The Timing of Family Meetings in the Medical Intensive Care Unit

Abstract

Purpose

Family meetings in the medical intensive care unit can improve outcomes. Little is known about when meetings occur in practice. We aimed to determine the time from admission to family meetings in the medical intensive care unit and assess the relationship of meetings with mortality.

Methods

We performed a prospective cohort study of critically ill adult patients admitted to the medical intensive care unit at an urban academic medical center. Using manual chart review, the primary outcome was any attempt at holding a family meeting within 72 hours of admission. Competing risk models estimated the time from admission to family meeting and to patient death or discharge.

Results

Of the 131 patients who met inclusion criteria in the 12-month study period, the median time from admission to family meeting was 4 days. Fewer than half of patients had a documented family meeting within 72 hours of admission (n = 60/131, 46%), with substantial interphysician variability in meeting rates ranging from 28% to 63%. Patients with family meetings within 72 hours were 30 times more likely to die within 72 hours (32% vs 1%, P < .001). Of the 55 patients who died in the intensive care unit, 27 (49%) had their first family meeting within 1 day of death.

Conclusions

Family meetings occur considerably later than 72 hours and are often held in close proximity to a patient’s death. This suggests for some physicians, family meetings may primarily be used to negotiate withdrawal of life support rather than to support the patient and family.

Introduction

Intensive care unit (ICU) admissions account for over 5.7 million patient admissions each year in the United States.¹ Families are important in the care of ICU patients since many ICU patients lack decisional capacity requiring family members to act as alternative decision-makers.² These families may personally encounter negative effects when their relative is in the ICU, as over 50% may experience significant stress due to the patient’s illness.³ Family members also are at risk of developing post-traumatic stress symptoms.⁴

Structured communication with families may be a part of solution to improving caregiver stress. A recent randomized control trial showed structured ICU communication leads to increased surrogate ratings of quality of communication, patient and family centeredness of care, and decreased length of ICU and hospital stay.⁵ A 2011 systematic review also found communication interventions in the ICU may benefit patient families by reducing family distress and improving family comprehension of medical care.⁶

Despite the above evidence that structured communication may improve patient and family outcomes, little is known about how family meetings are actually used in the ICU. We hypothesize some physicians are not regularly conducting family meetings, which may result in poor communication with families and poor patient-centered care.

We hypothesized that the majority of family meetings in the ICU occur later than 72 hours into an ICU admission and typically occur shortly before patient death. Therefore, in this study, we aimed to determine when ICU physicians hold family meetings after patient admission and the relationship between family meetings and death in the ICU.

Materials and Methods

Study Population

We collected an observational prospective cohort of all adult patients admitted to the medical ICU at an urban academic medical center for 1 year from January 1, 2017 to December 31, 2017. Inclusion criteria were continuous mechanical ventilation for at least 24 hours and Acute Physiology and Chronic Health Evaluation II (APACHE II) score ≥ 25. This inclusion criteria identified patients with high predicted ICU mortality >55%⁷ who likely lacked decisional capacity and thus required family meetings to determine a plan of care respecting patient values. It excluded patients who maintained or quickly recovered decisional capacity for whom family meetings may not be needed.

Chart Abstraction and Determination of a Family Meeting

Our primary study outcome was whether a family meeting occurred within 72 hours of admission, which we defined as a “timely family meeting.” Recent guidelines recommend routine family meetings without setting a specific time frame.⁸ We chose 72 hours as we hypothesize this time frame allows clinicians to complete an evaluation of the patient, attempt to reach the patient’s health-care surrogate, and ensure timely assessment of patient values regarding care, such as a previous request to be do not resuscitate.

Families were defined broadly as any alternative decision-makers or related persons involved in patient care. Family meetings were defined as any documented discussion between any clinician and families regarding patient care, excluding brief bedside updates. These meetings could include discussion of topics such as code status and patient values regarding medical care. Clinicians could include physicians such as medical trainees and attendings from any specialty, nurses, social workers, and chaplains. We manually evaluated all notes in the electronic medical record for any discussion of a family meeting. We then recorded the time from ICU admission to when a family meeting was stated to have occurred. Documented attempts at family meetings were counted as actual meetings. Attempts at family meetings required documentation that someone from the medical team tried to contact the patient’s family.

