CORR ES POND ENCE

Correspondence expansion, as suggested by experiments in a murine model,2

and nothing is known about its eventual effect on post-
engraftment events, as noted by Rubinstein et al.

MANUEL N. FERNÁNDEZ, M.D., PH.D.

ISABEL MILLÁN, PH.D.
Universidad Autónoma de Madrid
28035 Madrid, Spain

ELIANE GLUCKMAN, M.D., PH.D.

Cord-Blood Transplants
To the Editor: In the excellent paper by Rubinstein et al.
(Nov. 26 issue)1 on cord-blood transplants, the number of
nucleated cord-blood cells that were transfused per kilo-
gram of the recipient’s weight emerged as the main influ-
ence on engraftment. The serologic status of the recipient
with respect to cytomegalovirus was irrelevant to engraft-
ment and survival.
We have recently evaluated the results of 133 cord-blood
transplantations listed in the Eurocord Registry up to April
1998. We found that the dose of nucleated cord-blood
cells (measured before the blood was frozen) ranged from
9.7 million to 552 million per kilogram (median, 43.5 mil-
lion per kilogram). Rates of engraftment were 80 to 90
percent for all patients taken together and for all subgroups
of patients analyzed, except those over 15 years of age
(failure rate, 32 percent; 8 of 25 patients) and those who
were over 15 years of age and who had received less than
37 million nucleated cord-blood cells per kilogram (failure
rate, 47 percent; 7 of 15 patients). Substantial delays in the
time to engraftment occurred only in patients who were
over 15 years of age and in those who received less than
37 million nucleated cord-blood cells per kilogram.
These data are in agreement with the conclusions drawn
by Rubinstein et al. that the dose of nucleated cord-blood
cells before freezing is a major factor in engraftment, the
first goal of successful cord-blood transplantation. There-
fore, to improve the outcome of cord-blood transplantation
for adults, blood banks should concentrate on collecting
large units of cord blood. The time to and probability of
engraftment may not be significantly improved by in vitro
Hôpital Saint-Louis
75475 Paris CEDEX 10, France
1. Rubinstein P, Carrier C, Scaradavou A, et al. Outcome among 562
recipients of placental-blood transplants from unrelated donors. N Engl
J Med 1998;339:1565-77.
2. Güenechea G, Segovia JC, Albella B, et al. Delayed engraftment of non-
obese diabetic/severe combined immunodeficient mice transplanted with
ex vivo-expanded human CD34(+) cord blood cells. Blood 1999;93:
1097-105.
The authors reply:
To the Editor: Dr. Fernández and colleagues correctly
note the association between the dose of cells in placental-
blood transplants and engraftment indexes, especially, as
we reported, the time to engraftment. It is noteworthy,
however, that very few adults in our series received doses
of »50 million white cells per kilogram, and we could de-
tect no association between overall survival and the dose
of cells that was independent of the age of patients in mul-
tivariate analyses. Thus, as is the case with ex vivo cell ex-
pansion, the overall benefit of administering larger doses
of cells in placental-blood transplantations in adult pa-
tients remains speculative.
Nevertheless, we believe that larger patients may benefit
from receiving as many cells as possible. Enlarging the re-
positories of placental-blood units available would give
larger patients more options. We do not believe, however,
that banks should concentrate on collecting larger units if
this means discarding smaller ones. Small units may pro-
vide perfectly suitable grafts for children. We routinely save
all units with a blood volume of at least 40 ml. Among

INSTRUCTIONS FOR LETTERS TO THE EDITOR

Letters to the Editor are considered for publication (subject to editing and abridgment) provided they do not contain material that has been
submitted or published elsewhere. Please note the following: •Your letter must be typewritten and triple-spaced. •Its text, not including
references, must not exceed 400 words (please include a word count). •It must have no more than five references and one figure or table.
•It should not be signed by more than three authors. •Letters referring to a recent Journal article must be received within four weeks of its
publication. •Please include your full address, telephone number, and fax number (if you have one). •You may send us your letter by post, fax,
or electronic mail.
Our address: Letters to the Editor • New England Journal of Medicine • 10 Shattuck St. • Boston, MA 02115
Our fax numbers: 617-739-9864 and 617-734-4457
Our e-mail address: letters@nejm.org
We cannot acknowledge receipt of your letter, but we will notify you when we have made a decision about publication. We are unable to
provide prepublication proofs. Please enclose a stamped, self-addressed envelope if you want unpublished material returned to you.
Financial associations or other possible conflicts of interest must be disclosed. Submission of a letter constitutes permission for the Massachu-
setts Medical Society, its licensees, and its assignees to use it in the Journal’ s various editions (print, data base, and optical disk) and in
anthologies, revisions, and any other form or medium.

