ORIGINAL RESEARCH

Opioid and Benzodiazepine Withdrawal Syndrome in

PICU Patients: Which Risk Factors Matter?
Paulo Sérgio Lucas da Silva, MD, MSc, Maria Eunice Reis, MD
Thais Suelotto Machado Fonseca, Med Student, and
Marcelo Cunio Machado Fonseca, MD, PhD

Key Words: benzodiazepine, fentanyl, midazolam, opiate, pediatric

Background and Aims: Although iatrogenic withdrawal syndrome
intensive care, withdrawal syndrome
(IWS) has been recognized in patients exposed to opioids and
benzodiazepines, very few studies have used a validated tool for (J Addict Med 2016;10: 110–116)
diagnosis and assessment of IWS in critically ill children. We sought
to determine IWS rate, risk factors, and outcomes of IWS patients.
INTRODUCTION
Methods: Prospective observational study conducted in a pediatric
intensive care unit. A total of 137 patients (31 with IWS and 106 with
no IWS) received a continuous infusion of fentanyl and midazolam
A nalgesic and sedative agents are frequently used in
pediatric intensive care units (PICUs) to provide con-
tinuous level of comfort to critically ill patients. A consensus
for 3 or more days. The Sophia Observation withdrawal Symptoms guideline on sedation has recommended the combination of
scale was repeatedly applied when children were weaned off seda- midazolam and fentanyl in continuous infusion as first choice
tion/analgesia. for PICU patients for sedation/analgesia (Playfor et al., 2006).
Results: The overall incidence of IWS was 22.6%. Of the 31 IWS However, the use of these drugs for more than a few days may
patients, 6 showed IWS with less than 5 days sedation or analgesia. lead to tolerance and physical dependence.
Logistic regression showed that the median peak dose of midazolam The reported incidence of iatrogenic withdrawal syn-
was associated with IWS development (odds ratio 1.4). Receiver- drome (IWS) in PICU patients receiving benzodiazepines
operating curve showed a cut-off value of 0.35 mg/kg/h for mid- ranges from 17% to 35% (Hughes et al., 1994; Fonsmark
azolam peak dose (sensitivity 96.7%, specificity 51%, positive et al., 1999; Dominguez et al., 2006), 57% in patients on
predictive value 36.6%, and negative predictive value 98.2%), with fentanyl (Arnold et al., 1990; Katz et al., 1994), and 49% to
area under the curve of 0.80. IWS patients had a longer time on 77% in patients getting both drugs (Franck et al., 2008; Franck
mechanical ventilation, prolonged pediatric intensive care unit, and et al., 2012; Ista et al., 2013; Amigoni et al., 2014).
hospital stays, and required prolonged period to have drugs discon- Importantly, most of our knowledge about iatrogenic
tinued. withdrawal in PICU patients is derived from researches whose
Conclusions: Although length of sedation/analgesia for at least5 assessment instruments were only validated for neonates
days has been widely proposed for monitoring IWS, our data suggest (Finnegan score) (Fernandez-Carrion et al., 2013), clinical
that initiating monitoring after 3 sedation days is highly recom- judgments (Dominguez et al., 2006), or other nonvalidated
mended. In addition, patients requiring infusion rates of midazolam
tools (Siddappa et al., 2003). There are 4 published instru-
ments for assessing withdrawal symptoms in children,
above 0.35 mg/kg/h should be considered at high risk for IWS.
namely: Sedation Withdrawal Score (SWS) (Cunliffe et al.,
2004), Opioid and Benzodiazepine Withdrawal Scale
(OBWS) (Franck et al., 2004), Withdrawal Assessment Tool
version 1 (WAT-1) (Franck et al., 2008), and Sophia Obser-
From the Pediatric Intensive Care Unit, Department of Pediatrics, Hospital do
Servidor Público Municipal, São Paulo, Brazil (PSLS); Hospital e Mater-
vation withdrawal Symptoms-scale (SOS) (Ista et al., 2009).
nidade Santa Joana, São Paulo, Brazil (MER); Faculdade de Medicina do Yet, only WAT-1 and the SOS have shown reliability and
ABC, Santo André, Brazil (TSMF); Health Technologies Assessment validity in detecting withdrawal symptoms.
Center – Federal University of São Paulo, São Paulo, Brazil (MCMF). Nevertheless, most studies about opioid and benzo-
Received for publication October 23, 2015; accepted January 5, 2016. diazepine IWS focus on characterizing symptoms in the
This study received no financial support, including any departmental funds.
The authors have not disclosed any potential conflict of interest. pediatric population, on developing screening and assessment
Send correspondence and reprint requests to Paulo Sérgio Lucas da Silva, tools, or testing treatment regimens. A recent systematic
Department of Pediatrics, Pediatric Intensive Care Unit, Hospital do review (Best et al., 2015) found few prospective studies
Servidor Público Municipal, Rua Castro Alves, 60, São Paulo 01532- providing data about opioid and/or benzodiazepine-related
900, Brazil. E-mail: [Link]@[Link].
Copyright ß 2016 American Society of Addiction Medicine
IWS risk factors. Moreover, a scarce number of studies used a
ISSN: 1932-0620/15/0901-0031 validated monitoring tool to diagnose IWS in critically ill
DOI: 10.1097/ADM.0000000000000197 children (Franck et al., 2012; Fisher et al., 2013; Ista et al.,

