In July 2000, the Joint Commission on Accreditation of Healthcare Organizations introduced a new objective to improve the quality of health care in the United States (1). Effective January 1, 2001, institutions wishing to be compliant became responsible for ensuring that pain would be assessed and managed in all patients. The Commission concluded that acute and chronic pain were major causes of patient dissatisfaction in our health care system, leading to slower recovery times, creating a burden for patients and their families, and increasing the costs to the health care system. With these factors in mind, new standards were developed to create higher expectations for the assessment and management of pain in hospitals and other health care settings. The American Pain Society and many other national organizations have endorsed this standard (2). This attention to a major and often avoidable public health problem has the objectives of: recognizing patient rights, assessing pain in all patients, recording the results, determining and ensuring staff competency, and establishing policies educating patients and families about effective pain management (3).
By January, 2001, our institution adopted the policy of “pain as the fifth vital sign.” All nursing staff and support personnel were required to document each patient's rating of their pain intensity using a numerical scale (0 = no pain, 10 = the worst pain imaginable), along with other vital signs, with every routine patient assessment. In addition, treatment of the patient's pain based on their stated numerical pain score was to comply with Numerical Pain Treatment Algorithm (NPTA) guidelines adopted from the National Comprehensive Cancer Network (4) as illustrated in Figure 1. After this institutional policy change, we compared patient satisfaction with pain control and the incidence of opioid-associated adverse drug reactions (ADRs) before and after implementation of the NPTA pain assessment and management standards.
After approval from our IRB, we reviewed patient satisfaction data and every opioid-related ADR reported to our Pharmacy and Therapeutics Committee in the 4 years before (January, 1997 to December, 2000) and 2 years after (January, 2001 to December, 2002) implementation of new pain management standards. According to pharmacovigilance recommended by the World Health Organization's (WHO) International Drug Code (IDC-9), (5) an ADR is defined as any observed response to a drug that is noxious and unintended occurring at doses used in humans for prophylaxis, diagnosis, therapy of disease, or modification of physiologic function. Therapeutic failures were excluded. Pharmacists, physicians, nurses, risk managers, and other hospital personnel report ADRs. The current reporting program at H. Lee Moffitt Cancer Center and Research Institute was initiated in 1993. The ADRs are collected and the data and reports are maintained by the pharmacy and hospital performance improvement departments. There is at least one decentralized clinical pharmacist assigned to each of the patient care areas, and these areas contain from 8 to 20 patients. We obtained the records of all opioid-related ADRs reported at our institution during the study period, and these were coded by opioid type and reaction. When an ADR pertained to oversedation or respiratory depression, the following data were extracted from the patient's medical record: age, diagnosis, opioid prescribed, prescribing physician, level of consciousness, respiratory rate, requirement for intensive care, requirement for mechanical ventilation, and hospital survival.
For the study period, we also obtained the total number of hospital patient admissions, inpatient patient days and Case Mix Index (CMI) using APR-DRG (All-Patient Refined Diagnostic Related Group), a widely used severity of illness and risk of mortality adjustment, to adjust for possible effects of these factors. The reports of patient satisfaction surveys were also obtained. Patients rated satisfaction with how well their pain was controlled while hospitalized from 1 = poor to 5 = excellent as part of a 55-item postdischarge survey conducted by paid employees of the Patient Relations Department. One-hundred complete responses were obtained each month. A Poisson regression model (6) was fit to period (pre- versus postpolicy change) using the GENMOD procedure of SAS statistical software, version 8.2 (Statistical Analysis Software, Cary, NC). Intergroup age, Medicare CMI, and patient rating of pain control (Student's t-test), gender distribution (Pearson's χ2 test), and incidence data (Pearson's χ2 or Fisher's exact test) were compared using appropriate statistical testing. Intergroup difference with a P value of 0.05 or less was considered significant.
