History of tolerance and withdrawal in medical practice

The problems of opioid dependency and withdrawal in neonates and infants were first recognized and studied in the 1970s and 1980s in a population of infants born to mothers with a history of drug addiction. Despite the difference in the origin of the problem, this experience and these studies provided valuable information for dealing with these issues in the PICU population. Experience with this neonatal population resulted in the development of various pharmacologic treatment regimens and provided the first scoring systems used to grade the severity of withdrawal and to evaluate the efficacy of the treatment regimens. ^,

Arnold and colleagues were among the first to recognize the problems of dependency and withdrawal following prolonged opioid administration in the PICU population. In a retrospective review of 37 neonates who required extracorporeal membrane oxygenation (ECMO) for respiratory failure and who had received intravenous (IV) fentanyl for sedation, the authors identified the signs and symptoms of what they termed neonatal abstinence syndrome (NAS) as well as risk factors for its occurrence. During ECMO, fentanyl infusion requirements to achieve the desired level of sedation increased from 11.6 ± 6.9 µg/kg per hour on day 1 to 52.5 ± 19.4 µg/kg per hour on day 8. In addition to the increased fentanyl requirements, there was an increase in the plasma fentanyl concentration required to achieve the same level of sedation, demonstrating that the tolerance was pharmacodynamic and not pharmacokinetic (related to increased metabolism of the opioid). The investigators reported that the incidence of NAS was related to total fentanyl dose and duration of the infusion. A cumulative fentanyl dose of 1.6 mg/kg or greater or an ECMO duration of 5 days or longer was identified as a risk factor for the development of NAS with odds ratios of 7 and 13.9, respectively. The same investigators subsequently prospectively evaluated fentanyl dosing requirements and plasma fentanyl concentrations in a cohort of eight infants placed on ECMO. Fentanyl infusion requirements increased from 9.2 ± 1.9 µg/kg per hour on day 1 to 21.9 ± 4.5 µg/kg per hour on day 6. As in their previous study, they also noted an increase in the plasma fentanyl concentration from 3.1 ± 1.1 ng/mL on day 1 to 13.9 ± 3.2 ng/mL on day 6. These reports from Arnold and colleagues at the Boston Children’s Hospital were followed in 1990 by the first report of the use of oral methadone for treatment of iatrogenic dependency and tolerance in the PICU patient. This initial report, which was anecdotal, involved only 3 infants, and used a single dose of methadone (0.1 mg/kg), has been followed by numerous other reports outlining the use of methadone in this clinical scenario (discussed later).

Although initially noted with the opioids and benzodiazepines, tolerance and withdrawal have been reported with all of the agents used for prolonged sedation in the PICU patient, including benzodiazepines, barbiturates, propofol, the inhalational anesthetic agents, and, most recently, dexmedetomidine. Regardless of the agent in question, the incidence of withdrawal increases with total amount of medication administered and duration of its use (see later discussion). The incidence does not seem to correlate with the maximum infusion rate. The incidence of withdrawal approaches 50% with infusions for 3 days and 100% when the medication is administered for 5 or more days.

Clinical signs and symptoms of withdrawal

The development of strategies to provide effective treatment of physical dependency and withdrawal requires the accurate identification and recognition of its signs and symptoms. Ongoing or associated conditions that can manifest clinical signs and symptoms similar to withdrawal must be investigated and ruled out before concluding that the patient’s symptoms are the result of withdrawal. In the PICU patient, these associated conditions include central nervous system insults (ischemia or hemorrhage) or infections, ICU psychosis, delirium, metabolic abnormalities, hypoxia, hypercarbia, and cerebral hypoperfusion from alterations in cardiac output or cerebral vascular disease. There has been considerable recognition of the potential confusion of the clinical signs and symptoms of withdrawal with those of delirium. Future work in the arena of delirium in the pediatric population as well as improved scoring systems for delirium and withdrawal are needed to differentiate these two problematic disorders. Some type of delirium scoring is recommended not only during the routine use of agents for sedation but also as these agents are weaned. It may be appropriate to rule out delirium as a confounding factor before withdrawal is treated.