110 J Addict Med  Volume 10, Number 2, March/April 2016

Copyright © 2016 American Society of Addiction Medicine. Unauthorized reproduction of this article is prohibited.
J Addict Med  Volume 10, Number 2, March/April 2016 IWS Syndrome and Risk Factors

2013; Amigoni et al., 2014). Therefore, studies applying hours. Nonopioid analgesics were routinely given to decrease
validated IWS assessment tools (WAT-1 or SOS) are needed the amount of opioids administered (Playfor et al., 2006).
to identify which specific factors predispose PICU patients to The study drugs (midazolam and fentanyl) were started
develop IWS to help clinicians to prevent it (Best et al., 2015). immediately after the onset of mechanical ventilation. Level
The primary objective of this study was to evaluate the of sedation and analgesia were assessed by nurses every 4
incidence of withdrawal syndrome in a general pediatric hours using the COMFORT behavior scale (COMFORT-b,
PICU population applying a validated tool, the SOS scale. ranging from 6 to 30) (Ista et al., 2005) and the FLACC (Face,
Secondarily, we assessed the risk factors for IWS develop- Legs, Activity, Cry, Consolability – ranging from 0 to 10)
ment and patients’ outcome. scale (Merkel et al., 1997), respectively. Patients had the drugs
individually diluted with an infusion rate of 1 mL/h, corre-
METHODS sponding to 0.2 mg/kg/h midazolam and 2 mg/kg/h fentanyl,
respectively. Medications were started with an initial infusion
Study Design and Setting rate of 0.5 mL/h. Subsequently, the infusion rate was adjusted
This was a prospective observational study, conducted in steps of 0.5 mL/h to achieve and maintain a COMFORT-b
over 5 years (from January 2012 through December 2014) in score between 11 and 22 (Ista et al., 2005) and a FLACC scale
an 8-bedded PICU of a tertiary hospital. The PICU is not below 4 (Merkel et al., 1997). A COMFORT-b score of less
designed to handle specialist congenital heart surgery or burn than 11 implied oversedation, whereas a score greater than 22
cases, and bone marrow and solid organ transplants are not defined undersedation (Ista et al., 2005). If COMFORT-b
performed in the hospital. The Ethics Committee of the score was greater than 22 for 2 consecutive hours or in cases
hospital approved the study and waived the need for written of urgent need to avoid accidental extubation or central
informed consent. venous line dislocation, boluses of midazolam or fentanyl
equivalent to the hourly drug infusion amount were given
Inclusion and Exclusion Criteria alternately. Likewise, patients received supplementary doses
All consecutive patients aging between 1 month and 16 of fentanyl when FLACC scale was at least 4. Intravenous
years, requiring mechanical ventilation for at least 72 hours bolus doses could also be given before an anticipated noxious
and exclusively receiving fentanyl and midazolam as sedative/ stimulation such as chest physiotherapy, suctioning, or other
analgesic agents, were eligible for enrollment. Patients were procedures, to reduce patient–ventilator dyssynchrony and
not included if they had severe neurological injury or central for those patients who required intrahospital transport to the
nervous system impairment that could affect the assessment of radiology department.
the sedation level. Also, patients were excluded if they When the sedation score pointed to an increase in
received a neuromuscular blocking agent or any other sedative medication, the attending physician increased the infusion
drug (eg, ketamine, dexmedetomedine, clonidine, thiopental, rate of benzodiazepine or opioid, alternately, in increments of
thionembutal) during the study period, died, or were moved to 0.5 mL/h (midazolam: 0.1 mg/kg/h; fentanyl: 1 mg/kg/h). If
another PICU while receiving sedoanalgesia. Daily sedation analgesia and sedation were insufficient despite the maximum
interruption (DSI) is not a standard of care in our PICU. allowed midazolam (0.6 mg/kg/h) and fentanyl (10 mg/kg/h)
dosage (Playfor et al., 2006), the child was excluded from the
Sedation Protocol and Data Collection study and treated with other sedative/analgesic drugs (chloral
All mechanically ventilated patients received sedation hydrate or ketamine). On the contrary, when the COMFORT-b
and analgesia as previously recommended by the United score was less than 11, infusion rates were reduced in dec-
Kingdom Paediatric Intensive Care Society Sedation, Anal- rements of 0.5 mL/h (midazolam: 0.1 mg/kg/h; fentanyl: 1 mg/
gesia and Neuromuscular Blockade Working Group (Playfor kg/h). First, the opioid infusion was reduced and, after 15
et al., 2006). This multidisciplinary expert panel produced a minutes, if the COMFORT-b score still remained less than 11,
set of consensus guidelines comprising 20 key recommen- we decreased midazolam’s infusion rate. Patients were kept
dations to provide analgesia and sedation for critically ill within the desired sedation level (COMFORT-b score 11
children in the following areas: nonpharmacological inter- and 22) while on weaning from mechanical ventilation.
ventions (environmental factors, relaxation, distraction, pro- When the clinical condition had improved enough to judge
motion of sleep, and day–night orientation), pain assessment whether the patient was ready for extubation, we discontinued
and analgesic management, sedation assessment, sedative the drug infusion.
agents commonly used in the PICU, dosage, and route of The start of the weaning period was on the day that the
administration of analgesics and sedative, and withdrawal infusions of midazolam and fentanyl were gradually tapered
syndrome assessment prevention and management. Postop- and switched to lorazepam and methadone enterally. Con-
erative pain management initially included the use of sequently, the preweaning period was the period of midazo-
analgesics (nonsteroidal antiinflammatory drugs or acetami- lam and/or fentanyl treatment before weaning. Children were
nophen) for background pain on a planned intermittent basis. weaned from opioids by reducing the starting dose by10% to
Preemptive nonpharmacologic interventions (eg, consolation, 20% every 24 hours.
posture change, heat or cold-pack, or diaper change) and/or The demographic and clinical data included age, sex,
additional analgesia were provided for breakthrough pain. weight, severity of illness scores within 12 hours of PICU
Systemic analgesia with opioid was further administered if admission (PRISM, Pollack et al., 1988) and Pediatric Logis-
score assessment still suggested pain after 2 consecutive tic Organ Dysfunction (PELOD, Leteurtre et al., 2003),

