Effects of High Dose Opioids and Sedatives on Survival in Terminally Ill Cancer Patients

Transcription

1 282 Journal of Pain and Symptom Management Vol. 21 No. 4 April 2001 Original Article Effects of High Dose Opioids and Sedatives on Survival in Terminally Ill Cancer Patients Tatsuya Morita, MD, Junichi Tsunoda, MD, Satoshi Inoue, MD, and Satoshi Chihara, MD Seirei Hospice, Seirei Mikatabara Hospital, Shizuoka, Japan Abstract Concerns that high dose opioids and sedatives might shorten patient survival could contribute to insufficient symptom alleviation for terminally ill cancer patients. To examine the effects of opioids and sedatives prescribed in the final 48 hours on patient survival, a re-analysis of the prospectively collected data was performed on 209 hospice inpatients. Patient characteristics and clinical symptoms were prospectively recorded, and information about the use of opioids and sedatives in the last two days was collected by a chart review. Opioids were prescribed in 82% of the patients, with a median dose of 80 mg oral morphine equivalent (OME)/48 hours. Sixty percent received some sedative medications, mainly haloperidol (43% of total sample, 7.5 mg/48 hours), midazolam (23%, 23mg/48 hours), and hydroxyzine (15%, 50 mg/48 hours). There were no significant differences in survival between the patients who received different doses of opioids ( 240, , and 600 mg OME/48 hours) and of benzodiazepines (0, 1 59, and 60 mg parental midazolam equivalent/48 hours). Also, the survival of patients with haloperidol, hydroxyzine, and other sedative medications did not differ from those without. Furthermore, an addition of use of opioids and sedatives in the final 48 hours into the multiple regression model for survival prediction achieved no significant increase in predictability. In conclusion, opioids and sedatives used for symptom control in the last days are not associated with patient survival. They are safe and useful medications to palliate severe distress in the terminal stage of cancer when administered with a low initial dosage and adequate titration. J Pain Symptom Manage 2001;21: U.S. Cancer Pain Relief Committee, Key Words Advanced cancer, palliative care, sedation, sedative, opioid, survival Address reprint requests to: Tatsuya Morita, MD, Seirei Hospice, Seirei Mikatabara Hospital, 3453 Mikatabara-cho, Hamamatsu, Shizuoka , Japan. Accepted for publication: May 15, Introduction Sedation for symptom relief has been actively discussed in recent palliative care literature. 1 Many empirical studies have demonstrated that a considerable number of cancer patients required sedation to alleviate severe symptoms in the final days. 2 7 While some palliative care practitioners believe that sedation is a necessary part of basic palliative care, some clinicians have expressed an ethical dilemma and hesitation in the practice of sedation. 8,9 In a postal survey on physicians attitudes on sedation, 58% of experts in palliative medicine U.S. Cancer Pain Relief Committee, /01/$ see front matter Published by Elsevier, New York, New York PII S (01)

2 Vol. 21 No. 4 April 2001 Effects of High Dose Opioids and Sedatives on Survival 283 stated that the decision to use terminal sedation was somewhat or very difficult. 8 While there have been no systematic studies to investigate origins of such physicians dilemma, it is not difficult to image that concerns that sedation might shorten a patient s life would contribute to the hesitation. It is assumed that lack of established evidence about potential life-threatening effects of sedation would make it more difficult for many physicians to confidently make a decision to provide sedation, and might result in insufficient symptom palliation in the last stage of cancer. Thus, clarification of the effects of sedation on patient survival is of importance to adequately provide comfortable care to terminally ill cancer patients and discuss appropriateness of sedation therapy from the medical and ethical points of view. To date, one prospective observational study has suggested that sedation has no apparent life-shortening effect because target symptoms for sedation themselves reflected fatal organ dysfunction and patients were sedated in a way that minimized lifethreatening risk. 10 However, this study was designed as observational and had no control groups, so that the conclusion has only limited value. On the other hand, three preliminary studies have shown that there were no significant differences in survival between sedated and non-sedated patients in palliative care settings. 