All physicians attending in the medical ICU in 2017 were surveyed to confirm the association between family meeting documentation and their actual occurrence. Attendings were also surveyed on the primary reason they perform family meetings.

We also collected data including age, gender, race, ethnicity, preferred language, reason for admission, cancer diagnosis, indication for intubation, time to family meeting, time to discharge or death, labs, and vital signs.

Statistical Analysis

We calculated the rate of family meetings within 72 hours for all patients and determined the median time from admission to family meeting using competing risks analysis to account for recovery, discharge, and death before a family meeting. Differences in characteristics for patients with and without a family meeting were compared using Fisher exact test for categorical variables and the Wilcoxon rank sum test for continuous variables. We fit a mixed effects logistic regression model to identify physician level predictors of timely family meetings and quantify interphysician variation in meeting performance. We adjusted this model for APACHE II score, age, gender, race, cancer diagnosis, and indication for intubation. Using standard methodology for public reporting of hospital outcomes,^9–11 we used the model results to calculate the case-mix adjusted rate of timely family meetings for each physician.

To assess the relationship between family meetings and ICU mortality, we fit a Cox proportional hazards model and adjusted for APACHE II score, age, gender, race, cancer diagnosis, and reason for admission. We used a competing risks model to estimate the cumulative incidence of death, recovery, and discharge following a family meeting at any point during the ICU admission.^12,13 For patients who died in the ICU, we calculated the median time from family meeting to death. Figures were created using Stata software.

This human research study was approved by the University of Chicago Institutional Review Board. All analyses were performed using R 3.4.3, copyright The R Foundation for Statistical Computing 2017. Uncertainty of model estimates was quantified using 95% confidence intervals (CI). A P value of < .05 was considered statistically significant.

Results

Study Population Characteristics

During the 12-month study period, there were 2526 patients admitted by 18 attending physicians to the medical ICU at the University of Chicago. Of these patients, 370 (14.6%) were intubated during their admission. Patients were excluded if they did not remain intubated 1 day post ICU admission (n = 189) or if their APACHE II score was less than 25 (n = 50). A total of 131 severely ill intubated patients were eligible for analysis (Figure 1).

Figure 1.

Study population. To meet inclusion criteria for this study, patients needed to have received mechanical ventilation >24 hours and have APACHE II scores ≥25 on admission. Of the patients who met these criteria, 60 had a family meeting within 72 hours of admission and 71 did not. APACHE II indicates Acute Physiology and Chronic Health Evaluation II.

The average age of patients was 62 years and the majority (76%) of patients were African-American. The most common reason for admission was cardiac failure, the median duration of ventilation was 4 days, and the mean length of ICU stay was 6 days. The median APACHE II score was 29 (interquartile range: 26–33) and observed ICU mortality was 42% (Table 1). The mean length of stay in the ICU was 6 days for all patients, 8 days for patients who were discharged alive from the ICU, and 4 days for patients who died in the ICU.

Table 1.

Study Population Characteristics by Occurrence of Family Meeting Within 72 Hours.^a