ß 2016 American Society of Addiction Medicine 111

Copyright © 2016 American Society of Addiction Medicine. Unauthorized reproduction of this article is prohibited.
da Silva et al. J Addict Med  Volume 10, Number 2, March/April 2016

diagnosis, duration of mechanical ventilation, length of PICU for IWS. Variables clinically relevant and associated
and hospital stay, total duration of sedation, and cumulative (P < 0.10) with IWS in univariate analysis were used in the
and peak daily drugs dosage. All data were collected by a model. In the multivariable model, independent variables
single investigator (PSL). were eliminated from the highest to the lowest P value, but
Within 24 hours after tapering-off or cessation of mid- remained in the model if their P value was less than 0.05. Odds
azolam and/or opioids, all mechanically ventilated patients ratios (ORs) were estimated along with their 95% confidence
were assessed twice a day at 8 AM and 8 PM for IWS symptoms intervals (CIs). Finally, to find the cut-off value(s) for the risk
using the SOS-scale (SOS 4) (Ista et al., 2009). If clinically factor(s) identified in the logistic regression model, we per-
indicated, they were evaluated more frequently, for instance, formed a receiver-operating characteristic (ROC) analysis. To
every 4 hours, as in patients weaned off benzodiazepine/ assess discrimination, the ROC curve and its associated area
opioid after 3 or more days of continuous infusion and under the curve (AUC) were calculated. An AUC value of 0.5
until 72 hours after the last dose of midazolam or fentanyl. indicates no ability to discriminate, whereas an AUC between
The highest daily SOS score was used in the analyses. 0.70 and 0.79 represents an acceptable discrimination; a value
To describe analgesic and sedative administration in the of 0.8 is considered good, whereas 1 shows perfect discrimi-
PICU we calculated fentanyl and midazolam cumulative dose, nation (Fletcher et al., 2014). We used SPSS 16.0 statistical
mean daily dose, and peak daily doses from data retrieved packages (SPSS Inc., Chicago, IL) for all data entry and
from the clinical record of each patient. The cumulative dose analysis. Differences were considered statistically significant
was calculated by summing the daily dose of each agent over at P value less than 0.05.
the entire time at which the patient received intravenous
midazolam/fentanyl infusion. Unscheduled (pro re nata) RESULTS
doses of midazolam/fentanyl that were administered were During the study period, 257 mechanically ventilated
also included in the total daily benzodiazepine/opioid doses. patients were admitted to our PICU and were followed up for
Mean daily dose was calculated by dividing the cumulative the full duration of their admission (Fig. 1).
dose by the number of days that the specific drug was Nurses’ interobserver variability (Cohen kappa) were
administered to the patient. Peak dose of each drug was 0.77 (95% CI, 0.75; 0.79) and 0.86 (95% CI, 0.78–0.92) for
the highest dose administered in a 24-hour period, beginning SOS and FLACC scale, respectively.
at 07:00 hours. The drug therapy duration (d) was the number
of PICU days that a patient was treated with fentanyl and Primary Outcome
midazolam. The overall incidence of IWS symptoms was 22.6%. The
Tolerance was defined as the need of doubling, on day 2, characteristics of these patients are summarized in Table 1.
the initial dose, to achieve the same pharmacological effects Whereas more children older than 1 year developed IWS
observed at the therapy initiation (Anand et al., 2013). symptoms, there was no difference in the incidence of IWS
when older versus younger than 1-year patients were compared
Outcome Measures (22.9% vs 22%; P ¼ 0.894). Patients’ sex, admission diagnosis,
The primary outcome was the incidence of IWS. The and severity of illness scores were not associated with presence
secondary outcome measures were length of PICU stay, the of IWS.
duration of hospital stay, number of patients developing
tolerance, and time elapsed from sedation initiation to Secondary Outcomes
tolerance development. The IWS patients stayed 2 times longer on mechanical
ventilation (MV) (9 vs 5 d; P < 0.001), had longer PICU and
Statistical Analysis hospital stays, and required a prolonged period to have drugs
The enrollment goal was of 100 participants. A con- discontinued (Table 1). Although the median time elapsed for
servative assumption of an overall incidence of 30% IWS experiencing tolerance was similar between both groups, a
(Amigoni et al., 2014) was used to determine the sample size. higher proportion of IWS patients developed tolerance
Descriptive statistics were performed for all variables. Con- during treatment.
tinuous variables were reported as medians with interquartile
ranges (IQRs). For univariate analyses, the Mann-Whitney U Risk Factors
test was used for 2-group comparisons of continuous data and The median duration of therapy (Table 1) was longer in
the chi-square and Fisher exact test for nominal level data. IWS patients, and the median cumulative dose, peak dose, and
The median difference and 95% confidence interval (CI) were average daily doses of both fentanyl and midazolam were
calculated when appropriate. significantly higher for IWS patients (Table 2). Of the 31 IWS
The agreement between observers was performed using patients, 6 received continuous sedation/analgesia for less
unweighted Cohen kappa coefficient for sedation scores. than 5 days.
Kappa values greater than 0.75 are taken to represent excellent Only midazolam’s peak dose (OR 1.4) was associated
agreement, below 0.40 poor agreement, and values between with development of IWS in the logistic regression analysis
0.40 and 0.75 fair to good agreement (McHugh, 2012). A (Table 3). Finally, the ROC curve (Fig. 2) plotted for mid-
Cohen kappa below 0.65 was considered unsatisfactory azolam’s peak dose showed a cut-off value of 0.35 mg/kg/h,
(McHugh, 2012). A multivariable analysis using logistic with a sensitivity of 96.7% (95% CI, 83.3%–99.9%), speci-
regression was performed to identify independent risk factors ficity of 51% (95% CI, 41.1%–60.8%), a positive predictive

