Archives of Disease in Childhood 1994; 71: F218-F223

CURRENT TOPIC

The MacLean Centerfor Clinical MedicalEthics at theUniversity ofChicagoPritzker School ofMedicineJ D Lantos

Southwestern MedicalCenter at Dallas,Texas, USAJ E Tyson

Grace MaternityHospital, Halifax,Nova Scotia, CanadaA Allen

Department ofPediatrics, Center forMedical Ethics,University ofPittsburgh,Philadelphia, USAJ Frader

Case Western ReserveUniversity, Cleveland,Ohio, USAM Hack

University ofTennessee MedicalGroup, Memphis,Tennessee, USAS Korones

The Children'sHospital, Denver,Colorado, USAG Merenstein

Michigan StateUniversity, Michigan,USAN Paneth

Milton HersheyMedical Center,Hershey, USAR Poland

McMaster University,Hamilton, Ontario,CanadaS Saigal

Department ofNeonatology, StanfordMedical Center,California, USAD Stevenson

Harvard UniversityJoint Program inNeonatology,Children's Hospital,Boston,Massachusetts, USAR TruogL Van Marter

Correspondence to:Dr John D Lantos,La Rabida Hospital andResearch Center,E 65th Street andLake Michigan, Chicago, IL60649, USA.

Withholding and withdrawing life sustainingtreatment in neonatal intensive care: issues

for the 1990s

John D Lantos, Jon E Tyson, Alexander Allen, Joel Frader, Maureen Hack,Sheldon Korones, Gerald Merenstein, Nigel Paneth, Ronald L Poland, Saroj Saigal,David Stevenson, Robert D Truog, Linda J Van Marter

The transition from fetus to newborn isaccompanied by dramatic physiologicalchanges in the infant and by equally profoundchanges in emotional and moral attitudes, aswell as changes in legal status among parents,caregivers, and society. The fetus becomes ababy, adults cross the border to parenthood,and medical responsibility shifts from theobstetrician to the paediatrician. These radicalchanges create challenges even when the babyis healthy. But when the baby is critically ill,and decisions need to be taken quickly, themedical complexities amplify the moral, legal,and psychological concerns.During the 1 970s, ethical deliberations

about appropriate personal and societalresponses to critically ill newborns led to wide-spread agreementl 2 (though not completeconsensus3 4) that decisions should primarilyreflect the interests of the infant. Interests ofparents, other family members, or societyat large were secondary. Infants' interestsmight best be served by continuing or fore-going treatment. If the decision was made toforego treatment there was no legal or moraldifference between withholding and withdraw-ing treatment.5 Consensus on these issueshelped secure medical treatment for newbornswith disabling conditions, such as Down'ssyndrome or meningomyelocele.The 1980s brought a more restrictive view

of ethical options which would have prohibiteddiscontinuation of treatment unless infantswere either comatose or the treatmentconsidered 'futile' or 'inhumane'.6 Theseproposals were controversial, partly becauseconcepts such as futility or inhumanity werevague and difficult to put into practice, andalso because they went against the intuitions ofparents and medical professionals.7 In theUnited States attempts by the Reagan admini-stration to enforce these ethical guidelinescreated an atmosphere of distrust and fear.8Careful study and comparison of differentapproaches to end of life care for neonatesbecame difficult.At the same time advances in neonatology

tended to magnify the dilemmas, especiallywith respect to extremely low birthweightinfants, and created a new set of issues for

which guidelines from the 1970s were difficultto apply.9 10 Furthermore, changes in societalattitudes and public policies regardingtreatment decisions for dying adults created anincreasing disparity in the way these decisionswere handled for patients of different ages."While adults were gaining a 'right to die' thatincluded the right to forego fluid and nutrition,the parents of neonates were losing the rightto forego all but the most inefficacioustreatments.