	Study Population (N = 131)	No Family Meeting Within 72 Hours of Admission (n = 71)	Family Meeting Within 72 Hours of Admission (n = 60)	P Value
Indication for mechanical ventilation				<.001
Respiratory failure	33 (25%)	22 (31%)	11 (18%)
Airway compromise	8 (6%)	8 (11%)	0 (0%)
Cardiac failure	48 (37%)	21 (30%)	27 (45%)
Septic shock	21 (16%)	4 (6%)	17 (28%)
Other	21 (16%)	16 (23%)	5 (8%)
APACHE II at admission				.011
Median (IQR)	29 (26–33)	28 (26–32)	30 (28–35)
>35	25 (19%)	11 (15%)	14 (23%)
Duration of ventilation				.341
Median time to extubation (IQR)	4 (3–7)	4 (3–7)	4 (2–6)
Extubated within 72 hours	33 (25%)	22 (31%)	11 (18%)
Mechanical ventilation until death or discharge	48 (37%)	14 (20%)	34 (57%)
Time to death or discharge				<.001
Median (IQR)	5 (4–9)	7 (4–10)	4 (3–6)
In ICU more than 72 hours	100 (76%)	65 (92%)	35 (58%)
Outcome
Early death (<72 hours)	20 (15%)	1 (1%)	19 (32%)	<.001
Late death (>72 hours)	35 (27%)	15 (21%)	20 (33%)	.165
Discharged alive	76 (58%)	55 (77%)	21 (35%)	<.001
Race				.592
White	21 (16%)	13 (18%)	8 (13%)
Black	99 (76%)	51 (72%)	48 (80%)
Other	11 (8%)	7 (10%)	4 (7%)
Age				.015
Median (IQR)	62 (51–72)	58 (46–70)	66 (57–73)
>70 years	37 (28%)	15 (21%)	22 (37%)

Abbreviations: APACHE II, Acute Physiology and Chronic Health Evaluation II; ICU, intensive care unit; IQR, interquartile range.

^aTime-quantified in days. Participant characteristics are evaluated for the entire study population and the subsets of patients who received or did not receive a family meeting within 72 hours of admission. Continuous variables are reported as median (IQR) and categorical variables are reported as count (percentage). P values for comparisons of categorical variables were calculated using Fisher exact test. P values for continuous comparisons were calculated using 2-sample Wilcoxon rank-sum (Mann-Whitney) test.

Timing of Family Meetings

Within the first 72 hours of ICU admission, physicians held family meetings for 60 (46%) out of 131 critically ill patients. Patients with family meetings within 72 hours had on average higher APACHE II scores (30 vs 28, P = .011), greater likelihood of being admitted for septic shock (28% vs 6%, P < .001) or cardiac failure (45% vs 30%, P < .001), higher mortality within 72 hours of admission (32% vs 1%, P < .001), and older age (66 vs 58, P = .015; Table 1). Over the entire ICU admission, 90 (69%) out of 131 patients had a family meeting prior to being discharged from the ICU. Of these 90 meetings, 5 were documented attempts at meetings where surrogate decision-makers could not be found. All meetings evaluated were led by critical care medical trainees and attendings, with no documented meetings led by palliative medicine or other consulting clinicians.

Family meeting status is displayed in Figure 2. Accounting for the competing risks of recovery and discharge and death, the median time to from admission to family meeting was 4 days. Of the 43 patients who remained in the ICU over 7 days, 17 (40%) had no documented family meeting during their first week of admission despite their high severity of illness.

Figure 2.

Prevalence of family meetings in cohort of severely ill, intubated MICU patients over time. The relationship of timing of family meetings over time among surviving patients remaining in the ICU. Time is measured on the x-axis and evaluates at discrete time points the number of patients who remained in the ICU who had not received a family meeting or had received a family meeting. At day 3, 35 (35%) out of 100 patients who remained in the ICU had received a family meeting. After 1 week, 17 (40%) out of 43 patients who remained in the ICU had received a family meeting. Not until the end of day 5 were there more alive patients in the ICU who had received family meetings than those who did not. ICU, intensive care unit.

Interphysician Variability

During the study period, 18 physicians attended in the ICU including 6 professors, 5 associate professors, and 7 assistant professors/clinical instructors. Four (22%) physicians were female. Sixteen (89%) reported documentation of formal family meetings is always entered into the medical record (Online Figure 1 in the Supplemental Material). Seven (39%) reported the primary reason they arrange family meetings was to discuss treatment limitations or the withdrawal of life support (Online Figure 2 in the Supplemental Material). There was substantial interphysician variation in the holding of timely family meetings, with adjusted rates ranging from 28% to 63% (Figure 3). There was no relationship between attending rank or gender with the holding of timely family meetings (Online Table 1 in the Supplemental Material).

Figure 3.

