Teaching From Catastrophe: Using Therapeutic Misadventures From Hydromorphone to Teach Key Principles in Clinical Pharmacology

Core principles of clinical pharmacokinetics and pharmacodynamics that form the basis for safe drug practices are important to convey to medical learners. However, the discipline of clinical pharmacology as applied at the clinic and bedside in academic health centers has been in decline for the past 25 years.1 Although all medical students are provided with information on pharmacokinetics and pharmacodynamics of drugs in general and opioids in particular, this exposure occurs during their preclinical years. However, formal training of medical students in the core principles of clinical pharmacology should take place during their senior year to properly link the information conveyed directly to patient care to foster rational drug prescribing habits. This is recognized as a deficiency in medical education across the United States by the Association of American Medical Colleges (AAMC).2 A path forward in the medical school curriculum has been outlined by the AAMC.2 This has coincided with statements contained in the leading journals and organizations representing clinical pharmacology across the globe that underscore the importance for practicing academic clinical pharmacologists to be the leaders in this process of reform.3–10 Unfortunately, during this same period, the number of medical schools in the United States that offer fourth-year coursework in clinical pharmacology has declined from 20% in 1974 to 9% in 2007.11 This milieu provides the backdrop for the current situation whereby information in therapeutics is conveyed to medical students through an apprenticeship model by some combination of anecdotal experience of postgraduate physicians-in-training and/or faculty mentors.2,3,5,7 The series of patient cases in this report represents clear deviations in good clinical practice, resulting in medical catastrophe from the improper use of hydromorphone. As such, the most direct causes of these unnecessary deaths are mistakes on the part of the health care providers involved, leading to insufficient patient monitoring and incorrect hydromorphone administration. However, fundamental knowledge deficits typically underlie these more immediate causes of human error in third-order “why” internal quality investigations in hospitals to improve future patient safety.12 Indeed, had the fundamental principles of clinical pharmacology been properly understood, it is likely that these therapeutic misadventures would have been averted. Deaths from opioids, including hydromorphone, usually occur from parenteral administration13–16 and derive from root causes of errors in drug dosing and frequency,17 combined use with other sedatives,18,19 and inattention to underlying patient conditions that predispose to drug toxicity.18,20 This report adds to this literature by providing a series of patient deaths from hydromorphone through which the critically important principles of clinical pharmacology underlying these root causes are linked to each individual patient misadventure. A 23-year-old, 75-kg man with a history of sickle cell disease was admitted to the hospital for a pain crisis. He was given intravenous isotonic fluids and from hospital days 1 to 3 received morphine sulfate 10 mg intravenously (IV) every 2 hours as needed for pain. He received the drug regularly, except for 2 doses on the first hospital day. Since the pain was unrelieved by the current analgesic regimen, it was changed to hydromorphone 3 mg IV every 2 hours as needed for pain. Fifteen minutes after the third dose of hydromorphone, the patient was found unconscious with circumoral vomitus. Despite attempts at resuscitation, including aggressive suctioning with a difficult endotracheal intubation, and naloxone, the patient expired. The family refused a postmortem examination. A 35-year-old, 94-kg man came to the emergency department with increasing right lower quadrant pain over the past 4 days. He had a history of an appendectomy 15 years ago and chronic migraine headaches. Medications that he was receiving on admission included oxycodone 5 mg/acetaminophen 325 mg orally up to a maximum daily dose of 4 tablets in a 24-hour period for headache. Diagnostic evaluation in the emergency department included laboratory evaluation for routine causes of intra-abdominal pathology, surgical consultation, and computerized axial tomography of the abdomen and pelvis. No cause of his abdominal pain was identified by these methods, and the patient was admitted to the hospital for observation, serial abdominal examinations, gut rest, intravenous fluids, and analgesia. The patient was given hydromorphone 3 mg IV every 4 hours as needed for pain. The patient received 6 doses over the first 24 hours in the hospital. The patient's intravenous access was lost on the second day, and the analgesic regimen was changed to hydromorphone 3 mg intramuscularly (IM) every 3 hours to be given for a total of 8 doses as a standing (non-“prn”) order. One