3,6,7 The mean (median) survival from the reference to palliative care programs to death of the sedated and non-sedated were 9 5 days vs. 6 7 days, 25 days 23 days, and days vs days, respectively, in the Italian, South African, and U.K. studies. 3,6,7 However, these studies also have considerable limitations in the capacity to determine the effects of sedation on survival. That is, first, the definitions of sedation vary among different studies and its diagnostic validity has not been fully established. Second, the retrospective nature of these studies weakens the reliability of the results. Third, no studies evaluated prognostic factors for each patient and none used multivariate analyses to control for patients varied backgrounds in comparisons of survival. In this study, we intended to compare the survival of sedated and nonsedated patients receiving inpatient palliative care, with clear definitions on sedative treatment and using multivariate analytic techniques to adjust prognostic factors of each patient. Methods This is a re-analysis of data prospectively collected for another study, which was conducted to identify prognostic factors for terminally ill cancer patients. 11 In that study, consecutive cancer patients admitted to our palliative care unit from September 1996 to August 1997 and from October 1997 to April 1998 were enrolled. Patient characteristics, Palliative Performance Scale (PPS), 12 and clinical symptoms were prospectively recorded by attending physicians upon admission, and patients were followed up to six months. The PPS was a modified Karnofsky Performance Scale (KPS) which rated patients general condition as 0 (death) to 100 (normal). 12 The clinical symptoms investigated were reduced oral intake, edema, dyspnea at rest, and delirium. Our previous report had established that these were independent prognostic factors for terminally ill cancer patients. 13 Oral intake was scored as normal, moderately reduced (reduced but more than mouthfuls) and severely reduced (mouthfuls or less). If patients received total parenteral nutrition for bowel obstruction, they were included in a normal oral intake category. The other symptoms were rated as present or absent. Delirium was diagnosed based on the Diagnostic and Statistical Manual of Mental Disorders (4th edition), 14 and delirium caused by one medication was excluded because it did not always indicate general conditions. 15 The methods for data collection have been described elsewhere in detail. 11,13 For this study, additional information about the use of opioids and sedatives in the final 48 hours was collected by a chart review. The period 48 hours was determined on the basis of previous empirical reports, which had demonstrated the mean duration of sedation to be 1.3 to 3.9 days in palliative care settings. 5 7 Also, because definitions on sedation had not been established and the retrospective nature of this study did not allow to definitely clarify indications for sedative medications, this study examined all sedative psychotropics available in our practice. The sedatives investigated were benzo-

3 284 Morita et al. Vol. 21 No. 4 April 2001 diazepines (midazolam, flunitrazepam, bromazepam, diazepam), haloperidol, hydroxyzine, chlorpromazine, levomepromazine, barbiturate, and propofol. Sedatives prescribed to alleviate systemic convulsion and oral benzodiazepines were excluded, whereas haloperidol as an antiemetic was included because many patients received haloperidol for multiple indications in the latest terminal stage (e.g., nausea and delirium). For analyses, strong sedatives were defined as chlorpromazine, levomepromazine, barbiturate, and propofol. Opioid dosage was converted to an oral morphine equivalent (OME) following a standard ratio, 16 and benzodiazepine dosage was calculated as a parenteral midazolam equivalent (PME) following the equation: parenteral midazolam 