Decrease in Hospitalization and Mortality Rates among Children with Perinatally Acquired HIV Type 1 Infection Receiving Highly Active Antiretroviral Therapy

Abstract

Background. The impact of highly active antiretroviral therapy (HAART) on human immunodeficiency virus type 1 (HIV-1) disease progression in perinatally infected children is not well documented. This study aims to identify the effect of evolving antiretroviral therapy on the immunologic and virologic status of and hospitalization and mortality rates among perinatally infected children.

Methods. Children receiving outpatient care during 1994–2001 at 3 HIV clinics in southern California were observed longitudinally for CD4⁺ cell percentage, plasma HIV-1 RNA load, antiretroviral treatment, Pneumocystis jiroveci pneumonia (PCP) prophylaxis, and rate of hospital admissions.

Results. A total of 129 children were observed during the study period; 51% were girls, and 40.3% were Hispanic, 29.5% were African American, and 27.1% were white. The mean CD4⁺ cell percentage increased from 22.5% in 1994 to 31.2% in 2001 (P < .01), and the mean plasma HIV-1 RNA load decreased from 4.53 log₁₀ copies/mL in 1996 to 3.27 log₁₀ copies/mL in 2001 (P < .001). The use of HAART increased from 0% in 1994 to 93% in 2001 (P < .01), whereas the use of PCP prophylaxis decreased from 55% to 16% during this time (P < .001). The hospitalization rate decreased from 6.49 to 0.60 admissions per 100 person-years during 1994–2001 (P < .001). Acquired immunodeficiency syndrome—associated hospital admission rates decreased from 15.6% in 1994 to 0% in 2001 (P < .0001). Similarly, the admission rate for patients with Centers for Disease Control and Prevention category B decreased from 29.7% in 1994 to 5.9% in 2001 (P < .0001). Logistic regression analysis showed that CD4⁺ cell percentage and viral load were independently associated with risk of hospitalization. Survival was significantly longer for those who received HAART.

Conclusions. HIV-1—associated mortality and hospitalization rates decreased significantly between 1994 and 2001 in perinatally infected children. This correlated with an increase in CD4⁺ cell percentage and a decrease in HIV-1 RNA load concurrently with the expanded use of HAART.

With the introduction of protease inhibitor and nonnucleoside analogue combination antiretroviral therapy in 1996, HIV-1—associated morbidity and mortality decreased among adults receiving specialized HIV care in industrialized countries [1, 2]. This trend has been well documented in adult AIDS clinical trials and observational cohort studies and has correlated with increases in CD4⁺ lymphocyte counts and decreases in HIV-1 RNA loads [3–5]. A decrease in HIV-1—associated mortality was recently documented in infected children in 2 observational cohort studies in Italy and the United States [6, 7]. Gortmaker et al. [6] documented a mortality decrease of 5.3% to 0.7% between 1996 and 1999 among HIV-1—infected children, which correlated with an increase from 7% to 73% in the use of protease inhibitor combination antiretroviral therapy.

HIV-1—associated morbidity is following a similar trend. The incidence of AIDS-defining events and hospitalization rates had diminished among HIV-1—infected adults treated with HAART within the first 6–12 months that combination antiretroviral therapy was available [8–10]. Gebo et al. [11] also observed a decrease in hospitalization rates among HIV-1—infected adults between 1995 and 1996 that was associated with the introduction of combination antiretroviral therapy but noted an increase in the hospitalization rate for nonopportunistic infections in 1998, raising concerns about the durability of the beneficial effect of combination antiretroviral therapy on HIV-1—associated hospitalizations. Decreases in hospital admissions and morbidity rates associated with perinatally acquired HIV-1 infection have not been well documented in the literature on pediatric patients. The European Collaborative Study, which examined a cohort of perinatally HIV-1—infected patients observed prospectively during 1986–1997, documented 48 hospital admissions per 100 patient-years and estimated that 48% of HIV-1—infected children will be admitted by the age of 12 months [12]. The European study was conducted before the advent of combination antiretroviral therapy in children. Therefore, the objective of our study was to determine whether the effects of evolving antiretroviral therapy on immunologic and virologic status resulted in changes in hospitalization and mortality rates among perinatally HIV-1—infected children in 3 specialized HIV clinics in southern California.

