Adverse Sedation Events in Pediatrics: Analysis of Medications Used for Sedation

DISCUSSION

Our study demonstrates that children suffered drug-related adverse outcomes after administration of a wide variety of medications. The data suggest a relatively even distribution of adverse sedation events in children across the major drug classes (opioids, benzodiazepines, barbiturates, and sedative/hypnotics). The observation that negative outcomes were associated with all classes of drugs and all routes of administration is clinically important because it points out that these negative outcomes occur not because of the drugs themselves but rather because of drug administration practices (drug combinations, errors, and monitoring standards). These practices likely reflect the skills (or lack of skills) and knowledge (or a lack of knowledge) of the individuals who administered the drugs for procedural sedation. Many events related to medication errors have been well-characterized. Lack of knowledge of the drug, lack of patient information, failure to follow procedures, transcription errors, faulty dose dispensing, inadequate monitoring, and a variety of other causes have all been described and many of these were evident in out database.^13–18 Children in particular seen to be vulnerable; methods for prevention of medication errors in this group have also been described.¹⁹ The recent report from the Institute of Medicine has highlighted a number of initiatives that could be used to reduce medication errors.²⁰ Clearly the training, the understanding of the pharmacokinetics and pharmacodynamics of the drugs administered, and systems issues (such as drug limits, double-checking drug doses, computer checks of drug doses, improved patient informatics, the direct involvement of pharmacists, and limiting the medications used for certain types of procedures) are potentially important mechanisms for reducing medication errors.^20–32

Our observations are consistent with these and other reports concerning adverse drug events and medication errors. However, a number of adverse outcomes were related to systems issues, such as inadequate monitoring, lack of skills in cardiopulmonary resuscitation, inadequate recovery procedures, and others causes unrelated to the drugs administered. In addition, nearly one half of the adverse outcomes occurred in nonhospital-based facilities, where the usual hospital-based safety net of state-mandated regulations does not generally apply. Unfortunately there is no way of knowing the actual incidence of these sedation/medication-related events because these were voluntary reports and the data were collected retrospectively. However, it is very likely that the cases that we collected represent a gross underreporting especially because such reports are often used as a tool for measuring hospital/physician/organizational performance.^33–40 We agree with the Institute of Medicine report which suggests the need for a national mandatory reporting system that would be “afforded legal protections from data discoverability… devoted to analyzing and understanding the causes of errors to make improvements.”²⁰ Other proposed initiatives include voluntary reporting systems such as the Joint Commission on Accreditation of Health care Organization's sentinel event program, the Food and Drug Administration's MedWatch program.^41,42Such initiatives would help to focus on both drug-related and systems-related issues.

Our study found that there was no clear relationship between the route of administration and negative outcomes. The IV route was used most frequently and was the least associated with negative outcomes; however, compared with other routes of drug administration, the difference did not reach statistical significance. Several adverse events were successfully treated because of timely recognition and antagonism of drug effects. Although the data are somewhat soft, it suggests a possible advantage to the use of drugs that can be reversed if an adverse event should take place. Thus the use of an opioid or benzodiazepine alone or in combination with each other or combined with other sedating medications may be safer because at least one of the drugs has a specific antagonist that can reverse respiratory depression should it occur. The tendency toward a lower complication rate may also be related to the immediate availability of IV access for administration of resuscitative agents. Furthermore, we hypothesize that the tendency toward fewer adverse outcomes when medications are given intravenously may in part be related to the ability of titrate drugs to affect afforded by this route. Single large doses are commonly used when drugs are administered by other routes and there is no titration of drug. Another possibility is that these patients were monitored more closely. These hypotheses should be investigated further in future studies.

In many states, dental office practice certificates are based in part on the route of drug administration. Because our data clearly demonstrate that the route of drug administration is unrelated to outcome, certification of office practice based on route of drug administration does not adequately protect patients. Instead, certification should be based on practitioners' training in the use of these medications and on their sedation assessment, airway management, and resuscitation skills. In addition, the 2 Current Procedure Terminology codes recently established for sedation reimburse practitioners according to the route of drug administration (sedation administered by the IV, IM, and INH routes is reimbursed at 1 rate; medications delivered by the PO, PR, and IN routes are reimbursed at another rate).⁸ Many procedures are performed by physicians who are not trained to sedate children or who are uncomfortable sedating children. The new Current Procedure Terminology sedation codes only apply if the physician is performing the procedure himself or herself; this creates a disincentive for independent experienced physicians to participate in the sedation of children for other physicians. The cost to the practitioner or a medical facility to provide safe patient care during sedation for a procedure is related in part to the intensity and duration of the monitoring required. The routine use of an additional person whose only responsibility is to observe the patient certainly is an important factor in this cost. Reimbursement based on route of drug administration is fallacious and should be abandoned.