Given the different receptors that are occupied by sedative and analgesia agents (γ-aminobutyric acid, opioid, and α ₂ -adrenergic), there may be subtle differences depending on the specific medication. Additionally, as more than one agent is frequently used for sedation and analgesia, there may be withdrawal from both agents and, hence, an overlap in the clinical signs and symptoms. The time to the onset of withdrawal symptoms may vary depending on the half-life of the agent as well as the half-life of any active metabolites, which may be several times longer than the parent compound. In general, the signs and symptoms of withdrawal from sedative and analgesic agents include three end-organ systems: the central nervous system (CNS), the gastrointestinal (GI) tract, and the sympathetic nervous system. CNS manifestations include those of increased activity with irritability, including decreased sleep time, disturbed sleep patterns, tremulousness, hyperactive deep tendon reflexes, clonus, an inability to concentrate, frequent yawning, sneezing, and hypertonicity. In neonates and infants, additional signs of CNS overactivity and stimulation include a high-pitched cry and exaggerated Moro reflex. Seizures have been reported with withdrawal from various agents, including the opioids, benzodiazepines, barbiturates, propofol, and the inhalational anesthetic agents. Visual and auditory hallucinations have been described with opioid, benzodiazepine, barbiturate, and inhalational anesthetic agent withdrawal. GI manifestations include emesis, diarrhea, and feeding intolerance, which may be especially prominent in neonates and infants. When these problems occur in the absence of other signs and symptoms of withdrawal, they may be attributed to other problems and not withdrawal. Activation of the sympathetic nervous system with tachycardia, hypertension, dilated pupils, diaphoresis, and tachypnea are other prominent findings. Additional signs and symptoms of sympathetic hyperactivity include nasal stuffiness, sweating, and fever. Although the signs and symptoms of withdrawal may be fairly easy to recognize, it must also be noted that many of these signs and symptoms overlap with potentially life-threatening disorders, such as infections, respiratory insufficiency, GI ischemia, and cerebral hypoperfusion. As such, the diagnosis of physical dependency and withdrawal must always be a diagnosis of exclusion to avoid potentially missing a life-threatening disorder.

Limitation of the development of tolerance and physical dependency

Strategies to limit withdrawal are now focusing on techniques to delay or prevent its occurrence. Given that the incidence of withdrawal is related to the total amount of medication administered, strategies to limit total drug requirements have been implemented in many ICUs, including titration of the sedative or analgesic agents using clinical sedation scales, early mobilization strategies, nurse-driven sedation protocols, drug holidays, and rotational techniques to limit prolonged exposure to any one medication. All these techniques may help to limit the total amount of medication required.

One of the more popular tools that has been studied as a means to reduce iatrogenic withdrawal following sedation in the PICU has been the use of nurse-driven protocols. ^, Such protocols allow the bedside nurse to adjust the infusion rates of sedative and analgesic as needed to achieve the desired level of sedation. Although limited success has been noted when attempting to demonstrate a decreased length of mechanical ventilation or ICU stay, the initial results are promising in regard to the incidence of withdrawal and physical dependency. In a before-and-after protocol involving 337 medical PICU patients requiring sedation during mechanical ventilation, there was a decrease in the incidence of withdrawal from 23.6% to 12.8% following the initiation of a nurse-driven sedation protocol. The protocol used two different scoring systems, including the COMFORT-B and the Nurse Interpretation of Sedation Scores (NISS) as well as the bispectral index (BIS) to assess and adjust the depth of sedation. Based on the protocol, the dosing of sedatives and analgesics (morphine or fentanyl ± midazolam) was adjusted.