10mg parenteral flunitrazepam 2mg rectal bromazepam 6mg rectal diazepam 20 mg Of 245 cases in the previously obtained data, 36 cases were excluded because they received an overlapped assessment across two study periods (n 15), they were still alive and information about the final days could not be obtained (n 10), they died at home after discharge (n 9), and the records were incompletely described (n 2). Thus, 209 patients were analyzed in this study. The subjects comprised 112 males and 97 females, whose mean age was years. The average survival was days (median 26). Primary tumor sites were the lung (n 45); stomach (n 35); colon (n 22); rectum (n 14); breast (n 11); esophagus, liver, biliary system, pancreas (n 10); prostate (n 8); unknown (n 7); ovary, blood (n 5); neck (n 4); kidney, uterus, soft tissue (n 3); and others (brain, thymus, skin, bladder). Statistical Analyses First, to identify prognostic factors for the patients in this study, a multiple regression analysis was conducted using survival intervals as dependent variables, and PPS and the clinical symptoms as independent variables, followed by the initial screening by the Kaplan Meier method with log rank test. The Cox regression model was not adopted because every case experienced the terminal event and the proportional hazards assumption did not hold in several analyses. Survival was defined as the intervals from admission to death, and was converted to logarithm for analyses to archive normal distribution because the original data were considerably skewed. PPS was classified into three groups: PPS 10 20, PPS 30 50, and PPS 60. Because both analyses in which the PPS was calculated as a categorical or ordinal variable achieved the same solution, the PPS was represented as an ordinal variable converted to 2 (PPS 10 20), l (PPS 30 50), and 0 (PPS 60). Similarly, oral intake was presented as 2 (severely reduced), 1 (moderately reduced), and 0 (normal). Edema, dyspnea at rest, and delirium were categorized as absence (0) or presence (1). Second, to explore the effects of use of opioids and sedatives in the final two days on patients survival, the survival curves were compared among the patients with different opioid and sedative medications. The survival curves were calculated by the Kaplan Meier method and comparisons were based on the log rank test. Finally, to examine how opioids and sedatives influenced survival after adjustments of patients backgrounds, opioid, and sedative use in the last 48 hours were added as independent variables in the multiple regression model for survival prediction, and the change of variances explained (R-square) was investigated. Use of opioids was calculated as an ordinal variable converted to 2 ( 600 mg OME/48 hours), 1 ( mg OME/48 hours), and 0 ( 240mg OME/48 hours or not used); and benzodiazepine use was expressed as 2 ( 60 mg PME/48 hours), 1 (1 59 mg PME/ 48 hours) and 0 (not used). Use of haloperi- Table 1 Patient Performance Status and Symptom Prevalence (n 209) n % Palliative Performance Scale Oral Intake Normal Moderately reduced Severely reduced Edema Dyspnea at rest Delirium 63 30

4 Vol. 21 No. 4 April 2001 Effects of High Dose Opioids and Sedatives on Survival 285 Table 2 Opioids and Sedatives in the Final 48 Hours Frequency n (%) Mean Dose (mg/48 hrs) Median Dose (mg/48 hrs) Range (mg/48 hrs) Opioids 172 (82) a 80 a a Sedatives 126 (60) Benzodiazepines 57 (27) b 20 b b Midazolam 48 (23) Flunitrazepam 9 (4.3) Bromazepam 7 (3.3) Diazepam 4 (1.9) Haloperidol 89 (43) Hydroxyzine 31 (15) Strong Sedatives 11 (5.3) Chlorpromazine 6 (2.9) Levomepromazine 2 (0.96) Barbiturate 3 (1.4) c c c Propofol 3 (1.4) a Oral morphine equivalent. b Parenteral midazolam equivalent. c Phenobarbital 600 mg (n 1); amobarbital 2000 mg (n 1); amobarbital 600 mg and thiamylal 2400 mg (n 1). dol, hydroxyzine, or strong sedatives was categorized as absence (0) or presence (1). All analyses were performed using the Statistical Package for the Social Science (ver. 9.0). Results Patients physical conditions on admission are described in Table 1. The PPS was less than 60 in 85% of the patients, and 20 37% had some symptoms that indicated poor prognoses. Opioids were prescribed in 82% of the patients (n 172) with a