Many serious adverse events occurred when multiple drugs were administered despite the fact that each was administered in less than the maximum recommended dose. This suggests drug–drug interactions, the category most often associated with an adverse sedation event (n = 44) and a phenomenon that has been well-described in clinical studies.^43,44 Adverse outcome has been correlated with the number of drugs used to sedate children having computer-axial tomography scans.⁴⁵ Our study found a high association with death or permanent neurologic injury when 3 or more sedating medications were administered. Several children in our cohort suffered an adverse outcome even when appropriate doses of individual medications were administered in combination. This underscores the need for education regarding the potential for a greater than desired depth of sedation when combining sedating medications, even when each is administered within the recommended dosing limits. When administration of multiple drugs is planned, initial doses (mg/kg) should be lower than those for each drug given alone. Also, insertion of an IV line, perhaps after an initial nonparenteral dose of a drug, could facilitate titration of further sedative/analgesics.

Drug overdose was the second most common category of causes attributed to adverse sedation events (39 episodes in 34 patients). None of these involved a 10-fold overdose; this fact suggests that the errors were not simple multiplication/decimal point errors, but rather were caused by a lack of knowledge about drug dosing in children.¹⁹The 3 local anesthetic overdoses, an issue previously described in dental accidents, underscore the importance of double-checking all drug doses and of setting maximum mg/kg dose limits.^46,47Perhaps some of these adverse events resulted from the lack of pediatric labeling for nearly every medication used for sedation, analgesia, and amnesia in children.^48–52

Some children died or suffered permanent neurologic injury even when a single drug was administered in less than maximum recommended doses (n = 4). One of these was undergoing an echocardiogram and was sedated with chloral hydrate (60 mg/kg) by a technician; 1 was a 2-year-old sedated for a computer-axial tomography scan with methohexital (20 mg/kg, PR); another was a 3-year-old sedated by a parent with midazolam (.5 mg/kg, PO) who died in the car on the way to the dental office; and the fourth was a 19-year-old undergoing a therapeutic abortion in a clinic who received methohexital (<2 mg/kg, IV). Two of these 4 patients were not protected by the safety net of trained medical personnel. These cases illustrate the essential reason for having sedation guidelines that admonish against administration of sedating/anxiolytic medications at home or by those not qualified to provide skilled observation and rescue should an adverse event occur. Sedation guidelines allowing preprocedural drug administration at home should be modified to eliminate such practices.

Adverse events occurred in both hospital-based and nonhospital-based venues because of prescription or transcription errors. In 1 case, the pharmacy filled a chloral hydrate prescription for tablespoons instead of teaspoons and at twice the concentration, resulting in a dose 6 times higher than that intended by the prescriber. These types of events illustrate the need to have the same standard of care and vigilance regardless of the route of drug administration, drug class, or the venue in which the drug is administered. Because medication errors may occur at any time and at the hands of very skilled practitioners, sedation areas and delivery systems should be designed to prevent the occurrence of any errors.^20,53,54 For example, procedure policies could require that orders and prescriptions clearly indicate the child's weight, mg/kg, and the total dose to be administered. Further titration of medications should be specified. Because practitioners use a small number of medications for their particular procedures, precalculated drug dosage cards like those commonly used in emergency and critical care settings would be relatively easy to create. Further engineering of injury proof delivery systems should be encouraged.

