- Email: [email protected]
Discharge Against Medical Advice in the United States, 2002-2011
ORIGINAL ARTICLE
Discharge Against Medical Advice in the United States, 2002-2011 Kiara K. Spooner, DrPH, MPH; Jason L. Salemi, PhD, MPH; Hamisu M. Salihu, MD, PhD; and Roger J. Zoorob, MD, MPH Abstract Objective: To describe the national frequency, prevalence, and trends of discharge against medical advice (DAMA) among inpatient hospitalizations in the United States and identify differences across patient- and hospital-level characteristics, overall and in clinically distinct diagnostic subgroups. Patients and Methods: We conducted a retrospective, cross-sectional analysis of inpatient hospitalizations (18 years), discharged between January 1, 2002, and December 31, 2011, using the Nationwide Inpatient Sample. Descriptive statistics, multivariable logistic, and joinpoint regression were used for statistical analyses. Results: Between January 1, 2002, and December 31, 2011, more than 338,000 inpatient hospitalizations were discharged against medical advice each year, with a 1.9% average annual increase in prevalence over the decade (95% CI, 0.8%-3.0%). Temporal trends in DAMA varied by principal diagnosis. Among patients hospitalized for mental health- or substance abuse-related disorders, there was a 2.3% (95% CI, 3.8% to 0.8%) average annual decrease in the rate of DAMA. A statistically significant temporal rate change was not observed among hospitalizations for pregnancy-related disorders. Multivariable regression revealed several patient and hospital characteristics as predictors of DAMA, including lack of health insurance (odds ratio [OR], 3.78; 95% CI, 3.62-3.94), male sex (OR, 2.40; 95% CI, 2.36-2.45), and northeast region (OR, 1.91; 95% CI, 1.72-2.11). Other predictors included age, race/ethnicity, income, primary diagnosis, severity of illness, and hospital location/type and size. Conclusion: Rates for DAMA have increased in the United States, and key differences exist across patient and hospital characteristics. Early identification of vulnerable patients and preventive measures such as improved patient-provider communication may reduce DAMA. ª 2017 Mayo Foundation for Medical Education and Research
D
ischarge against medical advice (DAMA), or patient refusal of continued care,1 is a global health 2-9 care and public health concern. Approximately 1% to 2% of all hospital discharges in the United States are against medical advice (AMA).10-12 Compared with conventionally discharged patients, AMA discharges are likely to leave with deficient care, thereby increasing their risk of hospital readmission, morbidity, and mortality.12-16 Discharge against medical advice also imposes an increased burden on the health care system through disruption of patient care,12,15 disproportionate consumption of resources,17 and challenges to providers’ ethical obligations.18 Despite extensive research on this topic, there are a limited number of studies that have comprehensively examined national
n
Mayo Clin Proc. 2017;nn(n):1-11
DAMA-associated rates, temporal trends, and predictors. Previous research has either been disease-specific (eg, myocardial infarction)13,19 or focused on distinct settings (eg, trauma centers).10,15,16,20 Moreover, DAMA research on patients with mental health and substance abuse disorders has been largely based on data from psychiatric settings,21 with only a few using national data from general community hospital settings.22,23 Similarly, research on DAMA among pregnancy-related hospitalizations is scant,24-26 mainly due to their exclusion from large inpatient population analyses.11 Therefore, although evidence suggests that patients hospitalized for mental health-, substance abuse-, and pregnancyrelated disorders are at increased risk for DAMA,22-26 gaps exist in understanding potential differences in predictive factors
Mayo Clin Proc. n XXX 2017;nn(n):1-11 n http://dx.doi.org/10.1016/j.mayocp.2016.12.022 www.mayoclinicproceedings.org n ª 2017 Mayo Foundation for Medical Education and Research
From the Department of Family and Community Medicine, Baylor College of Medicine, Houston, TX.
1
MAYO CLINIC PROCEEDINGS
within these disparate diagnostic subgroups. In this study, we describe the frequency, prevalence, and 10-year temporal trends of DAMA among inpatient hospitalizations in the United States, as well as identify patient- and hospitallevel characteristics associated with DAMA, both overall and in several clinically distinct diagnostic subgroups. PATIENTS AND METHODS Study Design and Data Source We conducted a retrospective, cross-sectional analysis of data from the Nationwide Inpatient Sample (NIS) from between January 1, 2002, and December 31, 2011. Developed for the Healthcare Cost and Utilization Project (HCUP), and sponsored by the Agency for Healthcare Research and Quality, the NIS is the largest all-payer, publicly available database of inpatient hospitalization stays in the United States.27 With clinical and nonclinical data from more than 7 million hospitalizations annually (35 million weighted), it approximates a 20% stratified sample of all US nonfederal, nonrehabilitation, short-term community hospitals.27 Each year, HCUP hospitals are stratified by bed size, ownership, teaching status, urban/rural location, and US Census Bureau geographic region. The 2stage cluster sampling design for the NIS data set includes hospitals as the primary sampling units, and then all discharges from the selected hospitals.27,28 This study was deemed exempt by the Baylor College of Medicine Institutional Review Board. Study Population and Measures The study population comprised inpatient hospitalizations among adult (18 years) men and women who were discharged AMA. From these, we excluded records for those who died before discharge, and with missing data on discharge disposition or principal diagnosis; missingness was less than 1.5% for these variables. Considering the heterogeneity across all inpatient discharges, a hierarchical methodology was devised to classify the study population into mutually exclusive diagnostic subgroups. Principal diagnoses were identified using singlelevel Agency for Healthcare Research and Quality-developed Clinical Classifications Softwaredwhich classifies more than 14,000 2
Mayo Clin Proc.
n
International Classification of Diseases, 9th Edition, Clinical Modification diagnosis codes into smaller, clinically meaningful categories.29 These Clinical Classifications Software codes were then used to create 3 diagnostic subgroups. First, the “maternal” subgroup was identified by flagging hospitalizations among females with a major diagnostic category of 14 (“pregnancy, childbirth, and the puerperium”) and those in which an HCUP-created neonatal-maternal (NEOMAT) variable reflected maternal diagnoses and procedures. Then, the “mental health/ substance abuse” (MH/SA) subgroup was created to comprise nonmaternal admissions with a principal diagnosis of alcohol-, substance-, or mood-related disorders. Discharges not meeting the diagnostic criteria for the 2 above-mentioned groups were then included in the subgroup labeled “other.” The primary outcome of interest was DAMA. Patient-level covariates of interest included age, sex, race/ethnicity, income, severity of illness, primary payer/insurance, admission day, and length of stay. Hospitallevel covariates such as region, urban/rural location, teaching status, and bed size were also considered. The Table presents the categorization of each variable. Statistical Analyses The SAS program, version 9.4 (SAS Institute), and the Joinpoint Regression (JPR) program, version 4.1.1.1, were used to perform statistical analyses. A 5% type I error rate was assumed for all hypothesis tests (2-sided) and construction of CIs. The HCUP-supplied NIS discharge-level and trend weights were incorporated to account for the complex sampling design, ensure consistency in the trend analyses, and produce nationally representative estimates.28 Descriptive statistics were derived to estimate the frequency and rate of DAMA, overall and by patient- and hospital-level characteristics. Survey logistic regression was used to assess the associations between covariates and DAMA. Results from the multivariable regression models, adjusting for patient- and hospital-level covariates, were reported as odds ratios (ORs) with 95% CIs. We identified the principal diagnoses contributing to the largest burden of DAMA by ranking diagnoses according to the frequency of DAMA. To
XXX 2017;nn(n):1-11
n
http://dx.doi.org/10.1016/j.mayocp.2016.12.022 www.mayoclinicproceedings.org
Principal diagnosis CCS category Allc e (N ¼ 304,427,670) (nf ¼ 3,381,330)
n
http://dx.doi.org/10.1016/j.mayocp.2016.12.022
Characteristic
Annual ng
Overall Age (y) 18-29 30-39 40-49 50-59 60-69 70-79 80þ Sex Male Female Race/ethnicity NH-white NH-black Hispanic NH-otherk Unknown Household Incomel Highest quartile Third quartile Second quartile Lowest quartile Primary payerm Government Private Other Timing of admission Weekday Weekend Severity of illnessn Minor loss of function Moderate loss of function Major loss of function Extreme loss of function
338,133
MH/SA (N ¼ 17,428,727) (nf ¼ 713,387)
Maternal (Ne ¼ 43,440,025) (nf ¼ 119,032)
e
Otherd (N ¼ 243,558,919) (nf ¼ 2,548,911) e
%h
Ratei
AOR (95% CI)
Ratei
AOR (95% CI)
Ratei
AOR (95% CI)
Ratei
AOR (95% CI)
100
11.1
NA
2.7
NA
40.9
NA
10.5
NA
55,972 64,919 83,618 65,428 34,355 20,920 12,921
16.6 19.2 24.7 19.3 10.2 6.2 3.8
13.2 17.7 23.0 15.0 7.5 4.2 2.6
1 [Reference] 1.36 (1.33-1.40)j 1.40 (1.35-1.44)j 0.94 (0.91-0.98)j 0.43 (0.41-0.45)j 0.21 (0.20-0.22)j 0.13 (0.12-0.14)j
3.1 2.1 2.8 NA NA NA NA
1 [Reference] 0.99 (0.95-1.04) 1.16 (1.07-1.26)j NA NA NA NA
47.8 57.2 50.1 36.4 19.8 8.6 3.9
1 [Reference] 1.23 (1.17-1.30)j 1.08 (1.02 -1.15)j 0.83 (0.77-0.89)j 0.48 (0.44-0.52)j 0.22 (0.20-0.25)j 0.10 (0.09-0.12)j
25.2 23.9 20.0 13.4 7.2 4.1 2.6
1 [Reference] 1.03 (1.01-1.05)j 0.88 (0.87 -0.90)j 0.61 (0.59-0.62)j 0.28 (0.27-0.29)j 0.14 (0.13-0.14)j 0.09 (0.08-0.09)j
205,900 131,582
60.9 38.9
17.3 7.1
1 [Reference] 0.42 (0.41-0.42)j
NA 2.7
NA NA
52.2 28.4
1 [Reference] 0.63 (0.61-0.65)j
14.4 7.2
1 [Reference] 0.54 (0.54-0.55)j
