O'Brien KL, Selanikio JD, Hecdivert C, Placide M, Louis M, Barr DB, Barr JR, Hospedales CJ, Lewis MJ, Schwartz B, Philen RM, St. Victor S, Espindola J, Needham LL, Denerville K, for the Acute Renal Failure Investigation Team . Epidemic of Pediatric Deaths From Acute Renal Failure Caused by Diethylene Glycol Poisoning. JAMA. 1998;279(15):1175-1180. doi:10.1001/jama.279.15.1175
From the National Center for Infectious Disease (Drs O'Brien and Schwartz), National Center for Environmental Health (Drs Selanikio, D. B. Barr, J. R. Barr, Philen, and Needham), and Epidemic Intelligence Service, Epidemiology Program Office (Drs O'Brien and Selanikio), Centers for Disease Control and Prevention, Atlanta, Ga; Ministère de la Santé Publique et de la Population (Mss Hecdivert and Louis), Institut Haitien d'Enfance (Dr Placide), Hopital de l'Universite d'Etat d'Haiti (Drs St. Victor and Denerville), and Pan American Health Organization (Dr Espindola), Port-au-Prince, Haiti; and Pan American Health Organization/Caribbean Epidemiology Centre, Port of Spain, Trinidad and Tobago (Drs Hospedales and Lewis).
Context.— Contaminated pharmaceutical products can result in substantial morbidity
and mortality and should be included in the differential diagnosis of deaths
of unknown origin.
Objective.— To investigate an outbreak of deaths among children from acute renal
failure in Haiti to determine the etiology and institute control measures.
Design.— Case-control study, cohort study, and laboratory toxicologic evaluation.
Setting.— Pediatric population of Haiti.
Participants.— Cases were defined as Haitian residents younger than 18 years with idiopathic
anuria or severe oliguria for 24 hours or longer. Febrile hospitalized children
without renal failure were enrolled as control subjects.
Main Outcome Measure.— The odds of exposure to suspected etiologic agents among cases and controls.
Results.— We identified 109 cases of acute renal failure among children. The clinical
syndrome included renal failure, hepatitis, pancreatitis, central nervous
system impairment, coma, and death. Of 87 patients with follow-up information
who remained in Haiti for treatment, 85 (98%) died; 3 (27%) of 11 patients
transported to the United States for intensive care unit management died before
hospital discharge. A locally manufactured acetaminophen syrup was highly
associated with disease (odds ratio, 52.7; 95% confidence interval, 15.2-197.2).
Diethylene glycol (DEG) was found in patients' bottles in a median concentration
of 14.4%. The median estimated toxic dose of DEG was 1.34 mL/kg (range, 0.22-4.42
mL/kg). Glycerin, a raw material imported to Haiti and used in the acetaminophen
formulation, was contaminated with 24% DEG.
Conclusions.— An epidemic of severe systemic toxicity and deaths from DEG-contaminated
acetaminophen syrup occurred in Haiti. Good manufacturing practice regulations
should be used by all pharmaceutical manufacturers to prevent such tragedies.
ACUTE RENAL failure is an unusual cause of death among children. Where
infant and child mortality rates are high, the common causes of death are
acute respiratory illness, diarrhea, dehydration, malnutrition, and sepsis.1- 3 Increased rates of death
from other causes may not be recognized in these settings unless the diseases
are unusual or the outbreaks are large. Between November 1995 and May 1996, approximately
32 children with idiopathic acute renal failure were admitted to the University
General Hospital in Port-au-Prince, Haiti, whereas no children had been admitted
with this diagnosis in the preceding 5 years; 31 of these children died. An
intensive investigation involving members of the Ministry of Health of Haiti,
the University General Hospital in Port-au-Prince (HUEH), the Pan American
Health Organization (PAHO), the Centers for Disease Control and Prevention
(CDC), the Caribbean Epidemiology Center, and other local and international
organizations was initiated in June 1996 to determine the cause of the outbreak,
institute control measures, and evaluate their effectiveness.