We were particularly surprised by the number of children who suffered a negative outcome after the administration of chloral hydrate. This drug is widely used for infants and toddlers and has a long-standing reputation as a very safe medication with minimal effects on respiration.^55–58 Chloral hydrate is often used as a single sedating agent or in combination with other sedatives, particularly for dental and radiologic procedures.^59–63Thirteen of 60 cases resulting in death or permanent neurologic injury involved the use of chloral hydrate alone (n = 7) or in combination with other medications (n = 6). Adverse events after chloral hydrate included a number of medication-related factors: overdose, administration outside of the safety net of a medical venue (drug given at home), administration by nonmedically trained personnel (technician), and premature discharge from medical observation. One of the children who died after receiving an unintended chloral hydrate overdose was noted by the mother to have a rapid heart beat; this may have been a sign of chloral hydrate toxicity.^64–67 A technician rather than a medically trained individual supervised this child, and the potential clinical implications of the rapid heart rate were ignored. We do not know whether this child would have survived if a nurse or physician had intervened at that point. We suspect that the 2 children with congenital heart disease were vulnerable to the development of cardiac dysrhythmias. It is known that tonsillar and adenoidal hypertrophy and Leigh's encephalopathy are associated with airway obstruction in patients sedated with chloral hydrate.^68–70 Airway obstruction has also been associated with chloral hydrate administration to American Society of Anesthesiologists (ASA) physical status 3 children.⁷¹ We do not know whether any of our cases had similar airway-related problems, but it was most disturbing to find that 1 child who died had received only 60 mg/kg of chloral hydrate, which is well within recommended dose limits. Our study clearly points out that chloral hydrate is no exception to the rule that medications capable of causing depressed levels of consciousness should never be administered by nonmedical personnel in a nonsupervised medical environment.⁷²

Some children were injured in car seats on their way home after a procedure. A possible mechanism for the injury was the infant falling asleep with the rhythmic motion of the automobile and the head falling forward, thereby obstructing the upper airway. In the presence of residual drug effect, the child could be unable to arouse or unable to spontaneously reposition the head to relieve the airway obstruction. This sequence of airway obstruction and desaturation has been demonstrated in unsedated full-term neonates placed in car seats.⁷³ All 9 children who experienced an adverse event (8 of whom died or suffered permanent neurologic injury) at home or in an automobile after a procedure, received drugs known to have a long half-life in infants and children (chloral hydrate, promazine, promethazine, chlorpromazine, or phenobarbital [IM]; see “Appendix”). The active metabolite of chloral hydrate is trichloroethanol. In newborns the half-life of trichloroethanol is 27.8 ± 21.3 hours, whereas in toddlers it is 9.7 ± 1.7 hours.⁷⁴ Thus, although at the end of a procedure an infant or toddler may seem to have recovered from the sedative effects of chloral hydrate, residual drug, and active metabolite are still circulating and there is the potential for resedation once the child is no longer stimulated. Two other children who died or suffered permanent neurologic injury after discharge were sedated with the classic combination of DPT or chlorpromazine plus promazine. A pharmacodynamic study of DPT administered to pediatric patients in the emergency department has shown that it took a mean of 19 ± 15 hours for children who received this drug combination to return to normal behavior.⁷⁵ The half-life of a single dose of chlorpromazine in adults is 31 hours.⁷⁶ A half-life of 3.19 days has been demonstrated in newborns.⁷⁷ Children seem to have a shorter elimination half-life compared with adults (7.74 ± .65 hours) but this study was for IV not IM administration⁷⁸; the pharmacokinetics are likely different after IM administration attributable to delayed absorption and depot effect. Promazine and promethazine have a half-lives of 12.65 ± 4.7 hours⁷⁹ and 10 to 14 hours in adults, respectively.⁸⁰ Promethazine also has age-related differences in pharmacokinetics. The half-life is shorter in children (7.1 ± 2.3 hours) after oral administration, compared with adults (20 ± 4.1 hours).^81,82 The pharmacokinetics of promethazine may also be different after IM administration because of delayed absorption and depot effects. Another drug associated with death after discharge was pentobarbital (8 mg/kg, IM). The half-life of pentobarbital in children when administered intravenously is 25.5 ± 16 hours.⁸³ In many of these cases the known pharmacokinetic profile of these agents was apparently ignored because patients were discharged without complete recovery from sedation.

