- Email: [email protected]
An intensive communication intervention for the critically ill
An Intensive Communication Intervention for the Critically Ill Craig M. Lilly, MD, Dawn L. De Meo, MD, Larry A. Sonna, MD, PhD, Kathleen J. Haley, MD, Anthony F. Massaro, MD, Robert F. Wallace, MPH, Sean Cody, RN, MSN, MBA PURPOSE: We sought to determine the effects of a communication process that was designed to encourage the use of advanced supportive technology when it is of benefit, but to limit its burdens when it is ineffective. We compared usual care with a proactive, multidisciplinary method of communicating that prospectively identified for patients and families the criteria that would determine whether a care plan was effective at meeting the goals of the patient. This process allowed caregivers to be informed of patient preferences about continued advanced supportive technology when its continuation would result in a compromised functional outcome or death. MATERIALS AND METHODS: We performed a before-andafter study in 530 adult medical patients who were consecutively admitted to a university tertiary care hospital for intensive care. Multidisciplinary meetings were held within 72 hours of critical care admission. Patients, families, and the critical care team discussed the care plan and the patients’ goals and expectations for the outcome of critical care. Clinical “milestones” indicative of recovery were identified with time frames for their occurrence. Follow-up meetings were held to discuss palliative care options when continued advanced supportive technology was not achieving the patient’s goals. We measured length of stay,
mortality, and provider team and family consensus in 134 patients before the intensive communication intervention and in 396 patients after the intervention. RESULTS: Intensive communication significantly reduced the median length of stay from 4 days (interquartile range, 2 to 11 days) to 3 days (2 to 6 days, P ⫽ 0.01 by survival analysis). This reduction remained significant after adjustment for acute physiology and chronic health evaluation (APACHE) 3 score [risk ratio (RR) ⫽ 0.81; 95% confidence interval (CI), 0.66 to 0.99; P ⫽ 0.04). Subgroup analysis revealed that this reduction occurred in our target group, patients with acuity scores in the highest quartile who died (RR ⫽ 0.60; 95% CI, 0.38 to 0.92; P ⫽ 0.02). The intervention, which allowed dying patients earlier access to palliative care, was not associated with increased mortality. CONCLUSIONS: Intensive communication was associated with a reduction in critical care use by patients who died. Our multidisciplinary process targeted advanced supportive technology to patients who survived and allowed the earlier withdrawal of advanced supportive technology when it was ineffective. Am J Med. 2000;109:469 – 475. 䉷2000 by Excerpta Medica, Inc.
A
sition to palliative care affects patient comfort and critical care use. Dying patients consume a disproportionate amount of intensive care resources. In one study, high-use patients who died comprised 13% of intensive care unit (ICU) patients, yet consumed 32% of the resources (7). In our medical ICU in 1996, more than 50% of patient-days were spent providing advanced supportive technology for patients who did not survive. More important, because the support of critically ill patients is accompanied by substantial discomfort and deprivation, it is not humane to continue burdensome technology when there is no reasonable hope that it will be effective. The intravascular lines, support tubes, and restraints used in this process are inherently uncomfortable and are associated with reduced mobility, autonomy, and ability to communicate. When advanced supportive care is associated with healing and the return of function, these discomforts are almost universally acceptable, but they are difficult to justify for patients who die. For this reason, as many as 75% of hospitals now work with patients and their families to withdraw advanced supportive technology when it is not thought to be of benefit (4,8 –12). Like others (13,14), we believe that improved commu-
s much as 1% of the gross national product in the United States is used to provide care for critically ill adults (1). Access to critical care is valued not only by survivors but by the families of most of those who do not survive (2). The wider respect for patient autonomy and the advent of advance directives have allowed some dying patients to receive palliative, rather than aggressive, critical care (3). Despite the alternatives, many patients choose aggressive care, and 2% to 40% do not survive (4). Access to palliative care is often delayed for dying, critically ill patients (5,6), because it is difficult for providers to articulate, and for patients and families to accept, that advanced supportive technology has been ineffective or will not result in a functional outcome that is acceptable to the patient. The optimal timing of the tran-
From the Combined Program in Pulmonary and Critical Care Medicine, Department of Medicine, Brigham and Women’s Hospital and Harvard Medical School, Boston, Massachusetts; and US Army Research Institute of Environmental Medicine, Natick, Massachusetts. Supported by Brigham and Women’s Hospital. Requests for reprints should be addressed to Craig M. Lilly, MD, Respiratory Division, Brigham and Women’s Hospital, 75 Francis Street, Boston, Massachusetts 02115. Manuscript submitted November 22, 1999, and accepted in revised form June 5, 2000. 䉷2000 by Excerpta Medica, Inc. All rights reserved.
0002-9343/00/$–see front matter 469 PII S0002-9343(00)00524-6
Intensive Communication Intervention for the Critically Ill/Lilly et al
nication among the provider team, patient, and family, which affects the timeliness of decision making and length of stay for dying patients, will lead to better outcomes. Although the SUPPORT trial did not demonstrate that a communication intervention led by nurse facilitators affected the time that dying patients spent in the ICU, communication sessions facilitated by a proactive ethics consultation appear to be effective (14). We hypothesized that similar results could be achieved by the physicians and nurses caring for the patient. We therefore tested whether a process of intensive communication that consisted of attending physician– directed multidisciplinary sessions focusing on the goals and outcomes of advanced supportive technology would result in earlier transition to palliative care and reduced length of stay for dying patients, without increasing mortality.
MATERIAL AND METHODS Study Design In a prospective, nonblinded, change-of-practice intervention study, we measured ICU, hospital, and rehabilitation facility mortality rates; probability of remaining in the ICU as a function of time after admission; and survival to discharge from acute and rehabilitative care to home, skilled nursing, or chronic care facilities for 530 consecutive adult patients admitted to a general medical ICU in a tertiary care teaching hospital. We made our intervention in a 10-bed unit in which all patients are cared for by a team consisting of 1 (of a rotating group of 4) critical care– certified attending physician, 2 residents, 3 interns, and 45 nurses (in shifts). The same attending physicians provided care during both phases of this intervention. Admission decisions were made by a standard set of physiological criteria that were applied by a physician who was not part of the ICU team; discharge decisions were made by the critical care physicians. Clinical practice guidelines for the management of pain, dyspnea, anxiety, and delirium had been previously developed by the responsible attending physicians and did not change during the period of the study. The study was conducted for 12 months, beginning October 1, 1997, and was divided into a 3-month preintervention period that included 134 consecutive patients and a 9-month intervention (intensive communication) period that included 396 consecutive patients. The study was performed with the approval of the Human Subjects Committee of the Brigham and Women’s Hospital.
