   • View In Article
8. Johnson DH, Carriere KC, Houston S. Hospitalization for community-acquired pneumonia in Alberta patients with human immunodeficiency virus infection: a case control study. Can Respir J. 2003;10:265–270
   • View In Article
   • MEDLINE
9. Mundy LM, Auwaerter PG, Oldach D, Warner ML, Burton A, Vance E, et al. Community acquired pneumonia: impact of immune status. Am J Respir Crit Care Med. 1995;152:1309–1315
   • View In Article
   • MEDLINE
10. Christensen D, Feldman C, Rossi P, Marrie T, Blasi P, Luna C, et al. and the Community-Acquired Pneumonia Organization (CAPO) investigators HIV infection does not influence clinical outcomes in hospitalized patients with bacterial CAP: results from the CAPO International Cohort Study. Clin Infect Dis. 2005;41:554–556
    • View In Article
    • CrossRef
11. Ramirez JA. Fostering international multicenter collaborative research: the CAPO project. Int J Tuberc Lung Dis. 2007;11:1062–1065
    • View In Article
12. Niederman MS, Mandell LA, Anzueto A, Bass JB, Broughton WA, Campbell GD, et al. American Thoracic Society Guidelines for the management of adults with community-acquired pneumonia. Diagnosis, assessment of severity, antimicrobial therapy, and prevention. Am J Respir Crit Care Med. 2001;163:1730–1754
    • View In Article
    • MEDLINE
13. Fine MJ, Auble TE, Yealy DM, Hanusa BH, Weissfeld LA, Singer DE, et al. A prediction rule to identify low-risk patients with community-acquired pneumonia. N Engl J Med. 1997;336:243–250
    • View In Article
    • MEDLINE
    • CrossRef
14. Ewig S, de Roux A, Bauer T, García E, Mensa J, Niederman M, et al. Validation of predictives rules and indices of severity for community-acquired pneumonia. Thorax. 2004;59:421–427
    • View In Article
    • MEDLINE
    • CrossRef
15. Feldman C, Klugman KP, Yu VL, Ortqvist A, Choiu CC, Chedid MB, et al. International Pneumococcal Study Group Bacteraemic pneumococcal pneumonia: impact of HIV on clinical presentation and outcome. J Infect. 2007;55:125–135
    • View In Article
    • Abstract
    • Full Text
    • Full-Text PDF (160 KB)
    • CrossRef
16. Ramirez JA, Srinath L, Ahkee S, Huang A, Raff MJ. Early switch from intravenous to oral cephalosporins in the treatment of hospitalized patients with community-acquired pneumonia. Arch Intern Med. 1995;155:1273–1276
    • View In Article
    • MEDLINE
17. Centers for Disease Control and Prevention (CDC). 1993 revised classification system for HIV infection and expanded surveillance case definition for AIDS among adolescents and adults. MMWR Recomm Rep 1992; 41(RR-17):1–19.
    • View In Article
18. Gordon SB, Jarman ER, Kanyanda S, French N, Pridmore AC, Zijlstra EE, et al. Reduced interlukin-8 response to Streptococcus pneumoniae by alveolar macrophages from adults with HIV/AIDS. AIDS. 2005;19:1197–1200
    • View In Article
    • MEDLINE