MethodsTo reopen discussion of the issue of with-holding and withdrawing treatment in neonatalintensive care we asked a group of paedia-tricians, neonatologists, and intensive carephysicians to describe cases in which they ortheir colleagues would consider withholding orwithdrawing life sustaining treatment. Allof these physicians were familiar with thediscussions about ethical issues in neonatologyover the past two decades. In May 1993 webrought these physicians together to review thecases, and to determine whether they thought adecision to forego treatment would have beenmorally appropriate.

Thirty five cases were submitted. Mostbelonged to one of three categories. The firstcomprised cases of extreme prematurity. Atthe borderline of viability, decisions had to bemade whether to initiate treatment in thedelivery room or to discontinue treatment aftera very brief period of life support.The second group comprised cases in

which infants had developed severe, chronicproblems for which life prolonging treatmentwas possible but for which the outcome wasuncertain and the burdens of treatmenthigh. In such cases the issue was generallynot disagreement between families and profes-sionals. Rather, both had difficulty deter-mining what was best for a particular infant,and worked together to reach decisions.The final category consisted of cases in which

there was intractable disagreement betweendoctors and parents about the appropriatenessof further treatment. In most of these casesdoctors felt that death was inevitable in spite of

F218

Withholding and withdrawing life sustaining treatment in neonatal intensive care

continued treatment, but parents insisted ontreatment being continued. In some instancesparents requested that treatment be withdrawnand physicians insisted that treatment continue.Representative cases from each category arepresented as follows.

EXTREME PREMATURITY AND DECISIONS IN THEPERINATAL PERIODCase 1A 30 year old woman presented with rupturedmembranes at an estimated gestational ageof 22 weeks. A Gram stain of the amnioticfluid showed Gram positive cocci in chainformation. Because of the estimated gesta-tional age and the presumed infection withgroup B streptococci, the parents wereinformed that survival was impossible. Labourwas induced without fetal monitoring. Whendelivery occurred, the baby was gasping inroom air and had Apgar scores of 3 at oneminute and 1 at five minutes. A paediatricianwas called. She intubated the baby andbegan mechanical ventilation. The infantquickly achieved oxygen saturations of90-95% on relatively low ventilator settings.The parents had initially requested thattreatment be discontinued, but, on seeinghow well the baby was doing, changed theirminds. The baby did well for 10 days, thendeveloped a large intracranial haemorrhage.After discussion with the parents treatmentwas discontinued.

Case 2A 21 year old woman presented with rupturedmembranes at an estimated gestational age of21 weeks. Delivery occurred four hours later.There was no discussion with the motherregarding treatment options before delivery.The paediatrician attending the deliveryassessed the infant to be between 21 and 23weeks. Birth weight was 540 g. Apgar scoreswere 1 at one and five minutes. No oxygen ormechanical ventilation were provided. Theinfant died shortly after birth.

Case 3A 23 week old, 498 g girl was born to a 21 yearold mother with premature rupture ofmembranes and chorioamnionitis. Althoughgiven the option of withholding life support,the parents requested that everything be done.Over the first weeks of life, the baby had severemetabolic derangements, staphylococcal andcandida sepsis, bilateral grade III intraventric-ular haemorrhages, numerous pneumo-thoraces and a patent ductus arteriosus. Inspite of physicians' recommendations thattreatment be stopped, parents continued torequest aggressive treatment, although theyagreed to a Do Not Resuscitate order. By 3months of age, the baby was receiving 30%/oxygen, a combination of enteral andparenteral feeds, and her parents were pleasedwith her progress and making plans to take herhome.

Comments - Many academic medical centresand organisations of paediatricians are currentlydeveloping guidelines for the treatment ofextremely premature and low birthweightneonates. For example, The Canadian PediatricSociety comes out against treatment for babieswhose gestational age is under 23 completedweeks, recommends that parents be given theoption oftreatment for babies whose gestationalage is 23-24 completed weeks, and stronglyrecommends continued treatment for babieswhose gestational age is 25 or more completedweeks.1 la

Such guidelines reflect the general principlethat the treatment of newborns is obligatory ifit offers a reasonable chance of survival (withthe possible exception of situations in whichsurvival is possible only with severe medical orneurodevelopmental sequelae). When chancesfor survival are low, treatment may be withheldor withdrawn. In applying these principles,theoretical debate focuses on two issues. First,how accurately can we predict survival for aparticular baby? Second, assuming that wecan predict prognosis, what constitutes areasonable chance of survival?