Intensive Communication Intervention In the preintervention period, formal family meetings were usually held after the provider team had reached consensus that restoration of function or survival was unlikely. Patients and families also received information from care providers (most commonly nurses and house 470
October 15, 2000
THE AMERICAN JOURNAL OF MEDICINE威
officers) in a process that was informal and that occurred without explicit coordination among providers. In the intensive communication period, we introduced a proactive process of communication. An initial formal multidisciplinary meeting with the patient, family, or both was held within 72 hours of intensive care admission. The attending physician conducted this meeting for patients who had a predicted length of ICU stay longer than 5 days, a predicted mortality of greater than 25% as estimated by the attending physician, or a change in functional status that was potentially irreversible and sufficient to preclude eventual return to home. The clinical judgment of experienced critical care providers is known to rival or exceed the predictive power of APACHE 2 scores (15–18), and this method of selecting patients identified all patients in the cohort with an APACHE 3 predicted mortality of 25% or more. This meeting was preceded by an assessment of the patient’s medical condition, communication with the primary care physician to review and agree about a recommended care plan, and review of any known advance directives. The formal meeting included the attending physician, nurse, house officer, the patient’s family (including the health care proxy), and, whenever possible, the patient. The meeting had four primary objectives: 1) to review the medical facts and options for treatment; 2) to discuss the patient’s perspectives on death and dying, chronic dependence, loss of function, and the acceptability of the risks and discomforts of critical care; 3) to agree on a care plan; and, most important, 4) to agree on criteria by which the success or failure of this care plan would be judged. This last objective was accomplished by defining clinical events (termed “clinical milestones”) which would indicate that the care plan had been effective, such as resolution of hypotension or reduced need for ventilator support, and the time frame for their occurrence. The timing of subsequent formal meetings was set by this time frame and by the ability of the patient or family to process difficult medical facts. This approach is a variant of the time-limited trial of advanced supportive technology approach reported by Lee et al (19). As clinical milestones were met, new ones were established at subsequent meetings on the basis of the patient’s clinical course and perspectives on death and dying, chronic dependence, and loss of function and the acceptability of the discomforts of critical care. The patient and family were educated about and prepared for home care, rehabilitation, or chronic care, as appropriate. Failure to meet clinical milestones allowed the patient and family to know that advanced supportive technology had been ineffective; alternatives (including palliative care) were then explored. In addition, a weekly multidisciplinary case review was held with nurses, social workers, chaplains, rehabilitation facility providers, and critical care attending physicians.
Volume 109
Intensive Communication Intervention for the Critically Ill/Lilly et al
The goals of these sessions were to ensure a uniform understanding of the patient’s wishes and care plan; to enhance the care team’s ability to provide consistent information to the patient and family; to identify any barriers to the care plan that may have arisen; to review the requirements for successful rehabilitation with representatives of accepting facilities; to review patient outcomes; and to ensure that formal meetings had been held within 72 hours of admission for all patients who met entry criteria. Representatives of chronic care and rehabilitation facilities gave feedback to the ICU care team on the clinical status of patients who had been transferred to their facilities.
Measurement of Patient/Family and Provider Team Nonconsensus Every day, the provider team identified a long-term disposition goal for each ICU patient. These goals consisted of transfer to a floor service (with the intention of eventual discharge home), transfer to a rehabilitation or chronic care facility, or palliative care. The attending physician recorded as a nonconsensus day any day when there was disagreement on what the long-term disposition goal ought to be. Provider team nonconsensus was defined as disagreement among 2 or more involved nurses or physicians, including the primary care physician or specialist acting in this role. Family nonconsensus was defined as family disagreement with the provider team’s consensus recommendation. The attending physician’s daily assessment of consensus was reviewed each week at the multidisciplinary case reviews and independently by a nurse not directly involved in the patient’s care.
Measurement of Acuity and Use Patient acuity was estimated by calculation of the APACHE 3 score for the first critical care day. The primary indication for intensive care was similar to that of other medical ICUs, including management of respiratory failure (72% of patient-days), management of allcause hypotension or arrhythmia (19%), airway management or bronchodilator therapy (4%), electrolyte management (2%), and terminal care (2%). There were no significant differences in any of these categories between the preintervention and the intensive communication periods. APACHE 3 scores were obtained from chart review and determined by a single investigator (DD), who also recorded demographic and clinical characteristics, disposition status, length of stay, and vital status. This investigator was not involved in the study design and was not part of the care delivery team. Use was compared by dividing the number of intensive care days required by the number of preintervention or intensive communication patients served. Use by survival status was also determined.
Statistical Considerations Statistical analyses were performed with SAS 7.0 (Cary, North Carolina), SPSS 9.0 for Windows (Chicago, Illinois), and SigmaStat for Windows (Versions 1.2 and 2.0; Chicago, Illinois). Sex, race, lack of consensus, and whether a family meeting took place were compared by chi-square analysis. Age and APACHE 3 scores for the preintervention and intervention groups were compared by t test. Kaplan-Meier curves were constructed to describe length of stay in the ICU and analyzed by both the log-rank (Mantel-Cox) and the Breslow-Gehan-Wilcoxon tests. Cox proportional hazards analysis was used to assess the effects of covariates on ICU discharge rates. We performed univariate Cox regressions for each of the following variables: group (preintervention vs intervention), sex, season (fall, winter, spring, summer), age, and APACHE 3 score; intervention (P ⫽ 0.02) and APACHE 3 score (P ⫽ 0.09) met the criterion for inclusion (P ⬍0.1) in a multivariate model. Mortality rates by disposition were compared by Z test with Yates’ correction. Differences in mortality between the preintervention and intensive communication periods were compared by chisquare analysis or by Fisher’s exact test as well as after adjusting for APACHE 3 score in a logistic regression model using APACHE 3 score as a covariate. Risk and odds ratios are presented as the ratio of the value for the intensive communication group divided by the value for the preintervention group.
RESULTS The 134 patients in the preintervention group and the 396 patients in the intensive communication group had similar age, race, sex, and mean (Table 1) and median (Figure 1) APACHE 3 scores. A greater percentage of attending physician–led meetings were held in the intensive communication period than in the preintervention period [273 of 396 (69%) vs 58 of 134 (43%), P ⬍0.001]. Multidisciplinary meetings occurred with 270 (68%) of 396 intensive communication patients or their families within 72 hours of ICU admission, including 99% of the targeted group of 273 patients who had loss of independent function, an anticipated length of stay of greater than 5 days, or a predicted mortality greater than 25%. Of the 270 patients, 187 (69%) had an APACHE 3 predicted mortality of 25% or greater; the 83 remaining patients met the other criteria. We failed to meet with the families of 2 unconscious patients because they were never located; we were outside the 72-hour window for an additional patient whose identity was not known to us until the fourth hospital day. Including these initial meetings, 594 senior physician and nursing hours (1.5 hours per patient admitted to the ICU) were required to meet with intensive communication patients and their families.
October 15, 2000
THE AMERICAN JOURNAL OF MEDICINE威
Volume 109 471
Intensive Communication Intervention for the Critically Ill/Lilly et al
Table 1. Characteristics of the Patients before and during the Intensive Communication Intervention
Characteristic
Preintervention Group (n ⫽ 134)
Intensive Communication Group (n ⫽ 396)
P Value
Mean ⫾ SD, Median (25th–75th Percentile), or Number (Percent) Age (years) Male sex Race Caucasian African American Hispanic Asian/other Season of year upon admission Winter Spring Summer Fall APACHE 3 Score†
60 ⫾ 18, 63 (45–75) 67 (50)
58 ⫾ 18, 60 (45–72) 191 (48)
98 (73) 24 (18) 8 (6) 4 (3)
289 (73) 69 (17) 23 (6) 15 (4)
52 (39) 0 0 82 (61) 72 ⫾ 29, 68 (52–87)
99 (25) 134 (34) 143 (36) 20 (5) 67 ⫾ 30, 63 (46–81)
0.3, 0.3* 0.7 1.0
⬍0.001
0.1, 0.07*
* By Student’s t test, Mann-Whitney test. † Available for 387 of the patients in the intensive communication group.