Since 1993, annual reporting of all ADRs at H. Lee Moffitt Cancer Center and Research Institute as a percentage of admissions has a stable reporting frequency that did not vary significantly during the study period. Patient demographics did not change significantly during the study. However post-NPTA patient acuity, as measured by increase in the Medicare CMI, did undergo a small yet significant change. Patient-rated satisfaction with control of their pain increased (Table 1). There were 79 opioid ADRs reported for 117,672 inpatient hospital days from January 1997 through December 2000. Of these, 13 events involved patient oversedation or respiratory depression requiring discontinuation of the opioid or reversal with naloxone. From January 2001 through December 2002, 67 opioid ADRs were reported for 65,388 inpatient hospital days, with 16 involving patient oversedation or respiratory depression requiring discontinuation of the opioid or reversal with naloxone.
Summarized in Table 2 are data collected from the 13 oversedation events before implementation of the pain algorithm. Ages ranged from 33 to 87 yr, with a mean of 58.3 yr. There were 8 females and 5 males in the group. Nine of the events involved morphine and 2 of the events involved hydromorphone. Mean respiratory rate recorded in the patients' chart before the event was 20 breaths/min. Six patients received naloxone and 4 patients were transferred to the intensive care unit (ICU). Of these patients, 3 were tracheally intubated and 2 died.
Table 3 presents the 16 events that occurred after implementation of the guidelines. Patient ages ranged from 29 to 75 yr, with a mean of 60.6 yr. There were 9 females and 7 males. Four and 6 events involved morphine and hydromorphone, respectively. Other opioids noted were fentanyl, oxycodone, and intrathecal morphine. Respiratory rate averaged 18 breaths/min before the oversedation event. Eleven patients received naloxone during the event, 7 were transferred to the ICU, 3 required tracheal intubation for airway protection, and 1 patient died.
Comparing the opioid-related ADRs and cases of oversedation pre- and post-NPTA standards, we discovered a significant increase in the incidence of both events post-NPTA (P = 0.01 and P = 0.03, respectively). The overall rate of opioid ADRs increased by 49%, with a ratio of the rate of opioid-related ADRs post-NPTA to pre-NPTA of 1.49 (95% confidence interval [CI], 1.08–2.07). For sedation ADRs, the corresponding postimplementation to preimplementation ratio was 2.22, a more than 2-fold increase in the rate of opioid oversedation ADRs. The 95% CI was 1.07–4.60.
For opioid oversedation ADRs, the incidence increased in the post-NPTA period from 11.0 per 100,000 inpatient hospital days pre-NPTA to 24.5 per 100,000 inpatient hospital days post-NPTA (P < 0.001). In these patients, we found that 27 of the 29 had a decrease in their level of consciousness documented in the nursing progress notes during the 12 h before the event. In only 3 of the 29 patients was a decreased respiratory rate (<12 breaths/min) noted in the chart before the event. Numerical pain scores in the post-NPTA patients were variable and not predictive of ADR or oversedation event. No other variables measured in this study showed a statistically significant difference.
In addition to the number and type of ADRs that occurred, we also evaluated patient satisfaction with their pain management using our patient satisfaction survey. Because not all patients responded to all 4 questions, our data provide responses to between 326 and 377 patients on the 4 questions from before the implementation of the NPTA and between 1299 and 1354 patients for the years after the implementation. In the pre-NPTA period, patients responded to the 4 questions with “excellent” 46%, 56%, 58%, and 55% of the time compared with 55%, 66%, 68%, and 66% respectively for the post-NPTA period. Patients responded “poor” to the 4 questions 2%–3% of the time in the pre-NPTA period and 1% or less in the post-NPTA period. Wilcoxon's rank-sum test comparing these 2 time periods produced highly significant results (each P < 0.0001) indicating better responses in the post-NPTA period for all 4 questions.
Our compliance with NPTA pain management standards commencing January, 2001 was associated with a significant increase inpatient satisfaction with control of their pain, but coincidentally we observed a significant increase in the incidence of opioid ADRs and opioid oversedation or respiratory depression events. During this period, we observed no significant differences in age or gender distribution of the patient population, which may have impacted opioid-induced respiratory depression (7). The Medicare CMI increased by a small but significant amount, and it is not possible to determine if this contributed to the increased incidence of ADRs. Also, the change in CMI may be attributed to better coding procedures, as coding now affects hospital reimbursement, which may have resulted in both a substantial increase in the accuracy of coding and an incentive to code in higher-weighted DRGs.