112 ß 2016 American Society of Addiction Medicine

Copyright © 2016 American Society of Addiction Medicine. Unauthorized reproduction of this article is prohibited.
J Addict Med  Volume 10, Number 2, March/April 2016 IWS Syndrome and Risk Factors

FIGURE 1. Study flow chart.

value of 36.6% (95% CI, 26.2%–48%), a negative predictive children receiving exclusively midazolam and fentanyl as a
value of 98.2% (CI 95%, 90.2%–100%), and an AUC of 0.80 sedation regimen. Second, 20% of the patients developed IWS
(95% CI, 0.72–0.86) for the occurrence IWS. despite of a short period (<5 d) receiving continuous infusion
of benzodiazepine/opioids. This diagnosis was only possible
DISCUSSION by monitoring withdrawal symptoms within 24 hours after the
This study yielded several major findings. First, the beginning of the tapering-off or discontinuation of midazolam
incidence of IWS was 22.6% in mechanically ventilated and/or fentanyl. Finally, our exploratory analyses of risk

TABLE 1. Characteristics and Demographic Data of Study Patients

Variable Overall (N ¼ 137) No IWS (n ¼ 106) IWS (n ¼ 31) Median Difference (95% CI) P
Age, n (%)
1–12 mo 87 (63.5) 67 (63.2) 20 (64.5) 0.01 (0.20; 0.17) 0.89
1–4 y 21 (15.3) 12 (11.3) 9 (29) 0.17 (0.34; 0.00) 0.04
5–16 y 29 (21.1) 27 (25.4) 2 (6.4) 0.19 (0.07; 0.31) 0.02y
Weight, kg 7.5 (4.2–14.85) 7.15 (3.78–16) 8.3 (6–12.2) 1.4 (2.99; 0.90) 0.14z
Male, n (%) 84 (61.3) 65 (61.3) 21 (67.7) 0.06 (0.25; 0.12) 0.50
PRISM 13 (10–16) 13 (9–15.5) 13 (11–16) 1 (3; 1) 0.45z
PELOD 1 (1–3) 1 (1–3) 1 (1–2) 0 (0; 1) 0.35z
Admission diagnosis
Respiratory failure, n (%) 86 (62.7) 63 (59.4) 23 (74.1) 0.14 (0.32; 0.03) 0.10
Sepsis, n (%) 23 (16.8) 20 (18.8) 3 (9.6) 0.09 (0.03; 0.21) 0.28y
Trauma, n (%) 13 (9.5) 9 (8.5) 4 (12.9) 0.04 (0.17; 0.08) 0.49y
Postsurgical, n (%) 9 (6.5) 8 (7.5) 1 (3.2) 0.04 (0.03; 0.12) 0.68y
Others, n (%) 7 (5.1) 7 (6.6) 0 (0) 0.06 (0.01; 0.11) 0.35y
Time on sedation, d 5 (4–8.5) 5 (3.75–7) 8 (5–12) 3 (4; 1) <0.001z
SOS 1 (0–3) 0 (0–2) 5 (4–6) 4 (5; 4) <0.001z
Tolerance, n (%) 54 (39.4) 34 (32) 20 (64.5) 0.32 (0.51; 0.13) 0.001