Methods

Study population. We conducted a retrospective cohort study review of the medical records of children with perinatal HIV-1 infection observed by the Mother, Child, and Adolescent Program at the University of California—San Diego Medical Center, Pediatric Special Services at the University Medical Center (Fresno, CA), and the Maternal-Child Immunology Clinic at the University of California—Los Angeles, from January 1994 through December 2001. The study was approved by the institutional review board at each institution.

The following data were extracted: date of birth, sex, race, mean yearly CD4⁺ lymphocyte percentage, mean yearly log₁₀ plasma HIV-1 RNA load, and Pneumocystis jiroveci pneumonia (PCP) prophylaxis. Treatment was recorded as one of the following: no treatment, monotherapy, dual therapy, and HAART composed of ⩾3 drugs, including a protease inhibitor or a nonnucleoside reverse-transcriptase inhibitor. If patients received ⩾1 treatment regimen per year, the regimen that was recorded on most of the clinic visits in any given year was included in the analysis. The number of hospital admissions, the number of hospital days per admission, and the reasons for admission—opportunistic infection, other infectious processes, AIDS-associated neoplasms, AIDS-defining illnesses, diagnostic procedures, surgery and adverse drug events—were recorded. All HIV-1—associated diagnoses were recorded for any given admission, as were the date of death and immediate cause of death. Hematologic conditions leading to hospital admissions, such as anemia, thrombocytopenia, or pancytopenia requiring blood or platelet transfusion or intravenous immunoglobulin for the treatment of HIV-1—associated thrombocytopenia, were recorded. The following admissions were not recorded: those associated with pharmacokinetic or any research-related study, those associated with receipt of prophylactic intravenous immunoglobulin or pentamidine infusion, those for treatment of asthma, and those for trauma or other emergency surgery.

Statistical analyses. Patients in whom perinatal HIV-1 infection was diagnosed and who had at least 2 clinic visits in a calendar year were included. The hospitalization rate was calculated as the number of admissions per 100 patients per calendar year. Hospitalization admission rates were also computed as the number of hospitalizations per 100 person-years. Mean length of stay was defined as the total number of hospital days divided by the number of admissions. Hospital admissions were divided into 4 categories: (1) those for treatment of opportunistic diseases, (2) those for treatment of other infections or for Centers for Disease Control and Prevention clinical category B HIV-1 infection, (3) those for performance of therapeutic procedures (or surgery) and diagnostic procedures, and (4) those for treatment of adverse drug-related events. Data were analyzed using SAS, version 8.2 (SAS Institute). The χ² test, analysis of variance, and nonparametric tests were used to characterize population differences. Tests for trend were employed to assess temporal changes. Logistic regression was performed to identify covariates associated with risk of hospitalization. Linear regression was used to identify determinants of viral load and CD4⁺ lymphocyte percentage. Survival analysis using the Kaplan-Meier method was used to compute time from birth to death or last follow-up visit. Statistical significance was designated as a P value <.05 (2-tailed) or as 95% CIs that excluded 1.

Results

A total of 129 perinatally HIV-1—infected children were enrolled in San Diego (n = 61), Los Angeles (n = 46), and Fresno (n = 22) during 1994–2001. The number of patients increased temporally: there were 64 patients in 1994, 66 in 1995, 75 in 1996, 83 in 1997, 92 in 1998, 97 in 1999, 104 in 2000, and 101 in 2001. A total of 49% were male, and 40.3% were Hispanic, 29.5% were African American, 27.1% were white, and 3.1% belonged to other ethnic groups. Ethnic differences were observed between sites: nearly one-half of the patients in the San Diego and Fresno cohorts were Hispanic (47% and 57%, respectively), compared with nearly one-fourth in Los Angeles (24%; P = .0145). The age at the last follow-up visit ranged from 9 months to 21.3 years, and the mean age of 9.0 years did not differ significantly by site: 9.7 years each in Los Angeles and San Diego, compared with 7.0 years in Fresno (P = .083).