Achieving Consensus
Median (twenty-fifth to seventy-fifth percentile) length of stay in the ICU was reduced during the intervention
period from 4 days (2 to 11 days) to 3 days (2 to 6 days). Unadjusted survival curves (with death or transfer from the ICU as the endpoint) showed a reduction in the chance of remaining in the ICU during the intervention period (P ⫽ 0.01, Figure 2). The effect was most pronounced after the first 72 hours (when the intervention began). The risk of remaining in the ICU was significantly less in the intensive communication group after adjusting for APACHE 3 score [risk ratio ⫽ 0.81, 95% confidence interval (CI), 0.66 to 0.99, P ⫽ 0.04]. Subgroup analyses (Table 2) showed a statistically significant reduction in
Figure 1. Histogram of APACHE 3 scores. The median (twenty-fifth to seventy-fifth percentile) APACHE 3 scores were 68 (52 to 87) in the preintervention group (striped bars) and 63 (46 to 81) in the intensive communication group (black bars).
Figure 2. Proportion of patients remaining in the intensive care unit (ICU) by time. The probability of remaining for the intensive communication group (open circles) was significantly less than that for the preintervention group (solid circles) by survival analysis (log rank P ⫽ 0.01, Breslow-Gehan-Wilcoxon P ⫽ 0.01).
During the intervention period, the rate of provider nonconsensus days decreased from 65 days per 1,000 patientdays to 4 days per 1,000 patient-days, or from 0.56 to 0.02 days per patient. Similarly, the rate of family nonconsensus decreased from 171 days per 1,000 patient-days to 16 days per 1,000 patient-days, or from 1.7 to 0.09 days per patient, after the intervention (P ⬍0.001).
Effects on Length of Stay
472
October 15, 2000
THE AMERICAN JOURNAL OF MEDICINE威
Volume 109
Intensive Communication Intervention for the Critically Ill/Lilly et al
Table 2. Effects of Intensive Communication on Length of Stay Preintervention Group APACHE 3 Quartile (Score Range) Survivors I (11–47) II (48–64) III (65–82) IV (83–185) Patients who died I (11–47) II (48–64) III (65–82) IV (83–185)
Intensive Communication Group
N
Median (25th–75th Percentile) Length of Stay (Days)
Risk Ratio (95% Confidence Interval)*
P Value
3 (2–4) 3 (2–5) 5 (2–15) 5 (3–8)
104 68 67 20
3 (2–4) 3 (2–4) 4 (3–7) 4.5 (3–8)
0.9 (0.6–1.5) 0.9 (0.6–1.4) 0.9 (0.5–1.6) 1.1 (0.5–2.4)
0.8 0.7 0.8 0.8
8 (6–11) 8 (4–15) 10 (4–14) 5 (3–12)
4 24 31 69
0.9 (0.2–4.2) 1.1 (0.5–2.3) 1.0 (0.5–1.8) 0.6 (0.4–0.9)
0.9 0.9 0.9 0.02
N
Median (25th–75th Percentile) Length of Stay (Days)
19 29 16 9 4 9 17 31
5 (4–14) 6 (2–14) 7 (4–15) 3 (2–7)
*Effects of intervention on rate of discharge from intensive care unit, estimated using Cox proportional hazard models within APACHE 3 strata with treatment group as the sole independent variable.
length of stay only among those who had been principally targeted by our intervention, namely those in the highest quartile of APACHE 3 scores who eventually died. During the preintervention periods, 1,072 ICU beddays were required to care for 134 patients (8.0 bed-days per patient), compared with 2,349 ICU bed-days for 396 patients (5.9 bed-days per patient) during the intervention.
Effects on Mortality The intensive communication intervention did not increase mortality (Table 3). Rather, there was a decrease in unadjusted overall mortality (P ⫽ 0.02) and a trend toward reduced mortality during the ICU stay (P ⫽ 0.06). About one-third of the decrease in overall mortality was
accounted for by an increase in the proportion (from 17% to 28%; Table 4) of less severely ill patients (APACHE 3 scores in the lowest quartile) who were admitted to the ICU. This was also associated with a decrease in the ICU occupancy rate, from 98% to 92%, during the intensive communication period. The decrease in overall mortality remained after adjustment for APACHE 3 score (odds ratio ⫽ 0.61, 95% CI, 0.38 to 0.98, P ⫽ 0.04). Subgroup analyses revealed significant reductions in mortality among patients admitted to the ICU in the lowest and second-highest APACHE 3 quartiles (Table 4). The trend toward reduced ICU mortality was not the result of an increased rate of transfer of dying patients
Table 3. Patient Outcomes
Eventual Disposition
Intensive Communication Group (n ⫽ 396)
Preintervention Group (n ⫽ 134)
P Value
Number (Percent) Survived ICU stay and discharged* Directly from the ICU From a floor service From a rehabilitation facility Total Died In the ICU On a floor service In a rehabilition facility Total
3 (2) 42 (31) 28 (21) 73 (54)
8 (2) 157 (40) 97 (24) 262 (66)
0.8 0.1 0.5 0.02
42 (31) 12 (9) 7 (5) 61 (46)
90 (23) 39 (10) 5 (1) 134 (34)
0.06 0.9 0.02 0.02
* To home, an assisted living arrangement, or a skilled nursing facility. ICU ⫽ intensive care unit. October 15, 2000
THE AMERICAN JOURNAL OF MEDICINE威
Volume 109 473
Intensive Communication Intervention for the Critically Ill/Lilly et al
Table 4. Mortality Rates by APACHE 3 Score Preintervention Group
Intensive Communication Group
APACHE 3 Quartile
Died/Total
Percent
Died/Total
Percent
Odds Ratio (95% Confidence Interval)*
P Value
I II III IV All subjects†
4/23 9/38 17/33 31/40 61/134
17 24 52 78 46
4/108 24/92 31/98 69/89 134/396
4 26 32 78 34
0.2 (0.04–0.8) 1.1 (0.5–2.7) 0.4 (0.2–1.0) 1.0 (0.4–2.4) 0.6 (0.4–0.9)
0.03 0.8 0.04 1.0 0.02
* For mortality, comparing the intensive communication group with the preintervention group. † Includes 9 patients in the intensive communication group for whom APACHE 3 scores could not be calculated.
from the ICU, as 42 of the 61 patients (69%) who died in the preintervention period did so in the ICU, compared with 90 of the 134 patients (67%) who died in the intensive communication period. There was, however, a trend toward fewer transfers to a rehabilitation facility of patients who died before discharge to home: 11% of those who died in the preintervention period did so in a rehabilitation facility, compared with 4% in the intensive communication period (P ⫽ 0.08). This outcome was associated with a statistically significant reduction in mortality among those transferred to a rehabilitation facility: 7 of 35 such patients (20%) died in the preintervention period, and 5 of 102 such patients (5%) died in the intensive communication period (P ⫽ 0.02).