Interphysician variation in the performance of family meetings. Case-mix adjusted probability of holding a timely family meeting within 72 hours for the 18 attending physicians studied. There was substantial interphysician variability in holding family meetings with adjusted rates ranging from 28.5% to 64%. Error bars represent 95% confidence interval for each physician’s empirical Bayes estimate.

Relationship With ICU Mortality

Of the 55 patients who died during their ICU stay, 39 (71%) had a timely family meeting and 54 (98%) had a meeting at some point prior to their death (Table 1 and Online Figure 3 in the Supplemental Material). In contrast, of the 76 ICU survivors, 40 (53%) did not have a family meeting at any point during their ICU stay and only 21 (28%) had a timely family meeting. Patients who had a timely family meeting had an adjusted risk of ICU mortality over 3 times higher than patients who did not have a meeting within 72 hours (adjusted hazard ratio [HR]; 3.6, 95% CI: 1.8–7.3; Online Table 2 in the Supplemental Material). The HR comparing death to discharge from the ICU was 2 times higher for patients who had a family meeting (HR: 2.66, 95% CI: 1.65–4.29; Figure 4).

Figure 4.

Patient outcome following a family meeting. Cumulative incidence of ICU death and survival following a family meeting using a competing risks analysis. Within the first day of a family meeting, the cumulative incidence of death was 30% and the cumulative incidence of recovery and discharge was 7%. At 3 days post family meeting, the cumulative incidence of death was 42% and the cumulative incidence of recovery was 18% (P < .05 for all time point comparisons). If family meetings are being used for the primary goal of supporting patients and families, the cumulative incidence of both events should be similar. ICU indicates intensive care unit.

Of the 54 patients who had a family meeting and died during their ICU admission, 27 (49%) had their first family meeting on the same day as or day just prior to death. The most common primary diagnoses for these patients included 11 (41%) with cardiac failure, 7 (26%) with sepsis, and 4 (15%) with acute hypoxic respiratory failure. Only 16 (30%) of the 54 patients who had a family meeting and died during their ICU admission survived more than 3 days after their first family meeting (Online Figure 3 in the Supplemental Material).

Discussion

To our knowledge, this is the first study to examine both the timing of family meetings and their relationship to patient death in an ICU setting. We found family meetings often occur late in a patient’s ICU stay with only 46% of patients receiving a family meeting within 72 hours. Not surprisingly, our study also found there was substantial variability in the rate at which family meetings occurred based on attending physician. We found the median time to a family meeting to be 4 days, indicating over half of ICU patients were still awaiting a family meeting at day 5 of their admission. Finally, we found a strong association between the timing of family meetings and patient death. Patients who had timely family meetings were 30 times more likely to die within 72 hours of their admission and were more than 3 times as likely to die at some point during their ICU stay. Unfortunately, the time between family meetings and death was typically short, with about half of patients with timely family meetings dying within 1 day of their first family meeting.

The delay in performance of family meetings found in this study is concerning considering the known benefits of having structured communication in the ICU and that decision-incapacitated patients require family meetings to ascertain patient values.^5,6 Lack of ICU clinician knowledge of the benefits of family meetings and inadequate communication training may explain why family meetings occur late. Although communication skills training is required in pulmonary and critical care fellowships by the Accreditation Council for Graduate Medical Education and recommended by the American Thoracic Society, this training is not standardized and the breadth to which it occurs is not known.^14,15 Studies have shown ICU clinician communication can be inadequate with families perceiving conflicts with clinicians and having poor comprehension of both the patient’s illness and care plan.^16,17 Medical residents, who frequently are called upon by attending physicians to conduct family meetings in teaching institutions, report feeling unprepared to lead family conversations for acutely ill patients.^18,19 Physician practice plays a large role in when family meetings occur, which may be due to differences in communication training clinicians have received in residency and fellowship. Interestingly, we did not find an association between attending gender or rank and the timing of family meetings. Intensive care unit clinician deficiencies in communication likely cannot be overcome by the addition of palliative care trained communicators alone, as the inclusion of palliative care trained communicators in the ICU has not been found to improve family member emotional symptoms or satisfaction with care.^20,21 These findings suggest standardization and improved quality of communication training for ICU clinicians is needed to improve patient and family care.