hour following the sixth dose of hydromorphone, the patient's roommate told the nurse that he heard “loud snoring” coming from the patient. The nurse responded within 30 minutes and found the patient pulseless and not breathing. Subsequent efforts at resuscitation were unsuccessful. Postmortem examination did not find an anatomic cause for the death. The only drugs detectable were hydromorphone and its glucuronides. A 23-year-old, 65-kg man was admitted to the hospital for evaluation of a 10 × 5-cm thigh abscess that was sustained by an abrasion while in a swimming pool 1 week prior to presentation. The patient experienced a pulmonary embolism following fractures of his pelvis and left femur 2 years prior due to a motor vehicle accident. The only medication he was receiving was naproxen 500 mg, 3 times a day. Incision and drainage in the operating suite was performed on the first hospital day, and the patient was returned to the hospital ward at 1700. Postoperative orders included hydromorphone 2 mg IV every 1 hour as needed for pain. Over the ensuing 9 hours, from 1800 to 0330, the patient received 11 mg (over 9.5 hours). The nursing note at 0430 reads the patient was snoring loudly but that the vital signs were stable. Routine nursing check at 0630 found the patient unresponsive, pulseless, and not breathing. Advanced cardiac life support succeeded in reinitiating a blood pressure but no spontaneous respirations, and the patient was investigated for a pulmonary embolus by computerized angiography of the chest with bilateral lower extremity Doppler studies. These failed to reveal a clot. Three days later, the patient was determined to be brain dead, and ventilatory support was withdrawn. Postmortem examination showed no evidence of a pulmonary embolus with no anatomic cause of death identified. The only drugs detectable were hydromorphone and its glucuronides. A 21-year-old, otherwise healthy, 57-kg man was given a “flaming rum beverage” for his 21st birthday. He presented to the emergency department at 0030 with a blood alcohol level of 200 mg/dL. The patient was given hydromorphone 2 mg IV at 0100 and 0200 and then received hydromorphone 1 mg IV every 10 minutes during wound debridement and dressing (total of 8 mg given over 4 hours). The patient was discharged to home at 0500 with a dose of oxycodone 10 mg. He took this medication and went to sleep at 0530. His girlfriend found him dead in bed at 0630. The postmortem examination did not identify an anatomic cause of death. The only drugs detectable were oxycodone, hydromorphone, and their metabolites. A 52-year-old, 62-kg man awoke with exquisite right lower flank discomfort with hematuria that persisted for 12 hours prior to his presentation to the emergency department. The patient had 1 episode of nephrolithiasis 2 years prior without sequellae and was receiving no medications as an outpatient since that time. Computerized tomography confirmed a nonobstructing stone. The patient was admitted, made NPO, and given isotonic fluids and hydromorphone 2 mg IV every 2 hours as needed for pain, beginning at 2100. The patient subsequently received hydromorphone 8 mg IV over the ensuing 6 hours (last dose at 0300). During this interval, the patient was given zolpidem 10 mg orally at 0030. On routine check at 0330, the patient was found pulseless and not breathing, and resuscitation efforts were unsuccessful. Postmortem examination did not identify an anatomic cause of death. The only drugs detectable were hydromorphone, zolpidem, and their metabolites. Hydromorphone is a hydrophilic opioid available in oral and parenteral dosage forms for severe acute pain in ambulatory and hospitalized patients. Compared to morphine, hydromorphone is roughly 5 to 7 times more potent, with a slightly longer serum half-life (3 vs 2 h) and volume of distribution (4.5 vs 3 L/kg), respectively.21–24 Although there is overlap in the knowledge deficits in clinical pharmacology that led to errors in the administration of hydromorphone that is shared among these patients, the discussion will focus on a core principle per case that, if better understood, may have prevented the death of each patient. Table I summarizes the demographics of all 5 cases and categorizes each principle of clinical pharmacology that was violated to help organize the discussion thematically. Human factors principles help to clarify the underlying processes that led to the death of patient 1.25 Primary among these are that the ubiquity of pain in hospitalized patients tends to lull physicians into a false sense of familiarity with the therapeutics for such a common complaint. This human factors principle led to the underappreciation of the relationship between the potency of hydromorphone to the standard reference compound, morphine. This characteristic, coupled with the apparent lack of understanding involving the pharmacodynamic principle that drugs within a therapeutic class act via the same receptor, led to this patient's death. Although hydromorphone is generally considered to be roughly 7 times more