Before experiencing an opioid oversedation ADR, 94% of our patients exhibited an altered level of consciousness, yet the recorded numerical pain scores varied widely. These findings highlight an inherent patient safety concern when titrating opioid analgesia to a one-dimensional pain rating scale.
The International Association for the Study of Pain Subcommittee on Taxonomy defines pain as “an unpleasant sensory and emotional experience associated with actual or potential tissue damage” and “is always unpleasant and therefore also an emotional experience” (8). Research is consistent with the hypothesis that multiple components contribute to the pain experience. These include a physiological component (10), sensory, affective, and cognitive components (11), a behavioral component (12), and the sociocultural component (9–13). McGrath (14) has suggested that measurement of pain should be distinguished from the assessment of pain. The problem with the NPTA is that the psychological and emotional components are not measured by a one-dimensional rating, and they are poorly treated by opioids.
In addition to the demonstrated complexity of pain assessment, pain management by opioid titration protocols can be unsafe for other reasons. Clinical factors that will inevitably impact the patient's response to opioid titration must always be considered. Extremes in age, body weight, and comorbid conditions have been shown to predispose patients to opioid oversedation (15,16). Opioids depress ventilation and cause an increase in the partial pressure of arterial carbon dioxide. In high-risk patients, the classic awakening response of hypercarbia can be obtunded and sedation can progress to airway obstruction and respiratory arrest (15). Patients with a history of sleep apnea or chronic obstructed pulmonary disease are particularly prone to this phenomenon and should be treated judiciously (16). Also, patients with severe liver or kidney disease can exhibit an increased sensitivity to certain opioids with accumulation and potential for prolonged effects (17). Finally, demented patients or those who become confused after receiving opioids can report unreliable numerical pain scores and rapid opioid titration then becomes ill-advised.
We also examined the predictive value of respiratory rate in assessing a patient's propensity for opioid oversedation. Pharmacodynamics suggests that as a result of the physiologic response to the accumulation of carbon dioxide (i.e., peripheral and central chemoreceptor stimulation and eventual narcosis), respiratory rate and tidal volume should be reliable indicators of opioid-induced respiratory depression (18). In our study, however, we did not observe a predictable decrease in respiratory rate before opioid-induced respiratory arrest. Factors other than pharmacodynamics may explain why observed respiratory rate was not diminished before ventilatory failure. In particular, the respiratory rate may have been altered by arousing the patient to obtain other vital signs and pain scores or it may have been falsely equated with abdominal breathing movements, which might occur without air exchange in sedated patients experiencing airway obstruction. In this context, chest wall movements could occur without effective ventilation, and even monitoring respiratory rate with conventional transthoracic impedance would be ineffective. Pulse oximetry has been demonstrated to be a more reliable indicator of hypoventilation in patients who are not receiving supplemental oxygen (19). Therefore, pulse oximetry might be used effectively in patients at high risk to avoid opioid-induced respiratory depression and subsequent ventilatory arrest. The use of supplemental oxygen to reduce the severity of hypoxemia has also been studied by Stone et al. (20). According to their article, patients who received nocturnal oxygenation while receiving a morphine patient-controlled analgesia (PCA) had a less severe level of hypoxemia as compared with patients breathing room air. A recent article by Fu et al. (21), however, found that supplementation of inspired oxygen with an Fio2 >0.25 could mask potential hypoventilation and put a patient at risk for carbon dioxide narcosis before the Spo2 would decrease to less than 90%.
Overall, our findings of a recently increased incidence of opioid ADRs are particularly concerning because the previous nationwide trend had been toward a decrease in ADRs. Lazarou et al. (22) synthesized results from 39 studies conducted in the United States from 1964 to 1996 and discovered a mean incidence of serious ADRs per patient admissions of 6.7% (95% CI, 5.2%–8.2%) and of fatal ADRs of 0.32% (95% CI, 0.23%–0.41%) of patient admissions. These data suggest that in-hospital ADR ranked between the fourth and sixth leading cause of death in hospitalized patients in the United States. In 2003, Wiffen et al. (23) conducted a meta-analysis of 68 studies including 413,000 patients in which the incidence of ADR was indexed to hospital admissions and weighted by sample size. The mean ADR rate was also 6.7% (95% CI, 6.6–6.8), but when ADR rate was differentiated by studies published before and after 1985, the mean rates were 12.5% and 3.4%, respectively. Reviewing opioid ADRs, Looi-Lyons et al. (24) reported a study of 4,000 patients undergoing postoperative IV PCA in which only 9 patients experienced respiratory problems. This resulted in an incidence of opioid respiratory depression of 2.25 per 1000 patients treated. Considering an average of 40 percent of the patients admitted to our institution undergo IV PCA, the incidence of oversedation or respiratory depression would translate to 1.59 per 1000 patients before January, 2001 and 3.45 per 1000 patients after January, 2001. In addition, a recent study by Taylor et al. (25) found that patients receiving IV PCA for postoperative pain control reached a level of sedation similar to those who received “conscious sedation” for colonoscopy procedures. They concluded that patients might reach dangerous levels of sedation during the first 24 hours postoperatively.