We defined cases as idiopathic anuria or severe oliguria for 24 hours
or longer diagnosed on or after October 1, 1995, in a Haitian resident younger
than 18 years. We evaluated potential causes of renal failure by maternal
interviews and hospital chart information. We considered a case confirmed
if either a maternal interview or a hospital chart review was completed and
the case definition was fulfilled. If the chart review or the maternal interview
suggested a diagnosis that was not supported by historical, clinical, or laboratory
findings or that lacked sufficient detail to exclude a known etiology of renal
failure, we considered the case only possible. We also considered a case possible
if neither a maternal interview nor a chart review was completed for a child
who otherwise met the case definition. Possible cases were included in the
descriptive epidemiology to provide the most sensitive evaluation of the epidemic.
Cases were identified primarily through the recall of pediatricians
at HUEH. No admission or discharge logs were available, and there was no means
to systematically search medical records. Cases also were identified through
disease surveillance lists collected by the Ministry of Health of Haiti, notifications
from private pediatricians, and community-based reporting to field health
workers. Cases were included in the case-control study if either a maternal
interview or a chart review was completed on or before July 13, 1996, regardless
of onset date of illness or admission to the hospital.
A convenience sample of children hospitalized at HUEH between June 19
and July 2, 1996, was enrolled prospectively as control subjects if they were
younger than 18 years, had been hospitalized for any reason other than renal
failure, and had a history of fever during their current illness.
We collected detailed information by maternal interview and chart review
to determine the signs and symptoms of illness in the 2 weeks before admission,
the progression of disease, and all exposures in the 2 weeks before the onset
of anuria, including medications, creams, herbal remedies, baths, and tonics.
We also collected information on demographic characteristics of each study
subject and the subject's family.
Diethylene glycol (DEG) was detected in the contaminated syrups by several
separate confirmatory analytical techniques. The sodium adduct ([M+Na]+, mass-to-charge ratio, 129) of DEG was detected in the implicated
syrups and standards, but not in a control syrup, by continuous-infusion electrospray
ionization mass spectrometry with a magnetic sector mass spectrometer (model
70-4SE, Micromass, Manchester, England) using a 0.1% acetic acid–methanol
(1:1) solvent system. These results were confirmed by accurate mass fast-atom
bombardment mass spectrometry, high-performance liquid chromatography/electrospray
ionization mass spectrometry/mass spectrometry (HPLC/ESI-MS/MS), gas chromatography–mass
spectrometry, and proton and tagged stable carbon isotope nuclear magnetic
To quantify the percentage of DEG (vol/vol) in implicated syrups, we
developed a method for internal standard calibration using isotope dilution
technique and a triple quadrupole mass spectrometer (model TSQ700, Finnigan
MAT, San Jose, Calif). Samples and standards were prepared by dissolving 100
µL of syrup or 10 µL of deuterated DEG in 4 mL of methanol and
were analyzed by HPLC/ESI-MS/MS of the ammonium adduct ([M+NH4]+) of DEG. A calibration curve was constructed by plotting the response
factor against the percentage of DEG, and the equation from a linear regression
analysis was used to quantify the percentage of DEG in the acetaminophen preparations.
The limit of detection (3 SDs) of the method was determined to be 0.7% DEG
(vol/vol) in the formulation, and the average coefficient of variation of
the method calculated from repeat measurements was 16%.
The DEG dose per kilogram of body weight for individual patients was
estimated by multiplying the percentage of DEG in bottles provided by the
child's parent by the volume missing from the bottle and dividing by the weight
of the child as documented in the hospital admission record. For children
without a documented weight, we estimated age-appropriate weights by applying
the median weight-for-age z score from children whose
weight was documented.4 For children who reportedly
shared the contaminated medication with another person, we assumed that the
affected child consumed the entire quantity missing from the bottle. This
method provides an estimate of the maximum possible dose ingested; the actual
ingested dose could have been any amount less than the calculated dose.