The establishment of standardized protocols for family meetings may also play a role in improving care in the ICU. Protocolized communication with surrogates in the ICU can lead to improved surrogate perception of quality of communication and patient-centered care and reduce length of ICU stay.⁵ Although family meetings may become depersonalized if conducted based on a protocol, such protocols would ensure clinicians would learn about patient values and improve early support to families.^22,23

We found a close relationship between the timing of family meetings and mortality in the ICU. Clinicians were far more likely to have timely family meetings for patients who died in the ICU compared to patients who survived to ICU discharge. If family meetings were used primarily to support patients and families, patients should be no more likely to die after having a family meeting. Our results were not confounded by unknown patient identity or lack of surrogates as we included any attempt to find and meet with a patient surrogate as an actual meeting. The time between first family meeting and patient death was short. Among patients who died in the ICU, almost half had their first family meeting within 1 day of death. These findings raise concern that some physicians use family meetings primarily as a means to negotiate the withdrawal of life support rather than as a tool to comfort families and respect patient autonomy. Supporting this theory is that a plurality of attending physicians (39%) cited “discuss treatment limitations or withdrawal of life support” as the primary reason family meetings were arranged.

Limitations

Our study was from a single center and therefore may not be generalizable to other institutions. However, the interphysician variation we documented captured a range of physician practices increasing the generalizability of our study. Although patients evaluated were mostly English speaking, the patients in this prospective cohort had other diverse demographic factors including race, age, and primary diagnoses which are representative of many urban ICU populations. Our study only evaluated acutely ill patients who likely lacked decisional capacity, not patients with decisional capacity who may not require family meetings. Our study did not evaluate patients in other specialty ICUs such as the neurological or surgical ICU where patients may have longer length of stays and determination of patient prognosis may take longer to ascertain.

Although it is possible data from the electronic medical record such as patient demographics, vital signs, and laboratory values were incorrectly reported, inaccurate data likely had little impact on our results. As patient severity was determined using the APACHE II score which utilizes data from a variety of measures, it is unlikely a few incorrectly reported values would greatly alter the overall APACHE II score. Although it is possible inaccuracies in patient demographics existed, these likely are low due to institutional protocols requiring nursing staff in the ICU to reassess and document these demographics.

Because data were collected by chart review and not direct observation, there was likely heterogeneity in our primary outcome of a “timely family meeting.” We did not assess the quality of family meetings in any way in this study. It is possible we missed meetings not documented in the medical record; however, most physicians reported documentation of meetings in the medical record was standard practice. Inaccuracies in documentation of family meetings may have existed in the medical record, such as a clinician incorrectly documenting a meeting for a patient when it in fact had not occurred. We believe these inaccuracies likely are low, as many family meetings were documented by multiple clinicians, providing support that these meetings actually occurred on the day stated.

We used the time point to family meeting of 72 hours as this gives clinicians sufficient time to complete an evaluation of the patient to discuss with a patient’s surrogate; however, there is no consensus guideline recommending when a family meeting should occur. A recent positive randomized controlled study required structured family communication within 48 hours, which suggests early communication is effective.⁵ Our 72 hour time point is a more generous time frame than 48 hours, and we still found less than half of the patients we studied had family meetings in this period, which we believe is an important result.

Conclusions

In summary, family meetings occur considerably later than within 72 hours and a minority of attending physicians consistently hold meetings during this period. When meetings do occur, they often are held in very close proximity to the patient’s death, suggesting these meetings may be used primarily to negotiate the withdrawal of life support rather than to learn patient values and support families. Lack of standardized physician education in how best to have family meetings, inadequate physician knowledge of the known benefits of family meetings, and lack of institutional protocols for having family meetings in the ICU likely contribute to our findings. These results provide support for expanding communication skills training for clinicians in the ICU in addition to instituting a standard protocol for holding family meetings.