Assessment ofprognosis for an infant ofverylow birth weight relies on estimated gestationalage, birth weight, and the clinical status of theinfant. None of these factors can be preciselyestablished before birth. Despite rigorousdating of the pregnancy, gestational ageestimates are often off by one to three weeks.Furthermore, birth weight varies widely forbabies of the same gestational age. Forexample, at 24 weeks' gestation, birth weightsrange from 400-1000 g (Harvard MedicalSchool Joint Program in Neonatology; unpub-lished data). As a result, it is difficult to makedefinite plans for treatment or non-treatmentuntil the baby is born, and the neonatologisthas a chance to assess size, maturity, andclinical status.

Furthermore, the chances of survival for aninfant at a given birth weight or gestational agediffer depending on race, sex, the philosophyof obstetric and neonatal care, and otherunknown factors.'2 Thus non-treatmentpolicies which seek to standardise decisionmaking for infants with a similar prognosismust allow for wide local variation.

Focus on predicted survival may over-simplify the real issues, for several reasons.Precise prognoses are necessarily artificial inthe context of any particular case. In the realworld the chances for survival are alwaysdifficult to determine and may change rapidlyover time. Thus estimates are valuable onlyas a way of distinguishing types of clinicalsituations and changes in moral obligations atthe extremes of the spectrum of prognosis. Invirtually all cases a decision whether to initiateor forego treatment can only be made afterthe paediatrician or neonatologist assessesthe newborn and determines whether theneonate's initial condition is consistent withthe prenatal estimates of gestational age.

Because it is so difficult to give preciseprognostic assessments prenatally, communi-cation between obstetricians, neonatologists,

F219

Lantos, Tyson, Allen, Frader, Hack, Korones, et al

or paediatricians and parents is essential toavoid misleadingly absolute predictions ofviability or non-viability. Prenatal decisions toinitiate or withhold treatment should bereconsidered if the newborn is either more orless viable than anticipated. Shared anddynamic decision making, which mirrors theprocess for older children or competent adults,ensures that decisions incorporate parentalvalues, the infant's interests, and an optimalunderstanding of the facts and the inherentuncertainties in the child's clinical situation.

Prognosis will always depend on a com-bination of factors, and will always besomewhat uncertain for any particular baby.'3Currently, however, survival rates for babiesunder 24 weeks' gestation or less than 600 gbirth weight are predictably poor enough sothat parental choices regarding resuscitationof such babies ought to be encouraged andrespected. Decisions to respect parentalchoices in these situations acknowledge thatin order for parents to choose to forego lifesustaining treatment on their child's behalf,survival does not have to be impossibleor unprecedented - it only has to be veryunlikely. Prognostic inaccuracy and changingtechnological potential both preclude anattempt to define a boundary of probability -that is, 5%, 10% of some other per centchance of survival - for withholding treatmentmore precisely.

In situations where there is no time toinvolve parents in decision making (when awoman presents in premature labour anddelivery is imminent), the presumption shouldgenerally be in favour of treatment unlesssurvival is unprecedented.

CHRONIC DISEASE AND BURDENSOMETREATMENTCase 4A baby weighing 1100 g developed severebronchopulmonary dysplasia, requiring pro-longed assisted ventilation. Four attempts atextubation failed. He was finally weaned fromthe ventilator at 11 months of age, but quicklydeveloped respiratory distress and was reintu-bated. At the age of 12 months, his cognitivefunction was at the nine month level. By theage of 17 months, he was weaned from theventilator for a few weeks, but required inten-sive respiratory treatment and was failing tothrive. His respiratory physician recommendedlong term home ventilation. His parents, whowere very involved in his care, requestedthat he not be placed back on a ventilator andthat he be allowed to die.