potent than morphine,21 similar to other opioids, it has long been recognized that a wide interpatient variability exists in analgesic dose-response and pharmacokinetic characteristics.22,23 Indeed, clinical studies have reported a wide variability in dose equivalency, with a hydromorphone/morphine potency ratio as low as 4:10 to as high as 0.9:10, respectively.24,26 Therefore, particularly for drugs with a narrow therapeutic index as opioid analgesics, physicians-in-training should strictly limit the number of drugs they prescribe within a class, becoming thoroughly familiar with the pharmacokinetic and pharmacodynamic characteristics of each and the exact specifications of their therapeutic use.27 Hospital formulary systems that restrict drug prescribing to physician specialists for those narrow therapeutic drugs that are not routinely prescribed may also be useful.28 However, opioids tend to limit the utility of “hard-stops” or other similar mechanisms within computerized physician order entry systems due to the large dosage ranges required to treat pain between opioid-naive and opioid-tolerant patients. Fundamental misunderstandings in the relationship between redistribution and the termination of drug effect for opioids administered by intravenous bolus, coupled with knowledge deficits in the proportional relationship between drug accumulation and drug dose, dose interval, and both serum and transfer half-lives, were substantially causal in the deaths involved in patients 2 and 3. Particularly important for patient 3 was the inadequate time allotted for hydromorphone to reach maximal central nervous system (CNS) effect given that the dosing interval selected (1 hour) did not allow for the full effect of the drug to occur based on its 30-minute transfer half-life (150 minutes to reach full effect).29 This issue is not unique to hydromorphone but is shared with other hydrophilic opioids (eg, morphine with its combined transfer half-life of 10 hours, with its active morphine 6-glucuronide metabolite).17 In contrast, the lipophilic opioids have far shorter transfer half-lives (fentanyl, 5 minutes), thereby mitigating the toxicity of overdose from subsequent fixed doses since the maximal CNS effect will occur within 25 minutes of a given dose, despite its shorter duration of action due to its larger distribution volume compared with hydrophilic opioids.30 Taken together, these observations on fixed dosing of hydromorphone (whether on a “prn” or “non-prn”—exemplified by patient 2) form the basis for the conclusion that hydromorphone administered in this regimen was causal in these 2 fatalities. This serves to underscore the inadvisability of administering parenteral opioids in fixed doses, particularly given the high risk of respiratory events that occur during postoperative day 1, as exemplified in patient 3.31 This supports the added safety of patient-controlled analgesia (PCA) to patients without a history of opioid drug seeking. Indeed, in situations when PCA is managed with proper individual dose and per hour ceilings combined with lockout time intervals, the total opioid dose exposure to the patient is significantly reduced compared to fixed opioid doses (with accompanying anticipatory anxiety) adding to the perception of pain.32 Therefore, when a PCA system is available in a hospital setting, its exclusive use is strongly recommended for opioids by non-anesthesiologists outside of the operating suite, emergency department, and intensive care unit for reasons of patient safety and effectiveness.33,34 Figure 1 demonstrates a pedagogical tool to correlate the relationships between pharmacokinetics and pharmacodynamics by comparing 2 hypothetical drugs with identical clearances given by a single intravenous bolus. The use of such figures that superimpose the pharmacodynamic concept of minimum effective concentration overlying a pharmacokinetic graph effectively demonstrates the central impact that redistribution has on duration of drug action while simultaneously counteracting the common misperception of the importance of half-life in terminating the drug effect following a single bolus dose. . Illustration of how volume of distribution can influence elimination half-life. Identical doses of drugs A and B are given by rapid intravenous injection. Both have identical values of clearance. However, the total volume of distribution of drug A is 4 times that of drug B. As a consequence, the elimination half-life (t1/2) of drug A is 4 times longer than drug B. Further assume that the minimum effective concentration (MEC) is as shown; the plasma concentration must exceed the MEC for the drug to have clinical activity. As such, the duration of action of drug A, despite its longer half-life, is considerably shorter than that of drug B. Reprinted with permission: Greenblatt DJ. DJ. Pharmacokinetic principles in clinical medicine. J Clin Pharmacol. 