DISCUSSION We undertook this intensive communication intervention to improve care for critically ill adult medical patients by encouraging advanced supportive technology when it was effective for accomplishing patient-directed goals and limiting it when ineffective. We confirmed that critical care effectively restores life and function in the majority of critically ill adults (1,4,15–18), and we found that intensive communication accomplished this goal with greater concordance among patients, families, and providers and with fewer preterminal days of intensive care for dying patients. Early formal multidisciplinary meetings partnered physicians with nurses, chaplains, and social workers to facilitate the communication of consistent medical information. Our results demonstrate that caregivers spent more time meeting with and listening to patients and families about their views and values, a process that fostered agreement to continue advanced supportive technology when it was effective for meeting the goals of the patient and permitted an earlier transition to ICU-based palliative care when it was not. Dying patients spent fewer days in the ICU. There was a statistically significant reduction in overall median length of ICU stay that was attributable to a reduction in the length of stay of dying patients with the highest 474
October 15, 2000
THE AMERICAN JOURNAL OF MEDICINE威
APACHE scores, the very patients for whom our intervention was intended. Had the reduction in length of stay been caused by secular trends, we would have expected a reduction in length of stay in other subgroups as well. Furthermore, the reduction in length of stay began on the third ICU day, as would be expected for an intervention that occurred during the first 72 hours of intensive care admission. We cannot determine the relative importance of a formal consensus-building process, an increase in the uniformity of the content of our communications, a recognition of advanced supportive technology as ineffective when it was not working, or a commitment by the attending physician to offer the option of providing earlier access to palliative care for dying patients. We believe that each of these elements was beneficial in some circumstances. Our intervention was associated with a substantial reduction in ICU use: approximately 200 fewer days for every 100 critically ill patients. This reduction in use was not uniform for all patients; rather it was limited to highacuity patients who died (Table 2). A consequence of the reduction in length of ICU stay was an increase in bed availability, which permitted more patients with APACHE scores in the lowest quartile to be admitted to the ICU, thereby contributing to a lower overall mortality rate. We considered the possibility that the reduction in mortality was the result of systematic changes in our critical care practice or delivery system. However, the mortality reductions were not seen in all of the APACHE quartiles as might be expected from a secular trend, and the duration of the preintervention period was short. One of our concerns about this intervention was that it might increase mortality rates by introducing a bias toward palliative care when continued advanced supportive technology would have preserved life. The intervention, however, did not change mortality in our targeted group of patients in the highest quartile of APACHE 3 scores, despite a significant reduction in length of stay among patients in this subgroup who died. Unlike length of stay, mortality was reduced among patients in the first and third quartiles of APACHE 3 score. Although the reduc-
Volume 109
Intensive Communication Intervention for the Critically Ill/Lilly et al
tion in mortality among patients in the lowest (first) quartile of APACHE 3 score was probably the result of greater bed availability, reduced mortality in the third quartile may have been a consequence of our intervention. We speculate that agreeing to clinical milestones encouraged the provider team to support severely ill patients who had the potential to survive but were improving slowly. We also found a statistically significant reduction in mortality among patients transferred to rehabilitation facilities, which suggests that an indirect benefit of our weekly multidisciplinary case reviews was better selection and preparation of patients for the rehabilitation process. A major limitation of this study is that it was not randomized or blinded. We did not perform a randomized study for several reasons. First, patients and physicians would have needed to agree to being assigned to a group that received less communication. Second, enforcement or monitoring of communication would be necessary, because providers who viewed the communication process paradigm as a substantial advance in access to palliative care may have felt obligated to use these techniques for dying patients in the less intensive communication strategy. We were able to introduce some of the advantages of randomization and blinding. Because they were not involved in the ICU admission process, the investigators were prevented from selecting patients for ICU admission who were likely to require shorter intensive care stays. The small number of providers, the creation of guidelines, and the use of weekly multidisciplinary case reviews served as a protection against differences resulting from changing clinical practice and individual practice styles. In summary, we found that an attending physician– led, multidisciplinary process of early and frequent communication that identified patient goals and gauged the effectiveness of care delivered critical care effectively to surviving patients and permitted earlier withdrawal of advanced supportive technology when it was ineffective for restoring life or accomplishing the patient’s goals. The magnitude of the effects of our program on length of stay for dying patients suggests that it should be studied in a multicenter randomized trial of adult medical patients in the ICU, where a substantial number of days are spent caring for those who die.
ACKNOWLEDGMENT The authors would like to recognize the tireless efforts of Father George Winchester, Carol Smith, Barbara Collishaw, and the Brigham and Women’s Hospital medical intensive care unit nurses for their support of critically ill patients and families. This study was supported by the Brigham and Women’s Hospital.
REFERENCES 1. Halpern NA, Bettes L, Greenstein R. Federal and nationwide intensive care units and healthcare costs: 1986 –1992. Crit Care Med. 1994:22:2001–2007. 2. Danis M, Patrick DL, Southerland LI, Green ML. Patients’ and families’ preferences for medical intensive care. JAMA. 1988;260: 797– 802. 3. Danis M, Southerland LI, Garrett JM, et al. A prospective study of advance directives for life-sustaining care. N Engl J Med. 1991;324: 882– 888. 4. Prendergast TJ, Claessens MT, Luce JM. A national survey of endof-life care for critically ill patients. Am J Respir Crit Care Med. 1998;158:1163–1167. 5. Wong DT, Gomez M, McGuire GP, Kavanagh B. Utilization of intensive care unit days in a Canadian medical-surgical intensive care unit. Crit Care Med. 1999;27:1319 –1324. 6. Oye RK, Bellamy PE. Patterns of resource consumption in medical intensive care. Chest. 1991;99:685– 689. 7. Esserman L, Belkora J, Lenert L. Potentially ineffective care. A new outcome to assess the limits of critical care. JAMA. 1995;274:1544 – 1551. 8. Prendergast TJ, Luce JM. Increasing incidence of withholding and withdrawal of life support from the critically ill. Am J Respir Crit Care Med. 1997;155:15–20. 9. Brody H, Campbell ML, Faber-Langendoen K, Ogle KS. Withdrawing intensive life-sustaining treatment—recommendations for compassionate clinical management. N Engl J Med. 1997;336:652– 657. 10. Campbell ML, Carlson RW. Terminal weaning from mechanical ventilation: ethical and practical considerations for patient management. Am J Crit Care. 1992;1:52–56. 11. Gilligan T, Raffin TA. Withdrawing life support: extubation and prolonged terminal weans are inappropriate. Crit Care Med. 1996; 24:352–353. 12. Wilson WC, Smedira NG, Fink C, McDowell JA, Luce JM. Ordering and administration of sedatives and analgesics during the withholding and withdrawal of life support from critically ill patients. JAMA. 1992;267:949 –953. 13. The SUPPORT group of investigators. A controlled trial to improve care for seriously ill hospitalized patients. The study to understand prognoses and preferences for outcomes and risks of treatments (SUPPORT). JAMA. 1995;274:1591–1598. 14. Dowdy MD, Robertson C, Bander JA. A study of proactive ethics consultation for critically and terminally ill patients with extended lengths of stay. Crit Care Med. 1998;26:252–259. 15. Kruse JA, Thill-Baharozian MC, Carlson RW. Comparison of clinical assessment with APACHE II for predicting mortality risk in patients admitted to a medical intensive care unit. JAMA. 1988;260: 1739 –1742. 16. Meyer AA, Messick WJ, Young P, et al. Prospective comparison of clinical judgment and APACHE II score in predicting the outcome in critically ill surgical patients. J Trauma. 1992;32:747–754. 17. Marks RJ, Simons RS, Blizzard RA, Browne DR. Predicting outcome in intensive therapy units—a comparison of Apache II with subjective assessments. Intensive Care Med. 1991;17:159 –163. 18. Castella X, Artigas A, Bion J, Kari A. A comparison of severity of illness scoring systems for intensive care unit patients: results of a multicenter, multinational study. The European/North American Severity Study Group. Crit Care Med. 1995;23:1327–1335. 19. Lee DK, Swinburne AJ, Fedullo AJ, Wahl GW. Withdrawing care. Experience in a medical intensive care unit. JAMA. 1994;271:1358 – 1361.