Another commonly used technique to encourage the reassessment of depth of sedation on a daily basis in the adult population is the use of a daily drug holiday . This practice involves turning off sedative and analgesic agents until the patient responds and then restarting the infusion at a lower percentage of the previous infusion rate. Such practices are becoming commonplace in the adult population, as ongoing measures are being evaluated in an attempt to limit the incidence of withdrawal delirium. ^, Although this is a growing and common practice in the adult ICU population, there are currently limited data to support or refute the efficacy of this practice in the PICU setting. The rationale behind such practices remains sound and would seem to parallel those of using clinical sedation scores in that excessive infusion rates are avoided. With a drug holiday, if the patient is excessively sedated, the amount of medication administered is decreased to the appropriate level. However, physicians and bedside nurses may be hesitant to discontinue effective sedation and analgesia at a time when painful processes may be present in the critically ill patient. Additionally, despite evidence to the contrary in the literature and from current clinical practice, concerns have been raised that this practice may result in periods of excessive agitation in critically ill patients. Robust prospective trials in the pediatric population are needed to demonstrate not only the efficacy but also the safety of such practices.

Drug holiday schemes are being combined with other nonpharmacologic practices, including early mobilization even during the use of endotracheal intubation and mechanical ventilation. These techniques are now being brought into the PICU arena. The introduction of these techniques necessitated additions to the primary focus of the ICU, which had long been the maintenance of physiologic homeostasis as well as a shift in our way of thinking. The maintenance of these processes was assumed to require immobility, bed rest, sedation, and, at times, neuromuscular blockade to decrease the metabolic demand on an injured system to facilitate healing and recovery. Recognition of the benefits of early mobilization and its impact on the use of sedatives has resulted in the development of the ABCDEF bundle. This approach incorporates the following elements into daily ICU care: a ssessment of pain, b oth spontaneous awakening and breathing trial, c hoice of sedation and analgesia, d elirium monitoring and management, e arly mobility and exercise, and f amily engagement.

Other considerations, which also need further investigation, include rotating sedation regimens, intermittent versus continuous infusions of sedative/analgesic agents, and the role of other pharmacologic agents such as N -methyl- D -aspartate receptor antagonists (ketamine, methadone, magnesium) or opioid antagonists (naloxone) in preventing tolerance and dependency. Until further investigations provide additional insight into the factors controlling opioid dependency and ways of preventing or delaying it, PICU physicians will be faced with some patients who require specific actions to prevent the development of withdrawal symptoms.

Incidence of tolerance and physical dependency

Various risk factors related to sedative infusion have been identified that can be used to assess the incidence of withdrawal once sedative and analgesic agents are discontinued. ^, In a prospective trial of 23 infants and children who received fentanyl infusions for sedation during mechanical ventilation, Katz and associates determined the factors that could be used to identify the group at risk of withdrawal. Once sedation was no longer required, the fentanyl infusion was decreased by 50% every 24 hours times two and then discontinued. Withdrawal behavior was observed in 13 of 23 patients (57%). The total fentanyl dose and the duration of the infusion correlated with the risk of withdrawal, whereas the maximum fentanyl infusion rate did not. A total fentanyl dose of 1.5 mg/kg or greater or infusion duration of 5 days or longer was associated with a 50% incidence of withdrawal, whereas a total fentanyl dose of 2.5 mg/kg or greater or infusion duration of 9 days or longer was associated with a 100% incidence of withdrawal. These investigators were the first to report that the total duration of the infusion and total amount of drug administered could be used to determine the incidence of withdrawal while no relationship to maximum infusion rate was noted. Similar results have been noted with other sedative and analgesic agents, including benzodiazepines, inhalational anesthetic agents, barbiturates, and dexmedetomidine.

Fonsmark and colleagues retrospectively evaluated 40 patients receiving sedation during mechanical ventilation. Sedation was provided by continuous infusions of midazolam, pentobarbital, or a combination of the two. Withdrawal symptoms occurred in 14 of 40 patients (35%). Of the patients with withdrawal symptoms, eight had received both midazolam and pentobarbital, three received only midazolam, and three received only pentobarbital. A cumulative midazolam dose of 60 mg/kg or greater or a cumulative pentobarbital dose of 25 mg/kg or greater was associated with withdrawal, whereas the duration of the infusion was not. Sedation was gradually tapered in only 1 of 14 patients who experienced withdrawal.