160,068 67,162 35,742 16,893 58,268
47.3 19.9 10.6 5.0 17.2
9.7 20.1 13.7 12.2 8.8
1 [Reference] 1.25 (1.19-1.31)j 0.87 (0.80-0.95)j 0.92 (0.84-1.01) 0.95 (0.89-1.01)
2.1 6.7 2.2 2.0 2.7
1 [Reference] 1.49 (1.37-1.62)j 0.59 (0.52-0.65)j 0.67 (0.56-0.79)j 1.07 (0.94-1.22)
40.2 44.5 56.8 51.4 33.7
1 [Reference] 0.87 (0.79-0.97)j 1.07 (0.92-1.23) 0.98 (0.83-1.15) 1.01 (0.89-1.15)
8.7 20.2 16.2 13.0 8.1
1 [Reference] 1.37 (1.31-1.43)j 1.04 (0.95-1.13) 1.06 (0.98-1.15) 0.94 (0.88-1.01)
49,310 64,566 80,509 124,859
14.6 19.1 23.8 36.9
7.8 9.1 10.4 14.6
1 [Reference] 1.10 (1.05-1.15)j 1.22 (1.15-1.30)j 1.52 (1.42-1.62)j
1.2 2.0 2.8 4.6
1 [Reference] 1.33 (1.23-1.44)j 1.63 (1.48-1.79)j 2.22 (2.00-2.47)j
42.8 37.1 37.3 42.4
1 [Reference] 0.91 (0.84-0.97)j 0.94 (0.85-1.04) 1.05 (0.95-1.16)
7.0 8.6 9.9 14.1
1 [Reference] 1.14 (1.09-1.19)j 1.27 (1.20-1.35)j 1.59 (1.49-1.70)j
194,123 60,707 83,303
57.4 18.0 24.6
10.9 6.2 29.9
1 [Reference] 0.33 (0.32-0.35)j 1.26 (1.21-1.31)j
4.5 1.0 6.0
1 [Reference] 0.25 (0.23-0.27)j 1.49 (1.38-1.60)j
36.5 34.1 64.8
1 [Reference] 0.83 (0.77-0.89)j 1.45 (1.32-1.60)j
9.9 6.2 28.1
1 [Reference] 0.30 (0.29-0.31)j 1.04 (1.01-1.07)j
254,826 83,307
75.4 24.6
10.3 14.4
1 [Reference] 1.31 (1.30-1.33)j
2.7 3.1
1 [Reference] 1.12 (1.08-1.15)j
40.3 43.4
1 [Reference] 1.06 (1.03-1.09)j
9.6 14.3
1 [Reference] 1.38 (1.37-1.40)j
117,020 149,241 58,575 10,274
34.6 44.1 17.3 3.0
11.5 12.3 9.2 7.4
1 [Reference] 1.14 (1.11-1.17)j 0.96 (0.93-0.99)j 0.72 (0.69-0.75)j
1.4 4.2 9.6 11.8
1 [Reference] 2.63 (2.51-2.75)j 5.28 (4.87-5.72)j 5.94 (5.18-6.81)j
51.8 37.5 28.6 20.9
1 [Reference] 0.83 (0.77-0.91)j 0.81 (0.75-0.88)j 0.58 (0.52-0.64)j
12.4 10.9 8.5 7.0
1 [Reference] 1.02 (1.00-1.04)j 0.85 (0.83-0.87)j 0.64 (0.62-0.67)j Continued on next page
DISCHARGE AGAINST MEDICAL ADVICE IN THE UNITED STATES
Mayo Clin Proc. n XXX 2017;nn(n):1-11 www.mayoclinicproceedings.org
TABLE. Weighted Frequency, Prevalence, and Patient- and Hospital-Level Predictors of DAMA, 2002-2011a,b
3
4 TABLE. Continued Principal diagnosis CCS category Allc e (N ¼ 304,427,670) (nf ¼ 3,381,330) Characteristic
Mayo Clin Proc. n
XXX 2017;nn(n):1-11
Hospital region South Northeast Midwest West Hospital location/type Urban, teaching Urban, nonteaching Rural Hospital bed size Large Medium Small
Maternal (Ne ¼ 43,440,025) (nf ¼ 119,032)
MH/SA (N ¼ 17,428,727) (nf ¼ 713,387) e
Otherd (N ¼ 243,558,919) (nf ¼ 2,548,911) e
Annual ng
%h
Ratei
AOR (95% CI)
Ratei
AOR (95% CI)
Ratei
AOR (95% CI)
Ratei
AOR (95% CI)
115,789 103,445 59,387 59,512
34.2 30.6 17.6 17.6
9.9 17.1 8.4 10.5
1 [Reference] 1.99 (1.80-2.21)j 1.05 (0.96-1.13) 1.20 (1.10-1.31)j
2.6 4.1 2.8 2.0
1 [Reference] 1.92 (1.63-2.26)j 1.30 (1.10-1.52)j 1.25 (1.11-1.41)j
30.1 64.1 30.3 41.5
1 [Reference] 2.04 (1.69-2.46)j 1.01 (0.85-1.19) 1.36 (1.13-1.63)j
10.0 14.6 7.5 10.8
1 [Reference] 1.78 (1.60-1.98)j 0.99 (0.91-1.08) 1.20 (1.10-1.32)j
155,884 146,759 34,177
46.1 43.4 10.1
11.8 11.3 8.4
1 [Reference] 1.25 (1.16-1.35)j 0.88 (0.81-0.96)j
3.6 2.0 2.0
1 [Reference] 0.78 (0.70-0.87)j 0.52 (0.42-0.66)j
45.2 39.2 31.8
1 [Reference] 1.03 (0.89-1.18) 0.90 (0.75-1.08)
10.8 10.9 8.0
1 [Reference] 1.34 (1.24-1.45)j 0.91 (0.83-0.99)j
203,177 97,761 35,881
60.1 28.9 10.6
10.6 13.0 9.9
1 [Reference] 1.21 (1.11-1.32)j 1.01 (0.94-1.10)
2.8 2.8 2.0
1 [Reference] 1.04 (0.91-1.18) 0.76 (0.67-0.86)j
36.9 50.4 42.7
1 [Reference] 1.31 (1.12-1.53)j 1.17 (0.99-1.39)
10.1 12.0 9.2
1 [Reference] 1.19 (1.10-1.29)j 1.01 (0.93-1.10)
n
b
MAYO CLINIC PROCEEDINGS
http://dx.doi.org/10.1016/j.mayocp.2016.12.022 www.mayoclinicproceedings.org
AOR ¼ adjusted odds ratio; CCS ¼ Clinical Classifications Software; DAMA ¼ discharge against medical advice; MH/SA ¼ mental health/substance abuse; NA ¼ not applicable; NH ¼ non-Hispanic. All models adjusted for all variables listed in the table. c The entire study population, consisting of all inpatient hospitalizations during the study period, 2002-2011. d All other non-MH/SA or nonmaternal principal diagnoses (eg, infections, cancer, and injuries). e Total number of discharges within the respective group/column for all study years, 2002-2011. f Number of hospitalizations that ended in a DAMA across all study years, 2002-2011. g Calculated by determining the total number of inpatient hospitalizations ending in a DAMA (2002-2011) and dividing by the number of study years (10). h Percentages may not add to 100% for a given characteristic because of missing data. i Weighted DAMA prevalence rate per 1000 discharges. j P < .05. k Defined as non-Hispanic ethnicity, and race reported as Asian or Pacific Islander, Native American, or other/unknown race. l Defined by median household income for patient’s zip code. m Categorized as government (Medicare, Medicaid); private (commercial carriers, health maintenance organizations [HMOs], preferred provider organizations [PPOs]); and other (self-pay, no charge, ie, uninsured). n Refers to the All Patient Refined Diagnosis Related Group severity of illness subclasses. a
DISCHARGE AGAINST MEDICAL ADVICE IN THE UNITED STATES
the United States (an average of 338,133 per year), corresponding to a rate of 11.1 per 1,000 discharges. Rates for DAMA (per 1000 discharges) were highest among the uninsured (29.9), the 40 to 49 year age group (23.0), non-Hispanic (NH)-blacks (20.1), males (17.3), and northeast region hospitals (17.1) (Table). Figure 1 highlights the most common principal diagnoses among AMA discharges, based on both frequency and rate of DAMA.
identify patients at highest DAMA risk, we ranked principal diagnoses according to condition-specific DAMA rates. The JPR program was applied to examine temporal trends30 in DAMA rates. The JPR program first fits annual rate data to a straight line (null model with no joinpoints), followed by a Monte-Carlo permutation test to determine whether the addition of 1 or more joinpoints offers a statistically significantly better model than the null; an annual percent change and average annual percent change are computed to describe the rate changes during the study period.31 All the aforementioned descriptive, regression, and trend analyses were also conducted for each diagnostic subgroup.
Predictors Adjusted multivariable analyses on the full cohort revealed several patient- and hospitallevel characteristics as predictors of DAMA (Table). Compared with patients aged 18 to 29 years, we observed an increased odds of DAMA among patients aged 30 to 49 years and a decreasing likelihood of DAMA with increasing age among those 50 years or older. Women were 58% (95% CI, 0.41-0.42) less
RESULTS Between January 1, 2002, and December 31, 2011, approximately 3.4 million hospitalized men and women were discharged AMA in
Rank
Rank
240.5
1 2
238.7
2 1
Principal admission diagnosis 101.7 120.0
3
212.8
Mood disorders
20.4
5
119.0
Pregnancy-related, overall
2.7
112.5
Diabetes mellitus with complications
24.3
105.5
Substance-related disorders Nonspecific chest pain
28.9
4
139.5
Alcohol-related disorders
Skin and subcutaneous tissue infections
21.5
97.1
10.1
Congestive heart failure; nonhypertensive
94.9
9.9
Pneumonia (except that caused by TB or STD)
83.5
Pancreatic disorders (not diabetes)
30.2 4
62.2
49.1
3
34.5
59.9
Poisoning by other medications and drugs HIV infection
42.6
Poisoning by psychotropic agents
41.3
Sickle cell anemia
13.8
39.9
Screening/history of mental health, substance abuse
11.3
37.7
Liver disease; alcohol-related
38.1
Poisoning by nonmedicinal substances
30.9 24.8
6.4 1.0 5 Frequency of DAMA (in thousands)
49.0
Administrative/social admission
DAMA prevalence rate per 1000
FIGURE 1. Top ranked principal CCS diagnoses among AMA discharges by frequency and prevalence, 2002-2011. AMA ¼ against medical advice; CCS ¼ Clinical Classifications Software; DAMA ¼ discharge against medical advice; HIV ¼ human immunodeficiency virus; STD ¼ sexually transmitted disease; TB ¼ tuberculosis.
Mayo Clin Proc. n XXX 2017;nn(n):1-11 www.mayoclinicproceedings.org
n
http://dx.doi.org/10.1016/j.mayocp.2016.12.022
5
MAYO CLINIC PROCEEDINGS
likely to leave AMA than were men. In the unadjusted model, being of NH-black, Hispanic, and NH-other racial/ethnic minority was observed to be associated with increased odds of DAMA (data not shown). However, after adjustment, compared with NH-whites, only NH-blacks were significantly more likely to leave AMA (OR, 1.25; 95% CI, 1.19-1.31), and a protective association was observed for Hispanics (OR, 0.87; 95% CI, 0.80-0.95). The odds of self-discharge were 3 to 4 times higher among patients with government insurance (eg, Medicare and Medicaid) and the uninsured, respectively, than among those with private insurance. Patients’ household income was inversely associated with increased odds of DAMA. Compared with patients with minor severity of illness, those with moderate severity of illness were more likely to leave AMA (OR, 1.14; 95% CI, 1.11-1.17); however, decreased odds were observed among those with major and extreme severity of illness. Weekend hospital admission, compared with a weekday, was associated with an increased odds of DAMA. Hospital-level predictors included location in the Northeastern and Western US regions, urban hospitals (teaching and nonteaching status), and medium bed size. Stratified analyses on each diagnostic subgroup revealed similar predictors to those identified in the full cohort analysis (Table). Temporal Trends During the 10-year study period, we observed a 1.9% (95% CI, 0.8-3.0) average annual increase in the overall rate of DAMA (Supplemental Table, available online at http://www. mayoclinicproceedings.org). Subgroup analyses revealed statistically significant trends in DAMA rates, with a 2.9% average annual increase among the “other” subgroup, 2.3% average annual decrease in the MH/SA subgroup, and no statistically significant trend in the maternal subgroup (Figure 2). DISCUSSION We examined a nationally representative sample of inpatient hospitalizations to provide a comprehensive description of the prevalence, frequency, temporal trends, and characteristics associated with DAMA. Within the 10-year study period, on average, approximately 1.1% of all hospitalizations were discharged 6
Mayo Clin Proc.
n
AMA, with a 1.9% average annual increase in overall DAMA rates across the study years. Consistent with other national studies, we observed DAMA rates to be highest among patients who were younger, male, NH-black, low income, uninsured, Medicare/Medicaid recipients, and hospitalized for MH/SA disorders.10,11,15,19,20 We also observed high rates of self-discharge in urban hospitals and the Northeast region of the United States.10,11,19,22 Collectively, these findings highlight the intricacy of DAMA as it exists in general inpatient settings. Although a patient’s decision to leave AMA is likely multifaceted, evidence suggests that this choice generally arises from communication breakdowns,32-34 which tend to be rooted in a variety of provider- and patient-level characteristics.35,36 For instance, several patient factors known to influence perceptions of communication quality, including age, sex, race/ethnicity, income, and insurance status,37 were also identified as DAMA predictors. Although these factors are not modifiable in a clinical encounter, they serve as indicators for vulnerable patients who may require enhanced communication and other DAMA prevention strategies. Analogous to previous research, younger age was identified as a risk factor for DAMA.10,20,38 The high odds observed in the 30 to 49 year age group may reflect a necessary departure to attend to responsibilities34 typical of early adulthood. For example, the increased odds of DAMA among women of advanced maternal age (40-49 years) may be a consequence of challenges presented by a prolonged stay (eg, occupational, financial, or familial obligations). In addition, although we lack parity data on the women in our study, previous research purports an increased risk of DAMA with increasing parity.26 Likewise, males’ increased likelihood of DAMA may also be attributed to similar obligations.34 High DAMA rates among men may also be linked to the notion that men’s health behaviors (eg, propensity to not seek or continue care) are often predicted by their embodiment of masculinity and social norms.39-43 However, research that specifically examines sex differences in reasoning for leaving AMA is needed to further explicate these findings. Even after adjustment for confounders, NH-blacks in our study were more likely
XXX 2017;nn(n):1-11
n
http://dx.doi.org/10.1016/j.mayocp.2016.12.022 www.mayoclinicproceedings.org
DISCHARGE AGAINST MEDICAL ADVICE IN THE UNITED STATES
Mayo Clin Proc. n XXX 2017;nn(n):1-11 www.mayoclinicproceedings.org
n
50.0 45.0 Prevalence per 1000 discharges
than NH-whites to self-discharge. Although this is in agreement with some studies,10,15 reports of the association between DAMA and race/ethnicity are inconsistent.20,34 However, given that race is often a proxy measure for socioeconomic status, educational attainment, and health insurance coverage,44 it is possible that NH-blacks’ increased DAMA risk is partially due to residual confounding after incomplete adjustment for factors related to socioeconomic status. For example, uninsured and Medicare/Medicaid hospital stays accounted for a large proportion of patients discharged AMA,10,11,20 suggesting that concerns related to health services payment increases the risk of self-discharge. It is also probable that NH-blacks are more likely to leave AMA because of receipt of differential in-hospital care, historical race-related/ race-fueled mistrust, or perceptions of discrimination.10,45 Conversely, racial/ethnic minority status is not always an adverse risk factor for DAMA. Our observations of a protective effect against DAMA among Hispanics may be related to having increased social support,20 or cultural factors, such as being more accepting of health care providers as authority figures.10,46 Furthermore, among discharges with MH/SA diagnoses, being of NH-black race/ethnicity also had a protective effect, yielding a 13% lower odds of being discharged AMA than NH-whites. Despite being contradictory to findings for the full cohort, this observance may be a reflection of NH-blacks’ increased likelihood of initiating mental health and substance abuse services at later and more severe stages of the illness,47 subsequently making DAMA less likely. Although our findings are unable to elucidate the pathways by which race/ethnicity independently influences DAMA risk, they underscore the need for more race/ethnicity-specific quantitative and qualitative DAMA investigations. High DAMA rates were also observed among patients with a primary diagnosis of a MH/SA disorder. In corroboration with other studies, these findings are likely related to patients’ underlying psychological issues, addictive behaviors (ie, drug seeking), mistrust of providers, and perceived prejudice in delivery of care.34,38 They may also be partly explained by inadequate or lack of assessment
MH/SA02-11 = –2.3 (–3.8 to –0.8)
40.0 15.0 12.0
All02-11 = 1.9 (0.8 to 3.0) Other06-11 = 0.4 (–1.9 to 2.8)
9.0 Other02-06 = 6.0 (2.5 to 9.7)
6.0 3.0 Maternal02-11 = 1.9 (–0.4 to 4.3)
0.0 2002
2003
2004
2005
2007 2006 Year
2008
2009
2010
2011
FIGURE 2. Temporal trends in DAMA prevalence rates for principal CCS diagnostic groups, 2002-2011. All ¼ all discharges; CCS ¼ Clinical Classifications Software; DAMA ¼ discharge against medical advice; Maternal ¼ discharges with maternal-related diagnoses; MH/SA ¼ discharges with mental health/substance abuse-related principal diagnoses; Other ¼ discharges with non-MH/SA or maternal-related diagnoses. Markers indicate the observed annual rate, and solid lines represent the trend estimated by joinpoint regression. Values encased in boxes represent average annual percent change (AAPC), point estimate (95% CI); subscripted numbers represent the year range for AAPC estimates.
of patients’ decision-making capacity. Patients presenting with MH/SA disorders may have a diminished capacity to comprehend the gravity of the decision to leave AMA, and often make this decision hastily.48 Thus, policies, or closer adherence to guidelines that require a full assessment of the decision-making capacity for patients with MH/SA, may decrease the likelihood of this high-risk population from leaving AMA. In fact, the statistically significant 2.3% downward annual trend observed in the MH/SA population’s rates may have been influenced by a growing awareness of MH/SA patients’ vulnerability for DAMA and successful implementation of preventive strategies.38,49,50 In the general study population, we observed an inverse relationship with DAMA among patients with major and extreme severity of illness. Yet, in the maternal population, severity of illness was directly associated with a higher likelihood of DAMA. Although these findings could be construed as an anomaly, research suggests that pregnancy-related complications, combined with other
http://dx.doi.org/10.1016/j.mayocp.2016.12.022
7
MAYO CLINIC PROCEEDINGS
comorbidities, may unexpectedly prolong length of stay, thereby increasing DAMA risk.25 Together, these results suggest that the link between severity of illness and DAMA is complex and disease/conditiondependent. Hospital-level characteristics such as region, size, and teaching status were also found to predict DAMA. Consistent with other AMA studies, rates were highest in the northeast11,19 and medium-sized hospitals.10,23 The reasons for the observed regional differences are unclear and warrant additional research to investigate this phenomenon. Higher DAMA rates in hospitals of medium or large bed size could possibly be related to low patient satisfaction or limited provider attentiveness experienced by some patients in hospitals of larger sizes.51 Similarly, inadequate staffing or reduced continuity in care may also contribute to the increased likelihood of DAMA occurring among patients with weekend admissions. In comparison to rural hospitals, we observed increased odds of DAMA among inpatient stays at urban hospitals of both teaching and nonteaching status. Among the general study population receiving care in urban settings, the adjusted odds of DAMA tended to be lower in teaching versus nonteaching hospitals. This could be due to high specialization of teaching hospitals, or a consequence of being staffed by providers who embody patient-centeredness. The patient-centered approach to care not only respects the patient’s autonomy and provider’s recommendations but also supports good provider-patient communication52,53 and encourages shared decision-making54da process that offers the optimal opportunity to avoid DAMA.55 In the same vein, the low DAMA prevalence in rural hospitals may be due to increased familiarity and trust between patients and providers.56 Conversely, in the maternal population, DAMA was more likely to occur in urban teaching hospitals, compared with nonteaching hospitals. These findings may reflect previous hospitalizations marred by subpar health care quality, or a preference to continue treatment at home.57 Altogether, the above-mentioned observations offer additional support to the idea that DAMA rates 8
Mayo Clin Proc.
n
are reflective of measures of hospital quality,58 as well as patients’ perceptions of the quality of care10,57 and communication provided.37 In contrast, among the MH/SA subgroup, hospital size, location, and teaching status were not associated with risk of DAMA. It is likely that hospital characteristics’ effect on DAMA is neutralized once patients’ mental health/ substance use disorders are taken into account. Thus, more research is needed to better clarify the complex interplay between patient diagnosis, hospital teaching status, and DAMA risk. Study Limitations Despite the statistical power and precision offered by our large sample, the findings from this study should be interpreted within the context of several limitations. As expected in retrospective secondary analyses of administrative data, our results are subject to selection bias and misclassification/errors in coding. For example, mental health conditions (eg, depression) and substance abuse have been shown to be poorly identified when relying exclusively on International Classification of Diseases based codes.59,60 However, it is unlikely that these biases or errors account for our findings entirely, given the rigorous testing and quality control measures applied to NIS data to ensure their validity and reliability.61 Second, because the unit of analysis was the hospital discharge and not the patient, we were unable to discern whether multiple discharges were reflective of repeat hospitalizations for 1 person or initial hospitalizations for different patients.11 The inability to adjust for these occurrences may have resulted in DAMA-related estimates that are slightly different from those calculated from a patient-level data set with identifiers. Third, although we attempted to create subgroups on the basis of clinically meaningful categories, it is possible that patients’ primary admission diagnoses are confounding the associations observed between covariates and DAMA risk. Fourth, we recognize that the “other” diagnostic subgroup is heterogeneous. To address this heterogeneity, we ran a sensitivity analysis subdividing the “other” group by primary diagnosis into acute and chronic conditions. Our results were not
XXX 2017;nn(n):1-11
n
http://dx.doi.org/10.1016/j.mayocp.2016.12.022 www.mayoclinicproceedings.org
DISCHARGE AGAINST MEDICAL ADVICE IN THE UNITED STATES
substantively different when stratified by acute vs chronic subtypes (data not shown). In addition, given that previous studies have either excluded the maternal and MH/SA populations or studied them separately or in nonacute hospital settings, we considered it of value to report on both the overall and clinically distinct populations within a single study. Another notable limitation was that the NIS database does not capture patients’ motivation or reasons for being discharged AMA. Thus, we are unable to ascertain the impact of potential influences such as provider contributions (eg, suboptimal communication), patients’ external obligations or psychosocial characteristics (eg, perceptions and risk behaviors). These limitations aside, a notable strength of our study is that it is one of few investigations using nationally representative data to provide a comprehensive assessment of current DAMA rates, trends, and associated patient, hospital, and diagnostic predictors. CONCLUSION An increased understanding of the factors associated with patients’ decision to leave AMA is essential. It allows the opportunity to identify those at highest risk and thus, intervene early to reduce the subsequent morbidity, mortality, and additional health care costs associated with DAMA.38 Moreover, although data are lacking to explain the cause of the observed overall increasing DAMA rates, it is likely multifactorial; patient, societal, economic, changing health insurance patterns, and health care quality factors may all be influential. Furthermore, although the inability to reach consensus on the need for continued care20 is not always due to poor communication, enhancements in patient-provider communication and shared decision making have the potential to avert the occurrence of DAMA, achieve a more optimal quality of care, and reduce health disparities.52 SUPPLEMENTAL ONLINE MATERIAL Supplemental material can be found online at http://www.mayoclinicproceedings.org. Supplemental material attached to journal articles has not been edited, and the authors take responsibility for the accuracy of all data. Mayo Clin Proc. n XXX 2017;nn(n):1-11 www.mayoclinicproceedings.org
n
Abbreviations and Acronyms: AMA = against medical advice; DAMA = discharge against medical advice; HCUP = Healthcare Cost and Utilization Project; JPR = Joinpoint Regression; MH/SA = mental health/substance abuse; NEOMAT = Neonatal-Maternal; NH = non-Hispanic; NIS = Nationwide Inpatient Sample; OR = odds ratio Grant Support: This research was supported by grant T32 HP10031 from the Health Resources and Services Administration, an agency of the US Department of Health & Human Services. The funding source/study sponsor had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication. Correspondence: Address to Kiara K. Spooner, DrPH, MPH, Department of Family and Community Medicine, Baylor College of Medicine, 3701 Kirby Dr, Ste 600, Houston, TX 77098 ([email protected]).
REFERENCES 1. The Joint Commission. Discharge Disposition. Specifications Manual for Joint Commission National Quality Measures (Version 2016A). https://manual.jointcommission.org/releases/ TJC2016A/DataElem0537.html. Accessed November 28, 2016. 2. Hadadi A, Khashayar P, Karbakhsh M, Vasheghani Farahani A. Discharge against medical advice from a Tehran emergency department. Int J Health Care Qual Assur. 2016; 29(1):24-32. 3. Jimoh BM, Anthonia OC, Chinwe I, et al. Prospective evaluation of cases of discharge against medical advice in Abuja, Nigeria. Sci World J. 2015;2015:314817. 4. Jeong J, Song KJ, Kim YJ, et al. The association between acute alcohol consumption and discharge against medical advice of injured patients in the ED. Am J Emerg Med. 2016;34(3): 464-468. 5. Saia M, Buja A, Mantoan D, et al. Frequency and trends of hospital discharges against medical advice (DAMA) in a large administrative database. Ann Ist Super Sanita. 2014;50(4): 357-362. 6. Carron PN, Yersin B, Trueb L, Gonin P, Hugli O. Missed opportunities: evolution of patients leaving without being seen or against medical advice during a six-year period in a Swiss tertiary hospital emergency department. Biomed Res Int. 2014; 2014:690368. 7. Yong TY, Fok JS, Hakendorf P, Ben-Tovim D, Thompson CH, Li JY. Characteristics and outcomes of discharges against medical advice among hospitalised patients. Intern Med J. 2013; 43(7):798-802. 8. Kraut A, Fransoo R, Olafson K, Ramsey CD, Yogendran M, Garland A. A population-based analysis of leaving the hospital against medical advice: incidence and associated variables. BMC Health Serv Res. 2013;13:415. 9. Valevski A, Zalsman G, Tsafrir S, Lipschitz-Elhawi R, Weizman A, Shohat T. Rate of readmission and mortality risks of schizophrenia patients who were discharged against medical advice. Eur Psychiatry. 2012;27(7):496-499. 10. Ibrahim SA, Kwoh CK, Krishnan E. Factors associated with patients who leave acute-care hospitals against medical advice. Am J Public Health. 2007;97(12):2204-2208. 11. Stranges E, Wier L, Merrill CT, Steiner C. Hospitalizations in which patients leave the hospital against medical Advice (AMA), 2007. HCUP Statistical Brief #78. 2009. http://www. hcup-us.ahrq.gov/reports/statbriefs/statbriefs.jsp. Accessed January 1, 2016.
http://dx.doi.org/10.1016/j.mayocp.2016.12.022
9
MAYO CLINIC PROCEEDINGS
12. Southern WN, Nahvi S, Arnsten JH. Increased risk of mortality and readmission among patients discharged against medical advice. Am J Med. 2012;125(6):594-602. 13. Fiscella K, Meldrum S, Barnett S. Hospital discharge against advice after myocardial infarction: deaths and readmissions. Am J Med. 2007;120(12):1047-1053. 14. Glasgow JM, Vaughn-Sarrazin M, Kaboli PJ. Leaving against medical advice (AMA): risk of 30-day mortality and hospital readmission. J Gen Intern Med. 2010;25(9):926-929. 15. Olufajo OA, Metcalfe D, Yorkgitis BK, et al. Whatever happens to trauma patients who leave against medical advice? Am J Surg. 2016;211(4):677-683. 16. Weingart SN, Davis RB, Phillips RS. Patients discharged against medical advice from a general medicine service. J Gen Intern Med. 1998;13(8):568-571. 17. Saitz R, Ghali WA, Moskowitz MA. The impact of leaving against medical advice on hospital resource utilization. J Gen Intern Med. 2000;15(2):103-107. 18. Carrese JA. Refusal of care: patients’ well-being and physicians’ ethical obligations: “but doctor, I want to go home”. JAMA. 2006;296(6):691-695. 19. Kaplan GG, Panaccione R, Hubbard JN, et al. Inflammatory bowel disease patients who leave hospital against medical advice: predictors and temporal trends. Inflamm Bowel Dis. 2009;15(6):845-851. 20. Franks P, Meldrum S, Fiscella K. Discharges against medical advice: are race/ethnicity predictors? J Gen Intern Med. 2006; 21(9):955-960. 21. Brook M, Hilty DM, Liu W, Hu R, Frye MA. Discharge against medical advice from inpatient psychiatric treatment: a literature review. Psychiatr Serv. 2006;57(8):1192-1198. 22. Sclar DA, Robison LM. Hospital admission for schizophrenia and discharge against medical advice in the United States. Prim Care Companion J Clin Psychiatry. 2010;12(2):e1-e6. pii: PCC.09m00827. doi: 10.4088/PCC.09m00827yel. 23. Tawk R, Freels S, Mullner R. Associations of mental, and medical illnesses with against medical advice discharges: the National Hospital Discharge Survey, 1988-2006. Adm Policy Ment Health. 2013;40(2):124-132. 24. Letterie GS, Markenson GR, Markenson MM. Discharge against medical advice in an obstetric unit. J Reprod Med. 1993;38(5): 370-374. 25. Fiscella K, Meldrum S, Franks P. Post partum discharge against medical advice: who leaves and does it matter? Matern Child Health J. 2007;11(5):431-436. 26. Tucker Edmonds B, Ahlberg C, McPherson K, Srinivas S, Lorch S. Predictors and adverse pregnancy outcomes associated with antepartum discharge against medical advice. Matern Child Health J. 2014;18(3):640-647. 27. Healthcare Cost and Utilization Project (HCUP). Introduction to the HCUP Nationwide Inpatient Sample (NIS) 2011. November 2015. http://www.hcup-us.ahrq.gov/db/nation/nis/ NIS_Introduction_2011.pdf. Accessed January 1, 2016. 28. Houchens R, Elixhauser A. Final report on calculating Nationwide Inpatient Sample (NIS) variances for data years 2011 and earlier. HCUP Methods Series Report # 2003-02 online. December 14, 2015. http://www.hcup-us.ahrq.gov/reports/ methods/methods.jsp. Accessed January 1, 2016. 29. Elixhauser A, Steiner C, Palmer L. Clinical Classifications Software (CCS), 2015. 2015. http://www.hcup-us.ahrq.gov/toolssoftware/ ccs/ccs.jsp. Accessed February 13, 2017. 30. National Cancer Institute. Joinpoint Regression Program, Version 4.1.1.3, Statistical Methodology and Applications Branch, Surveillance Research Program. 2016. http://surveillance.cancer.gov/ joinpoint/. Accessed January 4, 2015. 31. Kim HJ, Fay MP, Feuer EJ, Midthune DN. Permutation tests for joinpoint regression with applications to cancer rates. Stat Med. 2000;19(3):335-351. 32. Windish DM, Ratanawongsa N. Providers’ perceptions of relationships and professional roles when caring for patients who
10
Mayo Clin Proc.
n
33.
34.
35.
36.
37.
38. 39.
40.
41.
42.
43.
44.
45.
46.
47.
48.
49.
50.
51.
52.
53.
leave the hospital against medical advice. J Gen Intern Med. 2008;23(10):1698-1707. Onukwugha E, Saunders E, Mullins CD, et al. A qualitative study to identify reasons for discharges against medical advice in the cardiovascular setting. BMJ Open. 2012;2(4). pii: e000902. doi: 10.1136/bmjopen-2012-000902. Onukwugha E, Saunders E, Mullins CD, Pradel FG, Zuckerman M, Weir MR. Reasons for discharges against medical advice: a qualitative study. Qual Saf Health Care. 2010; 19(5):420-424. Cooper LA, Roter DL, Johnson RL, Ford DE, Steinwachs DM, Powe NR. Patient-centered communication, ratings of care, and concordance of patient and physician race. Ann Intern Med. 2003;139(11):907-915. Street RL Jr, Gordon H, Haidet P. Physicians’ communication and perceptions of patients: is it how they look, how they talk, or is it just the doctor? Soc Sci Med. 2007;65(3): 586-598. Spooner KK, Salemi JL, Salihu HM, Zoorob RJ. Disparities in perceived patient-provider communication quality in the United States: trends and correlates. Patient Educ Couns. 2016;99(5):844-854. Alfandre DJ. “I’m going home”: discharges against medical advice. Mayo Clin Proc. 2009;84(3):255-260. Mahalik JR, Burns SM, Syzdek M. Masculinity and perceived normative health behaviors as predictors of men’s health behaviors. Soc Sci Med. 2007;64(11):2201-2209. Bertakis KD, Azari R, Helms LJ, Callahan EJ, Robbins JA. Gender differences in the utilization of health care services. J Fam Pract. 2000;49(2):147-152. Courtenay WH. Constructions of masculinity and their influence on men’s well-being: a theory of gender and health. Soc Sci Med. 2000;50(10):1385-1401. Grzywacz JG, Stoller EP, Brewer-Lowry AN, Bell RA, Quandt SA, Arcury TA. Gender and health lifestyle: an indepth exploration of self-care activities in later life. Health Educ Behav. 2012;39(3):332-340. O’Brien R, Hunt K, Hart G. ‘It’s caveman stuff, but that is to a certain extent how guys still operate’: men’s accounts of masculinity and help seeking. Soc Sci Med. 2005;61(3):503-516. Kawachi I, Daniels N, Robinson DE. Health disparities by race and class: why both matter. Health Aff (Millwood). 2005;24(2): 343-352. Blanchard J, Lurie N. R-E-S-P-E-C-T: patient reports of disrespect in the health care setting and its impact on care. J Fam Pract. 2004;53(9):721-730. Flores G. Culture and the patient-physician relationship: achieving cultural competency in health care. J Pediatr. 2000; 136(1):14-23. Cook BL, Zuvekas SH, Carson N, Wayne GF, Vesper A, McGuire TG. Assessing racial/ethnic disparities in treatment across episodes of mental health care. Health Serv Res. 2014; 49(1):206-229. Gerbasi JB, Simon RI. Patients’ rights and psychiatrists’ duties: discharging patients against medical advice. Harv Rev Psychiatry. 2003;11(6):333-343. Holden P, Vogtsberger KN, Mohl PC, Fuller DS. Patients who leave the hospital against medical advice: the role of the psychiatric consultant. Psychosomatics. 1989;30(4):396-404. Targum SD, Capodanno AE, Hoffman HA, Foudraine C. An intervention to reduce the rate of hospital discharges against medical advice. Am J Psychiatry. 1982;139(5):657-659. Tajeu GS, Kazley AS, Menachemi N. Lower satisfaction among hospitals that disproportionately treat black patients. J Health Hum Serv Adm. 2015;38(3):290-316. Institute of Medicine. Crossing the Quality Chasm: A New Health System for the 21st Century. Washington, DC: National Academy Press; 2001. Institute of Medicine. Envisioning the National Health Care Quality Report. Washington, DC: National Academy Press; 2001.
XXX 2017;nn(n):1-11
n
http://dx.doi.org/10.1016/j.mayocp.2016.12.022 www.mayoclinicproceedings.org
DISCHARGE AGAINST MEDICAL ADVICE IN THE UNITED STATES
54. Barry MJ, Edgman-Levitan S. Shared decision makingepinnacle of patient-centered care. N Engl J Med. 2012;366(9):780-781. 55. Alfandre D. Reconsidering against medical advice discharges: embracing patient-centeredness to promote high quality care and a renewed research agenda. J Gen Intern Med. 2013; 28(12):1657-1662. 56. Seaborn Moyse H, Osmun WE. Discharges against medical advice: a community hospital’s experience. Can J Rural Med. 2004;9(3):148-153. 57. Tabatabaei SM, Sargazi Moakhar Z, Behmanesh Pour F, Shaare Mollashahi S, Zaboli M. Hospitalized pregnant women who leave against medical advice: attributes and reasons. Matern Child Health J. 2016;20(1):128-138.
Mayo Clin Proc. n XXX 2017;nn(n):1-11 www.mayoclinicproceedings.org
n
58. Onukwugha EC, Shaya FT, Saunders E, Weir MR. Ethnic disparities, hospital quality, and discharges against medical advice among patients with cardiovascular disease. Ethn Dis. 2009; 19(2):172-178. 59. Fiest KM, Jette N, Quan H, et al. Systematic review and assessment of validated case definitions for depression in administrative data. BMC Psychiatry. 2014;14:289. 60. Pang JX, Ross E, Borman MA, et al. Validation of coding algorithms for the identification of patients hospitalized for alcoholic hepatitis using administrative data. BMC Gastroenterol. 2015;15:116. 61. Healthcare Cost and Utilization Project (HCUP). HCUP Quality Control Procedures. March 17, 2016. http://www.hcup-us. ahrq.gov/db/quality.jsp. Accessed June 1, 2016.
http://dx.doi.org/10.1016/j.mayocp.2016.12.022
11
Discharge Against Medical Advice in the United States, 2002-2011 Kiara K. Spooner, DrPH, MPH; Jason L. Salemi, PhD, MPH; Hamisu M. Salihu, MD, PhD; and Roger J. Zoorob, MD, MPH Abstract Objective: To describe the national frequency, prevalence, and trends of discharge against medical advice (DAMA) among inpatient hospitalizations in the United States and identify differences across patient- and hospital-level characteristics, overall and in clinically distinct diagnostic subgroups. Patients and Methods: We conducted a retrospective, cross-sectional analysis of inpatient hospitalizations (18 years), discharged between January 1, 2002, and December 31, 2011, using the Nationwide Inpatient Sample. Descriptive statistics, multivariable logistic, and joinpoint regression were used for statistical analyses. Results: Between January 1, 2002, and December 31, 2011, more than 338,000 inpatient hospitalizations were discharged against medical advice each year, with a 1.9% average annual increase in prevalence over the decade (95% CI, 0.8%-3.0%). Temporal trends in DAMA varied by principal diagnosis. Among patients hospitalized for mental health- or substance abuse-related disorders, there was a 2.3% (95% CI, 3.8% to 0.8%) average annual decrease in the rate of DAMA. A statistically significant temporal rate change was not observed among hospitalizations for pregnancy-related disorders. Multivariable regression revealed several patient and hospital characteristics as predictors of DAMA, including lack of health insurance (odds ratio [OR], 3.78; 95% CI, 3.62-3.94), male sex (OR, 2.40; 95% CI, 2.36-2.45), and northeast region (OR, 1.91; 95% CI, 1.72-2.11). Other predictors included age, race/ethnicity, income, primary diagnosis, severity of illness, and hospital location/type and size. Conclusion: Rates for DAMA have increased in the United States, and key differences exist across patient and hospital characteristics. Early identification of vulnerable patients and preventive measures such as improved patient-provider communication may reduce DAMA. ª 2017 Mayo Foundation for Medical Education and Research
D
ischarge against medical advice (DAMA), or patient refusal of continued care,1 is a global health 2-9 care and public health concern. Approximately 1% to 2% of all hospital discharges in the United States are against medical advice (AMA).10-12 Compared with conventionally discharged patients, AMA discharges are likely to leave with deficient care, thereby increasing their risk of hospital readmission, morbidity, and mortality.12-16 Discharge against medical advice also imposes an increased burden on the health care system through disruption of patient care,12,15 disproportionate consumption of resources,17 and challenges to providers’ ethical obligations.18 Despite extensive research on this topic, there are a limited number of studies that have comprehensively examined national
n
Mayo Clin Proc. 2017;nn(n):1-11
DAMA-associated rates, temporal trends, and predictors. Previous research has either been disease-specific (eg, myocardial infarction)13,19 or focused on distinct settings (eg, trauma centers).10,15,16,20 Moreover, DAMA research on patients with mental health and substance abuse disorders has been largely based on data from psychiatric settings,21 with only a few using national data from general community hospital settings.22,23 Similarly, research on DAMA among pregnancy-related hospitalizations is scant,24-26 mainly due to their exclusion from large inpatient population analyses.11 Therefore, although evidence suggests that patients hospitalized for mental health-, substance abuse-, and pregnancyrelated disorders are at increased risk for DAMA,22-26 gaps exist in understanding potential differences in predictive factors
Mayo Clin Proc. n XXX 2017;nn(n):1-11 n http://dx.doi.org/10.1016/j.mayocp.2016.12.022 www.mayoclinicproceedings.org n ª 2017 Mayo Foundation for Medical Education and Research
From the Department of Family and Community Medicine, Baylor College of Medicine, Houston, TX.
1
MAYO CLINIC PROCEEDINGS
within these disparate diagnostic subgroups. In this study, we describe the frequency, prevalence, and 10-year temporal trends of DAMA among inpatient hospitalizations in the United States, as well as identify patient- and hospitallevel characteristics associated with DAMA, both overall and in several clinically distinct diagnostic subgroups. PATIENTS AND METHODS Study Design and Data Source We conducted a retrospective, cross-sectional analysis of data from the Nationwide Inpatient Sample (NIS) from between January 1, 2002, and December 31, 2011. Developed for the Healthcare Cost and Utilization Project (HCUP), and sponsored by the Agency for Healthcare Research and Quality, the NIS is the largest all-payer, publicly available database of inpatient hospitalization stays in the United States.27 With clinical and nonclinical data from more than 7 million hospitalizations annually (35 million weighted), it approximates a 20% stratified sample of all US nonfederal, nonrehabilitation, short-term community hospitals.27 Each year, HCUP hospitals are stratified by bed size, ownership, teaching status, urban/rural location, and US Census Bureau geographic region. The 2stage cluster sampling design for the NIS data set includes hospitals as the primary sampling units, and then all discharges from the selected hospitals.27,28 This study was deemed exempt by the Baylor College of Medicine Institutional Review Board. Study Population and Measures The study population comprised inpatient hospitalizations among adult (18 years) men and women who were discharged AMA. From these, we excluded records for those who died before discharge, and with missing data on discharge disposition or principal diagnosis; missingness was less than 1.5% for these variables. Considering the heterogeneity across all inpatient discharges, a hierarchical methodology was devised to classify the study population into mutually exclusive diagnostic subgroups. Principal diagnoses were identified using singlelevel Agency for Healthcare Research and Quality-developed Clinical Classifications Softwaredwhich classifies more than 14,000 2
Mayo Clin Proc.
n
International Classification of Diseases, 9th Edition, Clinical Modification diagnosis codes into smaller, clinically meaningful categories.29 These Clinical Classifications Software codes were then used to create 3 diagnostic subgroups. First, the “maternal” subgroup was identified by flagging hospitalizations among females with a major diagnostic category of 14 (“pregnancy, childbirth, and the puerperium”) and those in which an HCUP-created neonatal-maternal (NEOMAT) variable reflected maternal diagnoses and procedures. Then, the “mental health/ substance abuse” (MH/SA) subgroup was created to comprise nonmaternal admissions with a principal diagnosis of alcohol-, substance-, or mood-related disorders. Discharges not meeting the diagnostic criteria for the 2 above-mentioned groups were then included in the subgroup labeled “other.” The primary outcome of interest was DAMA. Patient-level covariates of interest included age, sex, race/ethnicity, income, severity of illness, primary payer/insurance, admission day, and length of stay. Hospitallevel covariates such as region, urban/rural location, teaching status, and bed size were also considered. The Table presents the categorization of each variable. Statistical Analyses The SAS program, version 9.4 (SAS Institute), and the Joinpoint Regression (JPR) program, version 4.1.1.1, were used to perform statistical analyses. A 5% type I error rate was assumed for all hypothesis tests (2-sided) and construction of CIs. The HCUP-supplied NIS discharge-level and trend weights were incorporated to account for the complex sampling design, ensure consistency in the trend analyses, and produce nationally representative estimates.28 Descriptive statistics were derived to estimate the frequency and rate of DAMA, overall and by patient- and hospital-level characteristics. Survey logistic regression was used to assess the associations between covariates and DAMA. Results from the multivariable regression models, adjusting for patient- and hospital-level covariates, were reported as odds ratios (ORs) with 95% CIs. We identified the principal diagnoses contributing to the largest burden of DAMA by ranking diagnoses according to the frequency of DAMA. To
XXX 2017;nn(n):1-11
n
http://dx.doi.org/10.1016/j.mayocp.2016.12.022 www.mayoclinicproceedings.org
Principal diagnosis CCS category Allc e (N ¼ 304,427,670) (nf ¼ 3,381,330)
n
http://dx.doi.org/10.1016/j.mayocp.2016.12.022
Characteristic
Annual ng
Overall Age (y) 18-29 30-39 40-49 50-59 60-69 70-79 80þ Sex Male Female Race/ethnicity NH-white NH-black Hispanic NH-otherk Unknown Household Incomel Highest quartile Third quartile Second quartile Lowest quartile Primary payerm Government Private Other Timing of admission Weekday Weekend Severity of illnessn Minor loss of function Moderate loss of function Major loss of function Extreme loss of function
338,133
MH/SA (N ¼ 17,428,727) (nf ¼ 713,387)
Maternal (Ne ¼ 43,440,025) (nf ¼ 119,032)
e
Otherd (N ¼ 243,558,919) (nf ¼ 2,548,911) e
%h
Ratei
AOR (95% CI)
Ratei
AOR (95% CI)
Ratei
AOR (95% CI)
Ratei
AOR (95% CI)
100
11.1
NA
2.7
NA
40.9
NA
10.5
NA
55,972 64,919 83,618 65,428 34,355 20,920 12,921
16.6 19.2 24.7 19.3 10.2 6.2 3.8
13.2 17.7 23.0 15.0 7.5 4.2 2.6
1 [Reference] 1.36 (1.33-1.40)j 1.40 (1.35-1.44)j 0.94 (0.91-0.98)j 0.43 (0.41-0.45)j 0.21 (0.20-0.22)j 0.13 (0.12-0.14)j
3.1 2.1 2.8 NA NA NA NA
1 [Reference] 0.99 (0.95-1.04) 1.16 (1.07-1.26)j NA NA NA NA
47.8 57.2 50.1 36.4 19.8 8.6 3.9
1 [Reference] 1.23 (1.17-1.30)j 1.08 (1.02 -1.15)j 0.83 (0.77-0.89)j 0.48 (0.44-0.52)j 0.22 (0.20-0.25)j 0.10 (0.09-0.12)j
25.2 23.9 20.0 13.4 7.2 4.1 2.6
1 [Reference] 1.03 (1.01-1.05)j 0.88 (0.87 -0.90)j 0.61 (0.59-0.62)j 0.28 (0.27-0.29)j 0.14 (0.13-0.14)j 0.09 (0.08-0.09)j
205,900 131,582
60.9 38.9
17.3 7.1
1 [Reference] 0.42 (0.41-0.42)j
NA 2.7
NA NA
52.2 28.4
1 [Reference] 0.63 (0.61-0.65)j
14.4 7.2
1 [Reference] 0.54 (0.54-0.55)j
160,068 67,162 35,742 16,893 58,268
47.3 19.9 10.6 5.0 17.2
9.7 20.1 13.7 12.2 8.8
1 [Reference] 1.25 (1.19-1.31)j 0.87 (0.80-0.95)j 0.92 (0.84-1.01) 0.95 (0.89-1.01)
2.1 6.7 2.2 2.0 2.7
1 [Reference] 1.49 (1.37-1.62)j 0.59 (0.52-0.65)j 0.67 (0.56-0.79)j 1.07 (0.94-1.22)
40.2 44.5 56.8 51.4 33.7
1 [Reference] 0.87 (0.79-0.97)j 1.07 (0.92-1.23) 0.98 (0.83-1.15) 1.01 (0.89-1.15)
8.7 20.2 16.2 13.0 8.1
1 [Reference] 1.37 (1.31-1.43)j 1.04 (0.95-1.13) 1.06 (0.98-1.15) 0.94 (0.88-1.01)
49,310 64,566 80,509 124,859
14.6 19.1 23.8 36.9
7.8 9.1 10.4 14.6
1 [Reference] 1.10 (1.05-1.15)j 1.22 (1.15-1.30)j 1.52 (1.42-1.62)j
1.2 2.0 2.8 4.6
1 [Reference] 1.33 (1.23-1.44)j 1.63 (1.48-1.79)j 2.22 (2.00-2.47)j
42.8 37.1 37.3 42.4
1 [Reference] 0.91 (0.84-0.97)j 0.94 (0.85-1.04) 1.05 (0.95-1.16)
7.0 8.6 9.9 14.1
1 [Reference] 1.14 (1.09-1.19)j 1.27 (1.20-1.35)j 1.59 (1.49-1.70)j
194,123 60,707 83,303
57.4 18.0 24.6
10.9 6.2 29.9
1 [Reference] 0.33 (0.32-0.35)j 1.26 (1.21-1.31)j
4.5 1.0 6.0
1 [Reference] 0.25 (0.23-0.27)j 1.49 (1.38-1.60)j
36.5 34.1 64.8
1 [Reference] 0.83 (0.77-0.89)j 1.45 (1.32-1.60)j
9.9 6.2 28.1
1 [Reference] 0.30 (0.29-0.31)j 1.04 (1.01-1.07)j
254,826 83,307
75.4 24.6
10.3 14.4
1 [Reference] 1.31 (1.30-1.33)j
2.7 3.1
1 [Reference] 1.12 (1.08-1.15)j
40.3 43.4
1 [Reference] 1.06 (1.03-1.09)j
9.6 14.3
1 [Reference] 1.38 (1.37-1.40)j
117,020 149,241 58,575 10,274
34.6 44.1 17.3 3.0
11.5 12.3 9.2 7.4
1 [Reference] 1.14 (1.11-1.17)j 0.96 (0.93-0.99)j 0.72 (0.69-0.75)j
1.4 4.2 9.6 11.8
1 [Reference] 2.63 (2.51-2.75)j 5.28 (4.87-5.72)j 5.94 (5.18-6.81)j
51.8 37.5 28.6 20.9
1 [Reference] 0.83 (0.77-0.91)j 0.81 (0.75-0.88)j 0.58 (0.52-0.64)j
12.4 10.9 8.5 7.0
1 [Reference] 1.02 (1.00-1.04)j 0.85 (0.83-0.87)j 0.64 (0.62-0.67)j Continued on next page
DISCHARGE AGAINST MEDICAL ADVICE IN THE UNITED STATES
Mayo Clin Proc. n XXX 2017;nn(n):1-11 www.mayoclinicproceedings.org
TABLE. Weighted Frequency, Prevalence, and Patient- and Hospital-Level Predictors of DAMA, 2002-2011a,b
3
4 TABLE. Continued Principal diagnosis CCS category Allc e (N ¼ 304,427,670) (nf ¼ 3,381,330) Characteristic
Mayo Clin Proc. n
XXX 2017;nn(n):1-11
Hospital region South Northeast Midwest West Hospital location/type Urban, teaching Urban, nonteaching Rural Hospital bed size Large Medium Small
Maternal (Ne ¼ 43,440,025) (nf ¼ 119,032)
MH/SA (N ¼ 17,428,727) (nf ¼ 713,387) e
Otherd (N ¼ 243,558,919) (nf ¼ 2,548,911) e
Annual ng
%h
Ratei
AOR (95% CI)
Ratei
AOR (95% CI)
Ratei
AOR (95% CI)
Ratei
AOR (95% CI)
115,789 103,445 59,387 59,512
34.2 30.6 17.6 17.6
9.9 17.1 8.4 10.5
1 [Reference] 1.99 (1.80-2.21)j 1.05 (0.96-1.13) 1.20 (1.10-1.31)j
2.6 4.1 2.8 2.0
1 [Reference] 1.92 (1.63-2.26)j 1.30 (1.10-1.52)j 1.25 (1.11-1.41)j
30.1 64.1 30.3 41.5
1 [Reference] 2.04 (1.69-2.46)j 1.01 (0.85-1.19) 1.36 (1.13-1.63)j
10.0 14.6 7.5 10.8
1 [Reference] 1.78 (1.60-1.98)j 0.99 (0.91-1.08) 1.20 (1.10-1.32)j
155,884 146,759 34,177
46.1 43.4 10.1
11.8 11.3 8.4
1 [Reference] 1.25 (1.16-1.35)j 0.88 (0.81-0.96)j
3.6 2.0 2.0
1 [Reference] 0.78 (0.70-0.87)j 0.52 (0.42-0.66)j
45.2 39.2 31.8
1 [Reference] 1.03 (0.89-1.18) 0.90 (0.75-1.08)
10.8 10.9 8.0
1 [Reference] 1.34 (1.24-1.45)j 0.91 (0.83-0.99)j
203,177 97,761 35,881
60.1 28.9 10.6
10.6 13.0 9.9
1 [Reference] 1.21 (1.11-1.32)j 1.01 (0.94-1.10)
2.8 2.8 2.0
1 [Reference] 1.04 (0.91-1.18) 0.76 (0.67-0.86)j
36.9 50.4 42.7
1 [Reference] 1.31 (1.12-1.53)j 1.17 (0.99-1.39)
10.1 12.0 9.2
1 [Reference] 1.19 (1.10-1.29)j 1.01 (0.93-1.10)
n
b
MAYO CLINIC PROCEEDINGS
http://dx.doi.org/10.1016/j.mayocp.2016.12.022 www.mayoclinicproceedings.org
AOR ¼ adjusted odds ratio; CCS ¼ Clinical Classifications Software; DAMA ¼ discharge against medical advice; MH/SA ¼ mental health/substance abuse; NA ¼ not applicable; NH ¼ non-Hispanic. All models adjusted for all variables listed in the table. c The entire study population, consisting of all inpatient hospitalizations during the study period, 2002-2011. d All other non-MH/SA or nonmaternal principal diagnoses (eg, infections, cancer, and injuries). e Total number of discharges within the respective group/column for all study years, 2002-2011. f Number of hospitalizations that ended in a DAMA across all study years, 2002-2011. g Calculated by determining the total number of inpatient hospitalizations ending in a DAMA (2002-2011) and dividing by the number of study years (10). h Percentages may not add to 100% for a given characteristic because of missing data. i Weighted DAMA prevalence rate per 1000 discharges. j P < .05. k Defined as non-Hispanic ethnicity, and race reported as Asian or Pacific Islander, Native American, or other/unknown race. l Defined by median household income for patient’s zip code. m Categorized as government (Medicare, Medicaid); private (commercial carriers, health maintenance organizations [HMOs], preferred provider organizations [PPOs]); and other (self-pay, no charge, ie, uninsured). n Refers to the All Patient Refined Diagnosis Related Group severity of illness subclasses. a
DISCHARGE AGAINST MEDICAL ADVICE IN THE UNITED STATES
the United States (an average of 338,133 per year), corresponding to a rate of 11.1 per 1,000 discharges. Rates for DAMA (per 1000 discharges) were highest among the uninsured (29.9), the 40 to 49 year age group (23.0), non-Hispanic (NH)-blacks (20.1), males (17.3), and northeast region hospitals (17.1) (Table). Figure 1 highlights the most common principal diagnoses among AMA discharges, based on both frequency and rate of DAMA.
identify patients at highest DAMA risk, we ranked principal diagnoses according to condition-specific DAMA rates. The JPR program was applied to examine temporal trends30 in DAMA rates. The JPR program first fits annual rate data to a straight line (null model with no joinpoints), followed by a Monte-Carlo permutation test to determine whether the addition of 1 or more joinpoints offers a statistically significantly better model than the null; an annual percent change and average annual percent change are computed to describe the rate changes during the study period.31 All the aforementioned descriptive, regression, and trend analyses were also conducted for each diagnostic subgroup.
Predictors Adjusted multivariable analyses on the full cohort revealed several patient- and hospitallevel characteristics as predictors of DAMA (Table). Compared with patients aged 18 to 29 years, we observed an increased odds of DAMA among patients aged 30 to 49 years and a decreasing likelihood of DAMA with increasing age among those 50 years or older. Women were 58% (95% CI, 0.41-0.42) less
RESULTS Between January 1, 2002, and December 31, 2011, approximately 3.4 million hospitalized men and women were discharged AMA in
Rank
Rank
240.5
1 2
238.7
2 1
Principal admission diagnosis 101.7 120.0
3
212.8
Mood disorders
20.4
5
119.0
Pregnancy-related, overall
2.7
112.5
Diabetes mellitus with complications
24.3
105.5
Substance-related disorders Nonspecific chest pain
28.9
4
139.5
Alcohol-related disorders
Skin and subcutaneous tissue infections
21.5
97.1
10.1
Congestive heart failure; nonhypertensive
94.9
9.9
Pneumonia (except that caused by TB or STD)
83.5
Pancreatic disorders (not diabetes)
30.2 4
62.2
49.1
3
34.5
59.9
Poisoning by other medications and drugs HIV infection
42.6
Poisoning by psychotropic agents
41.3
Sickle cell anemia
13.8
39.9
Screening/history of mental health, substance abuse
11.3
37.7
Liver disease; alcohol-related
38.1
Poisoning by nonmedicinal substances
30.9 24.8
6.4 1.0 5 Frequency of DAMA (in thousands)
49.0
Administrative/social admission
DAMA prevalence rate per 1000
FIGURE 1. Top ranked principal CCS diagnoses among AMA discharges by frequency and prevalence, 2002-2011. AMA ¼ against medical advice; CCS ¼ Clinical Classifications Software; DAMA ¼ discharge against medical advice; HIV ¼ human immunodeficiency virus; STD ¼ sexually transmitted disease; TB ¼ tuberculosis.
Mayo Clin Proc. n XXX 2017;nn(n):1-11 www.mayoclinicproceedings.org
n
http://dx.doi.org/10.1016/j.mayocp.2016.12.022
5
MAYO CLINIC PROCEEDINGS
likely to leave AMA than were men. In the unadjusted model, being of NH-black, Hispanic, and NH-other racial/ethnic minority was observed to be associated with increased odds of DAMA (data not shown). However, after adjustment, compared with NH-whites, only NH-blacks were significantly more likely to leave AMA (OR, 1.25; 95% CI, 1.19-1.31), and a protective association was observed for Hispanics (OR, 0.87; 95% CI, 0.80-0.95). The odds of self-discharge were 3 to 4 times higher among patients with government insurance (eg, Medicare and Medicaid) and the uninsured, respectively, than among those with private insurance. Patients’ household income was inversely associated with increased odds of DAMA. Compared with patients with minor severity of illness, those with moderate severity of illness were more likely to leave AMA (OR, 1.14; 95% CI, 1.11-1.17); however, decreased odds were observed among those with major and extreme severity of illness. Weekend hospital admission, compared with a weekday, was associated with an increased odds of DAMA. Hospital-level predictors included location in the Northeastern and Western US regions, urban hospitals (teaching and nonteaching status), and medium bed size. Stratified analyses on each diagnostic subgroup revealed similar predictors to those identified in the full cohort analysis (Table). Temporal Trends During the 10-year study period, we observed a 1.9% (95% CI, 0.8-3.0) average annual increase in the overall rate of DAMA (Supplemental Table, available online at http://www. mayoclinicproceedings.org). Subgroup analyses revealed statistically significant trends in DAMA rates, with a 2.9% average annual increase among the “other” subgroup, 2.3% average annual decrease in the MH/SA subgroup, and no statistically significant trend in the maternal subgroup (Figure 2). DISCUSSION We examined a nationally representative sample of inpatient hospitalizations to provide a comprehensive description of the prevalence, frequency, temporal trends, and characteristics associated with DAMA. Within the 10-year study period, on average, approximately 1.1% of all hospitalizations were discharged 6
Mayo Clin Proc.
n
AMA, with a 1.9% average annual increase in overall DAMA rates across the study years. Consistent with other national studies, we observed DAMA rates to be highest among patients who were younger, male, NH-black, low income, uninsured, Medicare/Medicaid recipients, and hospitalized for MH/SA disorders.10,11,15,19,20 We also observed high rates of self-discharge in urban hospitals and the Northeast region of the United States.10,11,19,22 Collectively, these findings highlight the intricacy of DAMA as it exists in general inpatient settings. Although a patient’s decision to leave AMA is likely multifaceted, evidence suggests that this choice generally arises from communication breakdowns,32-34 which tend to be rooted in a variety of provider- and patient-level characteristics.35,36 For instance, several patient factors known to influence perceptions of communication quality, including age, sex, race/ethnicity, income, and insurance status,37 were also identified as DAMA predictors. Although these factors are not modifiable in a clinical encounter, they serve as indicators for vulnerable patients who may require enhanced communication and other DAMA prevention strategies. Analogous to previous research, younger age was identified as a risk factor for DAMA.10,20,38 The high odds observed in the 30 to 49 year age group may reflect a necessary departure to attend to responsibilities34 typical of early adulthood. For example, the increased odds of DAMA among women of advanced maternal age (40-49 years) may be a consequence of challenges presented by a prolonged stay (eg, occupational, financial, or familial obligations). In addition, although we lack parity data on the women in our study, previous research purports an increased risk of DAMA with increasing parity.26 Likewise, males’ increased likelihood of DAMA may also be attributed to similar obligations.34 High DAMA rates among men may also be linked to the notion that men’s health behaviors (eg, propensity to not seek or continue care) are often predicted by their embodiment of masculinity and social norms.39-43 However, research that specifically examines sex differences in reasoning for leaving AMA is needed to further explicate these findings. Even after adjustment for confounders, NH-blacks in our study were more likely
XXX 2017;nn(n):1-11
n
http://dx.doi.org/10.1016/j.mayocp.2016.12.022 www.mayoclinicproceedings.org
DISCHARGE AGAINST MEDICAL ADVICE IN THE UNITED STATES
Mayo Clin Proc. n XXX 2017;nn(n):1-11 www.mayoclinicproceedings.org
n
50.0 45.0 Prevalence per 1000 discharges
than NH-whites to self-discharge. Although this is in agreement with some studies,10,15 reports of the association between DAMA and race/ethnicity are inconsistent.20,34 However, given that race is often a proxy measure for socioeconomic status, educational attainment, and health insurance coverage,44 it is possible that NH-blacks’ increased DAMA risk is partially due to residual confounding after incomplete adjustment for factors related to socioeconomic status. For example, uninsured and Medicare/Medicaid hospital stays accounted for a large proportion of patients discharged AMA,10,11,20 suggesting that concerns related to health services payment increases the risk of self-discharge. It is also probable that NH-blacks are more likely to leave AMA because of receipt of differential in-hospital care, historical race-related/ race-fueled mistrust, or perceptions of discrimination.10,45 Conversely, racial/ethnic minority status is not always an adverse risk factor for DAMA. Our observations of a protective effect against DAMA among Hispanics may be related to having increased social support,20 or cultural factors, such as being more accepting of health care providers as authority figures.10,46 Furthermore, among discharges with MH/SA diagnoses, being of NH-black race/ethnicity also had a protective effect, yielding a 13% lower odds of being discharged AMA than NH-whites. Despite being contradictory to findings for the full cohort, this observance may be a reflection of NH-blacks’ increased likelihood of initiating mental health and substance abuse services at later and more severe stages of the illness,47 subsequently making DAMA less likely. Although our findings are unable to elucidate the pathways by which race/ethnicity independently influences DAMA risk, they underscore the need for more race/ethnicity-specific quantitative and qualitative DAMA investigations. High DAMA rates were also observed among patients with a primary diagnosis of a MH/SA disorder. In corroboration with other studies, these findings are likely related to patients’ underlying psychological issues, addictive behaviors (ie, drug seeking), mistrust of providers, and perceived prejudice in delivery of care.34,38 They may also be partly explained by inadequate or lack of assessment
MH/SA02-11 = –2.3 (–3.8 to –0.8)
40.0 15.0 12.0
All02-11 = 1.9 (0.8 to 3.0) Other06-11 = 0.4 (–1.9 to 2.8)
9.0 Other02-06 = 6.0 (2.5 to 9.7)
6.0 3.0 Maternal02-11 = 1.9 (–0.4 to 4.3)
0.0 2002
2003
2004
2005
2007 2006 Year
2008
2009
2010
2011
FIGURE 2. Temporal trends in DAMA prevalence rates for principal CCS diagnostic groups, 2002-2011. All ¼ all discharges; CCS ¼ Clinical Classifications Software; DAMA ¼ discharge against medical advice; Maternal ¼ discharges with maternal-related diagnoses; MH/SA ¼ discharges with mental health/substance abuse-related principal diagnoses; Other ¼ discharges with non-MH/SA or maternal-related diagnoses. Markers indicate the observed annual rate, and solid lines represent the trend estimated by joinpoint regression. Values encased in boxes represent average annual percent change (AAPC), point estimate (95% CI); subscripted numbers represent the year range for AAPC estimates.
of patients’ decision-making capacity. Patients presenting with MH/SA disorders may have a diminished capacity to comprehend the gravity of the decision to leave AMA, and often make this decision hastily.48 Thus, policies, or closer adherence to guidelines that require a full assessment of the decision-making capacity for patients with MH/SA, may decrease the likelihood of this high-risk population from leaving AMA. In fact, the statistically significant 2.3% downward annual trend observed in the MH/SA population’s rates may have been influenced by a growing awareness of MH/SA patients’ vulnerability for DAMA and successful implementation of preventive strategies.38,49,50 In the general study population, we observed an inverse relationship with DAMA among patients with major and extreme severity of illness. Yet, in the maternal population, severity of illness was directly associated with a higher likelihood of DAMA. Although these findings could be construed as an anomaly, research suggests that pregnancy-related complications, combined with other
http://dx.doi.org/10.1016/j.mayocp.2016.12.022
7
MAYO CLINIC PROCEEDINGS
comorbidities, may unexpectedly prolong length of stay, thereby increasing DAMA risk.25 Together, these results suggest that the link between severity of illness and DAMA is complex and disease/conditiondependent. Hospital-level characteristics such as region, size, and teaching status were also found to predict DAMA. Consistent with other AMA studies, rates were highest in the northeast11,19 and medium-sized hospitals.10,23 The reasons for the observed regional differences are unclear and warrant additional research to investigate this phenomenon. Higher DAMA rates in hospitals of medium or large bed size could possibly be related to low patient satisfaction or limited provider attentiveness experienced by some patients in hospitals of larger sizes.51 Similarly, inadequate staffing or reduced continuity in care may also contribute to the increased likelihood of DAMA occurring among patients with weekend admissions. In comparison to rural hospitals, we observed increased odds of DAMA among inpatient stays at urban hospitals of both teaching and nonteaching status. Among the general study population receiving care in urban settings, the adjusted odds of DAMA tended to be lower in teaching versus nonteaching hospitals. This could be due to high specialization of teaching hospitals, or a consequence of being staffed by providers who embody patient-centeredness. The patient-centered approach to care not only respects the patient’s autonomy and provider’s recommendations but also supports good provider-patient communication52,53 and encourages shared decision-making54da process that offers the optimal opportunity to avoid DAMA.55 In the same vein, the low DAMA prevalence in rural hospitals may be due to increased familiarity and trust between patients and providers.56 Conversely, in the maternal population, DAMA was more likely to occur in urban teaching hospitals, compared with nonteaching hospitals. These findings may reflect previous hospitalizations marred by subpar health care quality, or a preference to continue treatment at home.57 Altogether, the above-mentioned observations offer additional support to the idea that DAMA rates 8
Mayo Clin Proc.
n
are reflective of measures of hospital quality,58 as well as patients’ perceptions of the quality of care10,57 and communication provided.37 In contrast, among the MH/SA subgroup, hospital size, location, and teaching status were not associated with risk of DAMA. It is likely that hospital characteristics’ effect on DAMA is neutralized once patients’ mental health/ substance use disorders are taken into account. Thus, more research is needed to better clarify the complex interplay between patient diagnosis, hospital teaching status, and DAMA risk. Study Limitations Despite the statistical power and precision offered by our large sample, the findings from this study should be interpreted within the context of several limitations. As expected in retrospective secondary analyses of administrative data, our results are subject to selection bias and misclassification/errors in coding. For example, mental health conditions (eg, depression) and substance abuse have been shown to be poorly identified when relying exclusively on International Classification of Diseases based codes.59,60 However, it is unlikely that these biases or errors account for our findings entirely, given the rigorous testing and quality control measures applied to NIS data to ensure their validity and reliability.61 Second, because the unit of analysis was the hospital discharge and not the patient, we were unable to discern whether multiple discharges were reflective of repeat hospitalizations for 1 person or initial hospitalizations for different patients.11 The inability to adjust for these occurrences may have resulted in DAMA-related estimates that are slightly different from those calculated from a patient-level data set with identifiers. Third, although we attempted to create subgroups on the basis of clinically meaningful categories, it is possible that patients’ primary admission diagnoses are confounding the associations observed between covariates and DAMA risk. Fourth, we recognize that the “other” diagnostic subgroup is heterogeneous. To address this heterogeneity, we ran a sensitivity analysis subdividing the “other” group by primary diagnosis into acute and chronic conditions. Our results were not
XXX 2017;nn(n):1-11
n
http://dx.doi.org/10.1016/j.mayocp.2016.12.022 www.mayoclinicproceedings.org
DISCHARGE AGAINST MEDICAL ADVICE IN THE UNITED STATES
substantively different when stratified by acute vs chronic subtypes (data not shown). In addition, given that previous studies have either excluded the maternal and MH/SA populations or studied them separately or in nonacute hospital settings, we considered it of value to report on both the overall and clinically distinct populations within a single study. Another notable limitation was that the NIS database does not capture patients’ motivation or reasons for being discharged AMA. Thus, we are unable to ascertain the impact of potential influences such as provider contributions (eg, suboptimal communication), patients’ external obligations or psychosocial characteristics (eg, perceptions and risk behaviors). These limitations aside, a notable strength of our study is that it is one of few investigations using nationally representative data to provide a comprehensive assessment of current DAMA rates, trends, and associated patient, hospital, and diagnostic predictors. CONCLUSION An increased understanding of the factors associated with patients’ decision to leave AMA is essential. It allows the opportunity to identify those at highest risk and thus, intervene early to reduce the subsequent morbidity, mortality, and additional health care costs associated with DAMA.38 Moreover, although data are lacking to explain the cause of the observed overall increasing DAMA rates, it is likely multifactorial; patient, societal, economic, changing health insurance patterns, and health care quality factors may all be influential. Furthermore, although the inability to reach consensus on the need for continued care20 is not always due to poor communication, enhancements in patient-provider communication and shared decision making have the potential to avert the occurrence of DAMA, achieve a more optimal quality of care, and reduce health disparities.52 SUPPLEMENTAL ONLINE MATERIAL Supplemental material can be found online at http://www.mayoclinicproceedings.org. Supplemental material attached to journal articles has not been edited, and the authors take responsibility for the accuracy of all data. Mayo Clin Proc. n XXX 2017;nn(n):1-11 www.mayoclinicproceedings.org
n
Abbreviations and Acronyms: AMA = against medical advice; DAMA = discharge against medical advice; HCUP = Healthcare Cost and Utilization Project; JPR = Joinpoint Regression; MH/SA = mental health/substance abuse; NEOMAT = Neonatal-Maternal; NH = non-Hispanic; NIS = Nationwide Inpatient Sample; OR = odds ratio Grant Support: This research was supported by grant T32 HP10031 from the Health Resources and Services Administration, an agency of the US Department of Health & Human Services. The funding source/study sponsor had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication. Correspondence: Address to Kiara K. Spooner, DrPH, MPH, Department of Family and Community Medicine, Baylor College of Medicine, 3701 Kirby Dr, Ste 600, Houston, TX 77098 ([email protected]).
REFERENCES 1. The Joint Commission. Discharge Disposition. Specifications Manual for Joint Commission National Quality Measures (Version 2016A). https://manual.jointcommission.org/releases/ TJC2016A/DataElem0537.html. Accessed November 28, 2016. 2. Hadadi A, Khashayar P, Karbakhsh M, Vasheghani Farahani A. Discharge against medical advice from a Tehran emergency department. Int J Health Care Qual Assur. 2016; 29(1):24-32. 3. Jimoh BM, Anthonia OC, Chinwe I, et al. Prospective evaluation of cases of discharge against medical advice in Abuja, Nigeria. Sci World J. 2015;2015:314817. 4. Jeong J, Song KJ, Kim YJ, et al. The association between acute alcohol consumption and discharge against medical advice of injured patients in the ED. Am J Emerg Med. 2016;34(3): 464-468. 5. Saia M, Buja A, Mantoan D, et al. Frequency and trends of hospital discharges against medical advice (DAMA) in a large administrative database. Ann Ist Super Sanita. 2014;50(4): 357-362. 6. Carron PN, Yersin B, Trueb L, Gonin P, Hugli O. Missed opportunities: evolution of patients leaving without being seen or against medical advice during a six-year period in a Swiss tertiary hospital emergency department. Biomed Res Int. 2014; 2014:690368. 7. Yong TY, Fok JS, Hakendorf P, Ben-Tovim D, Thompson CH, Li JY. Characteristics and outcomes of discharges against medical advice among hospitalised patients. Intern Med J. 2013; 43(7):798-802. 8. Kraut A, Fransoo R, Olafson K, Ramsey CD, Yogendran M, Garland A. A population-based analysis of leaving the hospital against medical advice: incidence and associated variables. BMC Health Serv Res. 2013;13:415. 9. Valevski A, Zalsman G, Tsafrir S, Lipschitz-Elhawi R, Weizman A, Shohat T. Rate of readmission and mortality risks of schizophrenia patients who were discharged against medical advice. Eur Psychiatry. 2012;27(7):496-499. 10. Ibrahim SA, Kwoh CK, Krishnan E. Factors associated with patients who leave acute-care hospitals against medical advice. Am J Public Health. 2007;97(12):2204-2208. 11. Stranges E, Wier L, Merrill CT, Steiner C. Hospitalizations in which patients leave the hospital against medical Advice (AMA), 2007. HCUP Statistical Brief #78. 2009. http://www. hcup-us.ahrq.gov/reports/statbriefs/statbriefs.jsp. Accessed January 1, 2016.
http://dx.doi.org/10.1016/j.mayocp.2016.12.022
9
MAYO CLINIC PROCEEDINGS
12. Southern WN, Nahvi S, Arnsten JH. Increased risk of mortality and readmission among patients discharged against medical advice. Am J Med. 2012;125(6):594-602. 13. Fiscella K, Meldrum S, Barnett S. Hospital discharge against advice after myocardial infarction: deaths and readmissions. Am J Med. 2007;120(12):1047-1053. 14. Glasgow JM, Vaughn-Sarrazin M, Kaboli PJ. Leaving against medical advice (AMA): risk of 30-day mortality and hospital readmission. J Gen Intern Med. 2010;25(9):926-929. 15. Olufajo OA, Metcalfe D, Yorkgitis BK, et al. Whatever happens to trauma patients who leave against medical advice? Am J Surg. 2016;211(4):677-683. 16. Weingart SN, Davis RB, Phillips RS. Patients discharged against medical advice from a general medicine service. J Gen Intern Med. 1998;13(8):568-571. 17. Saitz R, Ghali WA, Moskowitz MA. The impact of leaving against medical advice on hospital resource utilization. J Gen Intern Med. 2000;15(2):103-107. 18. Carrese JA. Refusal of care: patients’ well-being and physicians’ ethical obligations: “but doctor, I want to go home”. JAMA. 2006;296(6):691-695. 19. Kaplan GG, Panaccione R, Hubbard JN, et al. Inflammatory bowel disease patients who leave hospital against medical advice: predictors and temporal trends. Inflamm Bowel Dis. 2009;15(6):845-851. 20. Franks P, Meldrum S, Fiscella K. Discharges against medical advice: are race/ethnicity predictors? J Gen Intern Med. 2006; 21(9):955-960. 21. Brook M, Hilty DM, Liu W, Hu R, Frye MA. Discharge against medical advice from inpatient psychiatric treatment: a literature review. Psychiatr Serv. 2006;57(8):1192-1198. 22. Sclar DA, Robison LM. Hospital admission for schizophrenia and discharge against medical advice in the United States. Prim Care Companion J Clin Psychiatry. 2010;12(2):e1-e6. pii: PCC.09m00827. doi: 10.4088/PCC.09m00827yel. 23. Tawk R, Freels S, Mullner R. Associations of mental, and medical illnesses with against medical advice discharges: the National Hospital Discharge Survey, 1988-2006. Adm Policy Ment Health. 2013;40(2):124-132. 24. Letterie GS, Markenson GR, Markenson MM. Discharge against medical advice in an obstetric unit. J Reprod Med. 1993;38(5): 370-374. 25. Fiscella K, Meldrum S, Franks P. Post partum discharge against medical advice: who leaves and does it matter? Matern Child Health J. 2007;11(5):431-436. 26. Tucker Edmonds B, Ahlberg C, McPherson K, Srinivas S, Lorch S. Predictors and adverse pregnancy outcomes associated with antepartum discharge against medical advice. Matern Child Health J. 2014;18(3):640-647. 27. Healthcare Cost and Utilization Project (HCUP). Introduction to the HCUP Nationwide Inpatient Sample (NIS) 2011. November 2015. http://www.hcup-us.ahrq.gov/db/nation/nis/ NIS_Introduction_2011.pdf. Accessed January 1, 2016. 28. Houchens R, Elixhauser A. Final report on calculating Nationwide Inpatient Sample (NIS) variances for data years 2011 and earlier. HCUP Methods Series Report # 2003-02 online. December 14, 2015. http://www.hcup-us.ahrq.gov/reports/ methods/methods.jsp. Accessed January 1, 2016. 29. Elixhauser A, Steiner C, Palmer L. Clinical Classifications Software (CCS), 2015. 2015. http://www.hcup-us.ahrq.gov/toolssoftware/ ccs/ccs.jsp. Accessed February 13, 2017. 30. National Cancer Institute. Joinpoint Regression Program, Version 4.1.1.3, Statistical Methodology and Applications Branch, Surveillance Research Program. 2016. http://surveillance.cancer.gov/ joinpoint/. Accessed January 4, 2015. 31. Kim HJ, Fay MP, Feuer EJ, Midthune DN. Permutation tests for joinpoint regression with applications to cancer rates. Stat Med. 2000;19(3):335-351. 32. Windish DM, Ratanawongsa N. Providers’ perceptions of relationships and professional roles when caring for patients who
10
Mayo Clin Proc.
n
33.
34.
35.
36.
37.
38. 39.
40.
41.
42.
43.
44.
45.
46.
47.
48.
49.
50.
51.
52.
53.
leave the hospital against medical advice. J Gen Intern Med. 2008;23(10):1698-1707. Onukwugha E, Saunders E, Mullins CD, et al. A qualitative study to identify reasons for discharges against medical advice in the cardiovascular setting. BMJ Open. 2012;2(4). pii: e000902. doi: 10.1136/bmjopen-2012-000902. Onukwugha E, Saunders E, Mullins CD, Pradel FG, Zuckerman M, Weir MR. Reasons for discharges against medical advice: a qualitative study. Qual Saf Health Care. 2010; 19(5):420-424. Cooper LA, Roter DL, Johnson RL, Ford DE, Steinwachs DM, Powe NR. Patient-centered communication, ratings of care, and concordance of patient and physician race. Ann Intern Med. 2003;139(11):907-915. Street RL Jr, Gordon H, Haidet P. Physicians’ communication and perceptions of patients: is it how they look, how they talk, or is it just the doctor? Soc Sci Med. 2007;65(3): 586-598. Spooner KK, Salemi JL, Salihu HM, Zoorob RJ. Disparities in perceived patient-provider communication quality in the United States: trends and correlates. Patient Educ Couns. 2016;99(5):844-854. Alfandre DJ. “I’m going home”: discharges against medical advice. Mayo Clin Proc. 2009;84(3):255-260. Mahalik JR, Burns SM, Syzdek M. Masculinity and perceived normative health behaviors as predictors of men’s health behaviors. Soc Sci Med. 2007;64(11):2201-2209. Bertakis KD, Azari R, Helms LJ, Callahan EJ, Robbins JA. Gender differences in the utilization of health care services. J Fam Pract. 2000;49(2):147-152. Courtenay WH. Constructions of masculinity and their influence on men’s well-being: a theory of gender and health. Soc Sci Med. 2000;50(10):1385-1401. Grzywacz JG, Stoller EP, Brewer-Lowry AN, Bell RA, Quandt SA, Arcury TA. Gender and health lifestyle: an indepth exploration of self-care activities in later life. Health Educ Behav. 2012;39(3):332-340. O’Brien R, Hunt K, Hart G. ‘It’s caveman stuff, but that is to a certain extent how guys still operate’: men’s accounts of masculinity and help seeking. Soc Sci Med. 2005;61(3):503-516. Kawachi I, Daniels N, Robinson DE. Health disparities by race and class: why both matter. Health Aff (Millwood). 2005;24(2): 343-352. Blanchard J, Lurie N. R-E-S-P-E-C-T: patient reports of disrespect in the health care setting and its impact on care. J Fam Pract. 2004;53(9):721-730. Flores G. Culture and the patient-physician relationship: achieving cultural competency in health care. J Pediatr. 2000; 136(1):14-23. Cook BL, Zuvekas SH, Carson N, Wayne GF, Vesper A, McGuire TG. Assessing racial/ethnic disparities in treatment across episodes of mental health care. Health Serv Res. 2014; 49(1):206-229. Gerbasi JB, Simon RI. Patients’ rights and psychiatrists’ duties: discharging patients against medical advice. Harv Rev Psychiatry. 2003;11(6):333-343. Holden P, Vogtsberger KN, Mohl PC, Fuller DS. Patients who leave the hospital against medical advice: the role of the psychiatric consultant. Psychosomatics. 1989;30(4):396-404. Targum SD, Capodanno AE, Hoffman HA, Foudraine C. An intervention to reduce the rate of hospital discharges against medical advice. Am J Psychiatry. 1982;139(5):657-659. Tajeu GS, Kazley AS, Menachemi N. Lower satisfaction among hospitals that disproportionately treat black patients. J Health Hum Serv Adm. 2015;38(3):290-316. Institute of Medicine. Crossing the Quality Chasm: A New Health System for the 21st Century. Washington, DC: National Academy Press; 2001. Institute of Medicine. Envisioning the National Health Care Quality Report. Washington, DC: National Academy Press; 2001.
XXX 2017;nn(n):1-11
n
http://dx.doi.org/10.1016/j.mayocp.2016.12.022 www.mayoclinicproceedings.org
DISCHARGE AGAINST MEDICAL ADVICE IN THE UNITED STATES
54. Barry MJ, Edgman-Levitan S. Shared decision makingepinnacle of patient-centered care. N Engl J Med. 2012;366(9):780-781. 55. Alfandre D. Reconsidering against medical advice discharges: embracing patient-centeredness to promote high quality care and a renewed research agenda. J Gen Intern Med. 2013; 28(12):1657-1662. 56. Seaborn Moyse H, Osmun WE. Discharges against medical advice: a community hospital’s experience. Can J Rural Med. 2004;9(3):148-153. 57. Tabatabaei SM, Sargazi Moakhar Z, Behmanesh Pour F, Shaare Mollashahi S, Zaboli M. Hospitalized pregnant women who leave against medical advice: attributes and reasons. Matern Child Health J. 2016;20(1):128-138.
Mayo Clin Proc. n XXX 2017;nn(n):1-11 www.mayoclinicproceedings.org
n
58. Onukwugha EC, Shaya FT, Saunders E, Weir MR. Ethnic disparities, hospital quality, and discharges against medical advice among patients with cardiovascular disease. Ethn Dis. 2009; 19(2):172-178. 59. Fiest KM, Jette N, Quan H, et al. Systematic review and assessment of validated case definitions for depression in administrative data. BMC Psychiatry. 2014;14:289. 60. Pang JX, Ross E, Borman MA, et al. Validation of coding algorithms for the identification of patients hospitalized for alcoholic hepatitis using administrative data. BMC Gastroenterol. 2015;15:116. 61. Healthcare Cost and Utilization Project (HCUP). HCUP Quality Control Procedures. March 17, 2016. http://www.hcup-us. ahrq.gov/db/quality.jsp. Accessed June 1, 2016.
http://dx.doi.org/10.1016/j.mayocp.2016.12.022
11