October 15, 2000
THE AMERICAN JOURNAL OF MEDICINE威
Volume 109 475
mortality, and provider team and family consensus in 134 patients before the intensive communication intervention and in 396 patients after the intervention. RESULTS: Intensive communication significantly reduced the median length of stay from 4 days (interquartile range, 2 to 11 days) to 3 days (2 to 6 days, P ⫽ 0.01 by survival analysis). This reduction remained significant after adjustment for acute physiology and chronic health evaluation (APACHE) 3 score [risk ratio (RR) ⫽ 0.81; 95% confidence interval (CI), 0.66 to 0.99; P ⫽ 0.04). Subgroup analysis revealed that this reduction occurred in our target group, patients with acuity scores in the highest quartile who died (RR ⫽ 0.60; 95% CI, 0.38 to 0.92; P ⫽ 0.02). The intervention, which allowed dying patients earlier access to palliative care, was not associated with increased mortality. CONCLUSIONS: Intensive communication was associated with a reduction in critical care use by patients who died. Our multidisciplinary process targeted advanced supportive technology to patients who survived and allowed the earlier withdrawal of advanced supportive technology when it was ineffective. Am J Med. 2000;109:469 – 475. 䉷2000 by Excerpta Medica, Inc.
A
sition to palliative care affects patient comfort and critical care use. Dying patients consume a disproportionate amount of intensive care resources. In one study, high-use patients who died comprised 13% of intensive care unit (ICU) patients, yet consumed 32% of the resources (7). In our medical ICU in 1996, more than 50% of patient-days were spent providing advanced supportive technology for patients who did not survive. More important, because the support of critically ill patients is accompanied by substantial discomfort and deprivation, it is not humane to continue burdensome technology when there is no reasonable hope that it will be effective. The intravascular lines, support tubes, and restraints used in this process are inherently uncomfortable and are associated with reduced mobility, autonomy, and ability to communicate. When advanced supportive care is associated with healing and the return of function, these discomforts are almost universally acceptable, but they are difficult to justify for patients who die. For this reason, as many as 75% of hospitals now work with patients and their families to withdraw advanced supportive technology when it is not thought to be of benefit (4,8 –12). Like others (13,14), we believe that improved commu-
s much as 1% of the gross national product in the United States is used to provide care for critically ill adults (1). Access to critical care is valued not only by survivors but by the families of most of those who do not survive (2). The wider respect for patient autonomy and the advent of advance directives have allowed some dying patients to receive palliative, rather than aggressive, critical care (3). Despite the alternatives, many patients choose aggressive care, and 2% to 40% do not survive (4). Access to palliative care is often delayed for dying, critically ill patients (5,6), because it is difficult for providers to articulate, and for patients and families to accept, that advanced supportive technology has been ineffective or will not result in a functional outcome that is acceptable to the patient. The optimal timing of the tran-
From the Combined Program in Pulmonary and Critical Care Medicine, Department of Medicine, Brigham and Women’s Hospital and Harvard Medical School, Boston, Massachusetts; and US Army Research Institute of Environmental Medicine, Natick, Massachusetts. Supported by Brigham and Women’s Hospital. Requests for reprints should be addressed to Craig M. Lilly, MD, Respiratory Division, Brigham and Women’s Hospital, 75 Francis Street, Boston, Massachusetts 02115. Manuscript submitted November 22, 1999, and accepted in revised form June 5, 2000. 䉷2000 by Excerpta Medica, Inc. All rights reserved.
0002-9343/00/$–see front matter 469 PII S0002-9343(00)00524-6
Intensive Communication Intervention for the Critically Ill/Lilly et al
nication among the provider team, patient, and family, which affects the timeliness of decision making and length of stay for dying patients, will lead to better outcomes. Although the SUPPORT trial did not demonstrate that a communication intervention led by nurse facilitators affected the time that dying patients spent in the ICU, communication sessions facilitated by a proactive ethics consultation appear to be effective (14). We hypothesized that similar results could be achieved by the physicians and nurses caring for the patient. We therefore tested whether a process of intensive communication that consisted of attending physician– directed multidisciplinary sessions focusing on the goals and outcomes of advanced supportive technology would result in earlier transition to palliative care and reduced length of stay for dying patients, without increasing mortality.
MATERIAL AND METHODS Study Design In a prospective, nonblinded, change-of-practice intervention study, we measured ICU, hospital, and rehabilitation facility mortality rates; probability of remaining in the ICU as a function of time after admission; and survival to discharge from acute and rehabilitative care to home, skilled nursing, or chronic care facilities for 530 consecutive adult patients admitted to a general medical ICU in a tertiary care teaching hospital. We made our intervention in a 10-bed unit in which all patients are cared for by a team consisting of 1 (of a rotating group of 4) critical care– certified attending physician, 2 residents, 3 interns, and 45 nurses (in shifts). The same attending physicians provided care during both phases of this intervention. Admission decisions were made by a standard set of physiological criteria that were applied by a physician who was not part of the ICU team; discharge decisions were made by the critical care physicians. Clinical practice guidelines for the management of pain, dyspnea, anxiety, and delirium had been previously developed by the responsible attending physicians and did not change during the period of the study. The study was conducted for 12 months, beginning October 1, 1997, and was divided into a 3-month preintervention period that included 134 consecutive patients and a 9-month intervention (intensive communication) period that included 396 consecutive patients. The study was performed with the approval of the Human Subjects Committee of the Brigham and Women’s Hospital.
Intensive Communication Intervention In the preintervention period, formal family meetings were usually held after the provider team had reached consensus that restoration of function or survival was unlikely. Patients and families also received information from care providers (most commonly nurses and house 470
October 15, 2000
THE AMERICAN JOURNAL OF MEDICINE威
officers) in a process that was informal and that occurred without explicit coordination among providers. In the intensive communication period, we introduced a proactive process of communication. An initial formal multidisciplinary meeting with the patient, family, or both was held within 72 hours of intensive care admission. The attending physician conducted this meeting for patients who had a predicted length of ICU stay longer than 5 days, a predicted mortality of greater than 25% as estimated by the attending physician, or a change in functional status that was potentially irreversible and sufficient to preclude eventual return to home. The clinical judgment of experienced critical care providers is known to rival or exceed the predictive power of APACHE 2 scores (15–18), and this method of selecting patients identified all patients in the cohort with an APACHE 3 predicted mortality of 25% or more. This meeting was preceded by an assessment of the patient’s medical condition, communication with the primary care physician to review and agree about a recommended care plan, and review of any known advance directives. The formal meeting included the attending physician, nurse, house officer, the patient’s family (including the health care proxy), and, whenever possible, the patient. The meeting had four primary objectives: 1) to review the medical facts and options for treatment; 2) to discuss the patient’s perspectives on death and dying, chronic dependence, loss of function, and the acceptability of the risks and discomforts of critical care; 3) to agree on a care plan; and, most important, 4) to agree on criteria by which the success or failure of this care plan would be judged. This last objective was accomplished by defining clinical events (termed “clinical milestones”) which would indicate that the care plan had been effective, such as resolution of hypotension or reduced need for ventilator support, and the time frame for their occurrence. The timing of subsequent formal meetings was set by this time frame and by the ability of the patient or family to process difficult medical facts. This approach is a variant of the time-limited trial of advanced supportive technology approach reported by Lee et al (19). As clinical milestones were met, new ones were established at subsequent meetings on the basis of the patient’s clinical course and perspectives on death and dying, chronic dependence, and loss of function and the acceptability of the discomforts of critical care. The patient and family were educated about and prepared for home care, rehabilitation, or chronic care, as appropriate. Failure to meet clinical milestones allowed the patient and family to know that advanced supportive technology had been ineffective; alternatives (including palliative care) were then explored. In addition, a weekly multidisciplinary case review was held with nurses, social workers, chaplains, rehabilitation facility providers, and critical care attending physicians.
