Analyzing clinical and system drivers of satisfaction in the intensive care unit as a component of high quality care

Abstract

Background:

Quality improvement in the intensive care unit has transitioned from focusing on mortality to improving care and reducing morbidity.

Objective:

This study prospectively investigated clinical and system drivers of family satisfaction in a large quaternary hospital ICU.

Methods:

A validated tool was distributed to family members and a registry chart analysis was conducted. The aims were to assess associations with high or low family satisfaction to evaluate unit-level satisfaction. Candidate predictors were selected from univariate logistic regressions and finalized in a multivariate model by a stepwise selection approach.

Results:

Overall, 75% (n=188) of respondents (n=250) indicated high satisfaction. Respondents with higher satisfaction had a Plan of the Day posted (OR = 3.3, 95% CI: 1.63, 6.89, p =0.001), and did not live with the patient (OR =0.5, 95% CI: 0.25, 0.96, p=0.044).

Conclusion:

This study indicates that communication and transparency of plans contributes to family satisfaction with ICU care.

INTRODUCTION

Intensive care units (ICUs) provide treatment of high-acuity illnesses and life-threatening disease to over five million patients annually in the United States, at a substantial emotional and financial burden for patients, families, and the US healthcare system[1–3]. In part due to technological advancements and efficiency of care, the survival of patients in ICUs has increased dramatically over the past two decades[4]. As a result, outcomes have transitioned from primarily mortality-focused to assessing the effectiveness of healthcare delivery and efforts towards value-based care. Value, traditionally outlined as outcomes per cost, has increasingly been recognized to also incorporate the patient’s or family’s perception of care[5, 6]. ICU Quality improvement (QI) and research collaboratives are increasingly integrating family input as a means to mitigate the negative impact ICU admissions have on families. [7]

To assess these efforts, satisfaction - the patient or family reported positive or negative experience of a given encounter with the healthcare system - has become an area of increased focus for healthcare systems [6, 8–10]. Understanding satisfaction data allows healthcare teams to provide compassionate care as families face challenging decisions, and offers the opportunity to insure family goals are shared for the trajectory of care. Studies conducted with ICU patients and families indicate higher patient and family satisfaction is correlated with lower in-hospital morbidity, lower readmission rates, and follow-up with specialists upon discharge[11–13]. However, there remains a gap in understanding risks of high or low satisfaction to families and patients. Specifically, assessing communication of the care team, or tools used to guide patients and families on their hospital course, will provide insight into what results in highly satisfied families. Attempts to improve communication with care teams, such as transparent daily plans (in our institution referred to as the “Plan of the Day”), have not been assessed whether they improve satisfaction.

Increasingly, reported satisfaction as an outward facing indicator of healthcare quality has been endorsed by leading professional organizations, including the American College of Medical Quality, and Society of Critical Care Medicine (SCCM) and the National Academy of Medicine, which proposed family satisfaction as “a key component for enhancement of the effectiveness of healthcare microsystems” [14–17]. Given the acuity and often altered state of critically ill patients, the SCCM has indicated in these statements that feedback from the family is an extension of patients and that the family view of care is “reflective of patient needs as the family is representative of the overall care to the patient” [14–17]. While the current body of literature signals potential use of satisfaction data as a process metric for downstream patient behavior change, patient and family satisfaction in itself represents an outcome of merit to provide excellent patient care, which remains understudied and ineffectively measured.

Satisfaction in the inpatient setting is a complex construct encompassing many domains, and challenges remain in identifying drivers of satisfaction, particularly in the ICU setting. Measuring ICU satisfaction to this point has been challenging. Patients often spend time on multiple floors or units in a single admission, yet hospital-wide satisfaction surveys to patients or families are based on the general hospital visit, without capturing the experiences of the ICU[18]. Due to such barriers, determinants of high or low satisfaction in the ICU are not known, inhibiting local improvement teams ability to intervene and advance the services provided to families.

The purpose of this study was to investigate drivers of family satisfaction specific to the ICU, which is currently not described. The factors assessed included system drivers, such as length of stay or communication with the care teams, and patient factors, such as severity of illness and type of diagnosis. Family satisfaction data was gathered at the time of transfer out of the ICU, in coordination with a focused review of the patient’s hospital course, to detect relationships between populations in the ICU and family satisfaction. We hypothesized populations with higher satisfaction would be those with lower severity, shorter length of stay, and presence of an outlined daily plan in their room.

METHODS

Over 2 years (August 2015 – November 2017), family satisfaction data was collected in a 32 bed surgical and cardiac ICU at a large quaternary hospital. Surveys were distributed at time of discharge or transfer from the ICU to the family member that was present with the patient at that time. We extracted system and clinical variables from patient medical records through a local ICU registry. The University of Utah Institutional Review Board reviewed and approved the study.

Instrument

The Family Satisfaction-ICU (FS-ICU) was utilized, and was then expanded with additional questions regarding unit specific practices. As a survey instrument, the FS-ICU was developed and since validated for use in the ICU setting [19–21]. Past evaluation of the FS-ICU shows high internal consistency and construct validity for assessing caregiver satisfaction in the ICU [22]. Over the past few years, it has been used widely in multicenter collaborations by the SCCM[23]. The survey consisted of 24 Likert-scale questions to assess satisfaction with domains of communication of care and medical decision-making, plus three additional open-ended questions regarding family opinions, and five demographic questions. The only question added by the authors to the widely used FS-ICU scale was one question assessing a communication tool posted in the room (“Plan of the Day”). The use of the “Plan of the Day” in the ICU was independent of this analysis which attempted to assess its impact. The survey is scored both in total and by each of the two domains. Scoring of the survey, as has previously been described, is divided into “High” and “Low” satisfaction, with high indicating an average score on questions of ≥ 4 (on a 1–5 Likert-scale). For cumulative scoring over time or across a unit, the cumulative score is the percent of respondents indicating “High” satisfaction.

Survey Distribution

Paper format surveys were distributed to family members with a consent cover letter explaining the purpose of the study, along with a postage paid return envelope. Upon distribution, survey forms were assigned a unique identifier with a separate securely maintained key. Trained research assistants sought to distribute surveys to families of all patients transferred out of the ICU to the floor, identified by the unit clerks. Returned surveys were then manually entered by the research assistants into REDCap using the unique identifier [24]. Research assistants were blinded to subsequent analysis.

Chart Review of Patient and System Variables

Once the family member completed and returned the FS-ICU, a detailed evaluation of their ICU stay was performed and analyzed to assess factors associated with high or low satisfaction. The University of Utah Surgical ICU Registry is a detailed, manually maintained ICU registry previously described and validated for research; variables are synced with the electronic medical record and audited to ensure precision of data upon data extraction. The registry was used to review two categories of data, “Patient Factors” and “System Factors”, with each of the following data points retrievable from the registry. Patient factors included: demographics, diagnosis category, severity of illness, need for mechanical ventilation (for anytime if non-operative or >24 hours if post-operative via endotracheal tube or tracheostomy, not including non-invasive positive pressure ventilation), ECMO use, and survival at 30 days and 365 days. Severity of illness was measured via the acute physiology and chronic health evaluation (APACHE II)[25]. System factors included: ICU length of stay (LOS), if the patient was an ICU readmission, the prior setting of the patient (operating room, floor, emergency room), and elective versus non-elective admission.

Plan of the Day

The Plan of the Day was a posted sheet detailing a short summarized list of objectives each day that was filled out by the ICU team during rounds, involving a conversation between the medical providers and nurses sharing what the anticipated plan was for the patient that day, and posted in each patient’s room. When surveying families, they were queried whether the plan of the day was posted and updated each day and on its utility (whether it was beneficial to the family’s understanding of their care plan, or not). The result of this question, whether the plan was posted, as well as whether the family found it helpful or not helpful, was an independent variable used to compared across satisfaction levels.

Statistical Analysis

Our primary outcome of family satisfaction was dichotomized for analysis to identify characteristics of families and patients with and without high satisfaction. An a priori power analysis produced a total sample size of 206 to detect a difference in satisfied and less satisfied families with 80% power at 0.05 significance level, based on prior studies analyzing satisfaction data in the ICU population [26, 27]. Continuous variables in both respondent and patient characteristics were summarized using mean and standard deviation (SD). Categorical variables were summarized as counts and percentages. Satisfaction was scored for the total instrument and the two domains (communication with care and decision-making) by averaging and transforming individual items for subjects who had a ≥70% item response rate within each domain. The ranges of the total instrument and the two domains were 0–100, where 100 indicated perfect satisfaction. Scores were dichotomized into high (≥80 on the 0–100 scale) versus low/normal satisfaction (<80 on the 0–100 scale), to identify the cohorts of patients whose families were not highly satisfied as a target of improvement, which facilitates comparisons to prior studies in which overall satisfaction was near 80% [12]. Demographics were summarized and compared across high versus low/normal satisfaction using Wilcoxon rank sum tests for continuous variables and chi-squared or Fisher’s exact tests for categorical variables. In effort to minimize error, each variable was independently compared across satisfaction levels, with only candidate variables with p < 0.05 were selected from univariate logistic regression models. Among the candidate set of predictors, we implemented forward and backward step-wise selection using the Akaike information criterion to obtain a final multivariate logistic regression model [28]. We reported odds ratios (ORs) and their 95% confidence intervals (CIs) and p-values from all models. Statistical significance was defined as p<0.05. Statistical analyses were implemented using R v. 3.6.0 and all tests were two-sided [29].

Free text comments were qualitatively analyzed with theme clustering into categories including “need for improvement” or “positive comments”, which reflect on the wording of the questions to the families. These comments were then quantitatively assessed by rates of positive or negative comments.

RESULTS

Over a 2-year period, 250 respondents completed the FS-ICU indicating satisfaction of care provided to 250 patients in the ICU (Table 1). The mean age of respondent was 52.8 years (standard deviation [SD]: 15.4) and mean patient age 57 years (SD 18.2), with the majority of patients self-identifying their ethnicity as White/Caucasian (N=211, 84%). Respondents were primarily female (N=180; 72%), while patients were more commonly male (N=169; 67.6%). The most common relationship of the respondent was the wife of the patient (N=95; 38.2%), followed by husband (N=25, 10%); other respondents including siblings, parents, children, and other-not-specified. Nearly 2/3 of respondents lived with the patient (N=158, 64.2%). Overall, 75% (N=188) of families reported high satisfaction (Table 2), a composite score for unit-level satisfaction.

Table 1.

Respondent, patient, and system characteristics

Clinical Variables (N=250)	Respondent Summary	Patient Summary
Age [mean (SD)]	52.8 (15.4)	57 (18.2)
Sex
Female	180 (72%)	81 (32.4%)
Male	70 (28%)	169 (67.6%)
Relationship
Brother	5 (2%)	-
Sister	11 (4.4%)
Daughter	42 (17%)	-
Son	16 (6.4%)
Father	9 (3.6%)	-
Mother	18 (7.2%)
Husband	25 (10%)	-
Wife	95 (38.2%)	-
Partner	5 (2%)
Other	23 (9.2%)
Race
White/Caucasian	-	211 (84.4%)
American Indian	-	11 (4.4%)
Asian or Pacific Islander		4 (1.6%)
Black/African American		3 (1.2%)
Hispanic/Latino		10 (4%)
Other		3 (1.2%)
Unknown		8 (3.2%)
Ventilated
Yes	-	149 (59.6%)
No	-	101 (40.4%)
Diagnosis Category
Cardiovascular	-	127 (50.8%)
GI	-	27 (10.8%)
Hematology	-	9 (3.6%)
Orthopedic	-	8 (3.2)
Infectious	-	12 (4.8%)
Oncology	-	3 (1.2%)
Renal		7 (2.8%)
Respiratory		18 (7.2%)
Trauma		21 (8.4%)
Other		18 (7.2%)
Survival
30 day	-	204 (81.6%)
365 day		156 (62.4%)
Admission Type
Elective postoperative		107 (42.8%)
Emergency postoperative	-	92 (36.8%)
Non-operative		51 (20.4%)
ICU Readmission		14 (5.6%)
Plan of the day (POD)
Posted		189 (75.6%)
Posted & Family found POD helpful		109 (57.7%)
Posted & Family found not helpful		80 (42.3%)

Missing values: respondent age (year) = 2, Relationship = 1

Table 2:

Satisfaction per clinical and system variable

Variable	Satisfied (N=188)	Not Satisfied (N=62)	P-value
Respondent age (year):
Mean (SD)	52.9 (15.7)	52.6 (14.5)	0.82w
Median (IQR)	55.0 (39.0, 65.0)	53.0 (42.0, 63.0)
Respondent Sex:
Female	138 (73.4%)	42 (67.7%)	0.39c
Male	50 (26.6%)	20 (32.3%)	-
Relationship:
Brother	5 (2.7%)	0 (0%)	0.50s
Daughter	34 (18.2%)	8 (12.9%)	-
Father	6 (3.2%)	3 (4.8%)	-
Husband	19 (10.2%)	6 (9.7%)	-
Mother	14 (7.5%)	4 (6.5%)	-
Other	17 (9.1%)	6 (9.7%)	-
Partner	5 (2.7%)	0 (0%)	-
Sister	10 (5.3%)	1 (1.6%)	-
Son	10 (5.3%)	6 (9.7%)	-
Wife	67 (35.8%)	28 (45.2%)	-
Does the respondent live with the patient:
No	72 (38.7%)	16 (26.7%)	0.09c
Yes	114 (61.3%)	44 (73.3%)	-
Residence:
Rural	82 (44.1%)	31 (51.7%)	0.34c
Distance (out of state)	80 (43%)	25 (41.7%)	-
Urban	24 (12.9%)	4 (6.7%)	-
Plan of the day posted and updated:
No	52 (27.7%)	28 (45.2%)	0.004c
Unknown	43 (22.9%)	18 (29%)	-
Yes	93 (49.5%)	16 (25.8%)	-
Patient age (year):
Mean (SD)	57.1 (18.2)	56.6 (18.4)	-
Median (IQR)	60.0 (45.0, 70.0)	58.0 (40.5, 72.0)	0.84w
Range	(18.0, 90.0)	(21.0, 96.0)	-
Patient race:
American Indian	6 (3.2%)	5 (8.1%)	0.24f
Asian or Pacific Islander	3 (1.6%)	1 (1.6%)	-
Black/African American	3 (1.6%)	0 (0%)	-
Hispanic/Latino	9 (4.8%)	1 (1.6%)	-
Other	3 (1.6%)	0 (0%)	-
Unknown	4 (2.1%)	4 (6.5%)	-
White/Caucasian	160 (85.1%)	51 (82.3%)	-
Patient gender:
Female	63 (33.5%)	18 (29%)	0.51c
Male	125 (66.5%)	44 (71%)	-
Severity Apache II Score:
Mean (SD)	15.7 (7.2)	14.9 (6.8)	-
Median (IQR)	15.0 (11.2, 19.0)	15.0 (9.2, 19.8)	0.65w
Range	(3.0, 47.0)	(0.0, 33.0)	-
Admission Type:
Elective Postoperative	82 (43.6%)	25 (40.3%)	0.47c
Emergency Postoperative	71 (37.8%)	21 (33.9%)	-
Non-Operative	35 (18.6%)	16 (25.8%)	-
Readmission:
No	178 (94.7%)	58 (93.5%)	0.75f
Yes	10 (5.3%)	4 (6.5%)	-
Diagnosis Category:
Cardiovascular	94 (50%)	33 (53.2%)	0.60f
GI	19 (10.1%)	8 (12.9%)	-
Hematology	6 (3.2%)	3 (4.8%)	-
Infectious	10 (5.3%)	2 (3.2%)	-
Oncology	3 (1.6%)	0 (0%)	-
Orthopedic	5 (2.7%)	3 (4.8%)	-
Other	14 (7.4%)	4 (6.5%)	-
Renal	7 (3.7%)	0 (0%)	-
Respiratory	16 (8.5%)	2 (3.2%)	-
Trauma	14 (7.4%)	7 (11.3%)	-
Ventilated:
No	70 (37.2%)	31 (50%)	0.08c
Yes	118 (62.8%)	31 (50%)

Missing values: respondent age = 1, 1; live with patient = 2, 2; residence = 2, 2; severity score = 18, 12

Plan of the Day

The presence of the POD posted in the room was found to associate with higher satisfaction with high statistical significance (p=0.004). Of satisfied respondents, 49.5% indicated the POD was posted, compared to 25.8% of dissatisfied respondents. In the multivariate regression model (Table 3), the odds of being highly satisfied when the POD was posted were 3.3 times the odds when it was not (OR = 3.3, 95% CI: 1.63, 6.89, p=0.001). Additionally, respondents that lived with the patient at the time of completing the FS-ICU were half as likely to be satisfied as compared to those who did not live with the patient (OR =0.5, 95% CI: 0.25, 0.96, p=0.044), adjusted for other variables.

Table 3:

Univariate and multivariate analysis of high satisfaction

	Univariate		Multivariate
Variables	OR (95% CI)	p-value	OR (95% CI)	p-value
Respondent age (year)	1.00 (0.98,1.02)	0.88	-	-
Respondent sex
Male	0.76 (0.41,1.44)	0.39	-	-
Relationship
Partner (not otherwise specified)	0.78 (0.44,1.39)	0.40	-	-
Do you live with the patient?
Yes	0.58 (0.30,1.08)	0.09	0.5 (0.25,0.96)	0.044
Location
Distance (out of state)	1.21 (0.66,2.24)	0.54	-	-
Urban	2.27 (0.80,8.18)	0.16	-	-
Plan of the day posted and helpful
Not posted	1.29 (0.63,2.66)	0.49	-	-
Posted and helpful	3.13 (1.57,6.43)	0.001	3.3 (1.63,6.89)	0.001
Patient age (year)	1.00 (0.99,1.02)	0.85	-	-
Patient race
White/Caucasian	1.23 (0.55,2.59)	0.59	-	-
Patient gender
Male	0.81 (0.43,1.50)	0.51	-	-
Severity Apache II Score	1.02 (0.97,1.07)	0.46	-	-
Admission Type
Emergency Postoperative	1.03 (0.53,2.01)	0.93	-	-
Admission Type
Non-Operative	0.67 (0.32,1.42)	0.28	-	-
Readmission
Yes	0.81 (0.26,3.06)	0.74	-	-
Diagnosis category
Cardiovascular	1.14 (0.64,2.03)	0.66	-	-
Ventilated
Yes	1.69 (0.94,3.02)	0.08	1.64 (0.89,3.02)	0.11

Severity of Illness

Half of all patients (50.8%) had a primary diagnosis of cardiovascular disease (Table 1). The majority of patients were mechanically ventilated (N=149, 59.6%). Only four patients in the cohort were placed on ECMO (three venoarterial, one venovenous), without correlations with satisfaction. Ventilated status was included in the multivariate model because it had a p-value of <0.1 (62.8% of families of patients that were ventilated were satisfied, versus 50% of families of non-satisfied patients), but it did not achieve statistical significance in the multivariate analysis.

Free Text Comments

Of the 250 respondents, there were 500 opportunities for free-text comments (250 inquiring on areas that could be improved and 250 seeking feedback on what went well). 164 of 250 (65.6%) respondents filled out at least one free-text response. Of all comments, 52.6% suggested items for improvement, while 47.4% were positive. Of the negative comments, the majority related to rounds or the plan for the patient (25% of negative comments), provider communication with the family (23.4%), or family member needs (22%). Example comments for further improvement included “plan of the day was helpful, but would have been more helpful if told about it earlier” and “I couldn’t figure out most of the [POD]; if the nurses talked through it I would have found it very helpful”.

DISCUSSION

We identified a potential intervention in daily ICU care that could drive higher satisfaction among respondents. Our main findings are: 1) One quarter of our patient population is not highly satisfied and offers opportunity for improvement with granular, patient-linked data, 2) a posted and updated “Plan of the Day” is associated with positive satisfaction, and 3) the satisfaction of respondents is associated with the living situation of the respondents to the patients. There was no association between clinical indicators of acuity or risk of mortality and satisfaction.

Unit-level Satisfaction

Over the past two decades, investment has been made to better understand how to measure the experience of ICU care and the impact high satisfaction has on the outcomes and trajectory of patients [20–23, 30]. However, it remains unknown what is an appropriate or standard level of satisfaction for families in the ICU or who is at risk of low satisfaction. Studies have demonstrated satisfaction in the range of 68–83%; yet, many of these have only focused on subset populations, and there is little insight into how to improve satisfaction for the remaining significant percentage of patients [12, 21, 31]. The broad adoption of Hospital Consumer Assessment of Healthcare Providers and Systems (HCAHPS) satisfaction surveys – introduced in 2006 as a means to potentially link Centers for Medicare and Medicaid Services reimbursement bundles with satisfaction – has enabled a potential comparison of hospitals’ service delivery [32–34]. While some form of hospital satisfaction measurement is ubiquitous across health systems, satisfaction in the ICU remains a key modifiable measure of care for the highest acuity, and often highest burdened, patients. Additionally, hospital level data does not necessarily reflect the experience of single units during a multi-unit hospital stay. This is compounded by the fact that many critically ill patients are not fully conscious during their ICU stay. The result is a diminished opportunity to cater improvement efforts towards the drivers of high or low satisfaction specific to the ICU. Local improvement teams, based on the concepts of microsystems the Joint Commission, drives improvement at the bedside level but is dependent on data from the bedside level to drive improvements in local systems, rather than broad hospital or health system-wide data [35]. The ability to measure ICU satisfaction at a granular level through patient proxies (such as the FS-ICU), and link results with clinical metrics offers local improvement teams the opportunity to intervene and improve experience of care for families. Furthermore, this approach permits satisfaction benchmarking, through widely used unit-specific tools, such as the FS-ICU, and multi-institutional clinical registries.

Our study adds to this literature in that it describes a process of determining and trending unit-level satisfaction. Trending satisfaction at the unit level allows for benchmarking in research and directed quality improvement to improve satisfaction. While it is encouraging that our patient population is consistent with these subset populations, a meaningful percentage of families had low satisfaction scores. Aside from using satisfaction data to broadly track over time a unit or hospital’s overall satisfaction, opportunity to advance care can be sought by focusing attention on the unsatisfied population (in our case, 25% of our ICU).

Plan of the Day

When assessing our results, the clear indicator of higher satisfaction was the posted POD. The POD was intentionally utilized to facilitate communication between the care team and the family and/or patient, and its association with satisfaction may be reflective of improved communication. The use of a posted, dynamically updated sign articulating daily goals created by an interdisciplinary team of provides, while not novel in itself, has not been well studied in critical care. The POD provides transparency to the family in the care team’s thought processes and creates a shared mental model for the expected trajectory of the day, and highlights a shared understanding of care among various team members. This shared baseline expectation can mitigate misunderstandings in communication, which themselves can lead to lower satisfaction[36].