Our discovery of several patients with negative outcomes who received sedation with these medications is particularly relevant considering the widespread and long-time use of these medications for outpatient sedation for procedures, particularly for infants. There are minimal data regarding the pharmacodynamics of any of these drugs in children, especially how age, maturation of hepatic or renal function, route of administration, or enzyme inhibition might alter drug elimination. These cases clearly point out the need for very rigorous recovery procedures and discharge criteria. Our data suggest that children should rest/recover in a quiet monitored area after the end of the procedure, even if they seem to be awake immediately after it is completed. A step-down unit for further observation after discharge from a standard recovery area may be of value in children who have received sedative medications with long half-lives. Patient discharge must only be made by qualified personnel (nurse, physician, and dentist) and not by a technician. We would make the further suggestion that because of the very unfavorable pharmacokinetics and the known drug–drug interactions,^84–86 the combination of DPT should be abandoned, particularly for outpatient procedures and those performed in infants. Although DPT was useful in its time, many other better options are now available to practitioners.

Eight events, of which 5 resulted in death or neurologic injury, occurred in situations with clear evidence that the practitioners did not understand the pharmacology of the drugs that they had administered. Examples of this included several patients who developed chest wall and glottic rigidity after opioid administration, could not be ventilated or oxygenated, and died or sustained severe neurologic injury. Administration of naloxone may have been life saving. Another example was an attempt to reverse a chloral hydrate associated event with naloxone because the practitioner thought that chloral hydrate was an opioid. A third example was local anesthetic toxicity that resulted in seizures and arrhythmias that were then treated with additional lidocaine (IV). These cases clearly underscore the importance of intimate knowledge of the pharmacology and the pharmacodynamics of the medications used for sedation/analgesia. If a physician/dentist is going to administer any medication, they must understand the basic pharmacology of that drug and how to effectively manage expected drug-related complications.

We do not know the reason why dental specialists were disproportionately represented. In fact, we excluded 10 dental cases who died because 9 had received alphaprodine which is no longer manufactured and 1 received a drug (a transdermal fentanyl patch) for postdental surgery pain. Despite these exclusions this specialty had the highest representation with 29/32 suffering death or permanent neurologic injury. There was a very strong relationship of adverse outcomes with nonhospital-based facilities and with dental practitioners (n = 23). In some states, the category of sedation called Anxiolysis in the Dental Office Setting permits the prescription or administration of pharmacologic anxiolytics with concomitant use of nitrous oxide, without requirements for a special dental office anesthesia permit, advanced training, or pulse oximetry monitoring.⁸⁷ Seven patients who died or sustained permanent neurologic injury were given promethazine, diazepam, or chloral hydrate combined with nitrous oxide administered by nonanesthesiologists for dental procedures. Continuous monitoring of level of consciousness is particularly important, because many dental procedures in themselves compromise the airway: abnormal head and tongue positions; foreign materials (cotton and rubber dams); and the presence of blood, increased secretions, and exogenous water. Negative outcomes were also associated with the patient receiving 3 or more sedating medications. Dental practitioners accounted for the majority of patients who received 3 or more sedating medications and all the patients who received nitrous oxide. Nitrous oxide is generally considered to have minimal effects on respiration and consciousness. However when nitrous oxide is combined with any other depressing medication, even when the sedating medication is administered in standard doses, a state of deep sedation and/or general anesthesia may occur.^46,88–91 It is possible that drug–drug interactions, particularly with nitrous oxide, contributed to the adverse outcomes in the dental setting. The category of Anxiolysis in the Dental Setting, which allows anxiolytics to be combined with nitrous oxide without a requirement for monitoring, should be abandoned; this level of sedation requires the same training in airway management and monitoring as that required for deep sedation/general anesthesia.⁷² In addition, the 1998 revision of the American Academy of Pediatric Dentistry (AAPD) sedation guidelines⁹² deviates substantially from guidelines published by the AAP⁷² and ASA.⁹³ The AAPD has divided the category of conscious sedation (equivalent to sedation/analgesia for the ASA guideline) into 3 categories. Conscious sedation level 3 is defined as a state of consciousness in which “repeated trapezius pinching or needle insertion in oral tissues elicits reflex withdrawal and appropriate verbalization (complaint, moan, crying).” The ASA document clearly states (and the AAP document will soon state) that reflex withdrawal is not considered to be a state of conscious sedation/ or sedation/analgesia but rather is consistent with a state of deep sedation/general anesthesia. These disparate definitions and approach to sedation monitoring and training need to be made consistent for all patients regardless of practice. We encourage all dental specialists to examine their practices and urge them to develop monitoring guidelines and training requirements similar to those adopted by the AAP and ASA.