Volume 109
Intensive Communication Intervention for the Critically Ill/Lilly et al
The goals of these sessions were to ensure a uniform understanding of the patient’s wishes and care plan; to enhance the care team’s ability to provide consistent information to the patient and family; to identify any barriers to the care plan that may have arisen; to review the requirements for successful rehabilitation with representatives of accepting facilities; to review patient outcomes; and to ensure that formal meetings had been held within 72 hours of admission for all patients who met entry criteria. Representatives of chronic care and rehabilitation facilities gave feedback to the ICU care team on the clinical status of patients who had been transferred to their facilities.
Measurement of Patient/Family and Provider Team Nonconsensus Every day, the provider team identified a long-term disposition goal for each ICU patient. These goals consisted of transfer to a floor service (with the intention of eventual discharge home), transfer to a rehabilitation or chronic care facility, or palliative care. The attending physician recorded as a nonconsensus day any day when there was disagreement on what the long-term disposition goal ought to be. Provider team nonconsensus was defined as disagreement among 2 or more involved nurses or physicians, including the primary care physician or specialist acting in this role. Family nonconsensus was defined as family disagreement with the provider team’s consensus recommendation. The attending physician’s daily assessment of consensus was reviewed each week at the multidisciplinary case reviews and independently by a nurse not directly involved in the patient’s care.
Measurement of Acuity and Use Patient acuity was estimated by calculation of the APACHE 3 score for the first critical care day. The primary indication for intensive care was similar to that of other medical ICUs, including management of respiratory failure (72% of patient-days), management of allcause hypotension or arrhythmia (19%), airway management or bronchodilator therapy (4%), electrolyte management (2%), and terminal care (2%). There were no significant differences in any of these categories between the preintervention and the intensive communication periods. APACHE 3 scores were obtained from chart review and determined by a single investigator (DD), who also recorded demographic and clinical characteristics, disposition status, length of stay, and vital status. This investigator was not involved in the study design and was not part of the care delivery team. Use was compared by dividing the number of intensive care days required by the number of preintervention or intensive communication patients served. Use by survival status was also determined.
Statistical Considerations Statistical analyses were performed with SAS 7.0 (Cary, North Carolina), SPSS 9.0 for Windows (Chicago, Illinois), and SigmaStat for Windows (Versions 1.2 and 2.0; Chicago, Illinois). Sex, race, lack of consensus, and whether a family meeting took place were compared by chi-square analysis. Age and APACHE 3 scores for the preintervention and intervention groups were compared by t test. Kaplan-Meier curves were constructed to describe length of stay in the ICU and analyzed by both the log-rank (Mantel-Cox) and the Breslow-Gehan-Wilcoxon tests. Cox proportional hazards analysis was used to assess the effects of covariates on ICU discharge rates. We performed univariate Cox regressions for each of the following variables: group (preintervention vs intervention), sex, season (fall, winter, spring, summer), age, and APACHE 3 score; intervention (P ⫽ 0.02) and APACHE 3 score (P ⫽ 0.09) met the criterion for inclusion (P ⬍0.1) in a multivariate model. Mortality rates by disposition were compared by Z test with Yates’ correction. Differences in mortality between the preintervention and intensive communication periods were compared by chisquare analysis or by Fisher’s exact test as well as after adjusting for APACHE 3 score in a logistic regression model using APACHE 3 score as a covariate. Risk and odds ratios are presented as the ratio of the value for the intensive communication group divided by the value for the preintervention group.
RESULTS The 134 patients in the preintervention group and the 396 patients in the intensive communication group had similar age, race, sex, and mean (Table 1) and median (Figure 1) APACHE 3 scores. A greater percentage of attending physician–led meetings were held in the intensive communication period than in the preintervention period [273 of 396 (69%) vs 58 of 134 (43%), P ⬍0.001]. Multidisciplinary meetings occurred with 270 (68%) of 396 intensive communication patients or their families within 72 hours of ICU admission, including 99% of the targeted group of 273 patients who had loss of independent function, an anticipated length of stay of greater than 5 days, or a predicted mortality greater than 25%. Of the 270 patients, 187 (69%) had an APACHE 3 predicted mortality of 25% or greater; the 83 remaining patients met the other criteria. We failed to meet with the families of 2 unconscious patients because they were never located; we were outside the 72-hour window for an additional patient whose identity was not known to us until the fourth hospital day. Including these initial meetings, 594 senior physician and nursing hours (1.5 hours per patient admitted to the ICU) were required to meet with intensive communication patients and their families.
October 15, 2000
THE AMERICAN JOURNAL OF MEDICINE威
Volume 109 471
Intensive Communication Intervention for the Critically Ill/Lilly et al
Table 1. Characteristics of the Patients before and during the Intensive Communication Intervention
Characteristic
Preintervention Group (n ⫽ 134)
Intensive Communication Group (n ⫽ 396)
P Value
Mean ⫾ SD, Median (25th–75th Percentile), or Number (Percent) Age (years) Male sex Race Caucasian African American Hispanic Asian/other Season of year upon admission Winter Spring Summer Fall APACHE 3 Score†
60 ⫾ 18, 63 (45–75) 67 (50)
58 ⫾ 18, 60 (45–72) 191 (48)
98 (73) 24 (18) 8 (6) 4 (3)
289 (73) 69 (17) 23 (6) 15 (4)
52 (39) 0 0 82 (61) 72 ⫾ 29, 68 (52–87)
99 (25) 134 (34) 143 (36) 20 (5) 67 ⫾ 30, 63 (46–81)
0.3, 0.3* 0.7 1.0
⬍0.001
0.1, 0.07*
* By Student’s t test, Mann-Whitney test. † Available for 387 of the patients in the intensive communication group.
Achieving Consensus
Median (twenty-fifth to seventy-fifth percentile) length of stay in the ICU was reduced during the intervention
period from 4 days (2 to 11 days) to 3 days (2 to 6 days). Unadjusted survival curves (with death or transfer from the ICU as the endpoint) showed a reduction in the chance of remaining in the ICU during the intervention period (P ⫽ 0.01, Figure 2). The effect was most pronounced after the first 72 hours (when the intervention began). The risk of remaining in the ICU was significantly less in the intensive communication group after adjusting for APACHE 3 score [risk ratio ⫽ 0.81, 95% confidence interval (CI), 0.66 to 0.99, P ⫽ 0.04]. Subgroup analyses (Table 2) showed a statistically significant reduction in
Figure 1. Histogram of APACHE 3 scores. The median (twenty-fifth to seventy-fifth percentile) APACHE 3 scores were 68 (52 to 87) in the preintervention group (striped bars) and 63 (46 to 81) in the intensive communication group (black bars).