Although not commonly used for sedation in the ICU setting, withdrawal from the sedative effects of volatile anesthetic agents has been noted. Arnold et al. reported their experience with isoflurane for sedation in a small cohort of pediatric patients ranging in age from 3 weeks to 19 years. These patients received isoflurane during endotracheal intubation and mechanical ventilation for an average of 245 hours (range, 29–769 hours). Although effective sedation was achieved, when the isoflurane was discontinued, 5 of the 10 patients, who had received greater than 70 minimum alveolar concentration (MAC)-hours of the agent, manifested signs and symptoms of withdrawal, including agitation and nonpurposeful movements. Again, this demonstrates the importance of length of administration as well as the total dose (in this case, expressed as MAC-hours) in determining the incidence of withdrawal.

Data on the risk factors for withdrawal from dexmedetomidine were compiled by Honey and associates in their evaluation over a 12-month period in 36 children who had received continuous infusions of dexmedetomidine. The median duration of the infusion was 20 hours; the range varied from 3 to 263 hours. Withdrawal was manifested by neurologic signs and symptoms, including decreased verbal communication, facial drooping, unilateral pupillary dilation, increased agitation, abnormal chewing movements, nonreactive pupils, slow rhythmic jerking movements, and abnormal head turning. Withdrawal behaviors related to dexmedetomidine were more likely when the cumulative dose was 8.5 µg/kg or greater.

Weaning from sedative and analgesic agents

The other issue concerns the optimal method and rate of weaning. Should the medication be weaned incrementally every day using the IV route, or should we change to oral medications for weaning? The rate at which the medication is weaned becomes a significant issue when the IV route is chosen. If the medication can be weaned by 20% every day, the infusion can be discontinued in 5 days. A more gradual weaning process would mandate prolonged IV access unless a transition is made to oral medications. Although it has been suggested that weaning by 10% to 20% is feasible, these studies demonstrate a relatively high incidence of withdrawal, suggesting that a more reasonable approach may be a 5% to 10% decrease per day, as has been suggested for adult patients and supported for some in the PICU population. It is possible that in some patients, a more rapid wean can be accomplished if the agents have been administered for fewer than 3 to 5 days.

After prolonged administration of 5 or more days, the weaning process will likely require a more protracted period to prevent withdrawal symptoms. Although the weaning process can be accomplished by slowly decreasing the IV infusion rate, this mandates the maintenance of IV access, ongoing hospitalization, and, at times, continued monitoring in the PICU. This is because, depending on hospital policies, certain medications (e.g., fentanyl, midazolam) cannot be administered by continuous infusion in settings other than the PICU. In these circumstances, options to consider include either switching to the subcutaneous route or, most commonly, the use of oral administration. If it is decided that the infusion can be tapered within a period of time that will not delay hospital discharge and that switching to oral medications will not expedite discharge home, the patient may be considered a candidate for subcutaneous administration. Although not a commonly used mode of administration, there are limited data suggesting the efficacy of this route in various clinical scenarios. Patients for whom transition to subcutaneous administration is considered are generally receiving moderate doses of fentanyl (5–10 µg/kg per h) or midazolam (0.1–0.3 mg/kg per h). The switch to the subcutaneous route allows the removal of central venous access, eliminates the need to maintain peripheral IV access, and, depending on individual hospital policies, may eliminate the need for ongoing care in the ICU setting. Both fentanyl and midazolam can be effectively administered via the subcutaneous route and the infusions slowly tapered to prevent symptoms of withdrawal. Concentrated solutions of fentanyl (25–50 µg/mL) and midazolam (2.5–5 mg/mL) are used so that the maximum subcutaneous infusion rate does not exceed 3 mL/h. The subcutaneous infusions are started at the same dose that is currently being used for IV administration. A topical dermal anesthetic cream can be placed over the site of anticipated subcutaneous cannulation. Several areas are suitable for subcutaneous administration, including the subclavicular region, abdomen, deltoid, or anterior aspect of the thigh. The site is cleaned, prepped with sterile antiseptic solution, and then either a standard 22-gauge IV cannula or a 23-gauge butterfly needle is inserted into the subcutaneous tissue. Before placement, the tubing and needle are flushed with the opioid/benzodiazepine solution. The insertion site is then covered with a transparent, bio-occlusive dressing. The site should be changed every 7 days or sooner if erythema develops. The same infusion pumps that are used for IV administration can be used for subcutaneous administration. The pressure limit may need to be adjusted to allow for subcutaneous administration. Alternatively, a syringe pump can be used. If symptoms of withdrawal develop, additional boluses can be administered subcutaneously if necessary. Several opioids can be administered subcutaneously, including the synthetic opioids, morphine, hydromorphone, and meperidine. Longer-acting agents, such as methadone and levorphanol, are not recommended because dose titration is not feasible given the long half-lives of these agents. Tissue reaction and erythema have been noted with methadone. Although there is limited experience with the use of subcutaneous infusions of opioids/benzodiazepines as a means of weaning patients and preventing withdrawal, the subcutaneous route has been used to treat chronic cancer-related pain as well as postoperative pain.