Time for developing tolerance, d 3 (3–5) 3 (3–5) 3 (2.5–4) 0 (1; 1) 0.58z
Weaning period, d 4 (0–7.5) 4 (0–6) 7 (4–11) 4 (6; 1) <0.001z
Mechanical ventilation, d 7 (4–10) 5 (4–9) 9 (6–12) 3 (4; 0.99) <0.001z
Hospital stay, d 21 (14–30) 20 (13–29) 25 (19–41) 7 (12, 2) 0.005z
PICU stay, d 12 (7–16) 10 (6–15.25) 15 (11–20) 4.5 (1.99; 7) 0.001z
Data expressed as median (interquartile range) or number (%).

Chi-square test.
yFisher exact test.
zMann-Whitney U test.
PRISM, pediatric risk of mortality; PELOD, pediatric logistic organ dysfunction.

ß 2016 American Society of Addiction Medicine 113

Copyright © 2016 American Society of Addiction Medicine. Unauthorized reproduction of this article is prohibited.
da Silva et al. J Addict Med  Volume 10, Number 2, March/April 2016

TABLE 2. Characteristics of Fentanyl and Midazolam Treatment

Overall (N ¼ 137) No IWS (n ¼ 106) IWS (n ¼ 31) Median Difference (95% CI) P
Fentanyl
Total cumulative dose, mg/kg 0.25 (0.14–0.49) 0.19 (0.12–0.36) 0.52 (0.29–0.95) 0.27 (0.42; 0.15) <0.001
Average daily doses, mg/kg/h 2 (1.4–2.9) 1.77 (1.3–2.5) 2.8 (2.2–3.5) 0.94 (1.39; 0.54) <0.001
Peak infusion rate, mg/kg/h 3 (2–4) 3 (2–4) 4 (4–5) 1.5 (2; 1) <0.001
Additional doses, n 0 (0–2) 0 (0–1) 1 (0–2) 0.0 (1; 0) 0.019
Duration (including taper), d 7 (4–10) 5 (4–9) 9 (6–12) 3 (4; 0.99) <0.001
Duration before weaning, d 4 (3–7.5) 6 (2.75–6) 7 (4–11) 3 (4; 1) 0.002
Duration of weaning, d 4 (0–10) 2 (0–7.25) 7 (7–14) 5 (7; 4) <0.001
Midazolam
Total cumulative dose, mg/kg 31.2 (19–62.3) 28.8 (15.75–46.2) 70.4 (41.7–106.2) 35.05 (51.41; 21.4) <0.001
Average daily doses, mg/kg/h 0.26 (0.17–0.33) 0.2 (0.16–0.31) 0.32 (0.29–0.55) 0.12 (0.15; 0.08) <0.001
Peak infusion rate (mg/kg/h) 0.4 (0.2–0.5) 0.33 (0.2–0.4) 0.5 (0.4–0.6) 0.2 (0.2; 0.1) <0.001
Additional doses, n 1 (0–3) 1 (0–2) 1 (0–3) 0.0 (1; 0) 0.488
Duration (including taper), d 6 (4–9) 5.5 (4–8) 9 (6–12) 3 (5; 2) <0.001
Duration before weaning, d 6 (4–9) 5 (3.75–7) 9 (5–12) 3 (4.99; 2) <0.001
Duration of weaning, d 4 (1–9.5) 2.5 (1–7) 7 (7–14) 5 (7; 3) <0.001
Data expressed as median (interquartile range).

