Natural History of Human Immunodeficiency Virus Disease in Perinatally

Objective. To describe the progression of human immunodeficiency virus (HIV) disease through clinical stages from birth to death among a large number of perinatally infected children. Methods. The Pediatric Spectrum of Disease (PSD) project, coordinated by the Centers for Disease Control and Prevention (CDC), has conducted active surveillance for HIV disease since 1988 in seven geographic regions. PSD data are collected from medical and social service records every 6 months through practitioners at each participating hospital clinic. We analyzed data from pennatally HIV-infected children born between 1982 and 1993. The natural history of HIV disease was divided into five progressive stages using the clinical categories in the CDC 1994 pediatric HIV classification system: stage N, no signs or symptoms; stage A, mild signs or symptoms; stage B, moderate signs or symptoms; stage C, severe signs or symptoms; and stage D, death. A five-stage Markov model was fitted to the PSD data. To compare the estimates from the PSD project with the published estimates, we also fitted an alternative Markov model using acquired immunodeficiency syndrome (AIDS; 1987 case definition) in place of stage C and also calculated standard Kaplan-Meier estimates. Results. A total of 2148 perinatally HIV-infected children were included in the analysis. The estimated mean times spent in each stage were: N, 10 months; A, 4 months; B, 65 months; and C, 34 months. We estimated that a child born with HIV infection has a 50% (95% confidence interval [CII, 40%-60%) chance of severe signs or symptoms developing by 5 years of age and a 75% (95% CI, 68%-82%) chance of surviving to 5 years of age. For a child in stage B, there is a 60% (95% CI, 49%-71%) chance of severe signs or symptoms developing within the next 5 years and a 65% (95% CI, 56%-73%) chance of surviving 5 more years. The estimated mean time from birth to stage C was 6.6 (95% CI, 5.7-7.5) years, From the *Depa ent of Biostatistics, Rollins School of Public Health, Emory University, and Division of HIV/AIDS, Centers for Disease Control and Prevention, Atlanta, Georgia; §New York City Department of Health; ilLos Angeles Department of Health Services; #{182}PuertoRico Department of Health, San Juan; #Massachusetts Department of Public Health, Jamaica Plain; ** e, s Department of Health, Austin; Children’s National Medical Center, Washington, DC; and 55Stanford University, Stanford, California. I, IlMembers of the Pediatric Spectrum of Disease Clinical Consortium are listed in “Acknowledgments.” Received for publication Apr 17, 1995; accepted Oct 2, 1995. Reprint requests to (H.X.B.) Department of Biostatistics, Rollins School of Public Health, Emory University, Atlanta, GA 30322. PEDIATRICS (ISSN 0031 4005). Copyright © 1996 by the American Academy of Pediatrics. and the estimated mean survival time from birth was 9.4 (95% CI, 8.1-10.7) years. From the alternative Markov model, the estimated mean time from birth to AIDS was 4.8 (95% CI, 4.5-5.2) years. Conclusion. Markov modeling using the revised pediatric classification system allowed us to describe the natural history of HIV disease in children before diagnosis of AIDS. On average, children progress to moderate symptoms in the second year of life and then remain moderately symptomatic for more than half of their expected lives, underscoring their need for clinical care before the onset of AIDS. The results from the Markov model are useful in family counseling, health care planfling, and clinical trial designs. Pediatrics 199697:710716; pediatric HIV and AIDS, natural history, Markov model. ABBREVIATIONS. HIV, human immunodeficiency virus; AIDS, acquired immunodeficiency syndrome; PSD, Pediatric Spectrum of Disease; CDC, Centers for Disease Control and Prevention; LIP, lymphoid interstitial pneumonia; CI, confidence interval. Initial reports of the clinical manifestations of human immunodeficiency virus (HIV) infection in pennatally infected children described a rapidly progressive disease with short survival.”2 However, because these observations were made early in the epidemic on small numbers of children whose HIV infection had been ascertained because of their severe illness, they were biased toward including HIV-infected children with the most rapidly progressive disease. Later descriptions were more comprehensive, because larger numbers of children had been observed, the duration of follow-up was longer, and the development of HIV antibody testing allowed asymptomatic infected children to be included.3 As a result, recent estimates of the time to acquired immunodeficiency syndrome (AIDS) and death in peninatally H1V-infected children are much longer than the iitial estimates. Until recently, no systematic method for descnibing the course of HIV infection in children was available. The 1987 dassification system for HP/-infected children categorized HP/-related diseases but was not designed to describe disease progression.4 However, the 1994 revision of this classification system allows children to be classffied into mutually exclusive categories based on severity of disease. With this by guest on October 30, 2017 http://pediatrics.aappublications.org/ Downloaded from ARTICLES 711 new classification system and more cumulative foblow-up time, an even more accurate description of the course of pediatric HIV infection, especially duning the early stages, is now possible. An accurate understanding of the timing of the progression of HIV infection in children has important clinical, research, and public health benefits. For clinicians, such knowledge contributes to prognostic information that can be communicated to patients and their families. For researchers designing clinical trials, knowledge of the rates of occurrence of vanous defined stages of disease assists in calculating sample sizes and planning trial durations. For health planners, the length of each stage of disease is impontant for estimating the resource needs of HIVinfected children. A description of the progression of pediatric HIV disease through the clinical stages can be developed from clinical data. Ideally, the clinical data would come from prospective clinical observations. However, most prospective clinical series have limited numbers of children under observation. The Pediatric Spectrum of Disease (PSD) project has collected longitudinal clinical data from more than 2000 HIVinfected children born to HP/-infected mothers. These children were diagnosed at various ages and followed for varying lengths of time. The Mankov model can be used to model these longitudinal data to describe the progression of HIV disease. This method complements prospective cohort studies by estimating progression times based on larger numbens of infected children using the newly designed classification system. In addition, the Markov model is uniquely suited to estimating progression time, because even in prospective studies, the exact time of development of signs and symptoms that indicate progression to a more severe stage is not precisely known and is dependent on the frequency of medical evaluations. For example, documentation of liver and spleen enlargement moves a child from no-signs on symptoms stage to the mild-signs and symptoms stage. However, the precise age at which the child entered the mild-signs and symptoms stage is often not known (it occurred at some point in time between the two clinical evaluations). Therefore, the age at transition between disease stages is best estimated by mathematical models. In this article, we describe the age-specific natural history of HIV disease from birth to death among children enrolled in the PSD project. To best describe the course of HIV disease among these lange numbers of HP/-infected children, we applied Markov modeling techniques that are specifically suited for modeling short portions of individual disease histories and for evaluating transitions between clinical stages, such as those in the new classification system.