The mean CD4⁺ cell percentage increased temporally from 22.5% in 1994 to 31.2% in 2001, and the increase was apparent at all 3 sites (P = .007) (figure 1). The mean plasma HIV-1 RNA load decreased temporally from 4.53 log ₁₀ copies/mL in 1996 to 3.27 log₁₀ copies/mL in 2001 (P < .001) (figure 1). Treatment and prophylaxis data are shown in figure 2. Temporal changes in antiretroviral use and prophylaxis for PCP were observed (P < .0001). In 1994, 14% of the cohort were not receiving antiretrovirals, 56% were receiving monotherapy, 30% were receiving dual therapy, and none were receiving HAART. At the same time, 55% of the patients were receiving PCP prophylaxis. By 1997, nearly one-half of the patients were receiving HAART, and by 2001, a large majority (93%) were receiving HAART, whereas only 2% were not receiving treatment, 1% were receiving monotherapy, 4% were receiving dual therapy, and 16% were receiving PCP prophylaxis. The mean plasma HIV-1 RNA load was significantly lower among HAART users (3.58 log₁₀ copies/mL), compared with non-HAART users (3.89 log₁₀ copies/mL; P = .003). Similarly, the mean CD4⁺ cell percentage was significantly higher among HAART users (30.2%), compared with non—HAART users (22.7%; P = .003).

Figure 1

Mean CD4⁺ cell percentages and log₁₀ plasma HIV-1 RNA loads during the study period.

Figure 2

Antiretroviral treatment and Pneumocystis jiroveci pneumonia (PCP) prophylaxis during the study period.

The percentage of hospital admissions decreased temporally; in 1994, almost one-third (32.8%) of the cohort was hospitalized, whereas in 2001, only 5.9% was hospitalized (figure 3). The hospitalization rate also decreased significantly from 6.49 admissions per 100 person-years in 1994 to 0.60 admissions per 100 person-years in 2001 (P < .001). The mean length of stay decreased from 8.2 days in 1994 to 2.8 days in 2001, but the difference was not statistically significant (P = .167) (figure 3). When stratified by indication for hospitalization, hospital admissions decreased temporally for AIDS-related events; 15.6% of the cohort was admitted for an AIDS-related event in 1994, but this decreased to 0% in 2001 (P < .0001) (figure 4). Similarly, the percentage of children hospitalized for Centers for Disease Control and Prevention category B HIV-1 infection decreased temporally from 29.7% in 1994 to 5.9% in 2001 (P < .0001). Hospital admissions for surgical and/or diagnostic procedures or for adverse drug events were few and did not change over time. Pneumonia (40 cases) and sepsis (24 cases) were the principal causes of hospitalization during the 1994–1995 study period, followed by hematological-based AIDS events (17 cases; anemia, thrombocytopenia, and pancytopenia). Pneumonia continued to be the leading cause of hospitalization during 1996–1997 (25 cases) and 1998–1999 (22 cases), but a dramatic decrease in such hospitalizations occurred during 2000–2001 (3 cases). During latter period, other bacterial infections (3 cases), antiretroviral toxicity (3 cases), and organ biopsy (3 cases) emerged as leading causes of hospitalization (table 1).

Figure 3

Hospitalization rates and length of stay during the study period.

Figure 4

Percentage of hospitalized children, by indication for hospitalization. CDC category B, Centers for Disease Control and Prevention category B HIV-1 infection.

Table 1

Cause of HIV-1—related hospitalization of pediatric patients during 1994–2001.

Logistic regression analysis showed that CD4⁺ cell percentage and viral load were independently associated with the risk of hospitalization, even after adjusting for HAART use, PCP prophylaxis, year, site, sex, and ethnicity (table 2). Linear regression showed that use of HAART, CD4⁺ cell percentage, year, and PCP prophylaxis were independently associated with viral load.

Table 2

Covariates associated with the risk of hospitalization.

