Ventilator Withdrawal: Procedures and Outcomes. Report of a Collaboration Between a Critical Care Division and a Palliative Care Service

Transcription

1 954 Journal of Pain and Symptom Management Vol. 26 No. 4 October 2003 Clinical Note Ventilator Withdrawal: Procedures and Outcomes. Report of a Collaboration Between a Critical Care Division and a Palliative Care Service Sean O Mahony, MB, BCH, BAO, Marlene McHugh, RN, FNP, Leah Zallman, BA, and Peter Selwyn, MD, MPH Montefiore Medical Center (S.O M., M.McH., L.Z., P.S.), Albert Einstein College of Medicine, Bronx, New York, and Columbia University School of Nursing (L.Z., M.McH.), New York, New York, USA Abstract To describe an institutional procedure for ventilator withdrawal and to analyze patient responses to terminal extubation, the medical records of 21 patients who underwent withdrawal of mechanical ventilation according to the process followed by an interdisciplinary palliative care team were retrospectively reviewed. The cohort was a convenience sample of sequentially treated patients in a 1048-bed urban universityaffiliated medical center. Sixteen of the 21 patients were on medical or surgical floors and five patients were in critical care units. Patients were assessed for discomfort, such as dyspnea, agitation, or anxiety. Sedative and analgesic medications were administered based on clinical parameters. Palliative care clinician observations of patient reports, tachypnea, use of accessory muscles, and signs of discomfort such as agitation or anxiety were recorded for the first 4 hours after extubation. Medication use and length of survival were recorded. Fifty-seven percent were symptomatic during the extubation process and required administration of either a benzodiazepine or opioid medication. The median survival of the 18 patients who died post-extubation was 0.83 hours (interquartile range ). Bolus doses of opioid or benzodiazepine medications were effective for management of symptoms in about two-thirds of patients. One-third of patients required continuous infusions. Eighteen patients died following extubation in the medical center, and three of these patients were transferred to an inpatient hospice unit. Three patients (14%) survived to discharge from the hospital. The procedure followed provides a foundation for collaboration between palliative care and critical care services to ensure continuity of care across clinical settings/units. J Pain Symptom Manage 2003;26: U.S. Cancer Pain Relief Committee. Published by Elsevier Inc. All rights reserved. Key Words Terminal extubation, palliative care, critical care, symptom management procedures, outcome Address reprint requests to: Sean O Mahony, MB, BCH, BAO, Montefiore Medical Center, Palliative Accepted for publication: March 1, U.S. Cancer Pain Relief Committee Published by Elsevier Inc. All rights reserved. Care Service, 3335 Steuben Avenue, Bronx, NY 10467, USA /03/$ see front matter doi: /s (03)

2 Vol. 26 No. 4 October 2003 Ventilator Withdrawal: Procedures and Outcomes 955 Introduction The decision to forego life-sustaining treatments occurs in 75% of deaths in U.S. hospitals. 1 Despite this, data regarding the management of symptoms after withdrawal of ventilator support are limited. Because of the often-anticipated death of patients following ventilator withdrawal, relief of distress is imperative; however, approaches to the management of such patients are often algorithmic. This descriptive study describes the role of an interdisciplinary palliative care service in the management of patients who underwent the withdrawal of ventilator support. Most reports on ventilator withdrawal in the literature describe the care of patients in intensive care units (ICU). 1 5 Such practices can be difficult to replicate on medical or surgical wards due to restrictions on the choice and method of administration of sedating medications or inability to administer certain agents such as propofol or midazolam by continuous infusion. Patient wishes about the use of life-prolonging treatments may be withheld by family members who anticipate a favorable response to aggressive care in a setting of uncertain prognosis. Perhaps because of this uncertainty, the wishes of patients may only emerge when it becomes apparent to families days or weeks after institution of ventilator support that aggressive care offers little or no possibility of recovery to the patient s previous status. Decisions to withdraw life-support treatments should be based on patients goals of care and their expressed wishes. In addition, there should be confirmation of poor prognosis by an independent critical care credentialed physician. Many of the patients placed on ventilators in acute emergent situations may not have had formal advance directives already in place. Palliative care services often help manage patients with a poor prognosis whose decisions for end-of-life care are usually unknown and whose families are reluctant to make decisions about end-of-life care. This type of team may be helpful in cases of ventilator withdrawal, to work with critical care clinicians and other services in the care of patients and families regarding issues of end-of-life care and postwithdrawal management. Given the difficulty of meeting the post-icu care needs of patients once they are discharged from the ICU, it may also be necessary to manage such patients outside of an ICU setting due to the absence of step down units in many