To determine if other hospitals had experienced a similar increase in ADRs, we examined the Medmarx database, which is an anonymous, Internet-accessible, standardized program used by hospitals nationwide to report and track medication errors. According to the database, there has also been an increase in opioid-related errors in each of the last 2 years compared with 2000. In fact, the percentage of morphine-related errors resulting in patient harm increased from 4% in 2000 to 6% in 2001, whereas concurrent error rates associated with insulin and heparin—typically associated with frequent error rates—were unchanged at 9% and 5%, respectively. In 2002, 4 of the top 10 medications with the most frequent error rates involving harm were opioids. Cumulatively, these opioids account for 12.4% of all errors in 2002, exceeding any other class of medication and up from 9.5% in 2001. These data suggest that our findings are consistent with a nationwide trend in increased opioid-related adverse drug reactions (26)
Limitations of this study include the reliance on a single institution's self-reporting of ADRs, the potential impact of increased pain vigilance on reporting, and the lack of patient acuity adjustment. Self-reporting typically captures only a fraction of the events that occur in a population. Thus, any change in reporting frequency has a major impact on the rate of events. Although it is possible that the reporting frequency increased after the new pain standards were implemented, we did not experience an overall increase in hospital-wide ADRs reported. This may be because the new pain standards primarily affected the activities of the nursing staff, whereas pharmacists more commonly complete ADR reports. The MEDMARX data from nearly 500 institutions suggest that the problem of opioid safety is not limited to our institution. Another potential limitation is the lack of data on the level of pain control achieved in our patient population. It is possible that in a cancer population such as ours, an increase in side effects may be an acceptable consequence of achieving a greater degree of pain relief. Indeed, Moffitt patient satisfaction surveys indicated an increase in patient satisfaction (Table 3) in the time periods after implementation of the NPTA. However, we believe that a modified pain treatment algorithm could allow the same pain control without adverse drug reactions. Pain control safety measures should operate independently of patient diagnosis or acuity.
Consequent to our findings, H. Lee Moffitt Cancer Center has adopted a new pain/consciousness rating scale, which reflects not only the subjective patient report of pain but also a clinical assessment of the patient's level of consciousness. The pain/consciousness rating includes a numerical scale (0 to 10; 0 = no pain, 10 = the worst possible pain) and simple nursing assessment of level of consciousness (A, B, C, or D; A = Awake and alert, B = sleeping But easily aroused by voice only, C = Consciousness impaired with arousal only by stimulation or Confused, D = Disoriented). A patient with a consciousness assessment of C or higher triggers a physician notification and placement of a pulse oximeter without supplemental oxygen before administration of any additional opioids. From this point, adjunctive medications, such as acetaminophen, nonsteroidal anti-inflammatory drugs (NSAIDs), anticonvulsants, transdermal local anesthetics, or even regional anesthetic blocks, must also be considered before further dose escalation with opioids. Opioids are the mainstay of cancer pain management, and adjuvant analgesics typically are considered after opioid therapy is optimized. However, adjuvants as first-line drugs are recommended both in WHO guidelines (27) and National Comprehensive Cancer Network guidelines (28). The principal rationale for using adjuvant medicines is that they provide an opioid-sparing effect with attendant reduction in adverse side effects, but evidence for this strategy is contradictory (29). Contemplation of using adjunctive drugs is recommended in all cases, but differential efficacy may inform a context-specific choice. For example, patients suffering boney metastases may benefit from the combination of opioid and NSAIDs (30). Individual differences in cancer pain manifestation and opioid/adjuvant effectiveness and in their side effects are major limitations for individualized pain treatment, and future research should clarify context-specific therapeutic advantages and disadvantages of opioids and adjuvants to facilitate evidence-based practice guidelines.