Vol ume 340 Numb e r 16 · 1287

The New England Journal of Medicine

Downloaded from nejm.org at PRINCETON UNIVERSITY on August 11, 2015. For personal use only. No other uses without permission.
Copyright © 1999 Massachusetts Medical Society. All rights reserved.
 The Ne w E n g l a nd Jo u r n a l o f Me d ic i ne
the patients in our study, 25 percent received units of less
than 60 ml, or of less than 700 million white cells.
PABLO RUBINSTEIN, M.D.
CLADD E. STEVENS, M.D., M.P.H.
New York Blood Center
New York, NY 10021
JOANNE KURTZBERG, M.D.
Duke University Medical Center
Durham, NC 27710
Deafness and Mutations in the Connexin
26 Gene
To the Editor: Morell et al. (Nov. 19 issue)1 demonstrate
that two mutations in the connexin 26 gene (GJB2) ac-
count for most of the cases of nonsyndromic recessive
deafness among Ashkenazi Jews. They suggested that since
the carrier rate of mutations in the GJB2 gene (4.76 per-
cent) is similar to that of other genes governing recessive
diseases in this population, such as Tay–Sachs disease,
Gaucher’s disease, and familial dysautonomia, Ashkenazi
Jews may elect to have their carrier status determined as
part of genetic screening.
Although the frequency of a disease within a population
is an important factor in the decision whether to begin
population screening, the first step in the decision should
be to define the aim of screening. For instance, screening
for phenylketonuria and hypothyroidism is done in order
to allow early treatment, whereas the goal of screening for
Tay–Sachs disease or thalassemia is to identify carriers and
prevent these diseases. Among the genetic diseases that are
relatively frequent among Ashkenazi Jews, some are very
severe, such as Canavan’s disease, familial dysautonomia,
and mucolipidosis IV, and therefore, like Tay–Sachs dis-
ease, are candidates for population screening as a means of
prevention.
With respect to deafness, screening for the relatively fre-
quent mutations in the GJB2 gene may be appropriate,
since identifying newborns who are homozygous may al-
low early treatment of deafness. In recent years, however,
the feasibility of detecting hearing defects in newborns or
other infants on the basis of either automated auditory
brain-stem responses or transiently evoked otoacoustic emis-
sions has been demonstrated; therefore, general screening
of the population is possible.2,3 Such a universal approach
not only would be more effective than screening for mu-
tations, since it allows the identification of all types of hear-
ing defects, but also would avoid the discriminative aspects
associated with screening only Ashkenazi Jews.
The medical importance of the identification by Morell
et al. of the specific mutations responsible for a large pro-
portion of hearing defects in the Ashkenazi Jewish com-
munity is that it will allow the cause of the hearing defect
in a child to be diagnosed relatively easily and will identify
the need for genetic counseling for the family.
JOEL ZLOTOGORA, M.D., PH.D.
Hadassah Medical School at Hebrew University
Jerusalem 91120, Israel
1288 · Apr il 2 2 , 19 9 9
The New England Journal of Medicine
Downloaded from nejm.org at PRINCETON UNIVERSITY on August 11, 2015. For personal use only. No other uses without permission.
Copyright © 1999 Massachusetts Medical Society. All rights reserved.
1. Morell RJ, Kim HJ, Hood LJ, et al. Mutations in the connexin 26 gene
(GJB2 ) among Ashkenazi Jews with nonsyndromic recessive deafness.
N Engl J Med 1998;339:1500-5.
2. Bantock HM, Croxson S. Universal hearing screening using transient
otoacoustic emissions in a community health clinic. Arch Dis Child 1998;
78:249-53.
3. Mason JA, Herrmann KR. Universal infant hearing screening by automat-
ed auditory brainstem response measurement. Pediatrics 1998;101:221-8.
The authors reply:
To the Editor: Dr. Zlotogora makes two valid points re-
garding the advisability of screening for GJB2 mutations
in the Ashkenazi Jewish population. The first is that tests
of automated auditory brain-stem responses or transiently
evoked otoacoustic emissions are more appropriate ways of
detecting congenital hearing impairments in newborns.
Such tests would more efficiently identify congenital hear-
ing loss at an early stage, since 50 percent of all cases have
a nongenetic cause. The use of hearing tests as a screening
tool also raises fewer ethical problems than does the use
of genetic tests. The National Institute on Deafness and
Other Communication Disorders convincingly makes the
argument for universal screening based on hearing tests of
newborns in a 1993 consensus statement.1
The second point is that by itself the prevalence of GJB2
mutations in a population does not justify genetic screen-
ing, in the absence of any consideration of the severity of
the condition being screened for or an effective interven-
tion for those who test positive. On this issue also we agree
with him. Genetic counselors should consider testing for
GJB2 carrier status if a patient so chooses. Our findings
make it clear that an appropriate test for Ashkenazi Jews
must be able to detect the 167delT allele in addition to
the 30delG allele. However, the decision to test is an indi-
vidual one, as is the decision involving what to do with the
information provided by testing. We are not advocating
routine screening to determine GJB2 carrier status.
ROBERT J. MORELL, PH.D.
THOMAS B. FRIEDMAN, PH.D.
National Institute on Deafness
and Other Communication Disorders
Rockville, MD 20850
1. Early identification of hearing impairment in infants and young chil-
dren. National Institutes of Health Development Conference Statement,
March 1–3, 1993. NIH Consensus Statement 1993;11(1):1-24. (Or see
http://odp.od.nih.gov/consensus/cons/092/092_statement.htm.)
Prognostic Score for Hodgkin’s Disease
To the Editor: The prognostic scoring system for Hodg-
kin’s disease proposed by Hasenclever and Diehl (Nov. 19
issue)1 has addressed a difficult challenge for clinicians. We
wish to raise two issues of concern regarding this article.
First, there was no histologic review, and it is uncertain
whether cases of the nodular lymphocyte-predominant
subtype were included. This form of Hodgkin’s disease is
different in both presentation and prognosis from the clas-
sic form of the disease. Only 3 percent of cases had lym-
phocyte-predominant Hodgkin’s disease; distortion of the
analysis would have been slight, but it would be incorrect
to apply the conclusions of the study to this small and un-
representative subgroup.
 CORR ES POND ENCE

1. Hasenclever D, Diehl V. A prognostic score for advanced Hodgkin’s dis-

TABLE 1. SURVIVAL CHARACTERISTICS OF A POPULATION-BASED
COHORT OF PATIENTS WITH HODGKIN’S DISEASE ACCORDING TO
THE PROGNOSTIC SYSTEM OF HASENCLEVER AND DIEHL.
ease. N Engl J Med 1998;339:1506-14.
2. Proctor SJ, Taylor P, Mackie MJ, et al. A numerical prognostic index for
clinical use in identification of poor-risk patients with Hodgkin’s disease at
diagnosis. Leuk Lymphoma 1992;7:Suppl:17-20.