medical centers. This case series illustrates the experience of seriously ill patients followed by a palliative care service whose ventilator support was withdrawn, and for whom death was the likely anticipated outcome. Methods All patients in this study underwent rapid cessation of mechanical ventilation, and aggressive symptom assessment and management were provided. Terminal weaning, another option for withdrawal of mechanical ventilation, 4 was not performed in this study. Patients were referred to the palliative care service by the critical care division, or by primary services. Some of the patients were referred following discussion with the primary attending on the recommendation of the ventilator task force, an interdisciplinary team including senior clinical administrators, respiratory therapy supervisors, critical care specialists, and the palliative care team. The setting for this study was a 1048-bed urban university-affiliated medical center. A retrospective review was conducted of the medical records of 21 patients who underwent withdrawal of mechanical ventilation as a probable terminal care process. The sequential sample was drawn from the population of patients followed by the palliative care service in During this year the service consulted on 447 patients. The team comprised an attending physician, a nurse practitioner, and a social worker. The decision for withdrawal of ventilator support was made following discussion of prognosis for recovery with the patient s surrogates. This discussion focused on evidence of the patient s previously expressed wishes and evidence of poor prognosis. One patient was able to participate in the discussion. Ten patients (48%) had health care proxies and four patients (19%) had living wills. Other patients who underwent the withdrawal of ventilator support were managed by primary services and other services such as the pulmonary service and geriatrics division. Sixteen of the 21 (74%) patients reviewed were on medical or surgical

3 956 O Mahony et al. Vol. 26 No. 4 October 2003 floors and 5 (24%) patients were in intensive care units. Data Collection The medical record was reviewed with a standard data collection instrument. Demographic data included age, sex, primary language, race, primary medical diagnosis, and comorbidities. Information was collected on the occurrence of meetings with patient families as well as the presence of advance directives. Information was also collected on patient location, reason for consultation request, and referral source. The chart was reviewed for evidence of patient distress during the withdrawal process, medication use, dosage, and frequency, including the use of continuous infusions. The level of consciousness of patients was also recorded. The length of patient survival post-withdrawal was recorded, as well as information on reintubation and outcome. Withdrawal Process The palliative care attending physician and nurse practitioner in this study followed an institutional procedure for evaluating whether mechanical ventilation should be withdrawn. The patient s condition, capacity, and poor prognosis were assessed and documented by an attending physician, neurologist, and other consultants as needed (e.g., pulmonary, critical care, oncology, and neurology). A health care agent or surrogate provided the required information about the patient s current or previously expressed wishes about withdrawal of ventilator support. If the patient lacked capacity and did not have a designated health care agent, administrative review of family letters was done by the legal department in conjunction with the medical director or his or her designee. This process required that there be evidence of clear and convincing evidence of the patient s wishes in regard to ventilator support in the event of poor prognosis for recovery, consensus within the family about the patient s wishes, and independent confirmation of the patient s poor prognosis by a credentialed critical care physician designee of the medical center s medical director. If withdrawal of ventilation was deemed appropriate, arrangements were made for the process. A respiratory therapist was present throughout the process. For the initial withdrawal of ventilator support, a clinician experienced in ventilator withdrawal was present. In all cases the palliative care physician or nurse practitioner was in attendance during the withdrawal. Before the process of ventilator withdrawal was begun, do-not-resuscitate and do-not-intubate orders were completed, and the order entered. Notes were made in the chart documenting the rationale for comfort care, and any discussions with the attending physicians from the neurology and/or pulmonary services, and with the family or family surrogate were recorded. The family and spiritual assessments were documented, and ongoing support, including pastoral care, was provided. The process was explained to the family, who were reassured that the patient s comfort was a primary concern, that oxygen and medication would be used as needed, and that while involuntary movements of the patient might occur, they were not an indication of distress. Anticipatory dosing was not routinely used: two patients were brain dead and, therefore, were not medicated, 8 patients had brain stem reflexes and no response to deep noxious stimulus, and an additional 8 patients had depressed levels of consciousness. The clinicians wanted to avoid potential iatrogenic