Figure 2. Proportion of patients remaining in the intensive care unit (ICU) by time. The probability of remaining for the intensive communication group (open circles) was significantly less than that for the preintervention group (solid circles) by survival analysis (log rank P ⫽ 0.01, Breslow-Gehan-Wilcoxon P ⫽ 0.01).
During the intervention period, the rate of provider nonconsensus days decreased from 65 days per 1,000 patientdays to 4 days per 1,000 patient-days, or from 0.56 to 0.02 days per patient. Similarly, the rate of family nonconsensus decreased from 171 days per 1,000 patient-days to 16 days per 1,000 patient-days, or from 1.7 to 0.09 days per patient, after the intervention (P ⬍0.001).
Effects on Length of Stay
472
October 15, 2000
THE AMERICAN JOURNAL OF MEDICINE威
Volume 109
Intensive Communication Intervention for the Critically Ill/Lilly et al
Table 2. Effects of Intensive Communication on Length of Stay Preintervention Group APACHE 3 Quartile (Score Range) Survivors I (11–47) II (48–64) III (65–82) IV (83–185) Patients who died I (11–47) II (48–64) III (65–82) IV (83–185)
Intensive Communication Group
N
Median (25th–75th Percentile) Length of Stay (Days)
Risk Ratio (95% Confidence Interval)*
P Value
3 (2–4) 3 (2–5) 5 (2–15) 5 (3–8)
104 68 67 20
3 (2–4) 3 (2–4) 4 (3–7) 4.5 (3–8)
0.9 (0.6–1.5) 0.9 (0.6–1.4) 0.9 (0.5–1.6) 1.1 (0.5–2.4)
0.8 0.7 0.8 0.8
8 (6–11) 8 (4–15) 10 (4–14) 5 (3–12)
4 24 31 69
0.9 (0.2–4.2) 1.1 (0.5–2.3) 1.0 (0.5–1.8) 0.6 (0.4–0.9)
0.9 0.9 0.9 0.02
N
Median (25th–75th Percentile) Length of Stay (Days)
19 29 16 9 4 9 17 31
5 (4–14) 6 (2–14) 7 (4–15) 3 (2–7)
*Effects of intervention on rate of discharge from intensive care unit, estimated using Cox proportional hazard models within APACHE 3 strata with treatment group as the sole independent variable.
length of stay only among those who had been principally targeted by our intervention, namely those in the highest quartile of APACHE 3 scores who eventually died. During the preintervention periods, 1,072 ICU beddays were required to care for 134 patients (8.0 bed-days per patient), compared with 2,349 ICU bed-days for 396 patients (5.9 bed-days per patient) during the intervention.
Effects on Mortality The intensive communication intervention did not increase mortality (Table 3). Rather, there was a decrease in unadjusted overall mortality (P ⫽ 0.02) and a trend toward reduced mortality during the ICU stay (P ⫽ 0.06). About one-third of the decrease in overall mortality was
accounted for by an increase in the proportion (from 17% to 28%; Table 4) of less severely ill patients (APACHE 3 scores in the lowest quartile) who were admitted to the ICU. This was also associated with a decrease in the ICU occupancy rate, from 98% to 92%, during the intensive communication period. The decrease in overall mortality remained after adjustment for APACHE 3 score (odds ratio ⫽ 0.61, 95% CI, 0.38 to 0.98, P ⫽ 0.04). Subgroup analyses revealed significant reductions in mortality among patients admitted to the ICU in the lowest and second-highest APACHE 3 quartiles (Table 4). The trend toward reduced ICU mortality was not the result of an increased rate of transfer of dying patients
Table 3. Patient Outcomes
Eventual Disposition
Intensive Communication Group (n ⫽ 396)
Preintervention Group (n ⫽ 134)
P Value
Number (Percent) Survived ICU stay and discharged* Directly from the ICU From a floor service From a rehabilitation facility Total Died In the ICU On a floor service In a rehabilition facility Total
3 (2) 42 (31) 28 (21) 73 (54)
8 (2) 157 (40) 97 (24) 262 (66)
0.8 0.1 0.5 0.02
42 (31) 12 (9) 7 (5) 61 (46)
90 (23) 39 (10) 5 (1) 134 (34)
0.06 0.9 0.02 0.02
* To home, an assisted living arrangement, or a skilled nursing facility. ICU ⫽ intensive care unit. October 15, 2000
THE AMERICAN JOURNAL OF MEDICINE威
Volume 109 473
Intensive Communication Intervention for the Critically Ill/Lilly et al
Table 4. Mortality Rates by APACHE 3 Score Preintervention Group
Intensive Communication Group
APACHE 3 Quartile
Died/Total
Percent
Died/Total
Percent
Odds Ratio (95% Confidence Interval)*
P Value
I II III IV All subjects†
4/23 9/38 17/33 31/40 61/134
17 24 52 78 46
4/108 24/92 31/98 69/89 134/396
4 26 32 78 34
0.2 (0.04–0.8) 1.1 (0.5–2.7) 0.4 (0.2–1.0) 1.0 (0.4–2.4) 0.6 (0.4–0.9)
0.03 0.8 0.04 1.0 0.02
* For mortality, comparing the intensive communication group with the preintervention group. † Includes 9 patients in the intensive communication group for whom APACHE 3 scores could not be calculated.
from the ICU, as 42 of the 61 patients (69%) who died in the preintervention period did so in the ICU, compared with 90 of the 134 patients (67%) who died in the intensive communication period. There was, however, a trend toward fewer transfers to a rehabilitation facility of patients who died before discharge to home: 11% of those who died in the preintervention period did so in a rehabilitation facility, compared with 4% in the intensive communication period (P ⫽ 0.08). This outcome was associated with a statistically significant reduction in mortality among those transferred to a rehabilitation facility: 7 of 35 such patients (20%) died in the preintervention period, and 5 of 102 such patients (5%) died in the intensive communication period (P ⫽ 0.02).