Mann-Whitney U test.

factors indicated that midazolam’s peak dose was the main Several studies have shown that higher cumulative dose
risk factor associated with IWS, and we were also able to show and longer duration of therapy influence the development of
that an infusion rate greater than 0.35 mg/kg/h of midazolam withdrawal symptoms with less evidence in relation to age
had a good performance to predict high-risk patients for IWS. and illness severity (Best et al., 2015).
The IWS rate found in our study is lower than those Duration of opioid receptor occupancy seems to be an
reported in studies using validated assessment tools (37%– important mechanism in developing tolerance (Anand et al.,
77%) (Franck et al., 2012; Fisher et al., 2013; Ista et al., 2013; 2010). In our study, we found that IWS patients spent sig-
Amigoni et al., 2014). However, we should point out that this nificantly more time receiving sedation/analgesia than
incidence is for a specific sedation/analgesia regimen with patients without IWS symptoms (8 vs 5 d). Likewise, they
midazolam and fentanyl, and that studies evaluating the were more likely to develop tolerance during their treatment.
incidence of withdrawal syndromes in general PICU popu- These findings corroborate previous studies demonstrating
lation are scarce and lack homogenization concerning the that IWS patients had a sedation time ranging between 5 days
sedation regimens, making it difficult to accurately estimate (Franck et al., 2012) and 8 days (Ista et al., 2013). Importantly,
IWS incidence and even compare the results. studies linking the duration of opioid therapy and IWS have
Although we have applied a weaning-off sedation pro- found a threshold of =5 days as predictive of IWS (Best et al.,
tocol, we could not prevent development of IWS in our 2015) and as a result there is a current recommendation to use
patients. In fact, even with a weaning protocol, rates of tapering regimen for those patients exposed to infusions of
withdrawal have been reported from 5% to 87% (Best opioids/benzodiazepines longer than 5 days (Fisher et al.,
et al., 2015), whereas IWS develops in 5% to 33% of patients 2013). Nevertheless, our finding revealing that 20% of
despite a prophylactic therapy approach (Best et al., 2015). To patients with IWS received sedation/analgesia infusion
our knowledge, a validated opioid or benzodiazepine weaning between 3 and 5 days highlights the importance of early
regimen does not exist, and weaning is largely part of the IWS monitoring to promptly start symptomatic treatment.
‘‘state of art.’’ Therefore, the questions as how to best prevent Whereas several studies have focused on the association
and treat pediatric benzodiazepines/opioids withdrawal between the prescribed opioid dose and IWS risk in neonates
remain unanswered. (Arnold et al. 1990; Franck et al., 1998; Lugo et al., 2001;

TABLE 3. Multivariate Analysis (Determinants Associated With Iatrogenic Withdrawal Syndrome)

Variable B SE Wald P Exp(B)
Age 0.001 0.007 0.031 0.860 1.001
PRISM 0.069 0.048 2.101 0.147 1.071
Time on sedation 0.122 0.105 1.363 0.243 1.130
Tolerance 0.342 0.624 0.300 0.584 1.408
Respiratory disease 1.066 0.605 3.102 0.078 0.344
Cumulative dose of fentanyl 1.169 1.560 0.561 0.454 0.311
Cumulative dose of midazolam 0.011 0.010 1.113 0.291 1.011
Peak of fentanyl dose 0.007 0.256 0.001 0.980 1.007
Peak of midazolam dose 0.009 0.003 13.23 0.002 1.400
PRISM, pediatric risk of mortality.

114 ß 2016 American Society of Addiction Medicine

Copyright © 2016 American Society of Addiction Medicine. Unauthorized reproduction of this article is prohibited.
J Addict Med  Volume 10, Number 2, March/April 2016 IWS Syndrome and Risk Factors