complications of opioid medications such as emesis or paradoxical agitation. Anticipatory dosing was used if patients were more alert or if there was a history of recent agitation. One alert patient declined anticipatory dosing prior to extubation. Patients already on opioid or benzodiazepine infusions had their infusions maintained and bolus doses equivalent to 10% of the total daily dose prior to extubation. The uncertain length of survival and of the patient s outcome was explained to the family, with assurance that support would continue to be provided. The family or surrogate was encouraged to stay at the bedside, in close touch with the patient. Previous orders for laboratory tests, radiographs, and medications were reviewed and were discontinued, unless required for comfort and symptom management. Intravenous lines or drips were maintained, if clinically indicated. Orders were entered prior to the extubation of the patient for as needed PRN sedating medications to ensure the availability of sedating medications at the bedside at the time of extubation.

4 Vol. 26 No. 4 October 2003 Ventilator Withdrawal: Procedures and Outcomes 957 Sufficient dosages of opioid and benzodiazepine medications were drawn up and available for administration at the bedside prior to extubation. During the management of ventilator withdrawal, the respiratory therapist and clinician made individualized recommendations for withdrawal, based on ventilatory parameters, the patient s neurologic status, and his or her wishes. The need for establishment of an oral airway was clinically evaluated, as each patient was extubated. The goal was to decrease patient suffering and aid in secretion removal. Also, the type of apparatus for oxygen delivery was evaluated on a case-by-case basis. Sedation and Analgesia Choices of sedating medications were similar to those outlined in other reports in the literature Sedation or analgesia, or both, were administered. If the patient showed signs of discomfort, such as dyspnea, agitation, or anxiety, either an opioid medication or a benzodiazepine, or both, was administered intravenously and titrated every 15 minutes to keep the patient s respiratory rate below 28 breaths/ minute. Sedating medications were administered if the patient s respiratory rate increased by 50% over baseline or if the patient used accessory muscles of respiration or had nasal flaring. If the patient required more than two bolus doses of either an opioid or benzodiazepine medication, a continual infusion was initiated. No ceiling dose was used. Anticipatory medications were used if concerns were high about the potential emergence of difficult to control symptoms. The team recommended the use of anticipatory dosing to more alert patients and families. Opioid medications were administered prior to the use of benzodiazepines. The usual medication selected was morphine unless the patient had a prior history of dose limiting toxicity on exposure to morphine, renal failure, or hepatic failure, in which event other shortacting opioids were administered (fentanyl, hydromorphone). In patients not already exposed to an opioid medication, the usual starting bolus dose was 2 4 mg. Most of the patients in this series were elderly. The available literature suggests that elderly patients achieve equivalent rates of analgesia with opioid medications at 50% of the doses required in younger patients. If an individual bolus dose was ineffective, the dose was doubled at the next administration. If a patient was already receiving opioid medications, the bolus dose was equivalent to 150% of the hourly dose. If patients were already receiving opioid infusions, the opioid infusion was increased in % increments for the relief of signs or symptoms suggestive of dyspnea. The choice of benzodiazepine agents included short-acting agents such as lorazepam or midazolam. In patients already receiving benzodiazepine medications, the bolus dose was equivalent to 150% the current hourly dose that a patient was receiving. Scopolamine patches (1.5 mg) transdermally and atropine 0.4 mg administered subcutaneously every 12 hours were used for the relief of congestion secondary to copious large airway secretions. Either the attending physician or nurse practitioner attended to all patients for the first hour. Documentation of clinical status, respiratory rate, dyspnea, and agitation was kept by the palliative care clinician at the bedside. Symptom assessment and management was done for the first four hours after the ventilator withdrawal by the attending or nurse practitioner involved in the case. This required the presence of a palliative care clinician in the patient s room for the first hour after withdrawal followed by close follow-up by the palliative team and primary service for the next several hours. Ongoing symptom management was done in collaboration with the primary team caring for the patient. Orders were placed for sedating medications, follow-up continued in conjunction with the primary service (by telephone), and patients were followed daily by the consult service for advice on further symptom management. Data Analysis Data were analyzed with simple descriptive methods including frequencies, medians and interquartile ranges. Differences in survival associated with the use or non-use of medications were analyzed with Student t-tests. Results Twenty-one patients were retrospectively reviewed following the withdrawal of ventilation.

