miologic information can be gleaned (Am J Epidemiol 1976; community physicians of the possibility of the exis-

104:527, Am J Epidemiol 1979; 109:433). Similarly, detailed tence of a cardiac problem was notable as evidenced by
experience with the various anatomic cardiac lesions was more sick newborn infants being admitted to regional
extracted and presented as a guideline for expected average centers for diagnostic evaluation and management.
experience for regions outside of New England.
The report covers a period of time when interven-
COMMENTARY tional procedures and early surgical repair of certain
lesions were just beginning to be undertaken. For

T
he report of the New England Regional Infant
Cardiac Program (NERICP) was a landmark example, this was before arterial switch operations
publication when it appeared as a lengthy sup- for transposition of the great arteries, balloon valvu-
plement to Pediatrics in 1980. For the first time, a region loplasty for critical pulmonary valve stenosis, and
had cooperatively and critically analyzed the care it the Norwood procedure for the hypoplastic left heart
was delivering to seriously ill infants with heart dis- syndrome. Nevertheless, each center was able to re-
ease. As such it became a model for subsequent at- view its results and compare them with other centers
tempts to assess all aspects of care delivery, not only for in the region. This interchange led overall to im-
those with heart disease but for other diseases as well. proved survival statistics and, in some instances,
This is particularly important at present, given the con- discontinuing surgical management at some of the
cerns that have been raised about outcome data, costs, centers.
access, and long-term results. Today, and for the future, each regional cardiovas-
When one reviews this supplement in 1998, the cular center must closely evaluate its total operation
breadth of the study still appears most impressive. in terms of patient volume, short-term and long-term
There was attention focused on such items as case outcome data, costs and case findings, communica-
finding, transportation to a cardiovascular center, tion, and parent patient education as was done in the
communications between community physicians NERICP. Although access to a center should be fa-
and the centers, cost of diagnostic and surgical pro- cilitated, the number of regional centers per popula-
cedures, education of parents, psychosocial ele-
tion base has to be critically assessed by impartial
ments, and outcome. All these items that were ad-
consultants. This approach is a necessity given the
dressed in the 1970s assume even more importance
as we approach the 21st century. current emphasis on access to specialized care, cost
The NERICP also served as the model from which containment, and results. Although extremely im-
other regional programs that focused on infants with portant, currently there is a strong likelihood that
cardiac disease could be studied and led to the Bal- this will assume even more importance in the future
timore-Washington and the Midwestern pediatric as well. Thus, this initial report of the NERICP has
cardiac programs. served admirably as the template for what now must
The report provided a wealth of epidemiologic data take place in all regions of the country. It was ahead
that have been extremely useful when current pro- of its time but fortunately has stood the test of time
grams are analyzed. Mortality rates declined with and, therefore, is deemed a worthy contribution to
fewer infants dying before being evaluated at a re- the exciting history of pediatric cardiovascular dis-
gional center. An increased awareness on the part of ease in the United States.

COMMENTARY

Reye’s Syndrome and Salicylate Use, by Karen M. Starko, MD, et al,

Comments by Ralph E. Kauffman, MD

ABSTRACT OF ORIGINAL ARTICLE (STARKO ET jects were evaluated for characteristics of the patients’
AL). During an outbreak of influenza A, seven patients prodromal illness and the control subjects’ illness and
with Reye’s syndrome and 16 ill classmate control sub- for medication usage. Patients during the prodrome and
control subjects had similar rates of sore throat, coyza,
cough, headache, and gastrointestinal complaints except
From the Office of Medical Research, Children’s Mercy Hospital, Kansas for documented fever which occurred significantly more
City, Missouri.
often in patients than in control subjects (P 5 .05). While
Received for publication Mar 19, 1998; accepted Mar 19, 1998.
Address correspondence to: Ralph E. Kauffman, MD, Office of Medical
medications which did not contain salicylate were taken
Research, Children’s Mercy Hospital, 2401 Gillham Rd, Kansas City, MO as frequently by patients as control subjects, patients
64108. took more salicylate-containing medications than did
PEDIATRICS (ISSN 0031 4005). Copyright © 1998 by the American Acad- control children (P < .01). All seven patients took salic-
emy of Pediatrics. ylate whereas only eight of 16 control subjects did so