When the more prolonged administration of sedative and analgesic agents will be necessary, switching to the oral administration of long-acting agents such as methadone allows for earlier discontinuation or IV infusions as well as discharge from the ICU and, frequently, the hospital. This is especially true in patients who have received weeks of therapy and are on large doses of opioids, benzodiazepines, and dexmedetomidine. When considering opioid therapy, methadone is generally chosen as the oral agent. Advantages of methadone include its longer half-life, allowing for dosing two to three times per day, an oral bioavailability of 75% to 90%, and availability as a liquid. Although the first report regarding administration of methadone used a fixed starting dose of 0.1 mg/kg every 12 hours (similar to the analgesic doses that were commonly used), the three patients in the series were receiving relatively low opioid doses and therefore higher doses of methadone were not needed. The subsequent clinical experience has indicated that higher doses of methadone will be needed depending on the dose and duration of opioid administered. When considering the appropriate dose transition from IV fentanyl to oral methadone, thought should be given to the differences in the potency and half-life of the two medications as well as crossover tolerance.

After the initial reports regarding the use of methadone, several other authors have suggested variations in conversion ratios from fentanyl to methadone, the use of IV and then oral methadone, varying maintenance dosing intervals, and, most important, alterations in the weaning schedules. In clinical practice, it is useful not to dwell on the differences of these individual studies but rather to ensure that each institution has a formal policy for the initial dosing and subsequent weaning of methadone.

Given its advantages, methadone is now the common choice for the opioid-dependent patient in the PICU. However, stigma may remain concerning its use given its use in the adult opioid addiction potential. A thorough discussion with the parents is necessary to discuss why methadone is being used and to outline the differences between addiction and physical dependency. Because of these issues—as well as familiarity with long-acting morphine preparations, which are used in the treatment of children with chronic cancer-related pain—some physicians have used long-acting preparations of morphine. However, these agents are available only in tablets that cannot be crushed so that administration and subsequent weaning protocols are not feasible in younger patients and more problematic in older patients. Methadone, on the other hand, is available in a liquid formulation. Concern has been expressed regarding adults who are on maintenance methadone for drug addiction in terms of the potential for death from QT prolongation and arrhythmias. ^, To date, there are no reports from the pediatric population that demonstrate increased mortality. However, prolongation of the corrected QT interval has been demonstrated following the institution of methadone therapy in the PICU population. These concerns have led to the consideration of obtaining periodic electrocardiograms prior to and after instituting therapy with methadone. Given the multitude of medications used in the PICU patient, avoidance of interactions with other agents that can potentially prolong the QT interval is suggested. A final issue with methadone is its metabolism by the P-450 isoenzyme system of the liver, making alterations in metabolism possible based on genetic factors and the coadministration of other medications. These factors should be considered when methadone is started or other medications are added to the patient’s regimen.