5 958 O Mahony et al. Vol. 26 No. 4 October 2003 Demographic variables for this sample are indicated in Table 1. Patient Characteristics Most patients were located on medical and surgical floors. Fewer than one-quarter (n 4) of the patients were in intensive care units. Fiftytwo percent (n 11) of patients had advance directives prior to consultation of which half (n 6) were unavailable at the time of intubation. Fifty-nine percent (n 12) of patients had formally designated health care proxies. Twenty-nine percent (n 6) of the patients were referred for palliative care consultation after the primary team elicited information regarding knowledge of patient wishes to forego life-sustaining treatments. All patients had family members who provided evidence of patient wishes in relation to end-of-life care. Most patients had profound neurological injury and depressed level of consciousness. All of the patients had been intubated with oroendotracheal tubes which were removed in all cases with or without the use of an oral airway for suctioning. Patients were managed by a primary medical service and followed on a consultation basis by a Palliative Care Service. Patients had been intubated for a median of 8 days (range: 1 17) and had been hospitalized for a median of 9 days prior to consultation (range: Table 1 Patient Demographics (n 21) n (%) Mean age (yrs) (46 89) Sex Female 10 (48) Male 11 (52) Race Caucasian 10 (48) African-American 6 (29) Hispanic 4 (19) Other 1 (5) Mortality at discharge 18 (86) Survival (hrs) Referral source Private attending physician 7 (33) Service attending physician 14 (67) Patient location Medical ward 11 (52) Surgical ward 2 (10) Intensive Care Unit 5 (24) Neurology ward 1 (5) Cardiac Care Unit 0 (0) Pulmonary ward 2 (10) 1 72). The median time to extubation following a palliative care consult was 3 days (range: 0 13). All the cases reviewed had documented evidence of discussions with family members. Nine of the 21 (43%) patients reviewed had spent an average of 11.1 (median: 11, range: 2 36) days in intensive care units during their admissions. Illness Severity and Neurological Status The primary diagnosis was usually anoxic encephalopathy. Comorbidities and primary diagnoses at the time of extubation are listed in Table 2. Patients had a mean of 3.7 ( 1.7) comorbidities. Seventy-six percent (n 16) of patients had depressed levels of consciousness. Two patients (9.5%) were evolving to brain death. Three patients (14%) were alert and responsive, one of whom was able to communicate her wishes. Medication Use and Patient Outcome Sixty-two percent (n 13) of patients received sedating medications during the withdrawal process. Fifty-seven percent (n 12) of patients were tachypneic, agitated, or dyspneic during the extubation process and required the use of either an opioid medication alone (morphine, fentanyl, hydromorphone) or with a benzodiazepine (lorazepam or midazolam). Table 2 Comorbidities (n 21) n % Mean no. per person ( SD) Hypertension 8 38 Anoxic encephalopathy 8 38 Congestive heart failure 5 24 Chronic obstructive pulmonary 5 24 disease Cancer 4 19 CRI 4 19 Seizure 4 19 Pneumonia 3 14 ARF 2 10 Arrhythmia 2 10 CAD 2 10 Psychiatric illness 2 10 Dementia 2 10 Current diagnoses Anoxic encephalopathy 6 29 Respiratory failure 3 14 Other a a The following diagnoses were present in one or more test patient each: ARDS, brain dead, CML, COPD, cirrhosis, chronic renal insufficiency, GI bleed, hepatorenal syndrome, hypoxic encephalopathy, pleural effusion, pneumonia, psychiatric illness.