median of 80 mg/48 hours (Table 2). The dose of opioids was none or less than 240 mg OME/48 hours in 78% (n 164), mg OME/48 hours in 14% (n 29), and 600 mg OME/48 hours or more in 7.7% (n 16). Opioids prescribed were morphine (77%, n 133), buprenorphine (20%, n 35), and others (fentanyl, eptazocine; 2.3%, n 4). Administration routes were continuous-subcutaneous (52%, n 89), repeated or single subcutaneous and/or intravenous (19%, n 32), continuous-intravenous (17%, n 30), oral (6.4%, n 11), and rectal (5.8%, n 10). Sixty percent of the patients received some sedative medications in the last two days (Table 2). Sedatives frequently administered and their median dose were haloperidol (43% of total sample, 7.5 mg/48 hours), midazolam (23%, 23 mg/48 hours), and hydroxyzine(15%, 50mg/48 hours). Of 57 patients who received benzodiazepines, 40 patients (19% of the total sample) received less than 60 mg PME/48 hours and 17 (8.1%) received 60 mg PME/48 hours or more. Bromazepam and diazepam were used rectally and the other sedatives were administered parenterally. PPS, oral intake, edema, dyspnea at rest, and delirium were identified as significant prognostic factors by univariate analyses, and a multiple regression analysis revealed that they had an independent prognostic value (Table 3). R-square of this regression model was Figure 1 shows the survival curves of patients who received different doses of opioids ( 240, , and 600 mg OME/48 hours) and of benzodiazepines (0, 1 59, and 60 mg PME/ 48 hours). There were no significant differences in patient survivals between the groups. Table 3 Prognostic Determinants by the Multiple Regression Model Regression Coefficients Standard Error Palliative Performance Scale a Reduced oral intake b Edema c Dyspnea at rest c Delirium c a Coded as 2 (PPS 10 20), 1 (PPS 30 50), and 0 (PPS 60). b Coded as 2 (severe), 1 (moderate), and 0 (absent). c Coded as 1 (present) and 0 (absent). P

5 286 Morita et al. Vol. 21 No. 4 April 2001 Fig. 1. Survival of patients with opioids and benzodiazepines. As well, the survivals of the patients who received haloperidol, hydroxyzine, and strong sedatives did not differ from those who did not (Figure 2). Table 4 demonstrates the change of R-square by adding use of opioids and sedatives in the final two days as independent variables into the multiple regression model for survival prediction. Each addition achieved no significant increase in predictability of patient survival. Fig. 2. Survival of patients with different sedatives. Discussion Concerns that high-dose opioids and sedatives might hasten a patient s death can be a cause of physicians psychological conflicts in palliative care settings, and could contribute to insufficient symptom palliation in the latest stage of cancer. In this study, the effects of opioids and sedatives prescribed in the final 48 hours on patient survival were investigated. In general, opioids and sedatives used for symptom control in the final days were not associated with survival in our populations. By univariate analyses, the use of opioids, benzo-

6 Vol. 21 No. 4 April 2001 Effects of High Dose Opioids and Sedatives on Survival 287 Table 4 Effects of Addition of Use of Opioids and Sedatives in Multiple Regression Model for Survival Prediction 95% CI of Regression Coefficients P R-square a Opioid dose b Use of sedatives c Benzodiazepine dose d Use of haloperidol c Use of hydroxyzine c Use of strong sedatives c a R-square of the original model is b Coded as 2 ( 600 mg oral morphine equivalent/48 hours), 1 ( mg), and 0 ( 240 mg). c Coded as 1 (administered), and 0 (not administered). d Coded as 1 ( 60 mg parenteral midazolam equivalent/48 hours), 1 (1 59 mg), and 0 (not used). diazepines, haloperidol, hydroxyzine, or strong sedatives (i.e., chlorpromazine, levomepromazine, barbiturate, and propofol) showed no significant influence on patient survival. Also, the addition of opioids and sedatives as independent variables into the multiple regression model for survival prediction achieved no significant increase in predictability. These results support findings of preliminary reports that performance of sedation had no effects on patient survival. 