A temporal reduction in the mortality rate was observed. In 1994, the annual mortality rate was 7.8%, whereas no deaths were observed in 2001. The mean age at death was 4.7 years and did not vary by site. Kaplan-Meier analysis showed that the survival rate at 10.5 years was 88.6%. When stratified by treatment, the survival rate was significantly higher for patients who had received HAART (survival at 10.5 years, 95.7%), compared with children who had never received HAART, for whom the probability of surviving to the age of 10.5 years was 52.9% (P < .001, by the log-rank test) (figure 5). The median survival rate did not differ by site or ethnicity (P = .768 and P = .327, respectively, by the log-rank test). Causes of HIV-1—related mortality in pediatric patients are shown in table 3. Among the 98 children who received care during 1994–1997, 9 (9.2%) died; the average annual mortality rate was 3.1%, with sepsis and pneumonia being the principal causes of mortality, followed by disseminated cytomegalovirus infection, HIV-1 encephalopathy, non-Hodgkin lymphoma, disseminated Mycobacterium avium complex infection, and cardiomyopathy. A total of 110 children received care during 1998–2001, 2 (1.8%) of whom died; the average annual mortality rate decreased to 0.5%, with pneumonia being the principal cause of mortality (table 3).

Figure 5

Survival curve showing time from birth to death or last follow-up visit, by HAART use.

Table 3

Causes of AIDS-related mortality among pediatric patients, 1994–2001.

Discussion

Data from this cohort of pediatric HIV clinics in southern California have shown a substantial decrease in HIV-1—associated morbidity and mortality during 1994–2001 concurrently with the increased use of combination antiretroviral therapy. In addition, with the increased use of combination antiretroviral therapy in our cohort, we have observed a significant increase in the mean CD4⁺ cell percentage and a decrease in the mean log₁₀ plasma HIV-1 RNA load, which was still evident up to April 2003 (data not shown). Gortmaker et al. [6] found a significant association between the effect of combination antiretroviral therapy and a decrease in mortality (from 5.3% in 1996 to 0.7% in 1999) among children enrolled in the Pediatric AIDS Clinical Trial Protocol 219, although specific causes of mortality were not noted. This decrease in the mortality rate is similar to that in our 1994–2001 cohort, with a 7.8% to 0% decrease over the study period. Simpson et al. [13], in their study of a New Haven cohort in the pre-HAART era, reported a 47.7% mortality rate among 44 perinatally HIV-1—infected children, 86% of whom were <3 years old. The leading causes of death in this cohort were HIV-1 encephalopathy, PCP, disseminated M. avium complex infection, wasting syndrome, esophageal candidiasis, and disseminated cytomegalovirus infection. Recently, Selik et al. [14] used death certificate data obtained from the National Center for Health Statistics for 1987–1999 to document a decrease in the proportion of deaths due to PCP and an increase in the proportion of deaths related to sepsis that were temporally associated with the advent of combination antiretroviral therapy. In their study, the proportion of deaths due to nontuberculous mycobacterial infection, cytomegalovirus disease, wasting syndrome, and HIV-1 encephalopathy remained stable over the study period. A limitation of this study is that, by using death certificate data, mortality rates over time could not be ascertained, because the total number of HIV-1—infected patients observed was unknown and the rate of HIV-1—associated comorbidities was not reported [14].

In our cohort, sepsis and pneumonia were the leading causes of death, followed by disseminated cytomegalovirus infection, HIV-1 encephalopathy, and non-Hodgkin lymphoma. This is in contrast with the causes of mortality in the pre-HAART era, in which PCP, HIV-1 encephalopathy, wasting syndrome, and disseminated M. avium complex infection were the leading causes of death [13]. Reasons for this difference may be explained by the improvements in antiretroviral management during our study period, compared with that in the pre-HAART era. In addition, the increase in CD4⁺ cell percentage over time reduced the risk of opportunistic infections in our patients. Recently, Chantry et al. [15] reported on the positive correlation between height-growth velocity and survival in symptomatic HIV-1—infected children in the pre-HAART era. This effect was independent of plasma viral load or CD4⁺ cell count. Improvements in ponderal and linear growth have been observed in children treated with HAART [16].

A recent study from a pediatric HIV-1 clinic in Madrid, Spain, that spanned 1987–2001 reported a survival benefit and a decrease in the rate of progression to an AIDS-defining illness in children receiving HAART, compared with children not receiving HAART, although specific cause(s) of hospitalization or mortality were not reported [17]. Similar to the Spanish cohort, we found an improved survival rate among children receiving HAART, compared with those who were not receiving HAART.

