Regular, periodic follow-up of a large proportion of survivors can provide one type of audit of an individual NICU’s performance. Because intensive care nurseries differ in such critical management areas as neonatal resuscitation, modes of assisted ventilation, treatment of ventriculomegaly, and use of parenteral nutrition, and also in such neonatal outcomes as mortality and prevalence of medical complications (e.g., bronchopulmonary dysplasia [BPD], intracranial hemorrhage), units may wish to compare their neurodevelopmental morbidity with the contemporary experience of similar nurseries (11). They also may wish to monitor their major disabling morbidities from year to year to detect any significant differences that might accompany further reductions in mortality or the introduction of new intensive care procedures or treatments. It must be recognized that follow-up at 1 or 2 years of age, although providing much useful information about the prevalence of major neurosensory impairment among survivors, is of insufficient duration to identify changes over time in more subtle aspects of brain function, such as learning and behavior.

Neurodevelopmental follow-up can provide important ongoing subspecialty care to at-risk children and families. Follow-up clinic personnel with a multidisciplinary approach will likely have the most experience and expertise in a given community concerning the unique developmental patterns of LBW premature infants. In general, the follow-up program should complement and serve as secondary or tertiary developmental consultants to the primary health care providers. The program will encourage and facilitate the establishment of a community-based medical home (e.g., private practitioner, public health clinic) for every medically complex survivor. Experience with biologically and environmentally vulnerable indigent populations, however, suggests that actual provision of primary health care, in addition to evaluation and case management services, may be necessary in some situations to prevent attrition and maintain contact with those children and families at greatest long-term risk (12). Although the appropriate approach to this issue of role definition is likely to vary with different populations and access to medical care in different settings, it obviously is of fundamental importance to the organization of follow-up clinics and also to the maintenance of mutual trust relationships with the primary care community.

The specific objectives of follow-up neurodevelopmental assessment activities may be grouped conveniently as follows: to provide cautious reassurance to anxious parents; to ensure early identification and intervention for persistent developmental abnormalities; and to recognize the natural history of transient developmental abnormalities and thereby avoid unnecessary, costly interventions. Maintaining an appropriate balance of diagnostic and reassurance functions is one of the greatest challenges for the contemporary high-risk follow-up program.

The follow-up clinic provides a marvelous setting for interdisciplinary developmental training. It is a clinical laboratory for the observation of the gradual recovery and normalization over time of most at-risk infants and, in other cases, the gradual evolution of a wide variety of permanent neurodevelopmental dysfunctions. Thus, the at-risk population offers a longitudinal training experience that spans the normal-abnormal development continuum. In many pediatric training programs, the follow-up clinic is the sole opportunity for pediatric residents to observe the outcomes of their own intensive care efforts. It would seem virtually impossible for physicians to be informed adequately about the ethicical debates and dilemmas surrounding neonatal intensive care without a first-hand follow-up experience. Likewise, other child development professionals (e.g., psychologists, physical therapists, communication disorders specialists) can use the follow-up clinic profitably as a diverse training base, particularly to broaden the range of normative development for their students. Obviously, these training objectives will apply primarily to university-affiliated tertiary care centers, with their numerous and varied trainee availability.

The university-affiliated follow-up program should be engaged actively in clinical research that contributes to the understanding of the neurodevelopmental and neurobehavioral outcomes of children who experienced neonatal intensive care. These studies may take the form of either descriptive observational reports or clinical trials of specific perinatal-neonatal interventions. For example, the University of Washington’s High-Risk Infant Follow-Up Program has published studies describing the outcome of infants weighing less than 800 g at birth (13,14,15), as well as studies evaluating the utility of procedures such as electronic fetal monitoring of premature labor and delivery (16) and treatments such as high-frequency mechanical ventilation (17). Although tremendous variability exists in the target populations, methodologies, and general scientific quality of the accumulated high-risk follow-up research, a growing consensus of valid outcome observations gradually has emerged over the last 30 years, and informative summary conclusions can be synthesized. Even though the ideal, population-based, nonrisk-controlled, longitudinal to school age study rarely is accomplished for a variety of practical reasons (e.g., cost, subject mobility, investigator discontinuity), individual follow-up investigations, carefully performed with a limited scope, continue to modify and refine overall knowledge and, in some cases, challenge assumptions.

This is not to say that broad, well-funded, collaborative follow-up efforts should not be pursued vigorously on both regional and national, and even international, levels. A recognized need for uniform population descriptions, standardized assessment protocols, common disability definitions, and adequate numbers of pooled subjects still exists. Threats to the interpretability and generalizability of small, local studies include population demographic bias, neonatal treatment differences, attrition of highest-risk (i.e., doubly vulnerable) subjects, and cross-sectional data analysis combining multiple age end points. A great deal has been learned about the short- and long-term prognoses of NICU survivors from hundreds of independent follow-up studies, but much more has yet to be clarified by enhanced research approaches (18).

The size and complexity of the neurodevelopmental follow-up team depend on the scope of the program and size of the patient population. For example, a level II to III community hospital with primarily developmental service objectives will likely employ a smaller team, follow for a shorter period of time, and administer fewer standardized measures than a university-affiliated tertiary care center with training and research responsibilities. In either case, certain key tasks must be accomplished. Probably the most critical role in terms of maximizing follow-up compliance and minimizing attrition is that of the follow-up coordinator, usually a program nurse. This person is the liaison between the NICU and follow-up clinic. The nurse coordinator can identify and meet eligible infants and families before they leave the nursery, participate in the discharge conference and transition plans, and, in some cases, make preliminary contact with the family by means of a home visit before the initial follow-up evaluation. This liaison function is particularly important in those programs that conduct high-risk follow-up at a separate site away from the intensive care nursery and in which none of the follow-up personnel is actively involved in the NICU.

Overall program direction is typically provided by aphysician or psychologist. This person ultimately is responsible both for meeting the broad programmatic objectives and also for day-to-day operations. The director of a university-affiliated follow-up program frequently must balance competing service, training, and research obligations while eclectically maintaining sufficient funding sources to ensure long-term program viability. The director certainly should be knowledgeable in terms of current follow-up literature and contemporary models of program structure and function.

Other follow-up roles of the interdisciplinary team include the following:

Cerebral palsy, of varying types and severities, remains the most prevalent major developmental disability encountered in premature infants; the prevalence in VLBW infants varies between 6% and 10%, and approximately 40% of all children with cerebral palsy were born prematurely (i.e., less than 37 weeks of gestation) (34). Although both spastic (i.e., pyramidal) and athetoid (i.e., extrapyramidal) types of cerebral palsy may be encountered in NICU graduates, the spastic cerebral palsy syndromes (i.e., diplegia, hemiplegia, and quadriplegia) are the neuromuscular disorders most commonly seen in LBW infants. One specific type, spastic diplegia, in which the legs are much more affected than arms, is so strongly associated with prematurity (i.e., at least two-thirds of all children with this disorder are born before 37 weeks of gestation) that for over a century it has been referred to as “the disease of immaturity” (35). Figure 60-2 illustrates the relationship between spastic
diplegia and gestational age. Most cases occur in a window of vulnerability in infants born between 28 and 34 weeks of gestation.

Despite the long consistency of the spastic diplegia-prematurity association, the exact etiologic factors involved often have been elusive and difficult to identify precisely prospectively (36). Neither the severity of perinatal-neonatal illness nor the presence or severity of intracranial hemorrhage reliably predicts spastic diplegia. Data derived primarily from studies correlating ultrasonographic, neuropathologic, and clinical information led to the conclusion that spastic diplegia is the clinical expression of periventricular leukomalacia and its variants (37). Periventricular leukomalacia appears to be caused, in large part, by hypoxic-ischemic injury to the periventricular white matter. The demonstration on serial cranial ultrasounds of initial extensive periventricular echodensities followed in days to weeks by large, bilateral cyst formation (i.e., periventricular white matter infarction) is highly predictive (80% to 85%) of permanent cerebral palsy, especially the spastic diplegia type (38,39). Many cases of symmetric cystic periventricular leukomalacia occur in infants with relatively benign clinical courses and are detected only by routine ultrasound screening. Premature infants born to mothers with prolonged rupture of membranes and/or chorioamnionitis seem to be at an increased risk in some studies (40) but not in others (41). Several investigators have implicated prenatal factors (e.g., intrauterine growth retardation) in the etiology of some cases of spastic diplegia. Hagberg (42) has postulated that the complex interaction of prenatal abnormalities (i.e., “fetal deprivation of supply”) with perinatal difficulties in the birth process and the adjustment to the extrauterine environment may constitute a common pathogenetic mechanism of spastic diplegia. Accordingly, the etiology of spastic diplegia frequently is multifactorial, and all LBW infants merit close neuromotor monitoring during the first 2 years of life, regardless of the severity of their nursery course.

In contrast, development of the more severe spastic quadriplegia type of cerebral palsy, in which all four extremities are equally affected, often can be predicted better in NICU graduates on the basis of specific perinatal or neonatal events, including asphyxia, marked bilateral intraventricular hemorrhage with ventriculomegaly, prolonged neonatal seizures, and central nervous system infection. Although most premature children with spastic diplegia have average or near average mental abilities, children with spastic quadriplegia are far more likely also to have serious cognitive impairments. Spastic hemiplegia, in which only one side is affected, with the arm usually more than the leg, often is heralded by the ultrasonographic appearance of a unilateral, periventricular hemorrhagic infarction with subsequent cystic transformation that occurs in association with and presumably as a result of substantial asymmetric intraventricular hemorrhage (37).

Cerebral palsy typically presents over time in a developmental manner. Thus, very early neurologic signs and symptoms may prove to be transient in nature and not indicative of eventual cerebral palsy. Conversely, infants may initially appear asymptomatic with a relatively normal neurologic examination at the time of nursery discharge and even for several months thereafter, particularly in the cases of spastic diplegia and spastic hemiplegia, only to manifest clearly evident cerebral palsy by 1 year of age. Premature infants with evolving cerebral palsy reveal increasing neuromotor abnormalities of muscle tone, movement, posture, and reflex activity, particularly between 6 and 18 months of corrected age, in combination with increasingly delayed motor milestones.

Mental retardation, as defined by a standardized intelligence or developmental quotient consistently more than two standard deviations below the test mean for corrected age, often occurs in conjunction with one or more of the other major handicaps, especially cerebral palsy. In fact, severe mental retardation and severe cerebral palsy share associated perinatal-neonatal risk factors. Evidence suggests some increase in the prevalence of severely multihandicapped children after increased VLBW survival (28). Mental retardation occurs in 4% to 5% of VLBW infants followed longitudinally to school age. Isolated mental retardation, without cerebral palsy, is a reported consequence of severe BPD, particularly in cases of greatly prolonged duration of mechanic ventilation and oxygen administration (43,44).

NICU graduates are at increased risk for both sensorineural and conductive hearing loss. Although the risk of sensorineural loss sufficient to require hearing aids, special education, and nonvocal communication strategies (60 to 100 dB) usually is estimated to be 2% to 3% for VLBW infants, some investigators have reported prevalence estimates between 5% and 9% coincident with the increased survival of more vulnerable infants (45). Exposure to ototoxic drugs, infections, hypoxia/ischemia, and hyperbilirubinemia are among the interacting and cumulative factors contributing to the risk of sensorineural loss.
The duration and extent of hyperbilirubinemia in VLBW infants has been examined carefully. de Vries and colleagues (46) found bilirubin levels in excess of 14 mg/dL to be associated with a high risk of deafness in VLBW infants but not in healthy premature infants with a birth weight greater than 1,500 grams. Others also have emphasized the potential ototoxicity of hyperbilirubinemia in VLBW infants incombination with hypoxia, acidosis, and prolonged administration of multiple ototoxic medications such as the aminoglycoside antibiotics and furosemide. These investigators conclude that the additive effects of protracted illness plus its associated treatments, independent of specific diagnostic categories, constitute important risk factors for permanent hearing loss in this population (47).

There is ample evidence that infants of all birth weights who sustain severe persistent pulmonary hypertension of the newborn comprise a particularly high-risk subgroup for sensorineural hearing loss, with prevalence estimates ranging from 20% to 40% (48). In some cases, the loss is progressive during the first 3 years of life. The exact mechanism of insult remains unclear in this population of infants who typically experience prolonged hypoxia, severe acute and chronic lung disease, and multiple aggressive interventions. Another concern has been the potential deleterious effect of prolonged incubator noise on hearing function. Abramovich and associates (49) found no evidence for this hypothesis in VLBW infants. Many of the risk factors associated with hearing impairment also are associated with cerebral palsy, and these two disabilities often occur together in the same child.

Mild and moderate (25 to 59 dB) sensorineural hearing losses, sufficient to contribute to delayed language development but compatible with oral communication, also occur with increased frequency (6% to 8%) in LBW infants. Previously unrecognized unilateral sensorineural hearing losses, with adverse language and learning consequences, may become apparent in the older child (50). A high prevalence (20% to 30%) of chronic otitis media with middle ear effusion and fluctuating, conductive hearing loss greater than 25 dB is reported in LBW premature infants (51). Suggested mechanisms for this relationship focus on probable eustachian tube dysfunction initiated by a combination of dolichocephalic head shape, muscular hypotonia, and prolonged nasotracheal intubation.

There have been important advances in the hearing assessment of LBW infants. Two techniques in particular, electrophysiologic auditory brainstem response (ABR) audiometry and behavioral visual reinforcement audiometry, have made early, reliable detection of hearing loss in the NICU graduate clinically feasible. Centers that routinely screen high-risk, LBW infants with ABR before nursery discharge report a false-positive rate of 8% to 10% compared to follow-up testing at 4 months of age (52). Conversely, the unanticipated appearance of severe sensorineural hearing loss in high-risk survivors of neonatal intensive care after having passed an initial ABR screening test in the newborn period has been reported (53). It also must be recognized that ABR tests only the high sound frequencies (i.e., 2,000 Hz and above) and will not detect hearing losses confined to the lower frequencies. Thus, clinicians must remember that determinations of the adequacy of hearing made only with ABR test data before nursery discharge are subject to error. Visual reinforcement audiometry is an operant conditioning technique that reliably can provide auditory thresholds for infants who are functioning at a developmental age of approximately 6 months or older. It has great utility in the high-risk follow-up clinic. A third and newer audiologic procedure, evoked otoacoustic emissions, is of use in newborn screening in conjunction with ABR (54).

The major cause of visual loss in LBW infants is retrolental fibroplasia, now included under the rubric of ROP. With controlled oxygen administration, ROP was relatively rare until the last 15 years or so, when significant numbers of extremely premature infants began to survive. The name ROP recognizes that immaturity at birth is the single largest risk factor for this disease. For all practical purposes, this is a disorder of the VLBW infant. Virtually no retinal detachment and little retinal scarring is described in larger premature infants. For the entire VLBW population, current prevalence estimates range from 20% to 25% with early-stage, regressed ROP; 5% to 10% with more advanced- stage, scarred ROP; and 2% to 4% with major visual impairments, including legal blindness and requiring special educational assistance. The distribution of visually impaired infants, however, is skewed heavily toward those weighing 1,000 grams or less at birth. In these ELBW infants, regressed ROP occurs in 40% to 50% of survivors, scarred ROP in 10% to 25%, and major visual impairments in 5% to 10%. Figure 60-3 shows the overall prevalence of ROP by birth weight. These estimates continue to apply with a greater than 80% prevalence for infants of 26 or fewer weeks’ gestational age.