3,6,7,10 The opioid dose in our patients corresponded with that found in previous reports A recent survey from Israel has demonstrated that the proportion of patients who required morphine at a dose of 300 mg/day or more was 12%, 20 and the proportion in this study was 7.7%. Likewise, the doses of sedatives administered for our patients in the final days were similar to or less than those shown in earlier reports (Table 5). For example, the mean dose of midazolam in this study was 26 mg/day and the maximum dose was 100 mg/day, whereas many observations in palliative care settings have demonstrated that the mean dose of midazolam prescribed in the final days was mg/day with a maximum of 240 mg/day. 3,6,23 27,29,30 Also, haloperidol, a major drug for terminal delirium, was administered to our populations at a dose of mg/day, while previous reports have revealed that it effectively palliated delirious symptoms of cancer patients at a dose of 5 24 mg/day in palliative care and mg/day in a psychiatric setting. 6,8,38,39 Also, other sedatives such as flunitrazepam, chlorpromazine, levomepromazine, barbiturate, and propofol were used at the doses within reported ranges. 6,8,31 37 These findings suggest that sedatives were effective at a relatively low dose for symptom palliation in the final days and larger dosages were rarely required. This study has several limitations. First, the control group of this study is not a definite control that is, they are not patients who suffered from refractory severe symptoms but did not choose sedatives. Therefore, it is difficult to analytically separate the effects of sedative medications on patient survival from those of the underlying symptoms themselves for whom sedatives and opioids were administered. We believe, however, that as a randomized controlled trial with such a control group cannot Table 5 Dose of Sedatives in Final 48 Hours Mean Dose (mg/day) Median Dose (mg/day) Range (mg/day) Reference Benzodiazepines Midazolam ,6,23 27,29, ,28 Flunitrazepam a ,32 Haloperidol ,8 Chlorpromazine ,33 Levomepromazine ,8 Barbiturate Phenobarbital ,34 Amobarbital Propofol ,36,37 a Usual dose.

7 288 Morita et al. Vol. 21 No. 4 April 2001 be accepted for ethical reasons, this study design would be the second best way. Next, as available sedatives vary among different institutions and settings, these findings could not be generalized to other situations where sedatives that were rarely used in our practice are frequently prescribed. Finally, as the subjects studied were limited to Japanese cancer patients, the results could not be generalized to other populations with different fatal diseases. Nonetheless, this study provides an important suggestion that opioids and sedatives are safe and useful medications to palliate severe distress in the latest stage of cancer when they are administered with a low initial dose and adequate titration. Further studies should be focused on a multicenter prospective design to examine the effects of various sedative medications on patient survival, as well as intervention trials to develop valid strategies for palliative sedation therapy. References 1. Chemy NI, Portenoy RK. Sedation in the management of refractory symptoms: guidelines for evaluation and treatment. J Palliat Care 1994;10(2): Fainsinger R, Miller MJ, Bruera E, et al. Symptom control during the last week of life on a palliative care unit. J Palliat Care 1991;7(1): Fainsinger R, Landman W, Hoskings M, Bruera F. Sedation for uncontrolled symptoms in a South African hospice. J Pain Symptom Manage 1998;16: Lichter I, Hunt E. The last 48 hours of life. J Palliat Care 1990;6(4): Morita T, Inoue S, Chihara S. Sedation for symptom control in Japan: the importance of intermittent use and communication with family members. J Pain Symptom Manage 1996;12: Stone P, Phillips C, Spruyt O, Waight C. A comparison of the use of sedatives in a hospital support team and in a hospice. Palliat Med 1997;11: Ventafridda V, Ripamonti C, De Connno F, et al. Symptom prevalence and control during cancer patients last days of life. J Palliat Care 1990;6(3): Chater S, Viola R, Paterson J, Jardis V. Sedation for intractable distress in the dying a survey of experts. Palliat Med 1998;12: Fondras JC. Sedation and ethical contradiction. Eur J Palliat