Between June and September 1996, we enrolled in a longitudinal, prospective
cohort study any child who presented to HUEH without renal failure who was
younger than 15 years and had a history of consuming acetaminophen syrup from
a DEG-contaminated lot. Children were included in the cohort analysis if they
could provide a medication bottle with a legible lot number or expiration
date for verification. A detailed history and physical examination were conducted
for each child. If the last dose of contaminated medication was consumed less
than 14 days before the interview, a blood sample was drawn for renal, hepatic,
pancreatic, and hematologic function tests. At least 1 visit was scheduled
at 4 weeks after the last exposure to evaluate for evolving signs or symptoms
of toxic effects. Because the DEG concentration was not measured from the
bottle of each cohort child, the median DEG concentration from patient bottles
was used along with the measured ingested volumes for each child in the cohort
to determine the ingested DEG dose. Weights of children were measured or estimated
by applying the mean weight-for-age z score of the
weighed children in the cohort to those who were not weighed.
Potential risk factors were assessed using odds ratios (ORs) and 95%
confidence intervals (CIs). Only individuals who could state the brand name
of a medication or who could provide a medication bottle were considered exposed
to that medication. Means were compared using analysis of variance, and proportions
compared using Mantel-Haenszel χ2 test for significance.
One hundred nine children were identified who fulfilled the case definition;
87 were confirmed and 22 were possible cases. Cases were classified as possible
for the following reasons: neither a maternal interview nor a chart review
(7); insufficient information to include or exclude any etiology (8); and
another diagnosis stated in the chart but historical, clinical, or laboratory
findings did not support that diagnosis (typhoid fever , hemolytic uremic
syndrome , meningitis , and dehydration ). Sixty-five patients (60%)
were male. Only 1 set of siblings was affected; none of the other case patients
was related. Cases began to present in the fall of 1995 and reached a peak
in June 1996; the last case detected occurred in July 1996 (Figure 1). Children ranged in age from 1 month to 13 years; 87 affected
children (80%) were younger than 5 years and 45 (41%) were younger than 2
years. Cases were identified from 5 of 9 geopolitical units, termed departments, within Haiti; 69% of case patients resided
in the capital city of Port-au-Prince.
Sixty-three of 109 patients and 52 controls fulfilled inclusion criteria
for the case-control analysis. The remaining 46 patients were excluded from
the case-control analysis because information from the maternal interview
or the hospital chart review was not available by the date of the analysis,
July 13, 1996, as defined in the "Case Finding" section. Patients were similar
to controls in age and sex distribution, but had consumed more medications
in the 2 weeks before admission (Table 1).
In a univariate analysis, consumption of a locally manufactured acetaminophen
syrup preparation marketed under the brand names Afebril and Valodon was highly
associated with disease. The OR for exposure to either product in the 2 weeks
preceding the date of admission was 52.7 (95% CI, 15.2-197.2). No other medication
exposure was statistically associated with disease; however, the laboratory
analysis did reveal that other medications produced by the same manufacturer
were contaminated with DEG. When exposure was defined as including any DEG-contaminated
medication (by laboratory measure or appropriate brand and lot number), the
OR for exposure was 44.2 (95% CI, 13.4-156.6).
Parents of 86 case patients could describe the symptoms that led to
the use of the contaminated acetaminophen liquid; these included fever in
81 (94%), vomiting in 29 (34%), diarrhea in 25 (29%), cough in 19 (22%), and
abdominal pain in 3 (4%). Subsequent to ingesting the medication, parents
noted various symptoms in their children, including anuria in 80 (95%), edema
in 40 (48%), abdominal pain in 23 (27%), altered consciousness in 17 (20%),
oliguria in 10 (12%), and dyspnea in 9 (10%).
Signs and symptoms observed in the hospital were inconsistently noted
in the charts; therefore, the proportion of all patients presenting with each
sign or symptom could not be established. By definition, all the case patients
had acute renal failure. One patient was in polyuric renal failure at admission
and subsequently developed anuria; all other patients had either anuria or
severe oliguria at the time of admission. Other manifestations of toxic effects
included hepatitis, pancreatitis, and severe neurological manifestations (eg,
encephalopathy, optic neuritis, minimally reactive or fixed dilated pupils,
unilateral facial paralysis, respiratory failure requiring mechanical ventilation,
and coma) (Table 2). A detailed
description of the clinical manifestations, evaluation, and management of
these children is the subject of a separate report.