DISEASE-
SPECIFIC MEDIAN FIVE-YEAR DISEASE- To the Editor: Hasenclever and Diehl describe a prognos-
PATIENTS DEATHS FOLLOW-UP SPECIFIC SURVIVAL tic score for advanced Hodgkin’s disease based on seven
CHARACTERISTIC AT RISK OBSERVED OF SURVIVORS (95% CI)*
factors, among which four are laboratory measurements
no. (%) mo (i.e., serum albumin, blood hemoglobin, white-cell count,
and lymphocyte count). However, they do not mention the
Score «3 403 (88) 56 (82) 74 84.5 (80.8–88.2)
methods used to measure these factors. It would therefore
Score »4 56 (12) 12 (18) 35 78.4 (66.1–90.7)
be impossible for an independent team to reproduce their
Score »5 17 (4) 4 (6) 33 67.7 (40.3–95.1)
results (a basic principle of good science). It is not possible
Ann Arbor 91 (20) 22 (32) 55 72.2 (61.4–83.0)
stage IV
to compare the prognostic value of any particular labora-
Total 459 (100) 68 (100) 71 84.5 (80.8–88.2)
tory measurements if such measurements are made with
different techniques. For example, electrophoretic, colori-
*CI denotes confidence interval. metric, turbidimetric, and nephelometric procedures do not
yield identical results for serum albumin concentrations,
and an international standardization for the measurement
of serum protein has been routine practice in most labora-
Second, we have attempted to place the findings in a tories only since 1995.1 Such a lack of consistency among
population-based setting. There are 1281 patients with clas- methods may exist for most of the other laboratory tests
sic Hodgkin’s disease, all of whom were negative for the performed in their study.
human immunodeficiency virus, whose presentation and
follow-up data are registered in the Scotland and Newcastle JOSEPH WATINE, M.D.
Lymphoma Group data base. This registry represents most Hôpital Général
cases of Hodgkin’s disease diagnosed in our population of 12027 Rodez CEDEX 09, France
8.5 million since 1986. One hundred eighty-seven (14.6
percent) are at least 66 years old and 41 (3.2 percent) are 1. Ward AM, Committee on Plasma Protein. In: Proceedings of the 16th
under 15; these patients were excluded from the scoring International Congress of Clinical Chemistry, London, July 8–12, 1996:
35-6.
system developed by the authors. Of the remaining 1053
patients (82.2 percent), 459 had complete data for the
score. The rates of disease-specific survival for these patients
are presented in Table 1. We looked at disease-specific surviv- Dr. Hasenclever replies:
al rather than freedom from progression for three reasons: To the Editor: Watine is concerned about our using lab-
disease-specific survival is a definitive measure of failure; if oratory tests in a prognostic score without fully specifying
disease-specific survival is not related to presentation fea- the methods involved. Our project combined most of the
tures, the argument for intensifying initial treatment is di- prospectively documented trial data that were collected in
minished; and the assessment of progression is notoriously the 1980s by leading centers and study groups worldwide.
difficult in patients with Hodgkin’s disease in whom resid- In these data sets, individual normal ranges were not rou-
ual masses after therapy are common and may be inactive. tinely documented or were only documented for labora-
In only 17 of 459 patients (3.7 percent) was the prog- tory measurements in which the normal range varies con-
nostic score >4. Fifteen of these 17 had Ann Arbor stage IV siderably (e.g., lactate dehydrogenase and serum alkaline
disease. We conclude that the system proposed by Hasen- phosphatase). For these tests, only values standardized in
clever and Diehl identifies only a small proportion of pa- units of the upper bound of the normal range were used
tients with poor outcome, with nearly all disease-specific in the analysis. Concerning the other tests, in particular
deaths occurring in “low-risk” categories. We have long those incorporated into the score, the center-specific normal
been concerned with the identification of patients for whom ranges and the center-specific distributions of the values
conventional four-drug regimens are likely to fail and will appeared to overlap sufficiently to justify a joint evaluation.
continue to use the index of the Scotland and Newcastle In addition, since we used cutoff points chosen to demar-
Lymphoma Group 2 to select candidates for our eight-drug cate a clearly abnormal state qualitatively, the error due to
hybrid regimen. To date, no prognostic system for Hodg- differing methods of measurement should be negligible.
kin’s disease based on traditional factors has been entirely Nevertheless, we agree with Watine that the quality of data
satisfactory, and we agree with Hasenclever and Diehl that in prospective clinical trials can be improved by systemat-
efforts should be intensified in the search for new and more ically collecting information on methods of measurement
pathologically relevant markers of prognosis. and normal ranges for all laboratory tests.
Jack et al. correctly point out that the prognostic score
FERGUS R. JACK, M.R.C.PATH., M.R.C.P. applies to classic Hodgkin’s disease. There are no specific
BRIAN ANGUS, F.R.C.PATH. data on the validity of the score in the recently delineated
PENNY R.A. TAYLOR, M.B., B.S. very small subgroup of patients with the nodular lympho-
Royal Victoria Infirmary cyte-predominant subtype. Jack et al. applied the score to
Newcastle upon Tyne NE1 4LP, United Kingdom the Scotland and Newcastle Lymphoma Group data base