In addition to opioids, nonopioid agents have been used to treat opioid withdrawal. This practice was relatively common when treating the withdrawal of infants born to drug-addicted mothers. In many cases, paregoric or other crude opioid preparations were used. Alternatively, barbiturates such as phenobarbital were given to control the clinical signs and symptoms. This would appear less than optimal, as it seems to make physiologic sense when dealing with the problems of tolerance and physical dependency to replace the missing agent rather than to treat the resulting symptoms. Given these concerns, our current clinical practice is generally to use a medication in the same class as the one resulting in withdrawal. However, in specific clinical scenarios, the centrally acting α ₂ -adrenergic agonists, clonidine and dexmedetomidine, have been used to treat and prevent opioid and benzodiazepine withdrawal.

Similar considerations are necessary when switching from IV midazolam to oral lorazepam or from IV dexmedetomidine to oral clonidine. Lugo and associates, in a study evaluating enteral lorazepam to decrease midazolam requirements during mechanical ventilation, suggested starting at a lorazepam dose that was one-sixth that of the total daily dose of IV midazolam. Once the appropriate enteral/oral dose was determined and therapy started, the IV administration was tapered quickly, generally over a 48-hour period. In a more recent study conducted when transition from IV midazolam sedation to enteral lorazepam for weaning, a higher daily dose of enteral lorazepam was suggested. Warrington and colleagues recommended using 0.3 times the total daily dose of midazolam and then administering the total daily dose of enteral lorazepam divided into 6 daily doses, or every 4 hours.

With the significant increase in the use of dexmedetomidine in the PICU in various clinical scenarios, withdrawal from this agent is becoming a more commonly encountered clinical problem. It was initially and erroneously speculated that this novel agent might cause little or no tolerance with prolonged infusion and that the infusion might be abruptly discontinued even following prolonged administration. However, with increased use and clinical experience came reports of tolerance, physical dependency, and withdrawal. Prevention of such occurrences requires the initiation of strategies once the patient has received an infusion for more than 4 to 5 days. Options include a slow taper via the IV route (0.1 µg/kg per h every 12–24 h), switching to subcutaneous administration (see earlier discussion), or transition to clonidine either orally or through the transdermal route. In common clinical practice, most centers have adopted a regimen that involves the transition from IV dexmedetomidine to oral clonidine followed by a gradual taper of the oral clonidine dose. Various reports provide useful information regarding this practice. Despite limited evidence-based medicine for this practice, clinical experience has suggested that the transition from IV dexmedetomidine to oral clonidine can be accomplished over a 48-hour period with a clonidine dose of 3 to 5 µg/kg per dose every 6 to 8 hours.

As noted earlier, when transitioning from an opioid, benzodiazepine, or dexmedetomidine infusion to oral medications to prevent withdrawal in the PICU setting, it is useful not to dwell on the individual studies and their differences but rather to ensure that each institution has a formal policy for the initial dosing and subsequent weaning of these medications. This practice is suggested to avoid the natural variations that occur in physician practice, which may impact the success of such programs. ^, Regardless of the protocol used, close observation during the conversion period is necessary with the use of sedation and withdrawal scores to avoid adverse effects from oversedation or to recognize the early symptoms of withdrawal. Pharmacy-driven protocols may allow for less physician variation in practice and facilitate the efficient transition to enteral medications and their subsequent weaning.

Summary

With attention to the need for more aggressive sedation and analgesia during prolonged critical illness, the newer issues of physical tolerance and withdrawal have become more pervasive. Although preventing or at least delaying tolerance may be feasible using newer practices—such as drug holidays, early mobilization protocols, and nurse-controlled sedation regimens—treatment algorithms are needed to avoid the adverse physiologic effects of these problems and to limit their impact on patient well-being and length of hospitalization. Clinical studies show consistently that, regardless of the agent administered, withdrawal will occur once the sedative or analgesic medication has been administered by continuous infusion for more than 4 to 5 days. In those scenarios, the most commonly used option is switching to an orally equivalent agent. During this process and as the oral agent is weaned, ongoing monitoring with a withdrawal score is suggested. To facilitate care and avoid variations in practice, many institutions have developed specific protocols for the oral agent to be used, conversion from IV to oral doses, and tapering regimen. By aggressively managing the issues of tolerance, physical dependency, and withdrawal, their impact on the effective provision of sedation and analgesia during critical illness can be minimized.