Case SA full term newborn with arthrogryposis andomphalocele continued to require assistedventilation after repair of the omphalocele.After being told that the baby would bequadriplegic and suffer repeated painful bonefractures, her parents opted to discontinueventilator support. Unexpectedly, the infantdid not die. At age 2, she remains quadriplegic

but cognitively intact, and is developingnormal language skills.Comment - In these cases there was some

chance oflong term survival, but survival wouldmost likely have required prolonged depen-dence on medical technology. The childrenwould also be left with severe disabilities. Inboth cases cognitive function was nearly intact.

Individual physicians within our groupdisagreed about whether, under the circum-stances, they would have recommended con-tinuation of treatment. All agreed, however,that in such cases it would be morally permis-sible for parents and physicians together todecide to withdraw treatment. A number offactors supported these decisions.The children's suffering was a primary

concern. Long term ventilator treatment for atoddler may require sedation. Suctioning maybe painful. A life of constant respiratory distressand air hunger may be a life of constant fear andanxiety. In these circumstances to consider onlythe chances for survival was seen as inhumane.

Concerns were also raised about the effect ofcontinued treatment on the parents and otherfamily members. 14 Such concerns includednot only the economic cost of care, but theemotional demands of having a child whosecare entails a series of medical crises and whorequires ongoing constant attention andconcern. When treatment will clearly lead tothe child's survival and recovery, such consid-erations pale in relation to the obligations to dowhat is best for the sick child. However, whenit is unclear whether continued treatment is inthe child's interest, a more complex calculus offamily interests may be necessary.'5 16

Legally, treatment decisions rest with theparents until they have been judged neglectfulby a court of law. Such judgments rely onconclusions that parents are not acting in thechild's best interest. In these cases doctorscould not agree about the value of continuedtreatment or about whether stopping treatmentwas an obvious or egregious violation of thechild's interest. Even if a physician felt that aparental decision was unacceptable the properresponse would not be to override parentalwishes unilaterally. Instead, it would be tonotify parents that the law requires physiciansto seek a determination by child protectionagencies of whether the parents' request todiscontinue treatment was permissible. Suchan approach would correctly relocate theconflict as one between parents and that state'sinterest in child protection, rather than as aconflict between parents and paediatricians.Such difficult assessments can best be made

by people who have a long term relationshipwith the child. Thus parents, in consultationwith a physician who has provided ongoingcare to the child, are in the best position tomake the difficult decision to discontinuesupport. Paediatricians must be parents' alliesin difficult cases. Optimal decision making isfacilitated by the involvement of a primaryphysician who is highly knowledgeable,consistently involved in the care of the child,and who develops ongoing rapport withparents. Too often, such children fall through

F220

Withholding and withdrawing life sustaining treatment in neonatal intensive care

the cracks of the primary care system as theirproblems are so complicated that they requireprolonged tertiary care. In tertiary care settingsattending physicians change periodically, andno one may have primary responsibility forthe ongoing management of chronic medicalproblems and ethical decision making.The discussion raised interesting and seldom

addressed questions about informed consentin paediatrics. In situations involving chronicdisease and burdensome treatment parentsmay request that treatment be withdrawn. Ifsuch parents find a physician who agrees thatwithdrawal is appropriate, babies are allowed todie. Other physicians find withdrawal unaccept-able and may not heed parental requests todiscontinue treatment. Variations in physicians'practice styles and moral values can lead to anarbitrariness in available options for parents.

Physicians should be careful to separatetheir personal views, which may be in favour ofcontinued treatment, from current medical,legal, and moral standards of care for suchchildren. These standards give parents theright to discontinue treatment unless this righthas been overruled by a judge. Parents shouldbe informed of these rights. Physicians in suchsituations must try to help parents understandthe complexities of determining prognosis,and the changing nature of prognosticestimates over time, and may thus recommendcontinuing treatment. They should also informparents when discontinuation of treatment ispermissible.