Vol ume 340 Numb e r 16 · 1289

The New England Journal of Medicine

Downloaded from nejm.org at PRINCETON UNIVERSITY on August 11, 2015. For personal use only. No other uses without permission.
Copyright © 1999 Massachusetts Medical Society. All rights reserved.
 The Ne w E n g l a nd Jo u r n a l o f Me d ic i ne
and concluded that “the system identifies only a small pro-
portion of patients with poor outcome.” Their results are
fully compatible with the data on the score applied to sur-
vival presented in Figure 1B of our paper with regard to
both the five-year survival rates and the proportions of pa-
tients with a particular number of adverse factors.
Unfortunately, they did not use the main end point, free-
dom from progression of disease, but instead used disease-
specific survival. It should be stressed that the score was
constructed and optimized for freedom from progression
of disease. As discussed in our article, a score optimized
for disease-specific survival would have to give age a much
stronger influence, because age is a major prognostic fac-
tor for survival after relapse. A validation of the score in
the intended context would be preferable.
The score we reported was developed for advanced-stage
Hodgkin’s disease. Jack et al. appear not to have excluded
patients with early stages of the disease (stages I and II,
without any risk factors), who were probably treated with
therapy that was not as aggressive as the treatment for ad-
vanced stages. A recent analysis of the data of the German
Hodgkin’s Lymphoma Study Group on patients with early
stages of disease1 shows that the prognostic score also
works for these patients, although the adverse factor of
stage IV disease should be interpreted to encompass any
extranodal involvement (E stage). This finding deserves
further validation in independent data sets.
DIRK HASENCLEVER, PH.D.
University of Leipzig
D-04103 Leipzig, Germany
1. Lieberz D, Paulus U, Franklin J, Tesch H, Diehl V. Applicability of the
international prognostic score for advanced stage Hodgkin’s disease to early
and intermediate stages. Blood 1998;92:Suppl 1:86a. abstract.
Carbon Monoxide Poisoning
To the Editor: Ernst and Zibrak (Nov. 26 issue)1 state
that coma is an undisputed indication for hyperbaric-oxygen
therapy, but this claim has not been proved2 and might be
misleading. Neurocognitive sequelae can develop in coma-
tose patients with carbon monoxide poisoning who are
treated with hyperbaric oxygen,3 and patients with severe
carbon monoxide poisoning can have a normal functional
and cognitive recovery without hyperbaric oxygen.4
Interim analysis of an ongoing randomized clinical trial5
and one completed randomized clinical trial6 of the role
of hyperbaric oxygen therapy in acute carbon monoxide
poisoning have failed to demonstrate differences in out-
comes between patients treated with normobaric oxygen
and those treated with hyperbaric oxygen. Both of these
trials enrolled comatose patients with carbon monoxide
poisoning. We acknowledge that some authorities recom-
mend that such patients receive hyperbaric oxygen, but
there is no compelling data from clinical trials indicating
that they require hyperbaric oxygen. There are risks asso-
ciated with hyperbaric oxygen, including those related to
oxygen transport, barotrauma affecting the middle and
inner ear, and in cases of carbon monoxide poisoning, a
1 to 3 percent probability of a seizure induced by hyper-
baric oxygen.6,7
In an ongoing longitudinal follow-up study conducted
1290 · Apr il 2 2 , 19 9 9
The New England Journal of Medicine
Downloaded from nejm.org at PRINCETON UNIVERSITY on August 11, 2015. For personal use only. No other uses without permission.
Copyright © 1999 Massachusetts Medical Society. All rights reserved.
at our institution, approximately 30 percent of the patients
with acute carbon monoxide poisoning have neurocogni-
tive problems one year after poisoning. Of these patients,
approximately one third have the delayed neuropsychiatric
syndrome and two thirds have persistent neurocognitive
problems, primarily difficulties with memory and executive
function.5,8 Unfortunately, the clinical and laboratory find-
ings at presentation are not predictive of long-term out-
come. The effect of hyperbaric oxygen on long-term out-
come is still unknown. We agree that carbon monoxide
poisoning is common and may be associated with substan-
tial neurocognitive morbidity8 and that patients should be
treated with 100 percent oxygen and possibly with hyper-
baric oxygen.
LINDELL K. WEAVER, M.D.
RAMONA O. HOPKINS, PH.D.
GREGORY ELLIOTT, M.D.
LDS Hospital
Salt Lake City, UT 84143
1. Ernst A, Zibrak JD. Carbon monoxide poisoning. N Engl J Med 1998;
339:1603-8.
2. Tibbles PM, Perrotta PL. Treatment of carbon monoxide poisoning: a
critical review of human outcome studies comparing normobaric oxygen
with hyperbaric oxygen. Ann Emerg Med 1994;24:269-76.
3. Pracyk JB, Stolp BW, Fife CE, Gray L, Piantadosi CA. Brain comput-
erized tomography after hyperbaric oxygen therapy for carbon monoxide
poisoning. Undersea Hyperb Med 1995;22:1-7.
4. Weaver LK, Hopkins RO, Larson-Lohr V. Neuropsychologic and func-
tional recovery from severe carbon monoxide poisoning without hyperbaric
oxygen therapy. Ann Emerg Med 1996;27:736-40.
5. Weaver LK, Hopkins RO, Larson-Lohr V, Howe S, Haberstock D.
Double-blind, controlled, prospective, randomized clinical trial (RCT) in
patients with acute carbon monoxide (CO) poisoning: outcome of patients
treated with normobaric oxygen or hyperbaric oxygen (HBO2) — an interim
report. Undersea Hyperb Med 1995;22:Suppl:14. abstract.
6. Scheinkestel CD, Jones K, Cooper DJ, Millar I, Tuxen DV, Myles PS.
Interim analysis — controlled clinical trial of hyperbaric oxygen in acute
carbon monoxide (CO) poisoning. Undersea Hyperb Med 1996;23:Suppl:
7. abstract.
7. Hampson NB, Simonson SG, Kramer CC, Piantadosi CA. Central
nervous system oxygen toxicity during hyperbaric treatment of patients
with carbon monoxide poisoning. Undersea Hyperb Med 1996;23:215-
9.
8. Weaver LK, Hopkins RO, Howe S, Larson-Lohr V, Churchill S. Out-
come at 6 and 12 months following acute CO poisoning. Undersea Hyperb
Med 1996;23:Suppl:9-10. abstract.
To the Editor: The development of the carbon monoxide
detector is potentially the most important advance in the
prevention of carbon monoxide poisoning over the past 10
years. A recent study estimated that these detectors could
have helped save 78 lives between 1980 and 1995 in the
state of New Mexico alone.1 Even more lives might have
been saved in temperate climates. The use of chemical-
reagent detectors (with a threshold response of about 100
parts per million) should be discouraged in favor of elec-