outcomes such as length of mechanical ventilation, PICU,

and hospital stay.
It is noteworthy that the vast majority of PICU studies
that utilized a validated tool used WAT-1 scale (Franck et al.,
2012; Fisher et al., 2013; Amigoni et al., 2014). Although
Franck et al. (2008) have suggested that WAT-1 is better in
detecting opioid symptoms than those of benzodiazepine
withdrawal, SOS seems to be more sensitive than the WAT-
1 in assessing signs and symptoms of benzodiazepine with-
drawal (Ista et al., 2013). From the clinical perspective, we
can assume that our findings indicate a key role of benzo-
diazepines in IWS in the PICU setting. In fact, this deduction
is in agreement with Amigoni et al.’s study (2014) that has
linked the highest administered dose of benzodiazepine to
IWS development. Hence, since pediatric studies have been
focusing on the prevention and treatment of opioid withdrawal
(Anand et al., 2010; Best et al., 2015), our study along with
Amigoni et al.’s study (2014) highlight the need for strategies
to address benzodiazepine-associated IWS. Therefore, a clin-
ically useful approach would be to limit the infusion of
benzodiazepine to doses lower than 0.35 mg/kg/h and/or
adopt a weaning plan with special attention to benzodiazepine
FIGURE 2. Receiver operator characteristic (ROC) curve taper.
for the peak dose of midazolam that predicted iatrogenic Our study has several strengths that distinguish it from
withdrawal syndrome. previous pediatric studies (Franck et al., 2012;Fisher et al.,
2013; Amigoni et al., 2014). Contrary to aforementioned
Dominguez et al., 2003), only 1 found, among studies using a studies, we used a prospective design and did not include
validated assessment tool for IWS in PICU patients, an patients who used neuromuscular blockers or sedative/anal-
association between higher cumulative dose of fentanyl equiv- gesic drugs other than midazolam and fentanyl, as it would
alent and IWS (0.4 vs 0.17 mg/kg; P ¼ 0.004) (Franck et al., impact on withdrawal symptoms. Also, we began to monitor
2012). Likewise, patients with higher cumulative doses of IWS just after a short period (3 d) of continuous benzo-
midazolam (range 32–78 mg/kg) were more likely to develop diazepine/opioid. Therefore, we can speculate that our lower
IWS compared with patients without IWS (Franck et al., IWS incidence could be due to a strict sedation regimen and
2012; Ista et al., 2013); however, no correlation linking the use of a weaning-off sedative/analgesic protocol. Also,
cumulative doses of fentanyl and midazolam was demon- risk factors were determined in a multivariable analysis in
strated by other investigators (Amigoni et al., 2014). only a few studies (Ista et al., 2013; Amigoni et al., 2014).
Although we have found that IWS patients had higher However, our study may also be subject to some limita-
cumulative dose, daily peak infusion rate, and average daily tions. The concomitant administration of opioids and benzo-
doses of fentanyl and midazolam, the logistic regression diazepines limits the interpretation of our findings; however,
analysis showed that only midazolam’s peak infusion rate this sedation regimen reflects the current clinical practice.
was associated with IWS (OR 1.4). In addition, we found that Second, we did not use DSI, which could have resulted in
a cut-off value of 0.35 mg/kg/h had high sensitivity, low decreased total dose of sedatives (Gupta et al., 2012). Of note,
specificity, and good performance (AUC 0.80). A previous the impact of DSI strategy on the development of IWS is still
study (Hughes et al., 1994) of benzodiazepine-associated IWS unknown, and larger studies are required to determine its
(statistical analyses were not performed) found that an infu- feasibility and safety in children. In fact, a recent Cochrane
sion rate greater than 0.3 mg/kg/h (midazolam equivalents) Systematic Review has not found strong evidence that DSI
resulted in symptoms consistent with IWS. On the contrary, a affects several outcomes, including drug consumption, in
recent study using WAT-1/SOS scores found a cut-off value of adults (Burry et al., 2014).
0.42 mg/kg/h (sensitivity: 45.5%, specificity: 83.9%) and a
weak discriminatory ability (AUC 0.67) for the highest CONCLUSIONS
administered dose of benzodiazepine (Amigoni et al., 2014). Whereas a sedation time longer than 5 days has been
From another perspective, the negative predictive value widely proposed in the literature for the initiation of IWS
of 0.98 that we found confirms that SOS is outstanding in monitoring, our data suggest that starting monitoring as early
predicting which patients will not develop IWS (Ista et al., as 3 days of continuous benzodiazepine/opioid infusion would
2013). be more appropriate. In addition, patients requiring infusion
Also, in line with other studies (Franck et al., 2012; Ista rates of midazolam above 0.35 mg/kg/h are more likely to
et al., 2013), our study demonstrated that the presence of IWS development and should be closely monitored to diag-
withdrawal syndrome had a negative impact on other nose and promptly treat this adverse event.

ß 2016 American Society of Addiction Medicine 115

Copyright © 2016 American Society of Addiction Medicine. Unauthorized reproduction of this article is prohibited.
da Silva et al. J Addict Med  Volume 10, Number 2, March/April 2016