SUPPLEMENT 259
(P < .05) Patients took larger doses of salicylate than did Reye’s syndrome between December 21 and 25,
the entire control group (P < .01). When the eight control 1978.3 The purpose of Starko’s study was to assess
subjects who took salicylate were compared with the the possible contribution of medication to risk of
patients, the patients still tended to take larger doses developing Reye’s syndrome. The 7 cases were com-
(P 5 .08). Patients with fever took salicylate more fre- pared with 16 control children who were matched on
quently than control subjects with fever (P < .01). In basis of gender, age, and characteristics of their pro-
addition, salicylate consumption was correlated with se-
dromal illness. Type and quantity of all medications
verity of Reye’s syndrome (P < .05). It is postulated that
salicylate, operating in a dose-dependent manner, possi- taken by both case and control subjects during the
bly potentiated by fever, represents a primary causative prodromal illness were determined by interviewing
agent of Reye’s syndrome. parents within 4 weeks after the children were hos-
pitalized. On analysis, the only difference between
ABSTRACT OF ORIGINAL ARTICLE (ARROWSMITH case and control subjects was in the use of salicylate-
ET AL). The number of cases of Reye syndrome re- containing medications. All of the case subjects used
ported annually to the Centers for Disease Control and salicylates during their prodromal illness, whereas
Prevention declined markedly between 1980 and 1985. In only half of those in the control group did. Further-
this article we present pharmaceutical marketing re- more, the mean salicylate dose in case subjects was
search data that suggest sharp decreases in the use and almost five times that in the control children. In spite
purchase of children’s aspirin between 1980 and 1985. of the limited size of the study, these findings were
These trends appear to correspond to the decrease in
highly statistically significant.
reporting of Reye syndrome cases. Additionally, analysis
of physician mentions of aspirin and acetaminophen for Although previous authors had noted use of sa-
treating flu and chickenpox showed statistically signifi- licylates in association with Reye’s syndrome,5 Star-
cant trends toward increasing recommendations for use ko’s was the first controlled study to document this
of acetaminophen. Trends in wholesale purchases of as- putative association. However, the study suffered
pirin and acetaminophen by drug stores from 1979 from several important weaknesses that precluded
through 1985 demonstrated a significant decline for the definitive conclusions. The number of cases was
81-mg children’s aspirin tablet and an increase in pur- small, there were potential sources of bias such as
chases of children’s acetaminophen products. Many fac- differential recall of medications or treatment selec-
tors may influence physician and parents’ choice of tion based on severity of illness, and control subjects
analgesic/antipyretic medication, including information were not proven to have influenza, although their
about Reye syndrome. Data suggest that a continuing symptom profiles matched those of the case subjects.
decline in the use of aspirin for children may be accom-
Nevertheless, when published in 1981, the apparent
panied by a continuing decline in the reported number of
Reye syndrome cases. association of aspirin with risk of Reye’s syndrome
caught the attention of the medical community,
CDC, US Food and Drug Administration (FDA), and

I
n 1963, Reye and colleagues first described a dis- aspirin manufacturers. Starko’s paper was the fore-
tinct clinical and pathologic syndrome of enceph- runner of a series of studies to be conducted over the
alopathy and characteristic fatty degeneration of ensuing seven years, all of which essentially con-
the liver after a viral-like prodromal illness.1 During firmed and supported her original findings.
the ensuing years, the etiology of this sometimes During the 1979 to 1980 season, the Michigan De-
lethal syndrome remained a mystery, although mul- partment of Health conducted a survey of 25 Reye’s
tiple putative contributory causative factors were syndrome cases and 46 matched control subjects.
proposed, including viral infections, aflatoxin, pesti- This survey collected information on 73 potential
cides, and various medications. For reasons that re- exposures, including type of residence, medications,
main poorly understood, an increase in reporting of type of prodromal illness, dietary history, exposure
Reye’s syndrome cases occurred during the late to toxic substances, and animal exposures. The only
1970s and early 1980s. More than 650 cases were significant association found was with salicylate ex-
reported to the Centers for Disease Control and Pre- posure, ie, cases were more likely than were control
vention (CDC) in 1977 to 1978.2 subjects to have taken salicylates. During the 1980 to
The articles by Starko et al3 and Arrowsmith et al4 1981 winter, the Michigan investigators focused spe-
included among the historic articles reviewed in this cifically on medication use. Again, the only differ-
issue of Pediatrics anchor two interconnected aspects ence found was with salicylate use. All 12 cases
of the Reye’s syndrome and aspirin story. This is a received salicylate during their prodromal illness,
story that unfolded over an entire decade, culminat- compared with 13 of 29 control subjects, a highly
ing in a dramatic decline in incidence of Reye’s syn- significant difference. These companion studies were
drome coincident with a major and permanent published together in June 1982.6
change in physician prescribing and consumer use of During the same period (1978 to 1980), the Ohio
aspirin in children. Department of Health and the CDC conducted a
A cluster of Reye’s syndrome cases occurred in large case– control study in Ohio to assess the asso-
Arizona school children in 1978 during an outbreak ciation of infection and medication exposure with
of influenza A/Brazil illness. During this epidemic, risk of Reye’s syndrome. A total of 97 children with
Dr Karen Starko, then a young Public Health Service Reye’s syndrome were compared with 156 matched
Officer, and her colleagues conducted a retrospective control children over a 2-year period. Of the medi-
case-control study of seven children, 8 to 15 years of cations taken during the prodromal illness, only sa-
age, who were hospitalized with the diagnosis of licylate use occurred more frequently in case (97%)

