Conditions in Newborns at Risk for - NCBI

3 downloads 131 Views 2MB Size Report
Health and Nutrition Sciences, Brooklyn Col- lege, Brooklyn, NY. .... as age, gender, and year of graduation from medical school) ...... Columbia Law. Rev. 1993 ...
The Treatment of Non-HIV-Related Conditions in Newborns at Risk for HIV: A Survey of Neonatologists

Betty Wolder Levin, PhD, David H. Krantz, PhD, John M. Driscoll, Jr., MD, and Alan R. Fleischman, MD

Introduction

HIVInfection in Newboms

Since the start of the acquired immunodeficiency syndrome (AIDS) epidemic, there has been much attention paid to attitudes of providers concerning treatment for patients who test positive for the human immunodeficiency virus (HIV). The main focus has been on physicians' willingness to treat people with HIV and on the appropriate treatment for HIVrelated conditions. There has been little discussion, however, about the relevance of HIV for decision making about treatment of conditions not directly related to HIV infection. One group of patients for whom such decisions must be made are HIV-positive infants admitted to neonatal intensive care units. Infection with HIV in women of reproductive age is increasing dramatically.1 Factors associated with HIV infection include prematurity and other conditions requiring admission to neonatal intensive care units.23 Therefore, neonatologists, the physicians who care for critically ill newborns, increasingly confront questions concerning the appropriate care for HIV-positive infants with life-threatening conditions. We present data from a survey of neonatologists indicating that a substantial proportion of respondents recommended less aggressive treatment for HIV-positive infants, not only those known to be infected but also children of HIVpositive mothers who were merely at risk for HIV. We show that these recommendations are associated with low estimates of the infants' maximum quality of life. We discuss decision making about the aggressiveness of treatment for non-HIVrelated conditions for HIV-positive infants.

There has been substantial research on the epidemiology of perinatally acquired HIV. Before birth, it is not possible to predict which infants bom to HIVpositive mothers will be infected. Since HIV antibodies cross the placenta, virtually all such infants test positive for HIV, even though most are not actually infected; in North America, approximately one quarter of the infants born to HIVpositive women are themselves infected.4 Recent research has demonstrated that zidovudine (AZT) can reduce the rate of transmission.5 Sophisticated tests now enable identification of HIV in some infants soon after birth; often, however, definitive diagnosis is not possible for weeks or months.6 The natural history of pediatric HIV is variable and cannot be predicted for any one individual. While some infants become sick early and die within the first few months of life, others live for many years before becoming sick.7'8 Clinical studies suggest that prophylactic treatBetty Wolder Levin is with the Department of Health and Nutrition Sciences, Brooklyn College, Brooklyn, NY. David H. Krantz is with the Departments of Psychology and Statistics, Columbia University, New York, NY. John M. Driscoll, Jr., is with the Department of Pediatrics, Columbia University College of Physicians and Surgeons. Alan R. Fleischman is with the Department of Pediatrics and Department of Epidemiology and Social Medicine, Albert Einstein College of Medicine and Montefiore Medical Center, New York, NY. Requests for reprints should be sent to Betty Wolder Levin, PhD, Department of Health and Nutrition Sciences, Brooklyn College, City University of New York, 2900 Bedford Ave, Brooklyn, NY 11210-2889. This paper was accepted July 11, 1995. Editor's Note. See related editorial by Avery (p. 1484) in this issue.

American Journal of Public Health 1507

Levin et al.

ment can prevent Pneumocystis carinii pneumonia, the most common cause of early death in children infected with HIV,9 and that other treatments also extend and improve quality of life.'0'" However, HIV must still be considered an ultimately lethal condition. In addition, these infants, whether infected or not, may suffer from the consequences of their mothers' illness and, often, the illness of other family members; also, they frequently grow up in environments in which there are a myriad of social problems.'2-14

Decision Making aboutAggressiveness of Treatment For the last 2 decades, decision making about the treatment of catastrophically ill newborns has been a major bioethical concern.'5-17 A number of studies have demonstrated that pediatricians sometimes make decisions to limit the aggressiveness of treatment for infants with terminal conditions, high risk of serious impairment, or other problems expected to lead to a very poor quality of life.18-23 In some cases, treatments may be withheld for conditions expected to reduce quality of life. In other cases, treatment for a concurrent life-threatening condition may be withheld. A few studies have indicated that some physicians believe treatments for non-HIV-related conditions should be withheld from patients who are HIV positive. These treatments include lifesustaining treatments for end-stage renal disease24 and cardiac surgery25 and more routine diagnostic tests, treatments, and practices.2627 In 1989, we conducted a survey in six hospitals in New York City to investigate attitudes of neonatal care providers concerning HIV.28 That study revealed that a significant proportion of the respondents would recommend less aggressive treatment for non-HIV-related conditions for infants at risk for HIV in comparison with those not at risk. Many people found the results from the study disturbing.29 Since HIV is becoming more prevalent in many areas of the country3O and knowledge of HIV continued to increase, we decided to survey a national group of neonatologists.