The authors thank John B. Downs, MD for his editorial assistance and Mark W. Rolfe, MD for his clinical guidance.
1. Phillips DM. JCAHO pain management standards are unveiled. JAMA 2000;284:4–5.
4. The National Comprehensive Cancer Network Cancer Pain Guidelines (Version 1. 2002) The Complete Library of NCCN Clinical Practice Guidelines in Oncology [CD-ROM] (May 2003). Jenkintown, PA: © National Comprehensive Cancer Network, 2003. Also Available at: www.nccn.org
5. World Health Organization. International drug monitoring: the role of the hospital. Geneva: WHO; 1996. Technical support series, no. 425.
7. Dahan A, Sarton E, Teppema L, Olievier C. Sex-related differences in the influence of morphine on ventilatory control. Anesthesiology 1998;88:903–13.
9. Clark WC, Yang JC, Tsui SL, et al. Unidimensional pain rating scales: A multidimensional affect and pain survey (MAPS) analysis of what they really measure. Pain 2002;98:241–7.
10. Merskey H, Bogduk N, eds. Classification of chronic pain, 2nd ed. Seattle: IASP Press, 1994.
11. Melzack R, Wall PD. Pain mechanisms: A new theory. Science 1965;50:971–9.
12. Fordyce WE. Behavioural methods for chronic pain and illness. St. Louis: Mosby, 1976; 1–236.
13. Wolff BB. Ethnocultural factors influencing pain and illness behavior. Clin J Pain 1985;1:23–30.
14. McGrath PA. Assessment of pain in children. In Bond MR, Charlton JE, Woolf CJ, eds. Proceedings of the VIth world congress on pain. Amsterdam: Elsevier, 1991:509–26.
15. Miller RD. Anesthesia, 5th ed. Intravenous Opioid Anesthetics. New York: Churchill Livingstone, 2000:324–9.
16. Lewis KP, Stanley GD. Pharmacology. Int Anesthesiol Clin 1999;37:73–86.
17. Rand CW, Kuldav JM, Yost RL. Obesity and post-operative pain. J Psychosom Res 1985;29:43.
18. Mueller RA, Lundberg DB, Breese GR, Hedner J, et al. The neuropharmacology of respiratory control. Pharmacol Rev 1982;34:255–85.
19. Fu E, Neymour R, Downs JB. Routine supplemental oxygen is not necessary during post-anesthesia recovery. Anesth Analg 1999;88:S39.
20. Stone J, Cozine K, Wald A. Nocturnal oxygenation during patient-controlled analgesia. Anesth Analg 1999;89:104–10.
21. Fu E, Downs JB, Schweiger JW, et al. Supplemental oxygen impairs detection of hypoventilation by pulse oximetry. Chest 2004;126:1552–8.
22. Lazarou J, Pomeranz BH, Corey PN. Incidence of adverse drug reactions in hospitalized patients: a meta-analysis of prospective studies. JAMA 1998;279:1200–5.
23. Wiffen P, Gill M, Edwards J, Moore A. Adverse drug reactions in hospital patients: A systematic review of the prospective and retrospective studies. Bandolier Extra, 2002. Available at http://www.jr2.ox.ac.uk/bandolier/extra.html
. Accessed August 13, 2003.
24. Looi-Lyons LC, Chung FF, Chan VW, McQuestion M. Respiratory depression: an adverse outcome during patient controlled analgesia therapy. J Clin Anesth 1996;8:151–6.
25. Taylor S, Voytovich AE, Kozol RA. Has the pendulum swung too far in postoperative pain control? Am J Surg 2003;186:472–5.
29. Lussier D, Huskey AG, Portenoy RK. Adjuvant analgesics in cancer pain management. Oncologist 2004;9:571–91.
30. Viney RP, Hayne D, Ayra M, Patel HR. Pharmacological management of metastatic boney pain. Expert Opin Pharmacother 2004;5:1555–63.