INTRACTABLE DISAGREEMENTCase 6A newborn weighing 750 g with an estimatedgestational age of 24 weeks developed severerespiratory distress syndrome. He also hadtransient renal failure, seizures on the secondday of life, and prolonged direct hyperbilirubi-naemia. He subsequently developed severebronchopulmonary dysplasia, was ventilatedfor 55 days, and remained oxygen dependentfor 80 days. The parents frequently expressedconcerns about his quality of life andquestioned whether intensive care should becontinued. Physicians felt that prognosis forneurodevelopmental outcome was uncertainand recommended continued treatment. Theparents were hesitant, but agreed to ligation ofa patent ductus arteriosus and insertion of acentral venous line. Treatment was continuedand he was discharged home at 133 days ofage. At 41/2 years of age, the child is micro-cephalic, mentally retarded, cortically blindand has quadriplegic cerebral palsy. He isentirely dependent on his caretaker. Hisparents feel bitter that their repeated requestsfor treatment to be discontinued were ignored.

Case 7A girl weighing 589 g was born at 23 weeks'gestation. Apgar scores were 1 at one minuteand 5 at five minutes. She developed severerespiratory distress, electrolyte imbalances,hyperglycaemia, necrotising enterocolitis and

bilateral intraventricular haemorrhages withventricular dilation. Neonatologists recom-mended discontinuation of treatment.Although the mother seldom visited, when shedid, she insisted that everything be done tokeep her baby alive. The father showed littleinterest in either visiting or participating intreatment decisions. Maximal ventilatorysupport was continued until the infant died onday 1 5.

Case 8A full term baby, born after a pregnancycomplicated by oligohydramnios, developedcyanosis, grunting, and retractions shortlyafter birth. He was intubated at 10 minutes. Inspite of maximal ventilatory support, pCO2sremained in the 90-100 range with the pHvalues between 6-99 and 7-11. A presumptivediagnosis of pulmonary hypoplasia was madeand doctors recommended discontinuation oftreatment. Parents requested transfer to anotherhospital for a second opinion. Although thephysicians felt this was unnecessary, theyarranged the transfer. At the other hospital,a renal ultrasound scan revealed dysplastickidneys and the infant was removed from theventilator with parental approval.Comment - Participants agreed that cases of

intractable disagreement were relatively rarebut emotionally trying. There were two typesof cases. First, those in which parents insistedon treatment that physicians felt would notbe beneficial. Discussion of these cases arepart of the ongoing debate over moral obliga-tions in cases where treatment seems tobe futile.'7 Second, those in which parentsstrongly opposed recommendations for con-tinued treatment, but eventually acquiesced.These cases are the type that have received themost coverage in the lay press,'8 and haverecently been addressed.'9The most difficult 'futility' cases in neo-

natology are those in which the parents seem tobe emotionally unattached to the child but,nevertheless, demand continued treatment. Insuch cases, providers often feel called on toprotect infants from the pain and suffering ofuseless overtreatment demanded by inconsid-erate or even malevolent parents. In theory,physicians and other providers have theright to withdraw from a case or refuse toprovide treatment which they view as morallyobjectionable. However, the realities of mostclinical situations preclude the physicians fromexercising this option.Four options are currently available to the

physician in these circumstances. First, thephysician can continue to treat, allowing his orher own moral values to be outweighed by theparents' demands to continue life support.Second, the physician could unilaterally refuseto provide treatment and insist that parentsfind another physician. Legal concerns aboutabandonment would dictate that the physicianhelp the parents locate another physician.Third, the physician could seek a judicialdetermination that continued painful and non-beneficial treatment constitutes a forrn of child

F22 1

Lantos, Tyson, Allen, Frader, Hack, Korones, et al

abuse and should be withdrawn. However,courts generally rule in favour of continuedtreatment in such cases. Finally, institutionalpolicies defining futile care could preclude orreduce the need for ad hoc decision makingregarding individual cases. Policies could eitherbe substantive - mechanical ventilation shouldnot be provided to patients in a persistentvegetative state; or procedural - treatment maybe withdrawn over parental objections if anethics committee recommends it.

Cases of intractable disagreement, wherephysicians want to continue treatment andparents want to stop, raise different issues.These cases illustrate the discrepanciesbetween the rights that adults have torefuse medical treatment and the rights thatparents have in regard to their children.Decisions for neonates have life long impacton patients and their families, while neonato-logists are generally involved with familiesfor only a brief period of time. This longerterm perspective may temper enthusiasm fortreatments of uncertain efficacy, or thosewhich are likely to leave children with severelong term morbidity.

COSTSThere was heated disagreement within ourgroup about the appropriateness of usingestimates of the cost of treatment to guidepolicies regarding decisions to forego lifesustaining treatment. The unique vulner-ability of newborns was reflected in opinionsthat cost considerations should not be usedonly or primarily for allocation decisions inneonatology, but should compare the costeffectiveness of neonatal care with treatmentsused for patients at other stages of life. Inmany public policy debates neonatal care issingled out for rationing in a way that doesnot reflect the relative cost effectiveness ofneonatal care compared with intensive carefor the elderly.20

There was also concern that costs are anartificial and somewhat arbitrary figure,reflecting varying degrees of technologicaladvances, inept administrative procedures,inefficient use of resources, and greed amonghealth care professionals and lawyers.Elimination of these extraneous factors woulddecrease costs considerably and may decreasethe pressure to end lives because the bills areunmanageable.The urgency of economic considerations

varied widely, with Canadian physicianshaving a different view than US physicians,and physicians in public hospitals having adifferent view than those in private hospitals.Even those who recognised a need for fiscalaccountability had difficulty with proposalsthat would withhold treatment from particularpatients based on that patient's projectedexpenses. These discussions showed how anyattempt to ration must consider deeply heldprofessional moral imperatives which forbidthe withholding of accessible medical treat-ment, based on abstract notions of justice ortheoretical econometric models.

ConclusionTreatment withdrawal dilemmas in neonatalintensive care arise primarily at two points inthe life of a neonate. First, decisions mustbe made whether to initiate treatment forextremely premature babies. These decisionsreflect parental motivations, obstetric assess-ments of maternal risks, and evolving prognos-tic predictions of the likelihood of neonatalmorbidity or mortality. Guidelines for decidingwhen babies should be considered non-viable,when intensive care is optional, or whenintensive care is obligatory, should reflectrecent hospital specific data on morbidity andmortality for babies at different birth weightsand gestational ages.

Shared decision making among obstetri-cians, neonatologists, and parents is essential.Ethics committees have been helpful indescribing general guidelines for decisionmaking, but are rarely consulted for day today decision making. They may be most usefulin developing institutional policies for dealingwith categories of patients, such as theextremely premature babies, or in helping toarbitrate intractable disagreements.Tough decisions also arise for neonates who

have survived the first days but who havesevere congenital or iatrogenic problems. Inthese cases the probability of survival, theburdens of treatment, the likelihood of majormorbidity, and the impact of decisions onfamilies must all be considered.

Disagreement about decisions to withholdor withdraw treatment in neonates no longercentres on disagreement about moral prin-ciples or the types of broad guidelines whichshould guide such decisions. Instead, a wide-spread consensus has emerged that the child'sinterests are the central moral consideration.Other considerations are relevant only when itis unclear whether continued treatment is inthe child's interest.

However, the concept of best interest is onlyprecise enough to get us so far. In particularcases 'best interest' may be evaluated differ-ently by lawyers, judges, physicians, parents,philosophers or special interest advocacygroups.2' These differences will not beresolved by enunciating more finely tuned orrevised principles, but by showing howmoral principles must be supple enough toaccommodate ever changing clinical realities.