Methods Study Design In May 1991, through the auspices of the Executive Committee of the Perinatal Section of the American Academy of Pediatrics, we mailed an anonymous, 14-page survey instrument, along with a 1508 American Journal of Public Health

postcard on which to report participation, to each of the 1508 physician members of the section who lived in the United States, Puerto Rico, or Canada. This was followed by a second mailing to nonrespondents in June. Approximately half of the pediatricians practicing neonatal medicine in the United States were members of the Perinatal Section at the time. The survey instrument was developed to investigate attitudes, beliefs, and knowledge about infants at risk for AIDS. Part of the instrument required respondents to recommend treatments for three neonates described in hypothetical vignettes. They were asked to assume that each treatment, if given, would increase the baby's chance of survival and that the parents' views were the same as their own. The vignettes were identical except that, for one, respondents were instructed to assume that the infant's mother was known to be HIV positive but the infant's status was unknown; in another, respondents were asked to assume that there was a test such that they could know for sure on the first day of life that the infant was infected with HIV. Finally, respondents were asked about treatment for an infant with no known HIV risk. Specifically, the vignettes presented the following series of medical problems and asked respondents to indicate if they would give each of five treatments by circling "definitely yes," "probably yes," "probably no," or "definitely no" for each infant. The vignettes began, "Soon after is found to have duodenal birth, Baby atresia (an intestinal defect)." Respondents were asked if they would give (1) "intravenous fluids" and (2) "surgery to correct the intestinal defect." They were then asked to "suppose the baby is also found to have VSD (a cardiac defect)" and asked if they would recommend (3) "open heart surgery." Finally, respondents were asked to "suppose the heart lesion was corrected, but he suffered severe chronic kidney damage" and asked if they would give (4) "maintenance dialysis" and (5) "cardiac resuscitation if he arrests." Other sections of the survey instrument contained questions concerning factors we thought could be related to treatment choices. We asked respondents to estimate the percentage of infants born to HIV-infected women they believed would actually be infected and, on average, at what age infected infants would die of HIV-related disease. In addition, we asked respondents to fill in the blanks in the following phrases: "I would not be

surprised to learn that as few as _% or as many as _% of the children born to HIV-positive women are actually infected with HIV" and "I would not be surprised to learn that, on average, children born infected with HIV in 1991 lived as little as - years or as long as - years." These values indicated respondents' estimates of the highest and lowest plausible transmission rates and the shortest and longest plausible life expectancy for newborns infected with HIV. Quality of life ratings were obtained by asking respondents to circle a range on a 1 to 7 scale, with the bottom of the range corresponding to the potential lowest quality of life and the top of the range corresponding to the potential highest quality of life for each condition. There were also other structured and open-ended questions on HIV and other aspects of care in neonatal intensive care units. In addition, we requested personal and professional background information.

Characteristics ofRespondents Nine hundred fifty-one physicians (63.1%) completed and returned the survey. In addition, 36 (2.4%) of those who returned questionnaires said they no longer practiced or thought they should not respond to the survey for another reason. More than 90% of the respondents who returned the survey answered all of the questions. Surveys were received from 49 states, the District of Columbia, Puerto Rico, and Canada. Ninety-one percent of the respondents identified themselves as neonatologists; all were board certified in pediatrics. More than 90% reported having cared for at least one neonatal intensive care unit patient thought to be at risk for HIV, but only a third reported caring for as many as five whose mothers were known to be HIV positive. Table 1 describes additional respondent characteristics. At the request of the Journal's Editor, in order to investigate possible

selection bias, we sought demographic information (such as age, gender, and year of graduation from medical school) for the nonrespondents. Unfortunately such information was unavailable retrospectively. However, as discussed below, we note that survey responses pertaining to treatment recommendations varied little with such demographic variables. Statistical Methods Treatment decisions were dichotomized as treatment recommended ("defiNovember 1995, Vol. 85, No. 11

HIV and Newborns

TABLE 1-Characteristics of Respondents: Members of the Section on Perinatal Medicine, American Academy of Pediatrics