Of 98 children who remained in Haiti throughout their illness, 11 were
in extremis when taken from the hospital by their families and were lost to
follow-up. Of the remaining 87 children, 85 (98%) died. Eleven children were
transported to the United States for intensive care management, including
hemodialysis or peritoneal dialysis; of these, 1 died during air transport
and 2 died during hospitalization. Of the 8 surviving children, 7 recovered
full renal function; 1 child developed chronic renal failure requiring chronic
ambulatory peritoneal dialysis. The 11 children transported to the United
States were not different from the 98 who remained in Haiti with respect to
age, dose of DEG ingested, or interval between ingestion and presentation
to the hospital. They were selected based only on the availability and logistics
of transportation to the United States.
Samples of the acetaminophen syrup were analyzed to determine the toxic
substance contained therein. Diethylene glycol, a known human toxicant, was
identified in bottles from patients, unopened bottles purchased in pharmacies,
and retained quality control samples from the manufacturer. Among 36 contaminated
patient bottles tested, the median DEG concentration was 14.4% (range, 1.2%-19.6%).
Among 32 patients for whom a maximum possible ingested dose could be estimated,
the median estimated DEG dose consumed was 1.34 mL/kg (range, 0.22-4.42 mL/kg);
12 children (37.5%) consumed an estimated maximum DEG dose less than 1.0 mL/kg.
A detailed medication history or bottle of medication was available
for 82 of 109 affected children. A bottle of the acetaminophen syrup was collected
from 50 of these, and an additional 31 had exposure documented by history.
Only 1 child had no history of exposure to either medication. Among 54 children
for whom the timing of acetaminophen syrup consumption was clearly given,
the median time from the first dose to onset of oliguria or anuria was 6 days
(range, 1-12 days). Of 7 children who had a definite history of stopping the
medication before the onset of oliguria or anuria, the median interval between
stopping and noting a change in urinary frequency was 4 days (range, 2-8 days).
Although other medications manufactured by the same company were also contaminated
with DEG, only 1 patient consumed any other contaminated medication. The medication
contained only 1.2% DEG, and this patient also consumed the acetaminophen
syrup with a DEG concentration of 17.2%. Among control subjects, only 2 consumed
any medications contaminated with DEG. An 8-month-old control subject consumed
an estimated DEG dose of 0.4 mL/kg in the form of acetaminophen drops contaminated
with 5% DEG, and an 11-month-old control subject consumed an iron supplement
produced by the manufacturer that was contaminated with an unknown proportion
Forty-nine well children who ingested a DEG-contaminated lot of acetaminophen
syrup were enrolled in a prospective cohort study and followed up for a median
of 87 days (range, 19-175 days) after the last dose of acetaminophen syrup.
If the date of last ingestion was unknown, the follow-up period began with
the first clinical visit. The median age of children was 45 months (range,
1-154 months), and the median ingested DEG dose among 17 children was 0.67
mL/kg (0.05-2.48 mL/kg). All 49 children survived through the follow-up period,
and none developed overt signs or symptoms of DEG toxicity; however, numerous
children had laboratory evidence of subclinical toxic effects (Table 3).
One hundred thirty-eight children from both the case-control study and
the cohort group who consumed acetaminophen syrup from a DEG-contaminated
lot were included in an analysis of risk factors for disease. The mean age
of children who became ill was 39.5 months compared with 52.1 months among
children who did not (P=.05). Cases were more likely
to have taken the acetaminophen syrup for symptoms of diarrhea (OR, 4.3; 95%
CI, 1.7-10.9) or vomiting (OR, 8.8; 95% CI, 3.7-21.8) than were children who
did not develop symptoms of DEG toxicity. Among 49 individuals for whom we
could estimate the DEG dose ingested per kilogram, children who became ill
consumed a mean dose of 1.34 mL/kg, compared with 0.84 mL/kg among those who
did not (P=.04); however, there was considerable
overlap in the range of doses ingested (Figure
A traceback investigation at the manufacturer revealed that glycerin,
used in the formulation of these syrups, was contaminated with 24% DEG. The
glycerin had been imported to Haiti through distributors in Europe from a
manufacturer in China. It is unknown how and at which point the contamination
occurred or if other countries received DEG-contaminated glycerin.