tronic detectors.1 At least 68 cases of occult carbon mon-
oxide poisoning were uncovered by detectors in the first
three months after an ordinance mandating their installa-
tion was implemented in Chicago.2
Contrary to what Ernst and Zibrak state, national and
local standards do exist for electronic carbon monoxide de-
tectors. Underwriters Laboratories has published standards
used by manufacturers of carbon monoxide detectors since
1991.3 These detectors approved by Underwriters Labora-
 C ORR ES POND ENCE
tories are designed to sound an alarm when ambient carbon
monoxide levels are reached that would cause a carboxy-
hemoglobin level of 10 percent or greater in a person en-
gaged in work requiring heavy exertion. In 1994, Chicago
became one of the first large metropolitan areas to require
residential carbon monoxide detectors.4 St. Louis, Albany,
New York, and Fort Lee, New Jersey, are among the other
municipalities that have such ordinances. Also, the National
Fire Protection Association has published recommended
practices for the installation of household carbon monox-
ide–warning equipment.5 It is clear that electronic carbon
monoxide detectors are effective tools for ameliorating the
public health problem of carbon monoxide poisoning and
that they can help unmask “the silent killer.”
JERROLD B. LEIKIN, M.D.
JACK C. CLIFTON II, M.D.
PAUL K. HANASHIRO, M.D.
Rush–Presbyterian–St. Luke’s Medical Center
Chicago, IL 60612
1. Yoon SS, Macdonald SC, Parrish RG. Deaths from unintentional carbon
monoxide poisoning and potential for prevention with carbon monoxide
detectors. JAMA 1998;279:685-7.
2. Leikin JB. Carbon monoxide detectors and emergency physicians. Am
J Emerg Med 1996;14:90-4.
3. The standard for single and multiple station carbon monoxide detectors
UL 2034. Proposed first ed. Northbrook, Ill.: Underwriters Laboratory,
1991. Revised 1994.
4. City Council of Chicago, Meeting of March 2, 1994. Amendment of
Title 13, Chapter 64 of the Municipal Code of Chicago by addition of new
sections 190 through 300 requiring carbon monoxide detectors in various
buildings.
5. Publication no. 720. Quincy, Mass.: National Fire Protection Associa-
tion, 1998.
To the Editor: One clinical scenario was conspicuously
absent from the review article by Ernst and Zibrak: Physi-
cians may be called to assist in the care of patients who
have been exposed to carbon monoxide through the break-
down of anesthetic in desiccated carbon dioxide absorbents
during the delivery of inhaled anesthesia in closed or semi-
closed breathing circuits.1 Although most anesthetics are
stable in the presence of normally hydrated carbon dioxide
absorbents, improper care of machines used to deliver an-
esthesia may cause desiccation of the absorbents, which
can result in the formation of carbon monoxide through
chemical reactions involving difluoromethyl ethers,2 which
include such popular anesthetics as enflurane, isoflurane,
and desflurane. Because a period of 24 to 48 hours is re-
quired for desiccation of these absorbents, most cases of
intraoperative carbon monoxide poisoning occur during
the first delivery of general anesthesia through an anesthe-
sia machine on Monday mornings.
Improved care of anesthesia machines has been shown
to reduce the incidence of carbon monoxide exposure from
approximately 1 in 200 to 1 in 2000 first cases,3 but some
remote or seldom-used facilities may be at particularly
high risk. Exposure can be severe; carboxyhemoglobin con-
centrations over 30 percent have been documented in hu-
mans,4 and animals have been exposed to lethal concen-
trations of over 80 percent carboxyhemoglobin in clinical
scenarios.5 It is possible that most exposure goes undetected
because monitoring for carbon monoxide or carboxyhe-
moglobin is not routine in these circumstances, because
The New England Journal of Medicine
Downloaded from nejm.org at PRINCETON UNIVERSITY on August 11, 2015. For personal use only. No other uses without permission.
Copyright © 1999 Massachusetts Medical Society. All rights reserved.
the symptoms of carbon monoxide poisoning are masked
by the effects of general anesthesia, and because, after the
patient emerges from anesthesia, signs and symptoms remain
nonspecific.
HARVEY J. WOEHLCK, M.D.
Medical College of Wisconsin
Milwaukee, WI 53226
1. Fang ZX, Eger E II, Laster MJ, Chortkoff BS, Kandel L, Ionescu P.
Carbon monoxide production from degradation of desflurane, enflurane,
isoflurane, halothane, and sevoflurane by soda lime and Baralyme. Anesth
Analg 1995;80:1187-93.
2. Baxter PJ, Garton K, Kharasch ED. Mechanistic aspects of carbon mon-
oxide formation from volatile anesthetics. Anesthesiology 1998;89:929-41.
3. Woehlck HJ, Dunning M III, Connolly LA. Reduction in the incidence
of carbon monoxide exposures in humans undergoing general anesthesia.
Anesthesiology 1997;87:228-34.
4. Berry PD, Sessler DI, Larson MD. Severe carbon monoxide poisoning
during desflurane anesthesia. Anesthesiology 1999;90:613-6.
5. Frink EJ Jr, Nogami WM, Morgan SE, Salmon RC. High carboxy-
hemoglobin concentrations occur in swine during desflurane anesthesia
in the presence of partially dried carbon dioxide absorbents. Anesthesiology
1997;87:308-16.
To the Editor: As Ernst and Zibrak point out, carbon
monoxide poisoning accounts for about 600 accidental
deaths and 3000 suicides each year. Cerebral symptoms
are often prominent and may progress to brain death. Be-
cause cardiorespiratory symptoms and frank injury to the
heart have been observed and because of an early unsuc-
cessful attempt at heart transplantation,1 there has been a
reluctance to consider victims of carbon monoxide poison-
ing who have been declared brain-dead as potential organ
donors.
Several reports of such victims serving as successful do-
nors of kidneys,2 livers,3 hearts,4 and even a lung5 indicate
that careful evaluation of organ function in these victims
can identify organs that are suitable for transplantation. All
these reports came from outside the United States.
The waiting list of the United Network for Organ Shar-
ing on October 31, 1998, had 62,994 registrants (includ-
ing 41,544 waiting for kidneys, 11,601 waiting for livers,
4184 waiting for hearts, 3088 waiting for lungs, and 2235
waiting for pancreases or kidneys and pancreases). Many
of these patients are in urgent need of transplants and are
on life-support mechanisms. Therefore, the judicious eval-
uation of individual organ function of brain-dead victims
of carbon monoxide poisoning could lead to a slight eas-
ing of the critical shortage of organ donors.
H. MYRON KAUFFMAN, M.D.
United Network for Organ Sharing
Richmond, VA 23225-8770