Characteristic

Specialty, % Neonatologya Other Type of hospital, % Voluntary Private Public Other Professional activity, % Clinical only Research and clinical Research only Other No. infants cared for known to have an HIVpositive mother, % No infants 1-5 infants 6-99 infants 2 100 infants Sex, % Male Female Age, y Range Median Country of birth, % United States Other Country of medical education, % United States Other Race, % White Asian Hispanic Black Religion, % Protestant Catholic Jewish None Hindu Other Importance of religion in daily life, % Very important Fairly important Not too important Not at all important Political orientation, % Progressive Liberal Middle of the road Conservative Very conservative

Sample (n = 951)

E

91 9 62 18 13 7

52 42 1 5

18 49 33

80

CD a) -1-

CD

E E

0 0

70 60505

4030 30

C

20

p a)

10

0

26

IV Feeding

Intestinal Surgery

no known HIV

rsk

Cardiac Surgery

Dialysis

mother HIV +

Resuscitation

infected with HIV

vertical bars are 95%-confidence intervals

Note. Shown are the percentages of neonatologists recommending each of five treatments for infants in three vignettes differing in terms of HIV status, with binomial 95% confidence intervals for each percentage.

FIGURE 1-Neonatologists' recommendations about the treatment of infants in three vignettes, by infants' HIV status.

1

73 27

29-72 42 77 23 78 22

83 11 4 1 38 23 20 7 6 6 26 29 27 18 6 28 43 22 1

Note. A small number of respondents did not answer each of the questions. a85% were board certified in neonatology.

November 1995, Vol. 85, No. 1 1

100

nitely yes" and "probably yes" responses) and treatment not recommended ("probably no" and "definitely no" responses). We used NcNemar's chi-square test with continuity correction to examine differences in treatment recommendations by HIV status. We used linear and dichotomous regression to analyze factors affecting treatment choice. One way to assess the reliability of questionnaires is to examine departures from the logically expected orderings of certain responses. For example, if the responses concerning treatment were careless or subject to momentary fluctuations, an appreciable number of respondents would have recommended treatment for the infant infected with HIV but not for the infant with no HIV risk. For the three cases, there were theoretically 15 such reversals possible (three pairs of cases by five treatment options). However, 97% of the respondents gave decision profiles with not even one such reversal. An item with a much less familiar format was the rating of likely minimum and maximum quality of life. Yet, 95% of respondents rated both the minimum and maximum no higher for an infant infected with HIV than for an infant at risk (mother HIV positive), just as one would expect for a reliable response. The validity of questionnaire responses can always be challenged, since one cannot be certain how they relate to

actual behavior. Internal regularities in responses, however, can be used as an indication of validity. Responses to this survey showed many internal regularities that suggest high construct validity. For example, a respondent's questionnaire responses concerning the likely quality of life for HIV-infected infants were closely related to his or her treatment decisions (see Results section). The confidence intervals and hypothesis tests given in the Results section indicate the allowance that must be made for sampling error if the respondents were viewed as having been drawn randomly. Additional systematic differences may exist between the respondents and other members of the Perinatal Section and nonmember neonatologists who were outside the scope of this survey.

Results Recommendations about Treatment for Infants at Risk for HIV Figure 1 shows that recommendations varied both by treatment and by HIV status. For the infant with duodenal atresia and no known HIV risk, virtually all of the respondents recommended intravenous fluids, surgery to correct the intestinal defect, and surgery for a concomitant ventriculoseptal defect. Other treatments (which are seen by neonatoloAmerican Journal of Public Health 1509

Levin et al.

TABLE 3-Neonatologists' Highest Quality of Life Estimates and Treatment for an Infant of an HIV-Positive Mother

TABLE 2-Neonatologists' Estimates of Future Quality of Life for Infants with Different Conditions

Quality of Life Condition of Infant

Excellent

Poor

Mother HIV positive Infected with HIV Chronic dialysis Down syndrome 625gatbirth 625g and grade IV intraventricular hemorrhage Ventriculoseptal defect (repaired)

4 4 4

5 5 5 5 5 5

6 6 6 6 6

6

7 7 7 7 7 7

4

5

6

7

3 2 1 |jj2j3j 1 2 3 3 2 1 3 2 1 3 1 2

4

3

4 4

Factors Possibly Related to Treatment Decisions What accounts for the differences in treatment recommendations for infants at risk for HIV and those infected with HIV? We performed internal analyses to examine the relationship, if any, of treatment choice to respondents' beliefs conceming the perinatal HIV transmission rate, life expectancy, and quality of life. We used linear regression to examine the relation of these variables to the difference between the cases (HIV-positive mother/no HIV risk, HIV infected/no