The manufacturer's formulation of acetaminophen syrup calls for 50%
glycerin by volume, resulting in a predicted DEG concentration of 12% in the
final product. The DEG-contaminated glycerin was used to produce 3 lots of
the acetaminophen syrup for an estimated total volume of 1900 L of syrup;
these lots were packaged into 12000 to 15000125-mL bottles of DEG-contaminated
syrup with acetaminophen syrup. Based on the manufacturer's production records,
we estimate that 38 lots of 15 other liquid preparations were also prepared
over the 4-month period when the DEG-contaminated glycerin was in use. The
concentration of glycerin in these 15 products ranged from 0.5% to 53%, resulting
in predicted DEG concentrations of the marketed products from 0.1% to 12.7%.
Infrared spectroscopy is one in a panel of tests recommended in the US Pharmacopeia to characterize glycerin. Infrared spectroscopic
tests conducted by the US Food and Drug Administration (FDA) on samples of
DEG-contaminated glycerin failed, when used alone, to detect the DEG. Application
of the entire test panel recommended by the US Pharmacopeia, however, would have detected the contamination. Both gas chromatography
and mass spectrometry alone were able to distinguish the glycerin as contaminated
On June 22, 1996, the Haitian Ministry of Health issued a public health
warning about the association of the acetaminophen syrup with acute renal
failure and prohibited the sale of the 2 products, Afebril and Valodon. Public
information campaigns were conducted through radio, television, newspapers,
flyers to schoolchildren, and notices to medical societies. The manufacturer
issued a recall of all liquid medications that it produced. In the week following
the public announcement, 7 additional children were identified with acute
renal failure; all had consumed the acetaminophen syrup prior to the announcement.
Only 3 additional cases were identified in the ensuing weeks (Figure 1).
This investigation identified DEG-contaminated acetaminophen syrup as
the cause of a large outbreak of acute renal failure deaths among children
in Haiti.5 Diethylene glycol, a known human
toxicant, contaminated a shipment of glycerin imported to Haiti from China
through Europe. The glycerin was used in numerous locally manufactured liquid
pharmaceutical products, including acetaminophen syrup marketed under 2 brand
names, Afebril and Valodon; consumption of either of these products was highly
associated with disease. Withdrawal of these products from the market resulted
in an abrupt cessation of cases.
Although 17 liquid medications were likely contaminated with DEG, only
DEG-contaminated acetaminophen syrup was epidemiologically associated with
illness. There may be several reasons for this finding. Acetaminophen may
provide an additive or potentiating effect for DEG toxicity since both are
toxic to the liver and may alter the metabolism of DEG. Some of the children
for whom we did not identify an exposure to either proprietary formulation
of acetaminophen syrup may have been exposed to other DEG-contaminated medications;
however, we were unable to document this by bottle collection or by the patient's
history. The DEG concentration in the acetaminophen syrup was greater than
that of 14 other products and equivalent to the concentration in the remaining
2 contaminated products (polyvitamin and iron preparations), which would likely
have been consumed in smaller amounts than acetaminophen. Finally, children
who developed clinical toxic effects were more likely than those without clinical
toxic effects to have taken the acetaminophen syrup for diarrhea or vomiting.
Preceding gastroenteritis may contribute to DEG absorption or enhance toxic
effects through other means.
Although acute anuric renal failure was recognized as the hallmark sign
of DEG ingestion, exposure resulted in severe systemic toxic effects. Therefore,
dialysis alone would likely have been insufficient to avert mortality. Because
renal failure from DEG is usually reversible, children may survive this toxic
effect when provided with multisystem intensive care support, including mechanical
ventilation, parenteral nutritional, and fluid management, in addition to
Through a cohort study of DEG-exposed children, we identified individuals
who had laboratory-based evidence of toxic effects but no clinical findings.
Although the cohort study involved a limited number of patients, the doses
ingested by these children were in a similar range as those ingested by the
children who experienced DEG toxicity, indicating that there are other risk
factors or significant individual variation that would predict the development
of disease. This observation agrees with the findings of the only other cohort
study of exposed individuals, which also noted that the fatal and nonfatal
dose ranges overlapped considerably.6 Age may
be 1 risk factor since it was independently associated with disease. However,
because the children in the cohort study were a self-selected rather than
a randomly selected group of all exposed children, a selection bias may have
The mechanism of DEG toxicity in humans is not well characterized, and
minimum toxic dose ranges have not been well established. A DEG dose of 1
mL/kg has been suggested as the minimum toxic dose7;
however, there is little evidence to support this. This study documented that
toxic doses are often less than 1 mL/kg; therefore, clinical outcome should
not be predicted based on this cutoff value. The intervals from the first
DEG exposure and the last DEG exposure to onset of illness indicate that,
if disease is going to occur, it will present within a short time after exposure.
The public health impact of a rapid investigation and resultant interventions
described in this report emphasizes the importance of disease surveillance
and early identification of outbreaks. From the company's manufacturing records,
visits to randomly selected pharmacies, and an inventory of the acetaminophen
syrup bottles turned over by pharmacies to the national police, we estimate
that at least 15000 bottles of DEG-contaminated acetaminophen syrup were produced,
and 40% of bottles on the market at the time of the investigation were from
contaminated lots. Approximately 60% to 70% of the contaminated lots probably
had already been sold; therefore, the 109 identified cases represented 60%
to 70% of the expected cases. By these estimates, between 45 and 75 additional
cases of DEG toxicity are estimated to have been averted by the intervention.
Although this outbreak was an unusual occurrence, it is not unique and
could occur again. Poisoning with DEG has most frequently been observed in
outbreak situations rather than as sporadic cases and has usually been observed
in association with contamination of ingestible pharmaceutical products (Table 4).8- 14
The first outbreak described was the Massengill disaster of 1937 in the United
States, when DEG was used as the excipient in a liquid sulfanilamide preparation.6,15 At that time DEG was not known to
be highly toxic to humans, and no safety evaluations were conducted before
marketing the product. One hundred five deaths, predominantly among adults,
resulted from ingestion. Recent outbreaks of acute renal failure have been
caused by DEG-contaminated acetaminophen liquid preparations in Nigeria and
Bangladesh.12,13 In both settings,
the contaminated raw material was propylene glycol. Glycerin contaminated
with DEG has been implicated in only 1 other outbreak; in 1986 orally administered
DEG-contaminated glycerin was used in an Indian hospital for control of intracranial
or intraocular pressure and resulted in 14 deaths.11
This outbreak highlights the challenges in developing countries where
there may not be adequate regulation, enforcement, or strict implementation
of current good manufacturing practice regulations in the pharmaceutical sector.
Not only should strict quality control procedures be required in all countries
where pharmaceutical products are manufactured, but these procedures must
be consistently and fully applied, otherwise an outbreak such as this could
occur even in countries where quality control procedures are usually strictly
applied. The Haitian government and local industry representatives are working
to propose and enact legislation and industry norms to avert such a disaster
in the future. As noted, infrared spectroscopy alone may not always identify
DEG contamination of glycerin. For this reason, it is important that manufacturers
verify the test results on certificates of analysis received from their suppliers.
Special problems exist in developing countries where the technology
to apply US Pharmacopeia standards may not be available
or may be too costly to use. To address this problem, the FDA and the World
Health Organization (WHO) have worked to develop and adapt a low-cost, safe,
simple-to-use thin-layer chromatography technique appropriate for identification
of ethylene and DEG in raw materials such as glycerin and in finished liquid
products.16 The technique can detect DEG to
a concentration of less than 1% and is undergoing field testing. In November
1996, a working group including representatives from FDA, WHO, PAHO, CDC,
and the pharmaceutical industry met to discuss this and other strategies aimed
at preventing disease outbreaks from contaminated pharmaceutical products
(M. McGeehin, MD, oral communication, February 8, 1998). It is likely that
disasters such as these will continue to occur until strict quality control
procedures are used consistently by all pharmaceutical manufacturers and until
countries around the world adopt and enforce regulations that ensure the safety
of pharmaceutical products.