1. Karwande SV, Hopfenbeck JA, Renlund DG, Burton NA, Gay WA Jr.
An avoidable pitfall in donor selection for heart transplantation. J Heart
Lung Transplant 1989;8:422-4.
2. Hébert M-J, Boucher A, Beaucage G, Girard R, Dandavino R. Trans-
plantation of kidneys from a donor with carbon monoxide poisoning.
N Engl J Med 1992;326:1571.
3. Verran D, Chui A, Painter D, et al. Use of liver allografts from carbon
monoxide poisoned cadaveric donors. Transplant 1996;62:1514-5.
4. Smith JA, Bergin PJ, Williams TJ, Esmore DS. Successful heart trans-
plantation with cardiac allografts exposed to carbon monoxide poisoning.
J Heart Lung Transplant 1992;11:698-700.
5. Shennib H, Adoumie R, Fraser R. Successful transplantation of a lung
Vol ume 340 Numb e r 16 · 1291
 The Ne w E n g l a nd Jo u r n a l o f Me d ic i ne
allograft from a carbon monoxide-poisoning victim. J Heart Lung Trans-
plant 1992;11:68-71.
To the Editor: In the excellent review article by Ernst and
Zibrak, we must point out that Figure 1 (Oxygen–Hemo-
globin Dissociation Curve) is mislabeled. This is readily
apparent if one considers that with 60 percent carboxyhe-
moglobin one cannot have an oxygen saturation greater than
40 percent; otherwise, the total hemoglobin saturation
would be greater than 100 percent. This is impossible, since
oxygen and carbon monoxide bind competitively to iron
atoms in hemoglobin.1 Perhaps the authors intended to label
the y axis “(Hemoglobin O2)÷([Hemoglobin O2]+[Red
Hemoglobin]) (%),” or equivalently, “(Hemoglobin O2)÷
([Hemoglobin]¡[Carboxyhemoglobin]) (%)” — i.e., the
oxygen saturation of the noncarboxylated hemoglobin.
DONAL P. RYAN, M.D.
ANTHONY M. COSENTINO, M.D.
St. Mary’s Medical Center
San Francisco, CA 94117
1. Roughton FJW. Transport of oxygen and carbon dioxide. In: Fenn WO,
Rahn H, eds. Handbook of physiology. Section 3. Respiration. Vol. 1.
Washington, D.C.: American Physiological Society, 1964:778-82.
The authors reply:
To the Editor: Our review of carbon monoxide poisoning
concentrated on the more common sources of production
that might be encountered by primary care and emergency
medicine clinicians. Treatment is often based on recom-
mendations, rather than evidence-based studies with con-
clusive results.
As pointed out by Dr. Woehlck, improperly maintained
anesthesia circuits may be a cause of carbon monoxide poi-
soning. This, fortunately, is a rare circumstance not com-
monly encountered by practicing clinicians. Appropriate and
diligent maintenance of anesthesia machines should allevi-
ate this problem.
Carbon monoxide detectors are useful but have not been
conclusively demonstrated to reduce morbidity and mor-
tality. The cited study by Yoon et al.1 is a descriptive analy-
sis that does not actually compare an intervention group
with a nonintervention group. We agree with Leikin et al.
that carbon monoxide detectors have the potential to de-
crease the incidence of carbon monoxide poisoning in resi-
dential settings, but they are a form of secondary prevention
and not a substitute for proper maintenance and appropri-
ate use of heating equipment.
We agree with Weaver et al. that “undisputed” may have
been a poor choice of words for describing indications for
hyperbaric-oxygen therapy in comatose patients. However,
we continue to believe strongly that the weight of clinical
empirical evidence supports this practice. Our review of the
literature concerning neurologic dysfunction as a conse-
quence of hyperbaric-oxygen therapy in patients with car-
bon monoxide poisoning fails to convince us of a uniform
negative effect. In fact, hyperbaric oxygen appears to modify
favorably the propensity of neurocognitive defects to de-
velop and is considered the standard of care by most author-
ities.2,3 Only further research can answer these questions
more definitively.
1292 · Ap r il 2 2 , 19 9 9
The New England Journal of Medicine
Downloaded from nejm.org at PRINCETON UNIVERSITY on August 11, 2015. For personal use only. No other uses without permission.
Copyright © 1999 Massachusetts Medical Society. All rights reserved.
We agree with Dr. Kauffman that victims of carbon mon-
oxide poisoning need to be considered as potential organ
donors. Carbon monoxide poisoning may lead to cellular
damage in a variety of organ systems, but such an effect
should not be considered an absolute contraindication to
organ transplantation. Several reports in the literature con-
firm the feasibility of this approach.4,5 This area also is in
need of further research and protocols should be estab-
lished to help alleviate the current shortage of organs.
Drs. Ryan and Cosentino correctly point out that the
ordinate of Figure 1 of our article is mislabeled. This axis is
intended to represent the relative oxygen saturation of the
residual hemoglobin molecules not bound to carbon mon-
oxide: 100¡Z, where Z is the percent of total hemoglobin
molecules bound to carbon monoxide. As suggested, the
correct label is that used by Roughton: 100 (Hemoglobin
O2)÷([Hemoglobin O2]+[Red Hemoglobin]). For a given
percentage of carboxyhemoglobin, this yields the ratio of
oxygen-bound hemoglobin to the sum of oxygen-bound
hemoglobin and reduced (unbound) hemoglobin. Thus
for a carboxyhemoglobin concentration of 60 percent (as
depicted in the figure), when all the remaining hemoglo-
bin is bound to oxygen, and red hemoglobin is therefore
0 percent, the expression yields: 100¬(40÷[40+0]), or
100 percent.
ARMIN ERNST, M.D.
JOSEPH ZIBRAK, M.D.
Beth Israel Deaconess Medical Center
Boston, MA 02215
1. Yoon SS, Macdonald SC, Parrish RG. Deaths from unintentional carbon
monoxide poisoning and potential for prevention with carbon monoxide
detectors. JAMA 1998;279:685-7.
2. Tibbles PM, Edelsberg JS. Hyperbaric-oxygen therapy. N Engl J Med
1996;334:1642-8.
3. Hyperbaric oxygen therapy: a committee report. Bethesda, Md.: Under-
sea and Hyperbaric Medical Society, 1992:12-3.
4. Smith JA, Bergin PJ, Williams TJ, Esmore DS. Successful heart trans-
plantation with cardiac allografts exposed to carbon monoxide poisoning.
J Heart Lung Transplant 1992;11:698-700.
5. Shennib H, Adoumie R, Fraser R. Successful transplantation of a lung
allograft from a carbon monoxide-poisoning victim. J Heart Lung Trans-
plant 1992;11:68-71.
Terbinafine and Fulminant Hepatic Failure
To the Editor: Terbinafine is an antifungal agent that is
widely prescribed for common skin infections.1 We describe
a patient who had taken terbinafine and in whom fulmi-
nant hepatic failure developed, requiring orthotopic liver
transplantation.
A 48-year-old woman took 250 mg of terbinafine daily
for five days for a fungal nail infection. Over the next four
weeks, fulminant hepatic failure developed. The patient
had been taking dothiepin (75 mg per day), which is a tri-
cyclic antidepressant, and propranolol (40 mg twice daily)
for more than 18 months. She had no risk factors for liver
disease, a minimal intake of alcohol (<40 g per week), and
no history of ingestion of acetaminophen or other analge-
sics. Serum acetaminophen levels were undetectable. Au-
toantibody screening and screening for serologic hepatitis
A, B, and C viruses were negative; abdominal ultrasonog-
raphy revealed a normal-sized liver with no evidence of sple-
 C ORR ES POND ENCE
nomegaly or ascites. The patient’s condition deteriorated;
encephalopathy increased, requiring ventilation and inten-
sive care. She subsequently underwent uncomplicated or-
thotopic liver transplantation. She remains well 18 months
after transplantation.
Histologic examination of the explanted liver revealed
panacinar submassive necrosis and nearly complete disap-
pearance of hepatocytes, with no evidence of chronic liver
disease. These findings are compatible with a drug-related
cause of disease.
In our patient, a presumed idiosyncratic (type B) drug
reaction to terbinafine may have been an important factor
in the development of fulminant hepatic failure. The two
other prescribed medications that she was taking have a
low reported potential for hepatotoxicity, and she had been
taking them for many months. Minor abnormalities in the
results of liver-function tests have been reported in up to
4 percent of patients during oral treatment with terbin-
afine,2 with two reports of predominantly cholestatic, re-
versible, terbinafine-associated hepatic injury.3-5
KOSH AGARWAL, M.R.C.P.
DEREK M. MANAS, F.C.S.(S.A.)
MARK HUDSON, F.R.C.P.
Freeman Hospital
Newcastle upon Tyne NE7 7DN, United Kingdom
1. Gupta AK, Scher RK. Oral antifungal agents for onychomycosis. Lancet
1998;351:541-2.
2. van der Schroeff JG, Cirkel PK, Crijns MB, et al. A randomized treat-
ment duration-finding study of terbinafine in onychomycosis. Br J Derma-
tol 1992;126:Suppl 39:36-9.
3. van’t Wout JW, Herrmann WA, de Vries R, Stricker BH. Terbinafine-
associated hepatic injury. J Hepatol 1994;21:115-7.
4. Lazaros GA, Papatheodoridis GV, Delladetsima JK, Tassopoulos NC.
Terbinafine-induced cholestatic liver disease. J Hepatol 1996;24:753-6.
5. Fernandes NF, Geller SA, Fong TL. Terbinafine hepatotoxicity: case re-
port and review of the literature. Am J Gastroenterol 1998;93:459-60.
Cost as a Barrier to Medical Care
To the Editor: In “Cost as a Barrier to Medical Care in
Relation to Unemployment Rates” (Nov. 26 issue),1 it
seems to me that Figure 1 is misleading. The y axes for the
two plotted variables have different points of origin, but
more important, different scales. The patterns of monthly
unemployment rates and the prevalence of the perception
that cost was a barrier to medical care cannot be compared
validly on such a misleading graph. Plotted correctly, the
two lines would not overlap or parallel each other as they
appear to.
MAXIM LEWKOWSKI
McGill University
Montreal, QC H3G 1L2, Canada
1. Nelson DE, Thompson BL, Bland SD. Cost as a barrier to medical care
in relation to unemployment rates. N Engl J Med 1998;339:1644-5.
The authors reply:
To the Editor: We appreciate Lewkowski’s comments.
Had we plotted our data using the same points of origin
The New England Journal of Medicine
Downloaded from nejm.org at PRINCETON UNIVERSITY on August 11, 2015. For personal use only. No other uses without permission.
Copyright © 1999 Massachusetts Medical Society. All rights reserved.
or the same scales, the two lines would not have over-
lapped. We disagree, however, that this is misleading.
Different points of origin were used on the y axes to
make it easier to visualize our data, since estimates of the
prevalence that cost was perceived as a barrier to medical
care were about twice as high as unemployment rates. This
approach is analogous to comparing cancer-incidence trends
according to age by graphing data on different scales.1 When
our data are displayed with the same scale width for both
y axes (not shown), the similarity of the two trend lines
remains comparable.
DAVID E. NELSON, M.D., M.P.H.
BETSY L. THOMPSON, M.D., M.S.P.H.
SHAYNE D. BLAND, M.S.
Centers for Disease Control and Prevention
Atlanta, GA 30341-3717
1. Harras A, ed. Cancer: rates and risks. Bethesda, Md.: National Institutes
of Health, 1996. (NIH publication no. 96-691.)
Extreme Hyperkalemia in Munchausen-
by-Proxy Syndrome
To the Editor: A 20-month-old girl was admitted to the
hospital because of a reported febrile convulsion. Her med-
ical history included recurrent simple febrile convulsions
and unexplained slight gynecomastia. Physical examina-
tion revealed a well-nourished and cheerful child, with
normal vital signs and physical and psychomotor develop-
ment, except for the slight gynecomastia. Repeated blood
tests from different puncture sites sent in different test
tubes to the hospital laboratory revealed extreme hyperka-
lemia, incompatible with life, and very high serum creati-
nine concentrations (Table 1, samples 1 through 3). The
electrocardiogram showed no signs of hyperkalemia. Urine
output was normal. The discrepancy between the clinical
findings and the laboratory results raised the suspicion of
factitious hyperkalemia and azotemia. Since it is common
practice in our medical center for samples to be delivered
to the laboratory by family members, another blood sample
from the girl was taken directly to the laboratory by hos-
pital staff. The results were normal (Table 1, sample 4). A
blood sample was drawn from a boy with normal kidney
function (sample 5). A portion of it was labeled with the
girl’s name and given to her mother to take to the labora-
tory (Table 1, sample 6). Once again, severe hyperkalemia
and high serum creatinine concentrations were found. Fur-
ther questioning revealed that all the initial samples from
the girl had been delivered to the laboratory by her mother.
We concluded that the mother was deliberately contami-
nating the samples, and therefore Munchausen-by-proxy
syndrome was diagnosed.
To explain the method by which the mother could have
altered the samples to yield these results, we calculated the
expected effect of adding urine, the only readily available
solution that could yield this constellation of laboratory
findings, to a blood sample. Because the hematocrit of the
mixed blood samples decreased by approximately 20 per-
cent, we calculated that the plasma volume was diluted by
30 percent. We then diluted the girl’s serum by one third
with urine from a normal adult (Table 1, sample 7) and
Vol ume 340 Numb e r 16 · 1293
 The Ne w E n g l a nd Jo u r n a l o f Me d ic i ne

TABLE 1. BIOCHEMICAL VALUES IN A GIRL WITH MUNCHAUSEN-BY-PROXY SYNDROME.

SOURCE METHOD OF
SAMPLE OFBLOOD TRANSPORT TO SERUM SERUM SERUM
NO. DAY TIME SAMPLE LABORATORY HEMATOCRIT SODIUM POTASSIUM CREATININE

% mmol/liter mg/dl*

1 1 5:00 p.m. Girl By girl’s mother 25 144 10.6 12.7

2 1 5:50 p.m. Girl By girl’s mother 25 138 10.3 12.0
3 1 9:00 p.m. Girl By girl’s mother 23 158 12.0 15.0
4 2 12:15 a.m. Girl By hospital staff 30 139 4.3 0.5
5 2 1:20 a.m. Boy By hospital staff 34 140 4.0 0.7
6 2 1:30 a.m. Boy By girl’s mother 30 136 17.0 20.0
7† 3 10:00 a.m. Girl By hospital staff — 151 12.8 13.6

*To convert values for serum creatinine to micromoles per liter, multiply by 88.4.
†A serum sample from the girl was diluted with urine from a normal subject.

obtained results that were very similar to the original, bi- be suspected whenever there is an unexplained discrepancy
zarre values. between clinical and laboratory findings.
The mother was confronted with the suspicion that she
was contaminating the blood samples. She denied doing so, DANIELLA MAGEN, M.D.
as is typically reported in Munchausen-by-proxy syndrome.1 KARL SKORECKI, M.D.
A hospital pediatric social worker and a psychiatrist were Rambam Medical Center
unable to elicit reasons for the mother’s behavior. The girl Haifa 31096, Israel
was discharged under the supervision of community social
services and a local child-welfare agency. 1. Schreier HA, Libow JA. Munchausen by proxy syndrome: a modern
pediatric challenge. J Pediatr 1994;125:S110-S115.
There are reports in the literature of Munchausen-by- 2. Meadow R. Münchausen syndrome by proxy: the hinterland of child
proxy syndrome in children in which their urine samples abuse. Lancet 1977;2:343-5.
were mixed with menstrual secretions,2 blood,3 feces,4 and 3. Jacobs JC. Factitious hematuria in two teenage boys. Am J Dis Child
other exogenous substances.5 It seems that in the case of 1979;133:550-1.
4. Reich P, Lazarus JM, Kelly MJ, Rogers MP. Factitious feculent urine in
Munchausen-by-proxy syndrome, the perpetrator may be an adolescent boy. JAMA 1977;238:420-1.
more creative than physicians can imagine. This case serves 5. Mitas JA II. Exogenous protein as the cause of nephrotic-range protein-
as a reminder that Munchausen-by-proxy syndrome should uria. Am J Med 1985;79:115-8.

©1999, Massachusetts Medical Society.

1294 · Ap r il 2 2 , 19 9 9

The New England Journal of Medicine

Downloaded from nejm.org at PRINCETON UNIVERSITY on August 11, 2015. For personal use only. No other uses without permission.
Copyright © 1999 Massachusetts Medical Society. All rights reserved.