1510 American Journal of Public Health

1-3 4

5 6 7

1

2

Note. Respondents were asked to indicate their best estimates of the range of future quality of life for infants having each one of a series of conditions by drawing an oval surrounding that range on a seven-point scale (1 = poor, 7 = excellent). The ranges portrayed extend from respondents' median lowest to median highest quality of life estimates for each condition.

gists as more aggressive) were recommended by a very large majority: 91% would recommend maintenance dialysis in cases of chronic kidney damage, and 85% would recommend resuscitation following cardiac arrest. For the infant at risk for HIV (mother infected), there were sharp decreases in the percentage of physicians recommending cardiac surgery, dialysis, and resuscitation following cardiac arrest. For these treatments, the odds of a positive treatment recommendation (percentage recommending divided by percentage not recommending) decreased by a factor of 4 to a factor of 7 for the infant at risk in comparison with the infant with no known risk. For the infant known to be HIV infected, in comparison with the infant merely at risk for HIV, a further sharp decline was seen in the percentage of physicians recommending each treatment. These declines were all highly reliable statistically (P < 10-8).

High Estimate of Infant's Quality of Life

HIV risk) in total number of treatments recommended (which, in principle, could range from -5 to 5). We used dichotomous regression techniques to look at the relation of specific treatment decisions to expectations about future quality of life. Estimates of the peninatal transmission rate of HIV Three variables expressed physicians' views of perinatal transmission: the best estimate of transmission rate, the lowest plausible rate, and the highest plausible rate. Slightly more than half gave best estimates that agreed with the contemporary studies indicating that one third or fewer children of HIVpositive women would be infected. Most thought that the highest plausible rate was at least 50% and that the lowest plausible rate would not be less than 20%. Estimates of life expectancy for infants infected with HIV When asked at what age they expected infants infected with HIV to die, 6% of respondents said 1 year or less, 28% said from 1 to 2 years, 40% said 2 to 4 years, 22% said 4 to 7 years, and 5% said more than 7 years. When asked about the shortest and longest life expectancy, almost three quarters (72%) thought that infants, on average, might live as little as 1 year or less; almost all of the remaining respondents (27%) indicated that they would not be surprised to learn that most children with HIV lived 5 years or less. When asked to indicate the longest life expectancy they would not find surprising, about one third (33%) said 5 years, about half (49%) said 5 to 10 years, and 14% said 12 to 20 years. Four percent said that they would not be surprised if, on average,

No. Respondents

% Recommending Cardiac Resuscitation

156 166 186 152 260

44 47 55 54 62

Note. Shown are the percentages of respondents who would recommend cardiac resuscitation for an infant who was the child of an HIV-positive mother and had suffered kidney damage requiring chronic dialysis following surgery for congenital defects in relation to respondents' estimates of the highest potential quality of life (1 = poor, 7 = excellent). The increase in percentage as a function of estimated quality of life was statistically reliable by linear logistic regression (P = .0004; the 95% confidence interval was .064 to .232 logits per unit quality of life).

infected children lived

more

than 20

years.

Expected future quality of life for infants at risk for HIV Most respondents expected the future quality of life for an infant born to an HIV-positive mother to be fairly poor whether the infant was infected with HIV or not. Table 2 summarizes the potential highest future quality of life and potential lowest future quality of life ratings for infants with a number of congenital conditions. The median quality of life ratings for an infant whose mother was infected with HIV were roughly between those for an uninfected child on chronic dialysis and an uninfected child with Down syndrome (see Table 2). However, 28% of the respondents did rate the highest future quality of life for the infant of an HIV-positive mother as 7, the highest scale rating. The median quality of life ratings for the HIV-infected infant were even lower, similar to those for an extremely premature infant who has suffered a grade IV intraventricular hemorrhage. By contrast, the quality of life ratings for an otherwise normal baby with a repaired ventriculoseptal defect (a cardiac defect corrected by open heart surgery) were close to the top of the scale. Analysis of treatment choices. Respondents who estimated a higher potential quality of life for infants at risk or infected

November 1995, Vol. 85, No.

11

HIV and Newborns

with HIV were more likely to provide such infants with more of the treatments they would provide for infants with no HIV risk. For the infant merely at risk for HIV, this effect of the highest future quality of life estimate was small; the correlation between the difference in number of treatments recommended and the estimate was merely -.13 (1.7% of variance). This was statistically reliable with the large number of respondents, and the regression coefficient for highest future quality of life remained stable and reliably negative when other factors were added to the regression model. Twelve other plausible explanatory variables were tested, including age and sex of the physician, lowest future quality of life estimate for infants at risk, judgments concerning the perinatal transmission rate, life expectancy of infected infants, and the quality of life estimates for infected infants. None of these variables had a statistically reliable effect singly or in combination, and all 12 together accounted for only another 1.8% of the variance.

For the infant infected with HIV, the highest future quality of life estimate had a more powerful effect on the difference in number of treatments recommended (correlation of -.31, or 9.6% of variance). Again, the regression coefficient was stable and reliably negative when the other factors were added to the model; these other factors accounted for an additional 3.5% of the variance. Dichotomous (logistic) regressions showed the same relations of highest future quality of life estimates to recommendations about particular treatments. A high estimate for an infant at risk produced a modest additional tendency to treat that infant, and a high estimate for an infected infant produced a stronger added tendency to treat (see Table 3 for an example of the effect of quality of life estimates on one treatment choice, cardiac resuscitation).

Responses to Open-Ended Questions In response to open-ended questions asking whether there was anything they would like to add after answering the structured questions, respondents provided a range of comments indicating reasons for their treatment choices. To explain why treatment should be provided, one respondent said: Unless it is clear that life expectancy will be short and quality of survival will be poor and painful in an individual case, we should not limit the initial intensive November 1995, Vol. 85, No. 11

treatment of HIV-suspect or HIVinfected newborns. We should stop later, however, if treatment appears futile or "worse than the disease."

Another said: A baby's HIV status does not change my therapies. With current (AZT) therapy, children with HIV at least live long enough to know love and joy. Besides, we will find a cure someday, and it could be in the child's lifetime.

The following response represented a qualified decision to treat: Although HIV+ status at present has poor prognosis ... outcome is not known to be 100% mortality. Thus my therapy would not necessarily be altered by knowing the HIV status of a fetus at birth. Invasive procedures (such as cardiac surgery) would be potentially hazardous to hosp. staff however. Also, to some extent parental wishes might (and should) enter into discussion re "heroic procedures."

In explaining reasons to limit treatment, some respondents emphasized the suffering of the infected children. For example, one wrote: In training we tortured terminal cancer patients prolonging their often inevitable death. Kids hate needles, doctors and hospitals! Why can we never simply take our dying children home and love them and comfort them? Now we do the same over treatment for children with AIDS. We need to be more realistic in America. We need to work on prevention and cure not prolongation.

Other respondents mentioned other factors in addition to the well-being of the child. One said: These ethical decisions are extremely difficult to make. My point of focus is usually basing the decision on what I anticipate the quality of life will be for the infant and his/her family. I realize the limitations of this position. Nevertheless, currently, given the fact that ... infant HIV disease [is] generally considered fatal, a non-aggressive approach seems appropriate. The other concern I have is utilizing maximal health care resources on patients in whom minimal benefit can be expected. I am becoming more and more convinced that our society's present philosophy of "treat all patients maximally" is both unethical and unfair. Explaining the withholding of treat-

ments for both infected and uninfected infants, one respondent wrote: The baby who has a chance that he is HIV infected should not go on maintenance dialysis for a chronic condition because we could funnel valuable resources into him, give him a not-verygood life on dialysis or as a transplant [patient] and then see him die of AIDS. It's a gamble but we need to be more

practical or our system will not stand up to the demands of AIDS.

Another said: If there were some solid hopeful information to mitigate the dismal outlook of HIV infection, I would consider it more of a treatable disease in resuscitation decisions. Though treatments could arise in a 10 year latent period that these children may face, I feel it is a terrible burden to a family and society in the meantime.

Others cited social criteria affecting choices: One factor that comes to mind in the HIV+ cases is always the social situations. Most of our HIV infected mothers are drug abusers with very unstable social situations. This would definitely influence me to be less aggressive with a baby who might need extensive longterm medical care.

Other respondents indicated additional factors that they would consider: In those case variations in which HIV was introduced into the story, a do nothing approach was chosen with the understanding that this disease is universally fatal. However, such decisions can be tempered by the parents' wishes ... because a cure might be found [or] ... because they want to take their child home and be with the baby as long as the active HIV process would allow. The final decision ... must arise from a collaborative team approach involving the doctor(s), the parent(s) and all other support staff. What is in the child's best interest must always prevail and should be the fundamental principle that must be at the foundation of all joint decisions.

Finally, one respondent made a plea for these decisions to be seen by the public within the broader context of medical decision making: AIDS is such an emotional topic-I hope the press does not claim that doctors discriminate against AIDS/ HIV patients by "withholding" treatment. Treatments have always been withheld based on likelihood of success, estimations of later impairments or life span, futility, parental wishes ... etc.... To not take into account HIV status would be folly. Anyway, whatever your results, please present them in a fashion that will be difficult for the press and various AIDS related organizations to misunderstand or misconstrue.

Discussion Many of the neonatologists who responded to our survey recommended less aggressive treatment for critically ill infants born at risk for HIV, or hypothetically known to be infected, than for

similar infants with no HIV risk. Research American Journal of Public Health 1511

Levin et al.

published after our survey has shown that the perinatal use of AZT reduces HIV transmission from mother to infant5 and that infected children may live longer than previously believed.31'32 Surprisingly, our data showed little or no associations between treatment recommendations and beliefs about the perinatal transmission of HIV or life expectancy. Respondents' current (post-survey) attitudes may or may not have been changed by the recent findings or other factors. In addition, there is no way to know if the members of the Section on Perinatal Medicine who did not respond to the survey, or neonatologists who were not members of the section, had or currently have different attitudes. Nevertheless, over 700 physician respondents said they would recommend withholding treatments from infants known to be infected with HIV, and over 400 said they would alter recommendations for infants merely at risk for HIV but not known to be infected. Therefore, even if the views of many respondents have shifted since the survey, we believe our results raise issues that are of continuing relevance. Infants who test positive for the HIV antibody at birth are a unique population. Because maternal antibodies cross the placenta, tests may indicate a positive serostatus, yet infants may not actually be infected. There are no adult patients who test positive for antibody to HIV yet are merely at risk for being infected. Therefore, HIV-positive infants admitted to neonatal intensive care units raise unique ethical issues about treatment choices. They are the only acutely ill patients for whom there is significant uncertainty as to whether they are, in addition, infected with HIV and therefore have this serious chronic illness, which is eventually and inevitably fatal. Many of the respondents said they would not recommend life-sustaining treatments for the infant whose mother was infected with HIV, although they would recommend the same treatments for an infant with no known risk; this occurred even though respondents knew that the infant was merely at risk and that most infants born to HIV-positive mothers would not actually be infected. According to the Child Abuse Amendments of 198433 and the Americans with Disabilities Act of 1990,34 withholding of care from newborns and other patients on the basis of disabilities, including HIV, may be considered illegal.29 These recommendations were consistent with the respondents' view of the 1512 American Journal of Public Health

diminished future quality of life of infants whose mothers are HIV positive. We noted with concern that most respondents believe that the potential quality of life of infants born to HIV-positive women, even those not necessarily infected, is substantially limited. The highest median expected quality of life for these infants was even lower than that for a premature infant weighing 625 g at birth. In addition, the median highest expected quality of life for an infant born to an HIV-positive mother was estimated to be the same as that for an infant with Down syndrome, yet the majority of HIV-positive infants will have no disease. Since infants who are merely seropositive but not infected do not suffer from any medical illness, treatments may be withheld primarily on the basis of presumptions about the infant's social situation and worth. Most infants born at risk are likely to lose their mothers while they are still young; however, such children may grow up with loving natural or foster families and may, in fact, enjoy a good quality of life. As the AIDS epidemic has evolved, people living with HIV are increasingly members of marginalized groups who have faced discrimination and reduced access to medical care because of their race, poverty, and life-style. In light of the new evidence concerning the effects of AZT on perinatal transmission and greater efforts to test pregnant women and newborns, it is likely that increasing numbers of infants will be identified as the children of HIV-positive mothers. Moreover, if more pregnant women and newborns take AZT, even a smaller proportion of the infants identified as at risk will actually be infected. When we conducted our survey, knowing at birth that an infant was actually infected with HIV was only hypothetical. Now, although it is still not possible to identify many infected infants in the immediate newborn period, some infants can be definitively diagnosed soon after birth.56 Therefore, cases of infants known to be infected, like those we presented as hypothetical, now represent real situations requiring decisions. Most of the respondents to our sunrey indicated that they would recommend withholding treatments from an infant if they knew for sure that the infant was infected with HIV. Not only did most indicate that they would withhold aggressive treatments such as dialysis and resuscitation from an infected infant who also suffered chronic kidney failure, but nearly half would not recommend open