Our study, to our knowledge, involves the first cohort of perinatally infected children in which morbidity and mortality are analyzed in a predominantly Hispanic population. The proportion of Hispanic HIV-1—infected children in other studies has ranged from 22% to 35% [6, 13, 18]. In our cohort, 40% were Hispanic, with the highest rate (57%) in Fresno, a clinic serving the rural central California valley. The mortality rate over the study period was no different in the urban clinics of San Diego (9.8%) and Los Angeles (6.7%), which were conducted in academic institutions, compared with that in the Fresno clinic (8.7%), which is closely supervised by an academic institution in San Francisco.

Our morbidity data is unique in that it reflects the diagnosis at hospital discharge during the transition from the era of HIV-1 monotherapy to the era of combination antiretroviral therapy for perinatally HIV-1—infected children who were already receiving specialized care. In our cohort, the principal discharge diagnoses in the pre-HAART era (1994–1995) were pneumonia, sepsis, and anemia. Pneumonia was also the leading cause of hospitalization in the European Collaborative study in the pre-HAART era [12]. Of interest, pneumonia continued to be the leading cause for hospitalization up to 1999 at our 3 clinics, but during 2000–2001, antiretroviral toxicity and organ biopsy emerged as leading causes of hospitalization, in addition to pneumonia and other bacterial infections. These more recent causes of hospitalization reflect the complexity of the current medical management of HIV-1 infection in pediatric patients, with a shift in morbidity- to treatment-related admissions. Amprenavir (Agenerase; GlaxoSmithKline)—associated rash, indinavir (Crixivan; Merck & Co.)—associated nephrolithiasis, and interstitial nephritis, as well as immune reconstitution syndrome associated with HAART (prompting renal and liver biopsies), are just a few examples of the shifts in morbidities associated with HIV-1 disease in children [19]. Dankner et al. [20] compiled pediatric HIV-1 opportunistic infections data from different pediatric AIDS Clinical Trials Protocols in the pre-HAART era and found that serious bacterial infection was the leading AIDS-related morbidity in patients who were already receiving specialized care for HIV-1 infection. In a subgroup analysis of patients with serious bacterial infections, pneumonia was the leading cause of morbidity, which is consistent with our study. Paul et al. [10], in an adult HIV cohort in New York, documented a 50% decrease in HIV-1—associated hospitalizations between 1995 and 1997. We observed a 10-fold decrease in hospitalizations during our 8-year study. In their study, Paul et al. [10] reported that bacterial pneumonia followed by PCP were the leading causes of hospitalization, with decreases in hospitalizations due to these causes over time. This is in contrast to our perinatal cohort, in which PCP was rarely seen. This discrepancy can be explained by the fact that the median CD4⁺ cell count in the adult cohort was very low, which is evidence that patients in that cohort were severely immunosuppressed. In contrast, the mean CD4⁺ cell percentage in our 1994 cohort was 22.5%, which is evidence of moderate immunosuppression. In addition, 100% of the children in our cohort with CD4⁺ cell counts or percentages considered to indicative of severe immunosuppression were receiving PCP prophylaxis (data not shown).

Our data suggest that HIV-1—related mortality and hospitalization rates for opportunistic infections and other infections decreased significantly during 1994–2001 in perinatally infected children. This decrease in morbidity and mortality has correlated with an increase in CD4⁺ cell percentage and a decrease in HIV-1 RNA load concurrent with the expanded use of HAART. Long-term longitudinal studies will be critical to fully describing the continuing benefits of HAART in HIV-1—infected children.

Acknowledgments

We thank Jenny Rohan, Camila Romero, Cameron Kaehler, and Antonio Turcios for medical records data extraction.

Conflict of interest. All authors: No conflict.

Presented in part: XIVth International AIDS Conference, Barcelona, Spain, July 2002 (abstract ThPeB7230); and Pediatric Academic Societies Annual Meeting, Seattle, May 2003 (abstract 1886).