Care 1996;3: Morita T, Tsunoda J, Inoue S, Chihara S. Do hospice clinicians sedate patients intending to hasten death? J Palliat Care 1999;15(3): Morita T, Tsunoda J, Inoue S, Chihara S. Survival prediction of terminally ill cancer patients by clinical symptoms: a development of a simple indicator. Japan J Clin Oncol 1999;29: Anderson F, Downing GM, Hill J. Palliative Performance Scale (PPS): a new tool. J Palliat Care 1996;12(1): Morita T, Tsunoda J, Inoue S, Chihara S. The Palliative Prognostic Index: a scoring system for survival prediction of terminally ill cancer patients. Support Care Cancer 1999;7: American Psychiatric Association. Diagnostic and Statistical Manual of Mental Disorders, 4th ed. Washington, DC: American Psychiatric Association, Breitbart W, Passik AD. Psychiatric aspects of palliative care. In: Doyle D, Hanks GCW, Mac- Donald N, eds. Oxford textbook of palliative medicine, 2nd ed. New York: Oxford Medical Publications, 1998: Hanks GCW, Cherny N. Opioid analgesic therapy. In: Doyle D, Hanks GCW, MacDonald N, eds. Oxford textbook of palliative medicine, 2nd ed. New York: Oxford Medical Publications, 1998: Dundee JW, Halliday NJ, Harper KW, Brogden RN. Midazolam. A review of its pharmacological properties and therapeutic use. Drugs 1984;28: Kasama T, Fujii Y, Aida Y, et al. Pharmacological action of bromazepam suppository. Folia Pharmacol Japon 1983;81: Moolenaar F, Bakker S, Visser J, Huizinga T. Biopharmaceutis of rectal administration of drugs in man IV. Comparative biopharmaceutics of diazepam after single rectal, oral, intramascular, and intravenous administration in man. Int J Pharmaceutics 1980;5: Bercovitch M, Waller A, Adunsky A. High dose morphine use in the hospice setting. A database survey of patient characteristics and effect on life expectancy. Cancer 1999;86: Coyle N, Adelhardt J, Foley KM, Portenoy RK. Character of terminal illness in the advanced cancer patient: pain and other symptoms during the last four weeks of life. J Pain Symptom Manage 1990;5: Lombard DJ, Oliver DJ. The use of opioid analgesics in the last 24 hours of life of patients with advanced cancer. Palliat Med 1989;3: Amesbury BDW, Dunphy KP. The use of subcutaneous midazolam in the home care setting. Palliat Med 1989;3: Bottomely DM, Hanks GCW. Subcutaneous midazolam infusion in palliative care. J Pain Symptom Manage 1990;5:

8 Vol. 21 No. 4 April 2001 Effects of High Dose Opioids and Sedatives on Survival Burke AL, Diamond PL, Hulbert J, et al. Terminal restlessness its management and the role of midazolam. Med J Aust 1991;155: de Sousa E, Jepson BA. Midazolam in terminal care. Lancet 1988;1: Gremaud G, Zulian G. Indications and limitations of intravenous and subcutaneous midazolam in palliative care center. J Pain Symptom Manage 1998;15: Holdsworth MT, Adams VR, Chavez CM, et al. Continuous midazolam infusion for the management of morphine-induced myoclonus. Ann Pharmacother 1995;29: McNamara P, Minton M, Twycross RG. Use of midazolam in palliative care. Palliat Med 1991;5: Ramani S, Karnad AB. Long-term subcutaneous infusion of midazolam for refractory delirium in terminal breast cancer. South Med J 1996;89: Sanders H, Smales L. Discomfort and pain associated with paratonia. Eur J Palliat Care 1996;3: Smales EA, Sanders HG. Flunitrazepam in terminal care [letter]. Lancet 1989;2: McIver B, Walsh D, Nelson K. The use of chlorpromazine for symptom control in dying cancer patients. J Pain Symptom Manage 1994;9: Stirling LC, Kurowska A, Tookman A. The use of phenobarbitone in the management of agitation and seizures at the end of life. J Pain Symptom Manage 1999;17: Greene WR, Davis WH. Titrated intravenous barbiturates in the control of symptoms in patients with terminal cancer. South Med J 1991;84: Mercadante S, DeConno F, Ripamonti C. Propofol in terminal care. J Pain Symptom Manage 1999;10: Moyle J. The use of propofol in palliative medicine. J Pain Symptom Manage 1995;10: Akechi T, Uchitomi Y, Okamura H, et al. Usage of haloperidol for delirium in cancer patients. Support Care Cancer 1996;4: Breitbart W, Marotta R, Platt MM, et al. A double-blind trial of haloperidol, chlorpromazine, and lorazepam in the treatment of delirium in hospitalized AIDS patients. Am J Psychiatry 1996;153: