Predictors of Patient-Reported Financial Hardship in Patients Hospitalized With Critical Illness and Its Association With Patient Outcomes

Abstract

BACKGROUND:

A high burden of financial hardship has been demonstrated in critically ill patients. Understanding the sociodemographic and clinical risk factors for financial hardship and its association with patient outcomes can help to guide future interventions to mitigate financial hardship in this patient population.

RESEARCH QUESTION:

What are the sociodemographic and clinical risk factors for financial hardship in critically ill patients and its association with patient-reported outcomes?

STUDY DESIGN AND METHODS:

This prospective cohort study enrolled adults with chronic, life-limiting illness, acute severe illness, or both hospitalized in the ICU. Twenty-six sociodemographic and clinical variables were measured using electronic health record or patient questionnaire data collected at either 1 or 3 months after ICU admission. Risk factors for patient-reported financial hardship were evaluated using linear regression. The association between financial hardship and 4 patient-reported outcomes was tested with probit and linear regression.

RESULTS:

We enrolled 171 patients with a median age of 58 years (interquartile range, 45, 70 years), including 64 female (37%) patients and 44 patients (26%) from racial or ethnic minority groups. Illness-related change in work status (β = 3.5; P = .02) and poorer self-reported health status (β = 2.4; P = .003) were associated with greater patient-reported financial hardship. Higher education level (β = −2.6; P = .001) was associated with less financial hardship. Higher financial hardship was associated with all 4 patient-reported outcomes (P < .001): feelings of depression (β = 0.2) and anxiety (β = 0.3), poorer quality of life (β = 0.04), and lower emotional preparedness for the future (β = 0.05).

INTERPRETATION:

This study demonstrated an association between patient-reported financial hardship and negative patient-reported outcomes in critically ill patients. Our findings also highlight important risk factors for financial hardship and suggest that sociodemographic factors, including education level, illness-related change in work status, and poorer self-reported health status, could contribute more than clinical factors. Our findings provide a foundation for future development of screening tools and interventions to mitigate financial hardship in critically ill patients.

Financial hardship is a multidimensional construct that encompasses the material burden, psychological responses, and coping behaviors that emerge as a consequence of illness and related medical care.¹ The occurrence and implications of financial hardship are important considerations for patients who experience critical illness. Secondary data analysis from a randomized controlled trial of patients with acute respiratory failure found that 42.5% of these patients reported financial hardship at 6 months from randomization²; patients’ ability to recover and return to work after acute respiratory failure also is impacted by financial hardship.^3,4 Although a high burden of financial hardship has been reported in family members and surrogate decision-makers of critically ill patients,^2,5 less is known about the sociodemographic and clinical risk factors that might contribute to financial hardship among critically ill patients themselves. Our first study objective was further delineation of these risk factors to promote early identification of at-risk patients, which can help to inform targeted, multilevel interventions to mitigate financial hardship in this population.

Additionally, associations between financial hardship and patient-reported outcomes such as depression, anxiety, and poor quality of life have been reported in patients with chronic illness,⁶ particularly in those with cancer.^7-9 Poor patient-reported outcomes are similarly well-documented for survivors of critical illness,^10-13 yet few data are available evaluating the association between financial hardship and such outcomes in this patient population. Our second study objective was to examine the relationship between financial hardship and other key patient-reported outcomes. Understanding the consequences of financial hardship among the critically ill can help to raise awareness of the importance of developing interventions to address this understudied stressor.

Study Design and Methods

Study Design and Data Source

This was an exploratory, prospective, longitudinal cohort study of critically ill patients hospitalized in the ICU conducted at 3 hospitals within a large academic health care system in the Pacific Northwest. Sociodemographic and clinical patient characteristics were obtained through electronic health record (EHR) data abstraction, along with patient questionnaires conducted at 1 or 3 months after ICU admission (e-Fig 2). Data collection for these questionnaires occurred from February 25, 2020, through March 8, 2023. Questionnaires were administered in person, via phone, by mail, or online. All study participants provided informed consent for study activities. This study was conducted under the approval of The University of Washington institutional review board (Identifier: STUDY00008703).

Study Population

Trained study staff identified study participants through EHR review of patients admitted to the ICU. Patients were eligible for study participation if they met the following criteria: (1) age 18 years or older, (2) English speaking, and (3) a diagnosis of chronic life-limiting illness suggesting median survival (without critical illness) of 2 years or an acute, severe illness with a hospital mortality risk of > 15%, or both. Chronic, life-limiting illness was defined by International Classification of Diseases, Tenth Revision, codes correlating with 9 chronic conditions outlined in the Dartmouth Atlas Project,^14,15 including poor prognosis cancer, chronic pulmonary disease, coronary artery disease, heart failure, peripheral vascular disease, end-stage liver disease, diabetes with end-organ damage, renal failure, and dementia. These conditions have been used in other recent studies of patients with serious illness^16-18 and reflect a patient population for whom health care spending is high.¹⁴ Acute, severe illness was defined by an Acute Physiology and Chronic Health Evaluation (APACHE) II score of ≥ 13, a diagnosis predicting mortality risk of > 15% during the index hospitalization, or both (e-Fig 1).^19,20 Patients who lacked decisional capacity, were incarcerated, had no address listed in the medical record, were admitted for suspected suicide attempt, or were pending withdrawal of life support were excluded from this study (e-Fig 2).

Exposure and Descriptive Variables

We tested 26 variables as potential risk factors for financial hardship, conceptually grouped as either sociodemographic or clinical factors (e-Fig 3). Six variables were based on patients’ questionnaire responses (e-Fig 4), coded for analyses as follows: level of education (0 = high school diploma or less, 1 = some education after high school, but less than a 4-year college degree, 2 = 4-year college degree or some graduate school, 3 = graduate degree), race and ethnicity (0 = White non-Hispanic; 1 = mixed or racial or ethnic minority group affiliation), any dependent children living at home (0 = no, 1 = yes), any change in work status related to illness or treatment (0 = no, 1 = yes), self-reported current health status (0 = excellent health, 1 = very good health, 2 = good health, 3 = fair health, 4 = poor health), and absence of an emotional support system (0 = social support available, 1 = no social support available). Twenty variables were abstracted from the EHR: sex (0 = male, 1 = female); age at the time of index ICU admission; health insurance coverage (3 dummy indicators: Medicare, Medicaid, or military or other type of insurance, with private or commercial insurance as the reference group); number of hospitalizations during the year before admission; documentation of each of 9 chronic, life-limiting illnesses during the 2 years before admission (0 = no diagnosis, 1 = diagnosis made); total number of chronic, life-limiting illnesses diagnosed in the 2 years before admission; APACHE II score during the first 24 hours after admission; number of index hospitalization days, index ICU days, and mechanical ventilation days before questionnaire completion; and whether palliative care consultation (0 = no, 1 = yes) or a goals-of-care discussion (0 = no, 1 = yes; identified with natural language processing methods)²¹ occurred during the index hospitalization and before questionnaire completion.

Several additional variables, drawn from patients’ questionnaire responses, were used for descriptive purposes only: specific racial and ethnic affiliations, current work status, and perception of frailty based on the fatigue level, difficulty walking up stairs, unintentional weight loss in the previous year, extent of frailty, and frailty’s impact on quality of life scale.²²

Outcome Variables

The primary outcome for this study was a measure of financial hardship, based on the standard financial toxicity score from the modified Functional Assessment of Chronic Illness Therapy Comprehensive Score for Financial Toxicity (FACIT-COST) instrument, a validated and widely used measurement tool for assessment of financial hardship in patients (permissions obtained from FACIT.org).^23,24 This instrument (e-Fig 5) was administered to patients as part of the questionnaire. If the patient responded to at least 1 of the items, we summed the valid responses, multiplied the sum by 11, and divided the result by the number of valid responses, resulting in a toxicity score ranging from 0 (maximum toxicity level) to 44 (no toxicity). The financial hardship measure used in our study subtracted the toxicity score from 44, thereby retaining the 0 to 44 range, but with higher scores indicating greater financial hardship. Two patients who lacked responses to all 11 items received no score and were excluded from the analyses.

We studied 4 other patient-reported outcomes to examine their association with financial hardship. Feelings of depression and anxiety were assessed using standard scores from the Hospital Anxiety and Depression Scale,^25,26 a validated measurement tool that has been used in studies of critically ill patients.^27,28 We used 2 composite measures from the Hospital Anxiety and Depression Scale: 1 based on 7 items measuring anxiety, and the other based on 7 items measuring depression. For each of the 2 measures, if the patient responded to at least 4 of the relevant items, we summed the valid responses, multiplied the sum by 7, and divided the result by the number of valid items. Patients who responded to fewer than 4 of the relevant items were excluded from the analysis. Lower emotional preparedness for the future was assessed with a Likert-scaled question from the Quality of Life at the End of Life Measure–Family Experience (QUAL-E Fam) questionnaire²⁹—“How prepared are you emotionally for the future, no matter what happens?”—using codes ranging from 0 (a great deal) to 4 (not at all). Finally, to measure poorer quality of life, we used a Likert-scaled item from the QUAL-E questionnaire—“How would you rate your overall quality of life?”—using codes ranging from 0 (excellent) to 4 (very poor).³⁰

Statistical Analyses

Associations between the 26 risk factors and patient-reported financial hardship were based on linear regression, estimated with robust maximum likelihood. Given the exploratory nature of our study, we first examined bivariate associations between each risk factor and financial hardship and then ran a multiple regression model containing all variables with P < .20 in the bivariate models. We chose this more conservative approach compared with the P value cutoff of < .25 proposed by Hosmer et al³¹ so as not to violate the general guidance of including no > 1 predictor for every 10 to 15 observations in the sample.

Tests of the association of financial hardship with each of the 4 preselected patient-reported outcome variables were based on regression models: linear regression, estimated with robust maximum likelihood, for the Hospital Anxiety and Depression Scale scores and probit regression, estimated with weighted least squares, for the Likert-scaled items. Each model included adjustment for the patient’s age, sex, racial or ethnic minority group, and APACHE II score during the 24 hours after ICU admission.

We used a P value threshold value of .05 in the multivariable models. Each regression model included only those patients with complete data on the independent, dependent, and adjustment variables. No evidence of multicollinearity was found, evidenced by variance inflation factor values ranging from 1.16 through 5.38. We used SPSS Statistics software (IBM) for descriptive information and Mplus software (Muthén & Muthén) for regression models.

Results

Sample Characteristics

We recruited 173 patients for this study, of whom 171 patients (99%) provided adequate information to compute a financial hardship score (Table 1). The median age of participants was 58 years (interquartile range [IQR], 45, 70 years); 64 patients (37%) were female and 44 patients (26%) reported a racial or ethnic minority group identity. Forty-eight patients (29%) had no education after high-school, 30 patients (18%) had dependent children living at home, 24 patients (14%) were employed, and 66 patients (40%) reported that they had experienced an illness-related change in work status at the time of survey completion. The median value for illness-related financial strain was 2 (ie, some of the time; IQR, 1, 3), median difficulty in meeting household expenses was 2 (ie, able to pay their bills, but only because of cutting back on things; IQR = 1, 3), and median self-reported current health was 3 (ie, fair; IQR = 2, 4). Of those for whom an APACHE II score was computed, the median score was 23 (IQR = 19, 28). The median number of diagnosed chronic, life-limiting illnesses was 1 (IQR = 0, 2).

TABLE 1 ].

Characteristics of Patient Participants

Characteristic	Valid Hardship Scorea		Total Sampleb
	Valid No.c	Statisticd	Valid No.c	Statisticd
Participantse	171	NA	173	NA
Age, y	171	58 (25)	173	59 (26)
Female sex	171	64 (37.4)	173	64 (37.0)
Race and ethnicityf	170		172
White non-Hispanic		126 (74.1)		127 (73.8)
Racial or ethnic minority group		44 (25.9)		45 (26.2)
Asian		7 (4.1)		7 (4.1)
Black		13 (7.6)		13 (7.6)
Hispanic (White)		3 (1.8)		3 (1.7)
Indigenous		5 (2.9)		6 (3.5)
Pacific Islander		2 (1.2)		2 (1.2)
Mixedg		14 (8.2)		14 (8.1)
Education level	167		168
Eighth grade or less		1 (0.6)		2 (1.2)
Some high school		8 (4.8)		8 (4.8)
High school diploma or equivalent		39 (23.4)		39 (23.2)
Trade school or some college		62 (37.1)		62 (36.9)
Four-year college degree		33 (19.8)		33 (19.6)
Some graduate school		5 (3.0)		5 (3.0)
Graduate or professional degree		19 (11.4)		19 (11.3)
Had dependent children at home	167	30 (18.0)	169	31 (18.3)
Current work statush	169		171
Employed full-time		17 (10.1)		18 (10.5)
Employed part-time		7 (4.1)		7 (4.1)
Unemployed		26 (15.4)		26 (15.2)
Disabled		46 (27.2)		46 (26.9)
Retired		45 (26.6)		46 (26.9)
Other		28 (16.6)		28 (16.4)
Any illness-related work status change	165	66 (40.0)	166	67 (40.4)
Extent current health was poori	169	3 (2, 4)	171	3 (2, 4)
Lack emotional support	163	8 (4.9)	165	8 (4.8)
Extent of difficulty paying billsj	164	2 (1,3)	164	2 (1, 3)
Extent of illness-related financial straink	169	2 (1, 4)	170	2 (1, 4)
Self-assessed frailty
Frequency of fatigue (past 4 wk)l	168	3 (2, 3)	169	3 (2, 3)
Any difficulty walking up stairs	157	75 (47.8)	159	77 (48.4)
Any unintentional weight loss in past year	160	103 (64.4)	162	105 (64.8)
Extent of frailtym	167	2 (1, 3)	169	2 (1, 3)
Frailty impact on quality of lifen	167	3 (1, 3)	168	3 (1, 3)
Insurance coverage	171		173
Private or commercial		68 (39.8)		69 (39.9)
Medicare		62 (36.3)		63 (36.4)
Medicaid		35 (20.5)		35 (20.2)
Military		1 (0.6)		1 (0.6)
Other		5 (2.9)		5 (2.9)
No. of inpatient stays in the year before index ICU admission	171	0 (0)	173	0 (0)
No. of EHR-documented chronic life-limiting conditions	171	1 (0, 2)	163	1 (0, 2)
Diagnoses documented during 24 mo before ICU admission
Cancer with poor prognosis	171	30 (17.5)	173	30 (17.3)
Chronic pulmonary disease	171	33 (19.3)	173	33 (19.1)
Coronary artery disease	171	26 (15.2)	173	27 (15.6)
Heart failure	171	35 (20.5)	173	35 (20.2)
Peripheral vascular disease	171	12 (7.0)	173	13 (7.5)
End-stage liver disease	171	14 (8.2)	173	14 (8.1)
Diabetes with end-organ damage	171	15 (8.8)	173	15 (8.7)
Renal failure	171	28 (16.4)	173	29 (16.8)
Dementia	171	4 (2.3)	173	4 (2.3)
APACHE II score	128	23 (19, 28)	128	23 (19, 28)
Days in hospital (index hospitalization)	171	18 (22)	173	18 (22)
Days in intensive care (index hospitalization)	171	6 (6)	173	6 (6)
Days of MV (index hospitalization)	171	2 (4)	173	2 (4)
Any palliative care consultation, index hospitalization	171	16 (9.4)	173	17 (9.8)
Any goals-of-care discussion, index hospitalization	171	70 (40.9)	173	71 (41.0)
Disposition setting	171		173
Private residenceo		103 (60.3)		105 (60.7)
Health care facilityp		65 (38)		65 (37.6)
Otherq		3 (1.7)		3 (1.7)

Data are presented as No., No. (%), or median (interquartile range). APACHE = Acute Physiology and Chronic Health Evaluation; EHR = electronic health record; NA = not applicable.

^aPatients for whom a valid financial hardship score could be computed when the patient completed the questionnaire.

^bAll patients who completed questionnaires.

^cNo. in the group named in the column for whom we had valid data for the characteristic named in the row.

^dUnless other specified in the row label, number of respondents (percent with valid data) having the characteristic.

^eThe total number of patients included in the sample named in the column.

^fObtained from the EHR if omitted from the questionnaire and available in the medical record.

^gIncludes combinations of racial and ethnic minority groups, as well as combinations of White non-Hispanic and ≥ 1 minority group categories.

^hReflects work status at the time of questionnaire completion.

ⁱMedian is based on responses to the question, “In general, would you say your health is … ,” with the following response codes: 0 (excellent), 1 (very good), 2 (good), 3 (fair), and 4 (poor).

^jMedian is based on responses to the question, “How would you describe your household’s financial situation right now?” with the following response codes: 0 (after paying the bills, I still have enough money for special things that I want), 1 (I have enough money to pay the bills, but little spare money to buy extra or special things), 2 (I have enough to pay the bills, but only because I have cut back on things), and 3 (I have difficulty paying the bills, no matter what I do).

^kMedian is based on responses to the Quality of Life at the End of Life Measure–Family Experience (QUAL-E Fam) question, “How much of the time do you feel financial strain related to your illness?” with the following response codes: 0 (none of the time), 1 (only a little of the time), 2 (some of the time), 3 (a good bit of the time), and 4 (a great deal of the time).

^lMedian is based on the question, “How much of the time during the past 4 weeks did you feel tired?” with the following response codes: 0 (none of the time), 1 (a little of the time), 2 (some of the time), 3 (most of the time), and 4 (all of the time).

^mMedian is based on the question, “Some people feel like they have lost strength and feel frail. How frail, if at all, do you feel?” with the following response codes: 0 (not at all frail), 1 (a little frail), 2 (somewhat frail), 3 (very frail), and 4 (extremely frail).

ⁿMedian is based on the question, “How much, if at all, has frailty interfered with your quality of life?” with the following response codes: 0 (not at all), 1 (only a little), 2 (somewhat), 3 (a good bit), and 4 (a great deal).

^oIncludes home or self-care, home with home health service, and home with hospice.

^pIncludes rehabilitation unit or hospital, skilled nursing facility, long-term care hospital, inpatient hospice facility, general hospital, psychiatric unit or hospital, and custodial or supportive care facility.

^qIncludes patients who left against medical advice to an unknown disposition setting.

Patients reported median financial hardship of 26 (IQR = 18, 33), median anxiety of 9 (IQR = 6, 12), and median depression of 9 (IQR = 5, 12). Median emotional preparedness for the future was 2 (ie, somewhat; IQR = 1, 3), and median quality of life was 2 (ie, fair; IQR = 1, 2) (Table 2).

TABLE 2 ].

Distribution of Financial Hardship Scores and Patient-Reported Outcomes

Outcome	Valid No.	Median (IQR)	Observed Range
Financial hardshipa	171	26 (18, 33)	0-44
HADS depression scoreb	171	9 (5, 12)	0-18
HADS anxiety scorec	171	9 (6, 12)	0-21
Emotional preparedness for the futured	168	2 (1, 3)	0-4
Current quality of lifee	169	2 (1, 2)	0-4

HADS = Hospital Anxiety and Depression Scale; IQR = interquartile range.

^aComputed as 44 minus the sum of 11 modified Functional Assessment of Chronic Illness Therapy Comprehensive Score for Financial Toxicity (FACIT-COST) items, weighted for the number of items with valid responses. Scores could range from 0 (lowest financial hardship) to 44 (highest financial hardship).

^bTraditional HADS depression scale score, weighted for the number of valid responses; missing if < 4 valid responses. Scores could range from 0 (no or very low symptoms of depression) to 21 (maximum symptoms).

^cTraditional HADS anxiety scale score, weighted for the number of valid responses; missing if < 4 valid responses. Scores could range from 0 (no or very low symptoms of anxiety) to 21 (maximum symptoms).

^dResponse to question, “How prepared are you emotionally for the future, no matter what happens?” (0 = great deal, 1 = good bit, 2 = somewhat, 3 = only a little, and 4 = not at all).

^eResponse to question, “How would you rate your overall quality of life?” (0 = excellent, 1 = good, 2 = fair, 3 = poor, and 4 = very poor).

Associations of Financial Hardship With Risk Factors of Interest

Of the 26 variables tested as potential risk factors for financial hardship, 13 variables showed a P value of < .20 in bivariate models and were included in the multiple regression model (Table 3). These 13 variables jointly explained almost 35% of the estimated variance in financial hardship. Three of the variables had notable independent associations with financial hardship. Illness-related change in work status (β = 3.54; P = .024) and poorer self-rated current health (β = 2.42; P = .003) were associated with greater financial hardship. Higher level of education (β = −2.56; P = .001) was associated with less financial hardship.

TABLE 3 ].

Association of Potential Risk Factors With Patient-Reported Financial Hardship

Exposure Variables	Single-Variable Models					Multivariable Modela
	Valid No.	β Coefficient	P Value	95% CI	R 2	β Coefficient	P Value	95% CI
Age	171	−0.253	< .001	−0.337 to −0.169	0.148	−0.114	.100	−0.250 to 0.022
Female sex	171	−1.975	.235	−5.236 to 1.287	0.008	NA	NA	NA
Racial or ethnic minority group	170	4.110	.017	0.740, 7.479	0.027	0.270	.889	−3.510 to 4.051
Educationb	167	−3.534	< .001	−5.088 to −1.980	0.097	−2.563	.001	−4.052 to −1.075
Had dependent children at home	167	1.434	.527	−3.005 to 5.873	0.002	NA	NA	NA
Any illness-related work status change	165	7.261	< .001	4.040, 10.482	0.103	3.543	.024	0.463, 6.622
Extent current health was poorc	169	1.800	.035	0.129, 3.472	0.029	2.419	.003	0.845, 3.993
Lacked emotional support	163	2.121	.669	−7.613 to 11.856	0.002	NA	NA	NA
Insurance coveraged	171	NA	< .001	NA	0.129	NA	.511	NA
Private or commercial		0.000	NA	NA	NA	0.000	NA	NA
Medicare		−5.169	.006	−8.877 to −1.461	NA	−1.190	.597	−5.596 to 3.216
Medicaid		5.151	.006	1.469, 8.834	NA	2.163	.265	−1.641 to 5.967
Military or other insurance type		4.453	.271	−3.474 to 12.380	NA	2.119	.633	−6.592 to 10.830
No. of inpatient stays in the year before ICU admission	171	−0.664	.233	−1.753 to 0.426	0.006	NA	NA	NA
No. of EHR-documented chronic illnesses	171	−1.830	.001	−2.930 to −0.730	0.053	−0.782	.538	−3.270 to 1.707
Specific diagnoses	171
Cancer		−5.036	.009	−8.794 to −1.279	0.031	−3.212	.260	−8.803 to 2.380
COPD		−1.543	.413	−5.238 to 2.152	0.003	NA	NA	NA
Coronary artery disease		−6.721	.001	−10.745 to −2.696	0.049	−2.990	.231	−7.887 to 1.906
Heart failure		−1.227	.546	5.216 to −2.761	0.002	NA	NA	NA
Peripheral vascular disease		−4.763	.095	−10.353 to 0.826	0.012	−1.637	.625	−8.207 to 4.934
End-stage liver disease		−3.661	.234	−9.693 to 2.371	0.008	NA	NA	NA
Diabetes with end organ damage		−2.951	.181	−7.278 to 1.376	0.006	−0.981	.726	−6.467 to 4.506
Renal failure		−3.126	.161	−7.499 to 1.246	0.011	0.707	.802	−4.826 to 6.240
Dementia		−6.101	.323	−18.191 to 5.989	0.007	NA	NA	NA
APACHE II score	128	−0.009	.948	−0.284 to 0.266	0.000	NA	NA	NA
Days in hospital before questionnaire	171	0.058	.215	−0.034 to 0.150	0.007	NA	NA	NA
Days in ICU before questionnaire	171	0.047	.630	−0.145 to 0.240	0.001	NA	NA	NA
Days of MV before questionnairee	170	0.174	.206	−0.095 to 0.443	0.008	NA	NA	NA
Palliative care consultation before questionnairee	169	−0.122	.964	−5.442 to 5.198	0.000	NA	NA	NA
Goals-of-care discussion before questionnairee	165	−2.345	.172	−5.711 to 1.021	0.011	−1.176	.461	−4.300 to 1.949

Results are based on linear regression models, estimated with robust maximum likelihood. The multivariable model includes all variables from the single variable models where P < .20. APACHE = Acute Physiology and Chronic Health Evaluation; EHR = electronic health record; MV = mechanical ventilation; NA = not applicable.

^aFor 156 participants included in the multivariable model, the 13 predictors jointly explained 34.7% (R² = 0.347) of the estimated variance in financial hardship.

^bEducation modeled as an ordinal variable: 0 ≤ high school diploma, 1 = some education after high school, 2 = 4-year college degree or some graduate school, and 3 = graduate degree.

^cCurrent health was modeled as an ordinal variable: 0 = excellent, 1 = very good, 2 = good, 3 = fair, and 4 = poor.

^dThe overall P value for insurance coverage was based on the Wald test of parameter constraints.

^ePatients excluded if they received some of this service during the index stay, but it whether it had occurred before questionnaire completion was indeterminate.

Associations of Financial Hardship With Negative Patient-Reported Outcomes

Financial hardship was associated with each of the 4 preselected patient-reported outcomes (P < .001 for all), with greater financial hardship suggesting greater depression (β = 0.21) and anxiety (β = 0.28), less emotional preparedness for the future (β = 0.05), and poorer quality of life (β = 0.04) (Table 4).

TABLE 4 ].

Association of Financial Hardship With Negative Patient-Reported Outcomes^a

Emotional State	Valid No.	B Coefficient	P Value	95% CI
Depression scoreb	127	0.213	< .001	0.149-0.278
Anxiety scorec	127	0.282	< .001	0.214-0.351
Lower emotional preparedness for the futured	124	0.051	< .001	0.033-0.070
Poorer quality of lifee	125	0.035	< .001	0.016-0.054

HADS = Hospital Anxiety and Depression Scale.

^aThe models reported in each row of the table were based on patients’ financial hardship score, which was treated as a linear variable, and their concurrent responses regarding each emotional state. The HADS depression and anxiety scores were defined as continuous variables and were analyzed with linear regression, estimated with robust maximum likelihood. The other outcomes were defined as ordered categorical variables and analyzed with probit regression, estimated with mean-adjusted and variance-adjusted weighted least squares. All models included adjustment for the patient’s age, sex, and racial and ethnic minority status, and the Acute Physiology and Chronic Health Evaluation II score during the 24 h after admission for the index ICU stay.

^bThe depression score (with potential range of 0-21) was computed as the sum of the responses to the 7 HADS depression items, weighted by 7 divided by the number of valid responses. Respondents with valid answers to < 4 of the items were excluded from the model.

^cThe anxiety score (with potential range of 0-21) was computed as the sum of the responses to the 7 HADS anxiety items, weighted by 7 divided by the number of valid responses. Respondents with valid answers to < 4 of the items were excluded from the model.

^dCodes for the patient’s response to the question, “How prepared are you emotionally for the future, no matter what happens?”: 0 = great deal, 1 = good bit, 2 = somewhat, 3 = only a little, and 4 = not at all.

^eCodes for the patient’s response to the question, “How would you rate your overall quality of life?”: 0 = excellent, 1 = good, 2 = fair, 3 = poor, and 4 = very poor.

Discussion

This exploratory study demonstrated several findings that help us to characterize risk factors for, and consequences of, financial hardship in patients hospitalized with critical illness. We found that illness-related change in work status and poorer self-rated current health were associated with more financial hardship among these patients. These findings build on previously demonstrated associations between health-related distress and work limitations resulting from chronic illness,³² along with associations between indices of financial stress and poor self-rated health.³³ Early identification of patients with risk factors for financial hardship after critical illness could facilitate targeted and timely interventions, such as financial navigation and education, which have been studied in patients with cancer.^34-36 Our study also demonstrated a strong association between higher education level and lower patient-reported financial hardship. Similar associations have been shown in patients with cancer^37-39 and diabetes⁴⁰; our study is the first to our knowledge to demonstrate a similar association in critically ill patients. Other factors potentially related to socioeconomic status, such as insurance coverage, were not associated with patient-reported financial hardship. One hypothesis for the lack of association between some of these characteristics and our primary outcome is that patients across diverse socioeconomic backgrounds are vulnerable to financial hardship. The experience of this may differ between patients and is an area for future research. Altogether, our findings suggest an important association between certain sociodemographic risk factors and patient-reported financial hardship.

Of note, and in contrast to findings reported from several patient populations with chronic, life-limiting illnesses⁴¹ such as cancer,^42-45 heart failure,⁴⁶ and chronic kidney disease,^47,48 we did not find chronic, life-limiting illness and specific chronic illness diagnoses to be associated significantly with patient-reported financial hardship. Our findings suggest that the lasting consequences of acute illness, such as change in work status and extent of poor health, may have a greater short-term impact on patient perception of financial hardship than one’s preexisting chronic disease burden. Additionally, clinical factors related to index hospitalization, including severity of acute illness, number of days of mechanical ventilation, duration of ICU stay, hospital length of stay, and prior hospitalization within the past year were not associated strongly with self-reported financial hardship. These findings suggest that sociodemographic factors may play a larger role than clinical factors in patient-reported financial hardship in this population.

Financial hardship was associated strongly with all 4 patient-reported outcomes, including greater depression and anxiety, lower emotional preparedness for the future, and poorer self-assessed quality of life. Survivors of critical illness are known to experience these poor outcomes,^10-13 and financial hardship previously was shown to be a mediator of greater symptoms of anxiety and depression in critically ill patients.² Our findings reinforce the growing awareness of the implications of financial hardship and highlight the importance of ongoing research to understand the patient experience better.⁴⁹ Such research has the potential to inform interventions focused on improving patient-reported outcomes after critical illness.

This study has several important limitations. First, we cannot assume causality in our findings of association based on this prospective cohort study design. Second, we used multivariable regression analysis to adjust for potential confounders of the relationship between financial hardship and patient-reported outcomes, with covariates selected based on prior studies in this realm. Our study design and small sample size precluded assessment of all potential confounders of interest; thus, it is possible that additional unmeasured confounders remain. Third, use of questionnaire data in this study raises the possibility of both selection and response bias, which could contribute to measurement error. Fourth, the direction of association cannot be concluded indisputably because financial hardship and the other patient-reported variables were measured at the same time point; therefore, we assumed the likely direction of association in conjunction with prior studies and a belief that we were adjusting for the measured variables most likely to have influenced both the exposure and outcome of interest. Related to this is the fact that we had no measure of financial hardship before hospital admission, and therefore cannot assess how much of the financial hardship detected via patients’ questionnaires preceded hospitalization. Fifth, EHR data are susceptible to measurement error and misclassification of outcomes; we mitigated this risk by using composite variables, such as APACHE II score, and choosing exposure variables that would be identifiable easily in the EHR. Finally, this study was conducted within a single academic health care system and geographic region, and our findings may not generalize to other health care systems or patient populations.

Interpretation

This exploratory prospective cohort study demonstrated an association between financial hardship and negative patient-reported outcomes. It also highlighted important risk factors for financial hardship in critically patients and suggested that sociodemographic factors may play a larger role than clinical factors. Our findings provide a foundation for future development of financial hardship screening tools and interventions to mitigate the impact of financial hardship on critically ill patients.

Take-Home Points.

Study question: What are the risk factors for financial hardship in critically ill patients and its association with patient-reported outcomes?

Results: Illness-related change in work status and poorer self-reported health status were associated with greater patient-reported financial hardship, while higher education level was associated with less financial hardship. Financial hardship was associated with feelings of depression and anxiety, poorer quality of life, and lower emotional preparedness for the future.

Interpretation: This study demonstrated that sociodemographic risk factors may contribute more than clinical risk factors to patient-reported financial hardship in critically ill patients. Our findings show that patient-reported financial hardship is also associated with negative patient-reported outcomes in this population.

ABBREVIATIONS:

APACHE

Acute Physiology and Chronic Health Evaluation

EHR

electronic health record

IQR

interquartile range