DISCUSSION We undertook this intensive communication intervention to improve care for critically ill adult medical patients by encouraging advanced supportive technology when it was effective for accomplishing patient-directed goals and limiting it when ineffective. We confirmed that critical care effectively restores life and function in the majority of critically ill adults (1,4,15–18), and we found that intensive communication accomplished this goal with greater concordance among patients, families, and providers and with fewer preterminal days of intensive care for dying patients. Early formal multidisciplinary meetings partnered physicians with nurses, chaplains, and social workers to facilitate the communication of consistent medical information. Our results demonstrate that caregivers spent more time meeting with and listening to patients and families about their views and values, a process that fostered agreement to continue advanced supportive technology when it was effective for meeting the goals of the patient and permitted an earlier transition to ICU-based palliative care when it was not. Dying patients spent fewer days in the ICU. There was a statistically significant reduction in overall median length of ICU stay that was attributable to a reduction in the length of stay of dying patients with the highest 474
October 15, 2000
THE AMERICAN JOURNAL OF MEDICINE威
APACHE scores, the very patients for whom our intervention was intended. Had the reduction in length of stay been caused by secular trends, we would have expected a reduction in length of stay in other subgroups as well. Furthermore, the reduction in length of stay began on the third ICU day, as would be expected for an intervention that occurred during the first 72 hours of intensive care admission. We cannot determine the relative importance of a formal consensus-building process, an increase in the uniformity of the content of our communications, a recognition of advanced supportive technology as ineffective when it was not working, or a commitment by the attending physician to offer the option of providing earlier access to palliative care for dying patients. We believe that each of these elements was beneficial in some circumstances. Our intervention was associated with a substantial reduction in ICU use: approximately 200 fewer days for every 100 critically ill patients. This reduction in use was not uniform for all patients; rather it was limited to highacuity patients who died (Table 2). A consequence of the reduction in length of ICU stay was an increase in bed availability, which permitted more patients with APACHE scores in the lowest quartile to be admitted to the ICU, thereby contributing to a lower overall mortality rate. We considered the possibility that the reduction in mortality was the result of systematic changes in our critical care practice or delivery system. However, the mortality reductions were not seen in all of the APACHE quartiles as might be expected from a secular trend, and the duration of the preintervention period was short. One of our concerns about this intervention was that it might increase mortality rates by introducing a bias toward palliative care when continued advanced supportive technology would have preserved life. The intervention, however, did not change mortality in our targeted group of patients in the highest quartile of APACHE 3 scores, despite a significant reduction in length of stay among patients in this subgroup who died. Unlike length of stay, mortality was reduced among patients in the first and third quartiles of APACHE 3 score. Although the reduc-
Volume 109
Intensive Communication Intervention for the Critically Ill/Lilly et al
tion in mortality among patients in the lowest (first) quartile of APACHE 3 score was probably the result of greater bed availability, reduced mortality in the third quartile may have been a consequence of our intervention. We speculate that agreeing to clinical milestones encouraged the provider team to support severely ill patients who had the potential to survive but were improving slowly. We also found a statistically significant reduction in mortality among patients transferred to rehabilitation facilities, which suggests that an indirect benefit of our weekly multidisciplinary case reviews was better selection and preparation of patients for the rehabilitation process. A major limitation of this study is that it was not randomized or blinded. We did not perform a randomized study for several reasons. First, patients and physicians would have needed to agree to being assigned to a group that received less communication. Second, enforcement or monitoring of communication would be necessary, because providers who viewed the communication process paradigm as a substantial advance in access to palliative care may have felt obligated to use these techniques for dying patients in the less intensive communication strategy. We were able to introduce some of the advantages of randomization and blinding. Because they were not involved in the ICU admission process, the investigators were prevented from selecting patients for ICU admission who were likely to require shorter intensive care stays. The small number of providers, the creation of guidelines, and the use of weekly multidisciplinary case reviews served as a protection against differences resulting from changing clinical practice and individual practice styles. In summary, we found that an attending physician– led, multidisciplinary process of early and frequent communication that identified patient goals and gauged the effectiveness of care delivered critical care effectively to surviving patients and permitted earlier withdrawal of advanced supportive technology when it was ineffective for restoring life or accomplishing the patient’s goals. The magnitude of the effects of our program on length of stay for dying patients suggests that it should be studied in a multicenter randomized trial of adult medical patients in the ICU, where a substantial number of days are spent caring for those who die.
ACKNOWLEDGMENT The authors would like to recognize the tireless efforts of Father George Winchester, Carol Smith, Barbara Collishaw, and the Brigham and Women’s Hospital medical intensive care unit nurses for their support of critically ill patients and families. This study was supported by the Brigham and Women’s Hospital.
REFERENCES 1. Halpern NA, Bettes L, Greenstein R. Federal and nationwide intensive care units and healthcare costs: 1986 –1992. Crit Care Med. 1994:22:2001–2007. 2. Danis M, Patrick DL, Southerland LI, Green ML. Patients’ and families’ preferences for medical intensive care. JAMA. 1988;260: 797– 802. 3. Danis M, Southerland LI, Garrett JM, et al. A prospective study of advance directives for life-sustaining care. N Engl J Med. 1991;324: 882– 888. 4. Prendergast TJ, Claessens MT, Luce JM. A national survey of endof-life care for critically ill patients. Am J Respir Crit Care Med. 1998;158:1163–1167. 5. Wong DT, Gomez M, McGuire GP, Kavanagh B. Utilization of intensive care unit days in a Canadian medical-surgical intensive care unit. Crit Care Med. 1999;27:1319 –1324. 6. Oye RK, Bellamy PE. Patterns of resource consumption in medical intensive care. Chest. 1991;99:685– 689. 7. Esserman L, Belkora J, Lenert L. Potentially ineffective care. A new outcome to assess the limits of critical care. JAMA. 1995;274:1544 – 1551. 8. Prendergast TJ, Luce JM. Increasing incidence of withholding and withdrawal of life support from the critically ill. Am J Respir Crit Care Med. 1997;155:15–20. 9. Brody H, Campbell ML, Faber-Langendoen K, Ogle KS. Withdrawing intensive life-sustaining treatment—recommendations for compassionate clinical management. N Engl J Med. 1997;336:652– 657. 10. Campbell ML, Carlson RW. Terminal weaning from mechanical ventilation: ethical and practical considerations for patient management. Am J Crit Care. 1992;1:52–56. 11. Gilligan T, Raffin TA. Withdrawing life support: extubation and prolonged terminal weans are inappropriate. Crit Care Med. 1996; 24:352–353. 12. Wilson WC, Smedira NG, Fink C, McDowell JA, Luce JM. Ordering and administration of sedatives and analgesics during the withholding and withdrawal of life support from critically ill patients. JAMA. 1992;267:949 –953. 13. The SUPPORT group of investigators. A controlled trial to improve care for seriously ill hospitalized patients. The study to understand prognoses and preferences for outcomes and risks of treatments (SUPPORT). JAMA. 1995;274:1591–1598. 14. Dowdy MD, Robertson C, Bander JA. A study of proactive ethics consultation for critically and terminally ill patients with extended lengths of stay. Crit Care Med. 1998;26:252–259. 15. Kruse JA, Thill-Baharozian MC, Carlson RW. Comparison of clinical assessment with APACHE II for predicting mortality risk in patients admitted to a medical intensive care unit. JAMA. 1988;260: 1739 –1742. 16. Meyer AA, Messick WJ, Young P, et al. Prospective comparison of clinical judgment and APACHE II score in predicting the outcome in critically ill surgical patients. J Trauma. 1992;32:747–754. 17. Marks RJ, Simons RS, Blizzard RA, Browne DR. Predicting outcome in intensive therapy units—a comparison of Apache II with subjective assessments. Intensive Care Med. 1991;17:159 –163. 18. Castella X, Artigas A, Bion J, Kari A. A comparison of severity of illness scoring systems for intensive care unit patients: results of a multicenter, multinational study. The European/North American Severity Study Group. Crit Care Med. 1995;23:1327–1335. 19. Lee DK, Swinburne AJ, Fedullo AJ, Wahl GW. Withdrawing care. Experience in a medical intensive care unit. JAMA. 1994;271:1358 – 1361.
October 15, 2000
THE AMERICAN JOURNAL OF MEDICINE威
Volume 109 475