260 SUPPLEMENT
than in control subjects (71%), again a statistically showed such a strong statistical association between
significant difference. This study was published in aspirin use and Reye’s syndrome that the Institute of
August 1982.7 Medicine advisory committee terminated the study
In response to concerns raised by Starko’s study early. The OR for risk of Reye’s syndrome with expo-
and the yet unpublished Ohio and Michigan studies, sure to aspirin was 40, many times higher than that
the CDC convened an expert panel in October 1981 found in the previous studies! Furthermore, there was
to review the studies and advise on appropriate ac- evidence of a dose–response effect; eg, not only were
tion. The FDA also formed a working group in 1981 cases more likely to have received aspirin, they also
to analyze and evaluate the available data indepen- took larger doses of aspirin than those control subjects
dently. In addition, the Aspirin Foundation of Amer- who were given aspirin.
ica, representing major manufacturers and distribu- Despite the striking results of the Public Health Ser-
tors of aspirin products, retained an independent vice study, concern remained, primarily among the
consulting firm, Biometric Research Institute, to con- aspirin manufacturers, that the apparent aspirin–
duct a detailed review of the studies.8 It is unlikely Reye’s syndrome association could be primarily attrib-
that any group of studies before or since has under- utable to bias in the study design. A study supported
gone the intense scrutiny to which these studies were by five major aspirin manufacturers was performed
subjected. Although the several groups recognized from November 1, 1986, through August 24, 1987.17 The
that sources of bias and methodologic errors could study, conducted by investigators at Yale and McGill
possibly explain at least part of the apparent associ- universities, was designed specifically to eliminate or
ation, a true association between salicylate use and control for potential sources of bias in its design. By the
risk of Reye’s syndrome could not be dismissed. time this study was initiated, the number of Reye’s
Because of this concern, efforts were made during syndrome cases had decreased to the extent that enroll-
the years 1981 to 1982 to advise health care providers ment was very slow. Twenty-four case subjects and 48
and the public of the possible association and to matched control subjects were eventually enrolled.
recommend against use of aspirin in children and When analyzed, the OR for developing Reye’s syn-
adolescents during a viral-like illness.9 –12 Neverthe- drome if exposed to aspirin during a viral prodromal
less, the debate raged on, with those challenging the illness was 35, virtually the same as that found in the
veracity of the studies on one side and those accept- Public Health Service study. The authors concluded
ing the association of salicylate use with risk of that bias could not explain the association, and this
Reye’s syndrome on the other.13,14 study essentially ended the debate.
In response to the controversy and criticism of the The second component of this story is presented in
three published studies, a US Public Health Service the paper by Arrowsmith, and colleagues.4 As early as
Task Force was formed to implement an additional 1982, initial studies suggesting an association between
study, which was designed to address the issues aspirin use and risk of Reye’s syndrome led to recom-
raised by the various critics. A committee of the mendations by news media, private organizations,
Institute of Medicine was appointed to oversee the health departments, and state and federal agencies to
protocol design and conduct of the study, and to avoid use of aspirin in children.9 –12 Arrowsmith and
perform scheduled interim review and analysis of colleagues at the FDA analyzed changes in aspirin use
the results. A pilot study was conducted from Feb- and number of cases of Reye’s syndrome reported an-
ruary 1984 through May 1984, primarily to deter- nually from 1980 to 1985, the period during which
mine feasibility and refine methodology for a larger information from the studies was being disseminated.
study. Enrollment was spread across 16 pediatric Arrowsmith found that physician recommendations to
referral centers in 11 states. Thirty case and 145 con- use aspirin for treating children decreased significantly
trol subjects were identified. Although the pilot from 1980 through 1985. Likewise, sales of children’s
study was not designed primarily to test the hypoth- aspirin products fell dramatically during the same pe-
esis of an association between aspirin and Reye’s riod. In contrast, sales of adult aspirin products did not
syndrome, when the Institute of Medicine Commit- change. The data show that the decline in use of chil-
tee reviewed the data, the association was so strong dren’s aspirin was apparent as early as 1981, preceding
that they decided to make the results public. The published warnings from the American Academy of
adjusted odds ratio (OR) for an increased risk for Pediatrics and the CDC and several years before the
Reye’s syndrome if exposed to aspirin during the warning in the labeling later required by the FDA. This
prodromal illness was 19, much higher than that in suggests that publicity arising from Starko’s paper had
any of the previous studies. This study was pub- a significant impact on physician and consumer behav-
lished in 1985.15 ior even before results of the subsequent studies were
The main Public Health Service Study was con- generally available. Similar changes in aspirin use to
ducted between January 1985 and May 1986 and pub- treat children were documented in local and regional
lished in April 1987.16 Enrollment was solicited from 70 studies.18 –19
pediatric referral centers throughout the United States. Coincident with the decreased use of aspirin, cases
Although the largest influenza B epidemic since 1969 of Reye’s syndrome reported annually to the Na-
occurred in 1985, only 33 cases of Reye’s syndrome tional Reye Syndrome Surveillance System declined
were identified, 27 of which were acceptable for anal- from a peak of 658 cases in 1980 to only 93 in 1985
ysis. The 27 case subjects were compared with 140 and have declined further during the past decade.
matched control subjects. The original target enroll- The decreased number of cases occurred in view of
ment was 100 to 200 cases. However, interim analysis intensified national surveillance for Reye’s syndrome

SUPPLEMENT 261
cases, major outbreaks of influenza in 1983, 1984, and 2. Centers for Disease Control and Prevention. Outbreaks of Reye syn-
drome—Utah, Arizona, Colorado. MMWR. 1979;28:39
1985, and no change in reported varicella cases dur- 3. Starko KM, Ray CG, Dominguez LB, et al. Reye’s syndrome and salic-
ing the 6-year period.20,21 ylate use. Pediatrics. 1980;66:859 – 864
Starko’s paper, and those that followed, dramati- 4. Arrowsmith JB, Kennedy DL, Kuritsky JN, Faich GA. National patterns
cally and permanently changed the choice of antipy- of aspirin use and Reye syndrome reporting, United States, 1980 to 1985.
retics for children in North America. Every study Pediatrics. 1987;79:858 – 863
5. Linnemann CC, Shea L, Partin C. Reye’s syndrome: epidemiologic and
supported the relationship between aspirin use and viral studies, 1963–1974. Am J Epidemiol. 1975;101:517–526
risk of Reye’s syndrome. Furthermore, the ORs in- 6. Waldman RJ, Hall WN, McGee H, et al. Aspirin as a risk factor in Reye’s
creased rather than decreased in the more rigorously syndrome. JAMA. 1982;247:3089 –3094
designed studies. A formal meta-analysis was not 7. Halpin TJ, Holtzhauer FJ, Campbell RJ, et al. Reye’s syndrome and
medication use. JAMA. 1982;248:687– 691
necessary to conclude that this was a real and prob- 8. Kauffman RE. The Reye syndrome/aspirin controversy 1986. Contemp
ably causal association. Fortunately, an acceptable Pediatr. 1986;Nov/Dec:16 –24
alternative to aspirin that was not associated with 9. American Academy of Pediatrics, Committee on Infectious Diseases.
increased risk of Reye’s syndrome was readily avail- Aspirin and Reye syndrome. Pediatrics. 1982;69:810 – 812
able. Both the case– control studies and marketing 10. Surgeon General’s advisory on the use of salicylates and Reye’s syn-
drome. MMWR. 1982;81:289 –290
data document the replacement of aspirin with acet- 11. Wall Street Journal, July 28, 1982;23
aminophen for childhood antipyretic treatment after 12. US Food and Drug Administration. Reye syndrome awareness pro-
1980. Today, acetaminophen products dominate the gram. FDA Drug Bull. 1982;12:24
pediatric antipyretic market, largely because of the 13. Glezen WP. Aspirin and Reye’s syndrome. Am J Dis Child. 1982;136:
971–972
documented association of aspirin with increased 14. Aspirin Foundation of America, Inc, RS Working Group. Reye syn-
risk of Reye’s syndrome. The shift from aspirin to drome and salicylates: a spurious association. Pediatrics. 1982;70:
acetaminophen in the early 1980s represents a dra- 158 –160
matic and rapid change in consumer and physician 15. Hurwitz ES, Barrett MJ, Bregman D, et al. Public Health Service study
behavior with respect to antipyretic use in response on Reye’s syndrome and medications, report of the pilot phase. N Engl
J Med. 1985;313:849 – 857
to public dissemination of information through a 16. Hurwitz ES, Barrett MJ, Bregman D, et al. Public Health Service study
variety of media sources. It is worth noting that of Reye’s syndrome and medications, report of main study. N Engl
implementation of public policy on this issue lagged J Med. 1987;257:1905–1911
far behind consumer and physician behavior. Nev- 17. Forsyth BW, Horwitz RI, Acampora D, et al. New epidemiologic evi-
dence confirming that bias does not explain the aspirin/Reye’s syn-
ertheless, careful and rigorous clinical research led to drome association. JAMA. 1989;261:2517–2524
changes in antipyretic use, in spite of economic and 18. Rahwan GL, Rahwan RG. Aspirin and Reye’s syndrome: the change in
political pressures, that reduced dramatically the in- prescribing habits of health professionals. Drug Intell Clin Pharm. 1986;
cidence of a devastating illness for which a specific 20:143–145
cause or treatment has not and may never be found. 19. Remington PL, Rowley D, McGee H, et al. Decreasing trends in Reye syn-
drome and aspirin use in Michigan, 1979 to 1984. Pediatrics. 1986;77:93–98
20. Centers for Disease Control and Prevention. Influenza—United States,
REFERENCES 1985. MMWR. 1986;35:66 –74
1. Reye RDK, Morgan G, Baral J. Encephalopathy and fatty degeneration 21. Barrett MJ, Hurwitz ES, Schonberger LB, et al. Changing epidemiology
of the viscera: a disease entity in childhood. Lancet. 1963;2:749 –752 of Reye syndrome in the United States. Pediatrics. 1986;77:598 – 602

COMMENTARY

Apgar Scores as Predictors of Chronic Neurologic Disability, by Karin B.

Comments by Gerald S. Golden, MD

ABSTRACT OF ORIGINAL ARTICLE. Apgar scores of 0 to 3 at 10, 15, or 20 minutes and survived, 12 (12%)
were recorded at one and five minutes for ;49 000 had later cerebral palsy; 11 of the 12 also were mentally
infants, and at 10, 15, and 20 minutes for babies who retarded (in 10, IQ <50) and half had seizure disorders.
did not achieve a score of 8 or higher at five minutes. Eight children who survived after having very low late
These children were followed to 7 years of age. Low Apgar scores and who did not have cerebral palsy had
Apgar scores were risk factors for cerebral palsy, but lesser but significant disabilities. Of the children who
55% of children with later cerebral palsy had Apgar had Apgar scores of 0 to 3 at 10 minutes or later and
scores of 7 to 10 at one minute, and 73% scored 7 to 10 survived, 80% were free of major handicap at early
at five minutes. Of 99 children who had Apgar scores school age.
COMMENTARY

T
Received for publication Mar 19, 1998; accepted Mar 19, 1998. he relationship between perinatal asphyxia
Address correspondence to: Gerald S. Golden, MD, National Board of
Medical Examiners, 3750 Market St, Philadelphia, PA 19104.
and prematurity, the condition of the infant at
PEDIATRICS (ISSN 0031 4005). Copyright © 1998 by the American Acad- birth, and motor and mental disability has
emy of Pediatrics. been of interest since 1861 when William John Little,

262 SUPPLEMENT