heart surgery, and some would not do such routine treatments as surgical repair of an intestinal defect. In general, the respondents expected infants infected with HIV to have a very poor quality of life. About two thirds expected most infected infants to live less than 4 years, and one third expected most to live less than 2 years. At the time the survey was conducted, there were already published studies indicating that many children who were infected with HIV had a better prognosis.35 Many of the first children with perinatally acquired disease are now of school age, and some are in their mid-teens. 6'37 Knowledge about HIV and the ability to care for infected children both medically and socially has improved during the first decade of the epidemic, changing perinatally acquired HIV from a disease seen universally as rapidly fatal to a chronic disease of uncertain incubation period and life expectancy.38 Our data raise concerns that neonatologists may not be making appropriate recommendations about the care of infected infants. Many people believe that newborns should be tested for HIV antibodies since identification offers the potential benefits of treatment for infants who are infected. Our data, reflecting attitudes rather than behavior, raise the possibility that infants labeled as HIV positive, whether infected or not, may suffer discrimination, with fewer and less aggressive treatments recommended for non-HIV-related illnesses. Knowing the HIV status of infants may become even more beneficial as more interventions become available to decrease infection, prolong the incubation period, or prevent complications. Nevertheless, when creating public policy concerning HIV testing, in addition to the obvious benefit, it is important to consider the potential negative impact created by labeling infants "at risk" or "infected." The HIV epidemic raises a plethora of ethical issues. As more people are known to be HIV positive, more physicians will have HIV-positive patients. Neonatologists will face unique issues because of the need to make recommendations for antibody-positive, yet possibly uninfected, patients. However, physicians in all medical specialties will confront questions concerning the appropriateness of various treatments for HIV-positive patients. O

Acknowledgments

This research was supported in part by Professional Services grants 14-126 and 3-FY92-0285

November 1995, Vol. 85, No. 11

HIV and Newborns

and Innovations and Research in Health Education and Service Delivery Program grant 3-FY93-0657 from the March of Dimes Birth Defects Foundation National Office; by grant 6-62196 from the PSC-CUNY Research Award Program of the City University of New York; and by project 5-MCH-P02027 from the Maternal and Child Health Bureau Program, Health Resources and Services Administration, Department of Health and Human Services. This research was presented in part at the meeting of the Society for Pediatric Research, May 1992, Baltimore, Md; the meeting of the American Academy of Pediatrics, October 1992, San Francisco, Calif; and the 121st Annual Meeting of the American Public Health Association, October 1993, San Francisco. We wish to thank all of the physicians who responded to the survey, the Executive Committee of the Perinatal Section of the American Academy of Pediatrics, and Ronald Bayer, Kate Brown, Stephen Burrows, John Colombotos, Jim Couto, Zola Golub, Sheldon Landesman, Bruce Levin, Kathleen Powderly, Ina Wallace, Katherine Wheeler, and Bob Wycoff, who assisted in the design and/or analysis of this study.

9.

10.

11.

12. 13. 14.

15.

References 1. Centers for Disease Control and Prevention. Tables 5 and 9, Figure 4. HIVIAIDS

Surveill. 1994;6(2):12,16,24. 2. Morse DL, Lessner L, Medvesky MG, Glebatis DM, Novick LF. Geographic distribution of newborn HIV seroprevalence in relation to four socio-demographic variables. Am J Public Health. 1991;

81(suppl):25-29. 3. Hand IL, Wiznia A, Checola RT, et al. Human immunodeficiencyvirus seropositivity in critically ill neonates in the South Bronx.PediatrInfectDisJ. 1992;11:39-42. 4. Krywanio ML, Fisher DG. Meta-analysis of neonatal HIV seroconversion. Pediatr AIDS HIVInfect FetusAdolesc. 1992;3:165169. 5. Centers for Disease Control and Prevention. Recommendations for the use of zidovudine to reduce perinatal transmission of human immunodeficiency virus. MA1UKR. 1994;43(RR-11):3. 6. Ammann AJ. Human immunodeficiency virus infection/AIDS in children: the next decade. Pediatrics. 1994;93(6, Part I):930935. 7. Tovo PA, de Martino M, Gabiano C, et al. Prognostic factors and survival in children with perinatal HIV-1 infection. Lancet.

1992;339:1249-1253. 8. Persaud D, Sulachni C, Rigaud M, et al. Delayed recognition of human immunode-

November 1995, Vol. 85, No. 11

16.

17.

18.

19.

20.

21.