Our findings that the use of a posted updated plan is associated with improved family satisfaction fits into the context of previous work demonstrating an association of ICU factors with patient outcome. In a large national study of ICUs, the use of a posted written plan of the day was associated with decreased mortality. While communication in the form of rounding has been extensively studied in critical care, the use of a plan of the day as a communication tool in the ICU has not been previously demonstrated to improve family satisfaction, as this study has asserted [37, 38]. Respondent provided insight into the POD in the form of free-text comments that supported its use, though suggested that family involvement in the use of the POD would further its benefit.

Family Caregivers

Family members living with patients reported lower overall satisfaction with care as opposed to family members who were not living with patients. While characteristics of the acute illness necessitating ICU care did not impact family satisfaction, including when assessing severity of illness, length of stay of the admission, or even survival, the family member residing with the patient did hinder satisfaction. This may indicate that there is a perceived burden the family member appreciates out of their responsibilities to the patient or the baseline expectations for care as compared to caregivers who do not live with the patient. Alternatively, awareness of the struggles patients face during ICU recovery may weigh larger among family members living with patients. The prevalence of post-traumatic stress symptoms in family members of ICU patients is known [39, 40]. While unclear from prior work if there is association between family members living with the patient, literature from pediatrics suggests parents are at greater risk, which may be in part due to the additional care burden of taking care of a post-critically ill patient at home [41]. Our findings indicate that independent of the unique patient circumstances, family members that bear the brunt of care for the patients, those that live with them may be feeling these stressors during the admission. Opportunities for identifying these families on a real-time basis could allow intervention early to limit the residual stressors that families experience from the burden of ICU care.

Free-Text Comments

While not statistically contributing to our findings, free-text comments from families reinforce our results: “patient was taken care of very well; as for me, I was put on the back burner. Someone should be assigned to family members’ needs and concerns.” Additionally, there were many free-text comments from families with encouraging messages. In an era of increasing provider burnout, dissemination of such appreciation to bedside providers offers an important feedback mechanism. The importance of interdisciplinary efforts to drive family satisfaction is clear through these comments, often recognizing nurses, respiratory therapists, and ancillary staff, and can foster a culture of inter-professional improvement. As free-text responses continue to be generated, there is opportunity in the future to implement natural language process techniques to analyze comments and better understand family needs and wants without the respondents being limited to feedback on multiple-choice Likert scale surveys.

Limitations

There are limitations to this study. As a single-center study, our findings may not be generalizable directly to other ICUs. Socioeconomic status was not available, and thus we could not assess the impact of financial burden on satisfaction. Additionally, we had a fairly homogenous Caucasian population, which could impact the relative percentage of patients who live directly with their family members who accompany them to the hospital. The lack of diversity was mostly reflected in ethnicity, with broad range of diverse age, diagnosis categories, and indication for ICU admission, and is reflective of the demographic makeup of the population served in the region. Finally, the plan of the day, though standardized during its use in our unit, is not a validated tool; studies using validated measures or checklists during rounds have alluded to success in communication, but future work is needed to indicate if such tools are widely beneficial [42]. Additionally, the assessment of its impact was based on family recall when reporting satisfaction.

Conclusion

This study offers an insight into the clinical and system drivers of satisfaction in a large academic ICU. In this population, presence of a Plan of the Day rounding tool transparently displayed in the room was associated with higher family satisfaction, while there was lower satisfaction among family members who live with patients at home. These findings indicate that communication with families may be a greater driver of satisfaction than characteristics of the disease or hospital admission itself. This provides an opportunity for improvement in ICU care. Importantly, as the shape of healthcare continues to evolve and health institutions adapt to best match expectations of families, recognizing clinical and system drivers of inpatient satisfaction in individual units will be crucial to advance the service we provide to families.

Figure 1:

Percent satisfaction per system or clinical variables

Abbreviations

FS-ICU

Family satisfaction in the intensive care unit

HCAHPS

Hospital Consumer Assessment of Healthcare Providers and Systems

POD

Plan of the day