1992;32:118–123. It then becomes straightforward to link lipophilicity of an opioid with its volume of distribution and correlate both inversely with the duration of action of a drug administered by intravenous bolus. Furthermore, since the comparative clearance between the 2 drugs in Figure 1 is identical, it becomes easy to show that volume of distribution and drug half-life are directly proportional. The linking together of these 2 pharmacokinetic principles (distribution volume and half-life) by reinforcing the primary, but inverse, correlation that distribution volume has on duration of action can then move the discussion to the impact that drug half-life directly has on drug accumulation by emphasizing the potential for serious toxicity from subsequent parenteral doses if given at fixed intervals. Indeed, the effect of accumulation from administering drugs at fixed dosing intervals at or less than the drug's half-life can readily be demonstrated by manipulating the equation determining drug concentration at steady state35: Teaching this simple relationship underscores the direct proportionality that both dose and half-life (t1/2) have and the inverse proportionality that the dosing interval (DI) has on steady-state plasma concentration (Cpss). Furthermore, the inverse relationship of the 0.693 factor to plasma concentration underscores the fact that administering a drug at its half-life will still cause the drug to accumulate with the administration of fixed doses. Opioids have CNS depressant effects, directly from action on mu opioid receptors and indirectly due to the resultant retention of carbon dioxide.36 Sedative hypnotic drugs act synergistically with opioids by acting on receptors and channels distinct from mu opioid receptors.37–39 Patients with underlying medical conditions, including obstructive sleep apnea and chronic obstructive pulmonary disease, also predispose to carbon dioxide retention, thereby placing these patients in an unfortunate positive feedback loop when opioids are administered to them at inappropriate doses or intervals.20,40,41 Patients 2, 4, and 5 emphasize these principles. In addition, patient 4 once again underscores the problem with accumulation of the hydromorphone in the CNS from too frequent of a dosing interval in relationship to transfer half-life.29 Unfortunately, the drug reached its peak effect in the CNS only after the patient was discharged, resulting in his death at home. Furthermore, patient 4 also demonstrates the synergistic effect of alcohol (decreasing hypoxic drive)37,38 while the concomitant administration of the opioid produced a rightward shift in the end-expiratory CO2 concentration.36 Therefore, only the conscious drive remained, resulting in apnea at the onset of sleep. Similarly, in patient 5, the benzodiazepine receptor agonist, zolpidem, which has been documented to augment the respiratory depressant effect of opioids,39 was administered midway between the dosing of 11 mg of intravenous hydromorphone given over an interval of 6 hours. Patient 2 exemplifies the pitfalls of opioid use in patients with obstructive sleep apnea (OSA) and underscores the importance of understanding the pharmacodynamic principles involved in such patients predisposed to hypoxemia and hypercarbia, as well as to the development and exacerbation of central sleep apnea and ataxic breathing.40 Specifically, patients with OSA are susceptible to a greater respiratory depressant effect of opioids and can suffer severe respiratory depression at plasma concentrations considered therapeutic in patients without the disorder (a shift to the left in the concentration-response curve).41 In each of these patients, a better understanding of the relationship between the specific mechanism that causes drug synergy, particularly for psychoactive compounds with added and highly complex pharmacokinetic relationships, must be taught to medical students and house officers. To maximize such benefit, such information ideally should be conveyed not in an auditorium or laboratory but during daily bedside teaching rounds on hospitalized patients. In addition, physician clinical pharmacologists are in a particularly advantageous position to synthesize all the relevant factors, linking clinical pharmacokinetics and pharmacodynamics with the knowledge of the overall patient status and disease process to maximize the quality of patient care, in the overall context of safe drug prescribing practices. A series of 5 patients are presented wherein serious errors were made by well-intentioned medical practitioners who obviously did not apply the most basic knowledge of clinical pharmacology that provides the underpinnings of safe drug practices to avert prescribing catastrophe. It is my hope that the deaths of these 5 patients did not occur in vain, and that is the primary driver for this article. It is my further hope that the core principles in clinical pharmacology that underlie these cases can be used in medical schools across the country that do offer electives in clinical pharmacology and in other schools within the curriculum at critical points, particularly during bedside clinical teaching rounds, to maximize their impact on the learning process.