6 Vol. 26 No. 4 October 2003 Ventilator Withdrawal: Procedures and Outcomes 959 Seven patients required the use of a benzodiazepine medication. One-third of patients (n 7) required an opioid medication by continuous infusion and 19% of patients (n 4) required a continuous infusion of a benzodiazepine medications. Three patients (14%) who were receiving either benzodiazepine medications or opioid medications for the management of agitated delirium prior to extubation had their medications continued at the same dose as prior to extubation as they did not display any signs of respiratory distress during the extubation process. For one patient who had been receiving midazolam for the treatment of agitated delirium secondary to hepatic encephalopathy, the midazolam infusion rate was subsequently reduced by 50% due to family concerns that the medication was preventing communication with the patient. The patient remained unresponsive and the infusion was subsequently increased to its previous rate due to subsequent respiratory distress. The median total doses of sedating medications required from the time of extubation until death or discharge for those patients requiring sedating medications were low at 8.0 mg of either midazolam or lorazepam (range: 4 196) and 12.0 mg of morphine equivalent (range: 2 48). The median dose of opioid medications in the seven patients receiving opioid infusions was 6 mg/hour (range: 1 50). The median dose of midazolam in the four patients receiving infusions was 4 mg/ hour (range: 2 72). Outcome Survival Time, Level of Consciousness and Medication Use. Three patients (14%) survived to be discharged from the hospital. The median survival of the 18 patients who died post-extubation was 0.83 hours (interquartile range ). Among those patients who were not brain dead, median survival time was 2.0 hours (interquartile , n 16). Two of the three patients (67%) who survived received only single doses of morphine sulfate (2 mg) during the extubation process for the relief of dyspnea. There was a trend to shorter survival for patients with depressed level of consciousness. The median survival time for patients with depressed levels of consciousness was 2.0 hours (interquartile range , n 14). Medication Use and Survival Time. Although there was a trend to shorter survival time for patients receiving greater than the median dose of opioid medication and infusions of opioid medications, these differences were not statistically significant. Among those patients who were not receiving infusions prior to extubation, median survival was 0.75 hours (interquartile range , n 15). Patients receiving both classes of sedating medications had median survival times of 0.90 hours (interquartile range , n 7). Those receiving either opioids or benzodiazepine medications died a median of 12.5 hours later (interquartile range , n 4). Those receiving neither medication died a median of 0.5 hours after extubation (interquartile range , n 7). These differences in survival times were not significant (t 0.597, P 0.58). Disposition Three of the 21 (14%) patients in this report survived to discharge and 18 patients died following extubation in the medical center. None of the 21 patients were reintubated; 3 of the patients who expired in the medical center following extubation died on an inpatient hospice unit in the medical center. The three patients who survived were subsequently transferred to nursing homes, one of whom was subsequently transferred to a palliative care facility. Discussion Unlike most reports, in which patients undergoing withdrawal of mechanical ventilation were treated in intensive care units, 1,3,4,6,8,10 patients in this study were treated primarily on medical/surgical floors. This is the first report of a series of patients followed by a palliative care service. Intensive care was unlikely to benefit more than half of the patients in this study at the time of intubation because of an assessment of poor prognosis. Most of the patients had either been intubated by emergency medical services in the field before admission or in the emergency room because the patient, family, or patient surrogate had made no end-of-life decisions. The presence of DNR orders at the time of extubation as well as the high acuity and poor