REFERENCES Franck LS, Harris SK, Soetenga DJ, et al. The Withdrawal Assessment Tool-1
(WAT-1): an assessment instrument for monitoring opioid and benzo-
Amigoni A, Vettore E, Brugnolaro V, et al. High doses of benzodiazepine diazepine withdrawal symptoms in pediatric patients. Pediatr Crit Care
predict analgesic and sedative drug withdrawal syndrome in paediatric Med 2008;9:573–580.
intensive care patients. Acta Paediatr 2014;103:e538–e543. Franck LS, Scoppettuolo LA, Wypij D, Curley MA. Validity and general-
Anand KJ, Willson DF, Berger J, et al. Tolerance and withdrawal from izability of the Withdrawal Assessment Tool-1 (WAT-1) for monitoring
prolonged opioid use in critically ill children. Pediatrics 2010;125: iatrogenic withdrawal syndrome in pediatric patients. Pain 2012;153:142–
e1208–e1225. 148.
Anand KJ, Clark AE, Willson DF, et al. Opioid analgesia in mechanically Gupta K, Gupta VK, Jayashree M, et al. Randomized controlled trial of
ventilated children: results from the multicenter Measuring Opioid Tol- interrupted versus continuous sedative infusions in ventilated children.
erance Induced by Fentanyl study. Pediatr Crit Care Med 2013;14:27–36. Pediatr Crit Care Med 2012;13:131–135.
Arnold JH, Truog RD, Orav EJ, et al. Tolerance and dependence in neonates Hughes J, Gill A, Leach HJ, et al. A prospective study of the adverse effects of
sedated with fentanyl during extracorporeal membrane oxygenation. midazolam on withdrawal in critically ill children. Acta Paediatr 1994;83:
Anesthesiology 1990;73:1136–1140. 1194–1199.
Best KM, Boullata JI, Curley MA. Risk factors associated with iatrogenic Ista E, van Dijk M, Tibboel D, et al. Assessment of sedation levels in pediatric
opioid and benzodiazepine withdrawal in critically ill pediatric patients: a intensive care patients can be improved by using the COMFORT ‘‘behav-
systematic review and conceptual model. Pediatr Crit Care Med ior’’ scale. Pediatr Crit Care Med 2005;6:58–63.
2015;16:175–183. Ista E, van Dijk M, de Hoog M, et al. Construction of the Sophia Observation
Burry L, Rose L, McCullagh IJ, et al. Daily sedation interruption versus no withdrawal Symptoms-scale (SOS) for critically ill children. Intensive
daily sedation interruption for critically ill adult patients requiring invasive Care Med 2009;35:1075–1081.
mechanical ventilation. Cochrane Database Syst Rev 2014;7:CD009176. Ista E, de Hoog M, Tibboel D, et al. Psychometric evaluation of the Sophia
Cunliffe M, McArthur L, Dooley F. Managing sedation withdrawal in children Observation withdrawal symptoms scale in critically ill children. Pediatr
who undergo prolonged PICU admission after discharge to the ward. Crit Care Med 2013;14:761–769.
Paediatr Anaesth 2004;14:293–298. Katz R, Kelly HW, Hsi A. Prospective study on the occurrence of withdrawal
Dominguez KD, Lomako DM, Katz RW, et al. Opioid withdrawal in critically in critically ill children who receive fentanyl by continuous infusion. Crit
ill neonates. Ann Pharmacother 2003;37:473–477. Care Med 1994;22:763–767.
Dominguez KD, Crowley MR, Coleman DM, et al. Withdrawal from lor- Leteurtre S, Martinot A, Duhamel A, et al. Validation of the paediatric logistic
azepam in critically ill children. Ann Pharmacother 2006;40:1035–1039. organ dysfunction (PELOD) score: prospective, observational, multicentre
Fernandez-Carrion F, Gaboli M, Gonzalez-Celador R, et al. Withdrawal study. Lancet 2003;362:192–197.
syndrome in the pediatric intensive care unit. Incidence and risk factors. Lugo RA, MacLaren R, Cash J, et al. Enteral methadone to expedite fentanyl
Med Intensiva 2013;37:67–74. discontinuation and prevent opioid abstinence syndrome in the PICU.
Fisher D, Grap MJ, Younger JB, et al. Opioid withdrawal signs and symptoms Pharmacotherapy 2001;21:1566–1573.
in children: frequency and determinants. Heart Lung 2013;42:407–413. McHugh ML. Interrater reliability: the kappa statistic. Biochem Med (Zagreb)
Fletcher RH, Suzanne W, Fletcher GS. In: Fletcher RH, Suzanne W, Fletcher 2012;22:276–282.
GS, editors. Clinical Epidemiology: the Essentials. 5th ed. Philadelphia, Merkel SI, Voepel-Lewis T, Shayevitz JR, et al. The FLACC: a behavioral scale
Wolters Kluwer: Lippincott Williams & Wilkins Health, 2014. for scoring postoperative pain in young children. Pediatr Nurs 1997;23:
Fonsmark L, Rasmussen YH, Carl P. Occurrence of withdrawal in critically ill 293–297.
sedated children. Crit Care Med 1999;27:196–199. Playfor S, Jenkins I, Boyles C, et al. Consensus guidelines on sedation and
Franck LS, Vilardi J, Durand D, Powers R. Opioid withdrawal in neonates analgesia in critically ill children. Intensive Care Med 2006;32:1125–
after continuous infusions of morphine or fentanyl during extracorporeal 1136.
membrane oxygenation. Am J Crit Care 1998;7:364–369. Pollack MM, Ruttimann UE, Getson PR. Pediatric risk of mortality (PRISM)
Franck LS, Naughton I, Winter I. Opioid and benzodiazepine withdrawal score. Crit Care Med 1988;16:1110–1116.
symptoms in paediatric intensive care patients. Intensive Crit Care Nurs Siddappa R, Fletcher JE, Heard AM, et al. Methadone dosage for prevention
2004;20:344–351. of opioid withdrawal in children. Paediatr Anaesth 2003;13:805–810.

116 ß 2016 American Society of Addiction Medicine

Copyright © 2016 American Society of Addiction Medicine. Unauthorized reproduction of this article is prohibited.