22.

ficiency virus infection in preadolescent children. Pediatrics. 1992;90:668-691. Centers for Disease Control. Guidelines for prophylaxis against Pneumocystis caninii pneumonia for children infected with human immunodeficiencyvirus.MMWR 1991; 40(RR-2):1-13. Hauger S, Nicholas S, Caspe W. Guidelines for the care of children and adolescents with HIV infection. J Pediatr. 1991; 119:2976-2981. Working Group on Antiretroviral Therapy, National Pediatric HIV Resource Center. Antiretroviral therapy and medical management of the human immunodeficiency virus-infected child. Pediatr AIDS HIV InfectFetusAdolesc. 1994;5:20-32. Heagarty MC. Pediatric human immunodeficiency virus revisited. Am J Dis Child. 1993;147:522-523. Cohen F, Durham JD, eds. Women, Children and HIVIAIDS. New York, NY: Springer Publishing Co; 1993. Michaels D, Levine C. Estimates of the number of motherless youth orphaned by AIDS in the United States. JAMA. 1992; 268:3456-3461. President's Commission for the Study of Ethical Problems in Medicine and Biomedical and Behavioral Research. Seriously ill newborns. In: Deciding to Forego LifeSustaining Treatment: Ethical Medical, and Legal Issues in Treatment Decisions. Washington, DC: US Government Printing Office; 1992:197-229. Hastings Center Research Group on the Care of Imperiled Newborns. Imperiled newborns. Hastings Center Rep. 1987;17(6): 5-32. Caplan AL, Blank RH, Merrick JC, eds. Compelled Compassion: Govemment Intervention in the Treatment of Cntically Ill Newboms. Totowa, NJ: Humana Press; 1992. Duff RS, Campbell AGM. Moral and ethical dilemmas in a special care nursery. NEnglJMed. 1973;289:890-894. Levin BW. Consensus and controversy in the treatment of catastrophically ill newborns. In: Murray TH, Caplan AL, eds. Which Babies ShallLive?HumanrsticDimensions of the Care of Imperded Newborns. Clifton, NJ: Humana Press; 1985:169-205. Guillemin JH, Holmstrom LL. Mired Blessings: Intensive Care for Newborns. New York, NY: Oxford University Press Inc; 1986. Kopelman LM, Irons TG, Kopelman AE. Neonatologists judge the "Baby Doe" regulations. N Engl J Med. 1988;318:677683. Levin BW. International perspectives on treatment choice in neonatal intensive care. Soc Sci Med. 1990;30:901-912.

23. Anspach R. Deciding Who Lives: Fatefiul Choices in the Intensive-Care Nursery. Berkeley, Calif: University of California Press; 1993. 24. Stevens A, Little J, Kerr S, Kilbane P, Doherty C. HIV testing in patients with end stage renal disease. BMJ. 1990;300:447449. 25. Condit D, Frater RWM. Human immunodeficiency virus and the cardiac surgeon: a survey of attitudes. Ann Thorac Surg. 1989;47:182-186. 26. Kass NE, Sugarman J, Goodman SN, Faden R. The influence of HIV serostatus on physicians' clinical decisions. AIDS Public Policy . 1994;9:93-99. 27. Sugarman J, Kass NE, Faden RR, Goodman SN. Catalysts for conversations about advance directives: the influence of physician and patient characteristics. J Law Med Ethics. 1994;22:29-35. 28. Levin BW, Driscoll JM, Fleischman AR. Treatment choice for infants in the neonatal intensive care unit at risk for AIDS. JAMA. 1991;265:2976-2981. 29. Crossley MA. Of diagnoses and discrimination: discriminatory nontreatment of infants with HIV infection. Columbia Law Rev. 1993;93:1581-1667. 30. Centers for Disease Control and Prevention. Tables 1 and 2. HIV/AIDS Surveill. 1994;6(2):7-9. 31. Italian Register for HIV Infection in Children. Features of children perinatally infected with HIV-1 surviving longer than 5 years. Lancet. 1994;343:191-195. 32. Frederick T, Mascola L, Eller A, O'Neil L, Byers B. Progression of human immunodeficiency virus disease among infants and children infected perinatally with human immunodeficiency virus or through neonatal blood transfusion. Pediatr Infect Dis J. 1994;13:1091-1097. 33. Pub L No. 98-457,98 Stat 1752-1755. 34. Pub L No. 101-336, 104 Stat. 327. 35. Blanche S, Tardieu M, Duliege A-M, et al. Longitudinal study of 94 symptomatic infants with perinatally acquired human immunodeficiency virus infection. Am JDis Child. 1990;144:1210-1215. 36. Meyers A. Natural history of congenital HIV infection. J Sch Health. 1994;64:9-10. 37. Grubman S, Gross E, Lerner-Weiss N, et al. Older children and adolescents living with perinatally acquired human immunodeficiency virus infection. Pediatrics. 1995; 95:657-663. 38. Meyers A, Weitzman M. Pediatric HIV disease: the newest chronic illness of childhood. Pediatr Clin North Am. 1991;38: 169-194.

American Journal of Public Health 1513