Practical guidance for decision making willemerge as general principles are tested throughanalysis of their applicability to particularcases. Guidelines, while crucial, will never besufficient, because they will always define notonly areas of right and wrong action but alsoareas of ambiguity. In such ambiguous areas,we will need to interpret guidelines by payingattention to the informed intuitions of parents,professionals, and concerned citizens whomust respond to particular babies in particularsitu3tions. This report attempts a first step inthat direction.The Center for Clinical Medical Ethics is supported by grantsfrom the Henry J Kaiser Family Foundation, the Andrew WMellon Foundation, and the Pew Charitable Trusts. Theconference on which this paper is based was supported by agrant from the Harris Foundation.

F222

Withholding and withdrawing life sustaining treatment in neonatal intensive care

1 Jonsen A. Critical tissues in newborn intensive care: Aconference report and policy proposal. Pediatrics 1975; 55:756-85.

2 President's Commission for the Study of Ethical Problemsin Medicine and Biomedical and Behavioral Research.Deciding to forego life-sustaining treatment. In: Seriouslyil newborns. Washington, DC: US Government PrintingOffice, 1983: 197-229.

3 Duff R, Campbell AGM. Moral and ethical dilemmas inthe special care nursery. N Engl J Med 1973; 289:890-4.

4 Todres ID, Krane D, Howell MC, Shannon DC.Pediatricians' attitudes affecting decision makingin defective newborns. Pediatrics 1977; 60: 197-201.

5 Robertson JA, Fost N. Passive euthanasia of defectivenewborn infants: legal considerations. Jf Pediatr 1976; 88:883-7.

6 Amendments to the Child Abuse Prevention and Treat-ment and Adoption Reform Act. Public Law 1984;98-457.

7 Kopelman LM, Irons TG, Kopelman AE. Neonatologistsjudge the 'Baby Doe' regulations. N Engl J Med 1988;318: 677-83.

8 Lantos JD. Baby Doe five years later: implications for childhealth. NEnglJMed 1987; 317: 444-7.

9 Caniano DA, Kanoti STD. Newborns with massive intesti-nal loss: Difficult choices. N Engl J Med 1988; 318:703-6.

10 Lantos JD, Frader J. Extracorporeal membrane oxygenationand the ethics of clinical research in pediatrics. N Engl JMed 1990; 323: 409-13.

11 Wanzer SH, Federman DD, Adelstein SJ, et al. The

physician's responsibility towards hopelessly ill patients. Asecond look. N Engl _j Med 1989; 320: 844-9.

hla Fetus and Newborn Committee, Canadian PediatricSociety; Maternal Fetal Medicine Committee, Societyof Obstetricians and Gynecologists of Canada.Management of the women with threshold birth of aninfant of extremely low gestational age. Can Med Assoc J1994; 151: 547-53.

12 Hack M, Fanaroff AA. Outcomes of extremely-low-birth-weight infants between 1982 and 1988. N Engl JfMed 1989; 321: 1642-7.

13 Fischer AF, Stevenson DK. The consequences of uncer-tainty. An empirical approach to medical decision makingin neonatal intensive care.3AAMA 1987; 258: 1929-3 1.

14 Hardwig J. What about the family? Hastings Cent Rep 1990;20: 5-11.

15 Strong C. The neonatologist's duty to patients and parents.Hastings Cent Rep 1984; 14: 1-4.

16 Lantos JD, Kohrman AF. Ethical aspects of pediatric homecare. Pediatrics 1992; 89: 920-4.

17 Schneiderman U, Jecker N. Futility in practice. Arch InternMed 1993; 153: 437-41.

18 Stinson A, Stinson P. The long dying ofbaby Andrew. AtlanticMonthly Press: New York, 1984.

19 Harrison H. The principles for family-centered neonatalcare. Pediatrics 1993; 92: 643-50.

20 Hadorn DC. Setting health care priorities in Oregon. Cost-effectiveness meets the rule of rescue. JAMA 1991; 265:2218-25.

21 Doyal L, Wilsher D. Towards guidelines for withholdingand withdrawal of life-prolonging treatment in neonatalmedicine. Arch Dis Child 1994; 70: F66-70.

F223