7 960 O Mahony et al. Vol. 26 No. 4 October 2003 neurological status of most of these patients correlated to a high expected mortality. Critical care services handling large caseloads often do not have the resources to provide aftercare for patients discharged from the ICU with poor prognosis, most of whom are not likely to benefit from care in a medical intensive care unit. These issues become even more complex when patients end-of-life wishes are unknown and their families are reluctant to address end-oflife care until relatively late in the patient s course. In these complicated cases, palliative care teams can provide expert clinical support to critical care services. The eventual ability of a family to divulge their knowledge of a patient s wishes in regard to end-of-life care often requires multiple meetings over several weeks and the reconciliation of differences of opinion between family members. This emphasizes the importance of the availability of an interdisciplinary team of experts in end-of-life care with knowledge of likely prognosis, clinical outcomes, family dynamics, and expertise in symptom management. Another important role for a palliative care service in ventilator withdrawals may be to ensure continuity of care for those patients who survive extubation. Despite anticipated poor prognoses, three patients survived several weeks after extubation, emphasizing the need to avoid presenting families with fixed prognostic estimates for anticipated patient expiration as well as the need to provide ongoing continuity care by palliative care providers to ensure that translation of palliative care plans into practice becomes a reality. Concerning the use of medications following ventilator withdrawal, fewer than two-thirds of patients required the administration of an opioid medication or benzodiazepine for the relief of distress. Most patients required only one or two bolus doses of sedating medications and only one-third of patients needed continuous infusions of sedating medications. However, the authors would recommend the use of sedating medications prior to extubation in more alert patients with anticipated short survival after extubation unless declined by patients. Patients were closely monitored for the presence of distress by a team including either an experienced palliative care physician or nurse practitioner. The median doses of both benzodiazepines and opioid medications used in this population were lower than other reports in the literature. However, the broad range in doses reported for both benzodiazepines and opioid medications emphasize the need to individualize sedation for different patients and the need for vigilance by clinicians experienced in the use of these medications. 4,8 In our institution, close collaboration between palliative care and critical care services has helped ensure appropriate care for critically ill patients whether or not they are likely to benefit from intensive care. This collaboration was facilitated by weekly rounds on all ventilated patients with senior clinical administrators, respiratory therapy leadership and the palliative care service. This type of collaboration could be a useful model for managing the withdrawal of ventilation and other futile interventions for dying patients in other institutions as well. As well as adding institutional legitimacy to the act of withdrawal, the presence of senior physicians and nurses helps provide house staff with role modeling in this difficult area of clinical practice. References 1. Prendergast TJ, Classens MT, Luce JM. A national survey of end of life care for critically ill patients. Crit Care Med 1998;158: Edwards MJ, Tolles SW. Disconnecting a ventilator at the request of a patient who knows that he will then die: the doctor s anguish. Ann Intern Med 1992;117: Wilson MC, Smedira NG, Fink C, et al. Ordering and administration of sedatives and analgesics during the withholding and withdrawal of life support from critically ill patients. JAMA 1992;267: Campbell ML, Bizek KS, Thill M. Patient responses during rapid terminal weaning from mechanical ventilation: prospective study. Crit Care Med 1999;27: Knaus WA, Draper EA, Wagner DP, et al. APACHE 2: a severity of disease classification system. Crit Care Med 1985;13: Faber-Langendoen K, Lanken PN. Dying patients in the intensive care unit: forgoing treatment, maintaining care. Ann Intern Med 2000;133: Faber-Langendoen K. The clinical management of dying patients receiving mechanical ventilation. Chest 1994;106:

8 Vol. 26 No. 4 October 2003 Ventilator Withdrawal: Procedures and Outcomes Daly BJ, Thomas D, Dyer MA. Procedures used in withdrawal of mechanical ventilation. Am J Crit Care 1996;5: Weatherhill GC. Pharmacologic symptom control during the withdrawal of life support: lessons in palliative care. AACN Clin Iss 1995;6: Rubenfeld GC, Crawford SW. Principles and practice of withdrawing life sustaining treatment in the ICU in managing death in the Intensive Care Unit. New York: Oxford University Press, 2001: Troug RD, et al. Recommendations for end-oflife care in the intensive care unit: The Ethics Committee of the Society of Critical Care Medicine. Crit Care Med 2001;12: