a b s t r a c t

Background: heart failure readmissions are a longstanding national healthcare issue for both hospitals and patients. Our purpose was to evaluate the efficacy of a structured, educational intervention targeted towards un- and under-insured emergency department (ED) HF patients.

Methods: HF patients presenting to the ED for care were enrolled between July and December 2015 as part of an open label, interventional study, using a parallel observational control group. Eligible patients provided informed consent, had an established HF diagnosis, and were hemodynamically stable. Intervention patients received a standardized educational intervention in the ED waiting room before seeing the emergency physician, and a 30-day telephone follow-up. Primary and secondary endpoints were 30- and 90-day ED and hospital readmission rates, as well as days alive and out of hospital (DAOH) respectively.

Results: Of the 94 patients enrolled, median age was 58.4 years; 40.4% were female, and 54.3% were African Amer- ican. Intervention patients (n = 45) experienced a 47.8% and 45.3% decrease in ED revisits (P = 0.02 & P b 0.001), and 60.0% and 47.4% decrease in Hospital readmissions (P = 0.049 & P = 0.007) in the 30 and 90 days pre- versus post-intervention respectively. control patients (n = 49) had no change in hospital readmissions or 30-day ED revisits, but experienced a 36.6% increase in 90-day ED revisits (P = 0.03). Intervention patients also saw a 59.2% improvement in DAOH versus control patients (P = 0.03).

Conclusion: An ED educational intervention markedly decreases ED and hospital readmissions in un- and under- insured HF patients.

(C) 2018

Introduction

In 2016, 13.1% of U.S. adults lacked any form of health insurance [1]. Even more possessed insurance but had such high out-of-pocket costs or deductibles relative to their income, that they were defined as under-insured [2]. Un- and under-insured patients receive less preven- tive care, are diagnosed at more advanced disease stages, and have higher mortality rates than their insured counterparts, due to fewer ac- cessible primary care services [3]. These issues are readily evident in chronic conditions, such as heart failure .

In 2011, HF patients accounted for 4.8% of 30-day hospital re- admissions across all insurance groups, and 7.3% of readmissions among Medicare patients. Accordingly, reducing HF readmissions has become a national priority [4,5]. Numerous educational efforts

* Corresponding author at: 1 Baylor Plaza, Houston, TX 77030, United States.

E-mail address: [email protected] (V. Asthana).

have been tested for this purpose, including nurse-led telephone follow-up, specialized HF clinics, home visits, multidisciplinary home-based interventions, and various remote-monitoring strate- gies. While the results of these efforts are mixed, most demonstrate no appreciable change in outcomes, including readmission rates [6-16].

The American Heart Association (AHA) currently recommends some form of HF education prior to patient discharge [17-19]. However, in many cases this education is unstructured, poorly suited for the target population, and administered immediately prior to discharge when the patient may be less receptive and eager to return home. As a result, current HF educational strategies are generally considered ineffective [20,21].

It is possible that a formalized educational intervention, timed to co- incide with a teachable moment, would be more effective than previous strategies. Accordingly, the purpose of our investigation was to evaluate the impact of an ED education-based intervention on outcomes in un- and under-insured HF patients.

https://doi.org/10.1016/j.ajem.2018.03.057

0735-6757/(C) 2018

Methods

Study participants

This was an open label, intervention study, using a parallel observa- tional control group. Eligible patients with a previous diagnosis of HF were enrolled in the waiting room of the Ben Taub urban county hospi- tal ED, located in Harris County, Texas, from June to December 2015. Inclusion criteria included a pre-existing diagnosis of HF in patients pre- senting to the ED for any cause. Exclusion criteria included patients that were 1) in unstable condition requiring immediate medical attention, 2) hemodynamically unstable (heart rate N 120 beats/min, respiratory rate N 25 breaths/min, or systolic blood pressure b 90 mm Hg), 3) b 18 years old, or 4) incarcerated.

Un- and under-insured patients were defined as patients who either pay for medical services entirely out-of-pocket, or possess Medicare, Medicaid, or County health insurance but have trouble paying out-of- pocket costs. Because the majority of patients in the Harris Health net- work are low-income individuals (200% below the federal poverty line), those patients lacking private insurance were considered un- or under-insured [22].

Study design

Study participants were identified using an Electronic Medical Re- cord (EMR)-generated report of patients waiting in the ED with a previ- ous diagnosis of HF (code ICD-9, keyword: Heart Failure). Report queries were generated upon availability of study personnel throughout the day. Patients listed in the report were randomly arranged, and screened for eligibility in the subsequent order that they appeared. Pa- tients of the intervention cohort were enrolled first, seen by one of two medical student volunteers, and provided personalized education regarding their condition, pre- and post-testing of their HF knowledge, and ED standard of care during the index visit. Patients were then contacted via telephone 30 days later for re-testing and reinforcement of previously learned material. Following completion of intervention group enrollment, control patients were selected over the same enroll- ment period as the intervention group. They received only ED standard of care and did not interact with study personnel. The consent process was waived for the control group from whom only de-identified data was obtained via EMRs. The study was single subject in design, with pa- tient data prior to the index visit serving as a reference baseline in both groups. The protocol was approved by the institutional review board of Baylor College of Medicine.

Experimental arms

Patients in the interventional arm were provided informed consent and a questionnaire. A pre-intervention quiz was then administered to probe patients’ understanding of their condition and self-management strategies (Supplementary text 1). Standardized information on the eti- ology of HF, mechanisms and side effects of medications prescribed, strategies for minimizing exacerbations, and warning signs of worsen- ing clinical status were then presented and provided as a take home packet (Supplementary text 2). Though the presented information cov- ered AHA guidelines, core material was simplified with pictorial repre- sentations to allow for easier comprehension by individuals with at least a 5th grade level of educational attainment. Additionally, the entire intervention was conducted orally to eliminate literacy as a barrier with translators available on site as needed. The pre-intervention quiz was then re-administered as a post-test to gauge patient’s understanding and retention, with correct answers provided at this time. Patients were then released to continue ED standard of care with no further index visit interactions.

Thirty days after the index visit, the same volunteer staff administered a follow-up questionnaire and repeat quiz by phone (Supplementary text

3). Educational reinforcement was also performed. Overall, volunteers spent approximately 1 h during the index visit and 30 min during tele- phone follow-up with each patient.

Patients in the control arm received only ED standard of care and were retrospectively followed via their EMR.

Outcome measures

Primary outcome measures were defined as the change in number of HF-specific ED revisits and hospital readmissions over 30- and 90-days, pre- versus post-index visit. Days alive and out of hospital (DAOH) over 365 days was a secondary outcome measure. Only HF ED revisits and re- hospitalizations, ascertained by chart review, were tallied. Data on out- come measures including mortality, number of ED revisits, and hospital- izations, as well as HF medications, and lab data in the 30- and 90-day periods before and after the index ED visit were collected by chart re- view in both groups.

Ben Taub Hospital is part of the Harris Health network and utilizes the Harris Health EMR system, which has access to all county facilities in the greater Houston area. Accordingly, patients presenting to alter- nate county sites in the region would be picked up via chart review.

Statistical analysis

ED revisits, hospital readmissions, and total length of hospitalization, were compared in the 30- and 90-days pre- and post-index visit within the intervention and control group using the non-parametric Wilcoxon Signed Rank Test. Stratified analysis with nonparametric covariable ad- justment (SANON) was used to compare ED and hospital readmissions at 30- and 90-days post-index visit between intervention and controls by adjusting for baseline deviations between the two arms [23]. DAOH post-index visit were compared between both groups using the Wilcoxon rank-sum test. A Kaplan-Meier time-to-event function was used to calculate the 365-day survival curve for HF-specific ED or hospi- tal revisits, and death from any cause. Event rate, absolute risk reduction (ARR), relative risk reduction (RRR), and number needed to treat (NTT) were determined by converting the numerical 30- and 90-day revisit values into binary data. Pearson’s chi-squared test was used to deter- mine statistical independence for all binary and categorical variables. To evaluate data collection bias between un-blinded study personnel, a blinded re-accession of chart data was performed in 15% of all patients upon the completion of data collection. Inter-rater reliability was calcu- lated using Cohen’s kappa [24-26].

The sample size was calculated using a significance level (?) of 5% and a power (1-?) of 80% and is based on an estimate of the relative spread of emergency room visits and hospitalizations among HF patients. Assuming a primary endpoint rate of 50% in the standard of care group and 10% loss to follow-up, we calculated that 45 patients would need to be enrolled per experimental arm for paired analysis. All statistical cal- culations were performed using JMP Pro 11 statistical software.

Results

Demographics and baseline characteristics are presented in Table 1. Of 94 enrolled patients, 92 were un- or under-insured. Median (IQR) age was 58 (53-64) years, with 40.4% female, and 54.3% African Amer- ican. There were 45 intervention and 49 control patients. Both groups had similar demographics, lab results, and ejection fractions, as well as number of ED visits and hospital admissions in the 30 and 90 days be- fore the index visit [27]. Medications as well as number of patients see- ing a primary care doctor/cardiologist, both of which are indicators of treatment adequacy, were similar between both groups. B-type natri- uretic (BNP), a marker of heart failure severity, was also similar between both groups [28]. Education level, perceived Patient adherence, and quiz scores during the index visit and on follow-up in the intervention cohort are reported in Supplementary Table 1. No patients were lost to follow-

Table 1

Patient demographics and baseline characteristics by treatment group.

	Intervention	Control
	(n = 45)b	(n = 49)e
Age	59 +- 10.3	57.8 +- 13.5
Male gender — no. (%) Race — no. (%)	27 (60)	29 (59.2)

Table 2

Change in revisit numbers pre- vs post-index visit.

ED revisits

30 Day

Intervention Control

Insurance status — no. (%)

African American	24 (53.3)	27 (55.1)
Hispanic	15 (33.3)	19 (38.8)
Caucasian	4 (8.9)	2 (4.1)

Private	1 (2.2)	1 (2)
Medicare	13 (28.9)	8 (16.3)
Medicaid	13 (28.9)	20 (40.8)
County	12 (26.7)	15 (30.6)
Uninsured	6 (13.3)	5 (10.2)

Comorbidities — no. (%)

Hypertension	37 (82.2)	41 (83.7)
Hyperlipidemia	22 (48.9)	18 (36.7)
Diabetes mellitus	23 (51.1)	22 (44.9)
Chronic kidney disease	14 (31.1)	19 (38.8)
Chronic obstructive pulmonary disease	6 (13.3)	6 (12.2)
Coronary artery disease	18 (40)	12 (24.5)

Medications — no. (%)

Furosemide	37 (82.2)	44 (89.8)
Any ?-blocker	36 (80)	42 (85.7)
Any ACE inhibitor and/or ARBa	28 (62.2)	31 (63.3)

Pre 0.51 0.69

Post 0.27 0.82

Pre vs Post P-value^b,d 0.02 0.39

SANON P-value^a,c 0.015

90 Day

Pre 1.42 1.45

Post 0.78 1.98

Pre vs Post P-value b0.001 0.03

SANON P-value 0.002

Hospital readmissions

30 Day

Pre 0.33 0.24

Post 0.13 0.27

Pre vs Post P-value 0.049 1

SANON P-value 0.11

90 Day

Pre 0.84 0.67

Post 0.44 0.69

Pre vs Post P-value 0.007 0.66

SANON P-value 0.021

Total length of hospitalization

30 Day

Pre 1.64 0.78

Post 0.56 0.94

Pre vs Post P-value 0.04 0.97

SANON P-value 0.11

90 Day

Pre 3.51 2.14

Post 2.04 2.96

Pre vs Post P-value 0.01 0.57

SANON P-value 0.014

Emergency department baseline revisits^d

Seeing a primary care physician — no. (%)	41 (91.1)	45 (91.8)
Seeing a cardiologist — no. (%)	23 (51.1)	29 (59.2)
BMI — kg/m2	33 +- 11.8	34 +- 10.4
Troponin I — ug/L	0.053 +- 0.09	0.19 +- 1.0
Left ventricular ejection fraction — % B-type natriuretic peptide — no. (%)c	39.7 +- 18.4	35.8 +- 21.5
b230 pg/mL	13 (36.1)	23 (48.9)
230-480 pg/mL	7 (19.4)	3 (6.4)
N480 pg/mL	16 (44.6)	21 (44.7)

30 days prior	0.51 +- 1.14	0.69 +- 1.84
90 days prior	1.42 +- 2.75	1.45 +- 3.1
Hospital baseline readmissions 30 days prior	0.33 +- 0.48	0.24 +- 0.63
90 days prior	0.84 +- 0.98	0.67 +- 1.26

^a Abbreviations: ACE, angiotensin-converting enzyme; ARB, angiotensin receptor blocker; BMI, body mass index.

^b Pearson’s chi-squared test was used for categorical variables; two-sample t-test was used for continuous variables.

^c B-type natriuretic peptide levels were broken down into risk categories based on ci- tation [35].

^d ED baseline revisits and hospital baseline readmissions reflect the average number of revisits/readmissions per patient during the specified interval.

^e All comparisons between experimental arms were non-significant (P N 0.05).

up, with vital status confirmed via the medical records at one year fol- lowing the index visit.

Primary outcome data is listed in Table 2. Compared to pre-interven- tion rates, the interventional arm had a 47.8% and 45.3% reduction in ED revisits in the 30 and 90 days post-intervention (P = 0.02 & P b 0.001 respectively). This is compared to controls who experienced a 36.6% in- crease in ED revisits over 90 days (P = 0.03). Finally, there was a 17.6% trend towards increasing 30 day ED revisits in control patients that was not found to be statistically significant (P = 0.39).

The intervention also decreased hospital readmissions by 60.0% and

47.4% in the 30 and 90 days pre- versus post-intervention respectively (P = 0.049 & P = 0.007). This is compared to controls whose hospital readmission values were unchanged over 30 and 90 days (P = 1.0 & P = 0.66). Overall, the interventional arm spent less time hospitalized over the 30- and 90-day period post- versus pre-intervention, corre- sponding to a 65.8% and 41.9% reduction respectively (P = 0.04 & P = 0.01). A corresponding change was not observed in the control group (P = 0.97 & P = 0.57).

When the intervention and control cohorts were compared directly after baseline adjustment using SANON, a statistically significant reduc- tion in 30- and 90-day ED revisits (P = 0.015 & P = 0.002), 90-day

^a Abbreviations: SANON, Stratified analysis with nonparametric co-variable adjustment.

^b Change in ED revisit and hospital readmission values as well as total length of hospi- talization in the pre- versus post-index period were compared using the Wilcoxon signed rank test.

^c Non-parametric baseline covariate adjustment was used to compare ED revisit, hos- pital readmission, and total length of hospitalization values between the experimental arms.

^d Statistical significance was set at P b 0.05.

hospital revisits (P = 0.021), and 90-day total length of hospitalization (P = 0.014) was found. Conversely, after adjustment, 30-day hospital readmissions as well as 30-day length of stay were no different between intervention and controls (P = 0.11 for both).

For secondary outcomes, patients in the interventional arm had a 59.2% improvement in DAOH versus the control arm (201 vs 127; P = 0.03). A Kaplan-Meir event-free survival curve was also generated using the DAOH data (Fig. 1).

By converting readmission values into binary events, absolute risk reduction (ARR), relative risk reduction (RRR), and number needed to treat (NNT) were calculated (Table 3). For 30- and 90-day ED revisits, patients experienced a clinically relevant and statistically significant ab- solute and relative risk reduction. Similar trends were seen for 30-day hospital readmissions, but did not reach significance until 90 days.

Five patients died during the study. Four in the treatment arm as a

result of worsening chronic kidney disease, cardiac arrest, sepsis, or re- spiratory arrest. One control patient died of unknown causes (P = 0.14). Lastly, inter-rater reliability was calculated and Cohen’s kappa de- rived with no evident data discrepancies (? = 1).

Discussion

Key findings

We present a previously un-explored educational intervention strat- egy designed to improve outcomes and reduce readmissions in un- and

Fig. 1. Event-free survival defined as time to first ED visit, hospitalization, or death over 365 days. For Wilcoxon comparison, P = 0.02.

under-insured patients. Our approach significantly reduced ED revisits, hospital readmissions, and total length of hospitalization over the 30- and 90-day period pre-versus post-intervention in patients presenting for any cause, thus expanding the treatable patient pool beyond individ- uals seeking care for HF only. Unlike patients in the intervention cohort, control patients remained either relatively consistent in hospital read- mission rates and total length of hospitalization, or experienced an in- crease in ED revisits.

When patients in the interventional arm were compared directly with their risk-matched control arm counterparts using SANON, the in- tervention demonstrated clinically relevant results in all metrics except change in 30-day total length of hospitalization and reduction in 30-day hospital revisits [29,30]. While education did not produce a change in 30-day number or total length of hospitalizations, it did have a discern- able effect over 90 days, suggesting that improvements in self-manage- ment may take longer to manifest as a reduction in hospitalizations.

In addition to improving primary outcomes, intervention patients spent more time alive and out of hospital than controls. The Kaplan- Meir event-free survival curve demonstrates that the majority of im- provement between the experimental arms occurs in the initial 10 to 60 days after the intervention, and then persists.

Innovation

The results of our pilot study are particularly surprising given the general notion that patients with chronic conditions (especially HF) are refractory to educational intervention and management strategies. We suspect this effect is in part the result of increased Patient engagement.

First, the target population represents a group that can benefit im- mensely from self-management strategies, and for whom emergent

care is the only non-cost-prohibitive form of treatment available. As a result, the intervention described here represents the first time many of these patients have the opportunity to engage in their own primary care.

Second, while the intervention covers the same AHA recommended material, it is more structured, and tailored to a 5th grade comprehen- sion level. By purposefully simplifying the presented material, patients can begin to understand the physiological basis of their disease, and more actively engage in its management [31,32].

Third, the intervention presented here targets patients earlier in the pipeline-upon presentation to the ED-where long wait times can be productively utilized. Studies have shown that when captured in the ED, where patients arrive out of necessity due to an acute worsening of their condition, patients are generally more keen to learn about their disease and its management to avoid such circumstances in the fu- ture [33-35].

Lastly, the administration of pre- and post-intervention quizzes also serves to engage patients in the learning process, and functions as a type of teach-back mechanism to improve patient understanding [36]. Data from our pilot study revealed a statistically significant improvement in post-intervention quiz scores that persisted to the 30-day follow-up period, demonstrating the efficacy of this approach (Supplementary Table 1).

Cost-benefit

The results and timing of our study are especially pertinent in the setting of today’s changing healthcare landscape. Readmissions for heart failure are one of the foremost metrics being evaluated by Medi- care & Medicaid, with poor outcomes translating into reduced reim- bursements [37-40]. Thus, potentially impactful interventions, such as

Table 3 Event rate — no. (%).
	Intervention	Control	P-valuec,e	RRR (95% CI)a	ARR (95% CI)	NNT (95% CI)d
ED revisitsb
30 Day	6 (13.3)	16 (32.7)	0.027	59.2% (4.80-82.5)	19.3% (2.86-35.8)	6 (2.79-35.0)
90 Day	15 (33.3)	29 (59.2)	0.012	43.7% (9.51-64.9)	25.9% (6.38-45.3)	4 (2.21-15.7)
Hospital readmissions 30 Day	5 (11.1)	10 (20.4)	0.22	45.5% (0-79.9)	9.3% (0-23.8)	11 (4.19-19.0)
90 Day	13 (28.9)	24 (50.0)	0.046	41% (0-65.6)	20.1% (0.82-39.4)	5 (2.54-121)

^a Abbreviations: CI, confidence interval; RRR, relative risk reduction; ARR, absolute risk reduction; NNT, number needed to treat.

^b Event rate was determined by converting the numerical 30- and 90-day revisit values into binary (i.e. any revisit value greater than zero was assigned a one for subsequent calculation).

^c Pearson’s chi-squared test was used to test for significance.

^d 95% CI for all risk estimates were confined from 0 to 100%.

^e Statistical significance was set at P b 0.05.

the one described here, would be in the best interest of the patient, care provider, healthcare institution, and insurer [41]. The presented ap- proach is also appealing because it can be integrated seamlessly into the current healthcare workflow without disrupting the standard ED patient workup, and is a simple and extremely cost-effective solution to a long-standing healthcare issue.

Ultimately, in a real-life clinical setting, the intervention will require staff to be trained to deliver this type of organized education, and spend a total of 1.5 h with each patient throughout the entire intervention (1 h on index visit and 30 min during the follow-up call). This represents a relatively small time and monetary investment for the potentially dra- matic reductions in Healthcare costs afforded by the intervention.

A study by Koelling, et al. from 2005 found that a one-hour HF train- ing session with a nurse educator could reduce readmission costs by

$2823 per patient relative to standard discharge protocol [19]. Given the N50% reduction in ED and hospital readmissions observed in the present study, we can hope to see a similar if not more pronounced cost saving with generalized implementation of the proposed approach.

Limitations

The improvements in outcome measures seen with our intervention should be interpreted in light of the study’s limitations. First, this was not a true randomized controlled study, as the control group was retro- spectively selected. Despite controlling for risk, our results may suffer from potential mismatches between experimental arms. Conversely, not consenting the control group provides a real world comparator, and limits the effect of motivational biases introduced by only enrolling patients with a desire to be study participants. While, the inability to fully blind study personnel may have also introduced bias, Cohen’s kappa analysis suggests this did not occur. Additionally, though the abil- ity to meet out-of-pocket expenses when determining patients’ un- and under-insured status was not directly assessed during the study, local demographic analysis suggests that the majority of individuals in the Harris Health network are low-income and would struggle to meet copay/deductibles.

Furthermore, despite the relatively small sample size and resultantly large 95% CI on all risk estimates, the pronounced improvement in out- come measures (often N50%) and associated P-values support the ro- bust nature of our findings. Moreover, while it would be helpful to know how many patients presented to the ED with HF or cardiac com- plaints, this data was not collected. Lastly, the study was limited in that it was restricted to a single hospital site. While hospitalizations outside of the network are certainly a possibility, we feel it is unlikely to repre- sent a significant contribution as the Harris Health EMR system has ac- cess to all county facilities in one of the geographically largest counties in the US.

Future directions

The results of the current pilot study strongly challenge the general notion that HF educational strategies are ineffective in the management of patients with chronic HF. To validate our findings, a larger, multicen- ter, randomized controlled trial will be required. Additionally, it would be informative to include patients of higher socioeconomic status with private insurance to determine if our intervention is effective in all HF patients, independent of their socioeconomic status, thereby validating the broad applicability of our interventional approach.

Conclusions

Our study found that the implementation of a novel, simple, low cost educational strategy in a busy urban county hospital ED significantly de- creased ED revisits and hospital readmissions, as well as DAOH, in un- and under-insured patients with a prior diagnosis of HF. If replicated in other environments, and in a fully insured population, the ease and

low cost of this strategy supports its potential to significantly impact general HF care.

Supplementary data to this article can be found online at https://doi. org/10.1016/j.ajem.2018.03.057.

Funding sources

This research did not receive any specific grant from funding agen- cies in the public, commercial, or not-for-profit sectors.

Disclosures

WF Peacock would like to report the following industrial relation- ships: Alere (significant; research grant), Banyan (significant; research grant), Beckman (modest; consultant/advisory board), Boehringer Ingelheim Pharmaceuticals (modest; consultant/advisory board), Cardiorentis (significant; research grant and consultant/advisory board), Comprehensive Research Associates (modest; ownership inter- est), Daichi-Sankyo (modest; consultant/advisory board), Emergencies in Medicine (modest; ownership interest), Instrumentation Laborato- ries (significant research grant), Janssen (significant; research grant and consultant/advisory board), Novartis (modest; consultant/advisory board), Prevenico (modest; consultant/advisory board), Relypsa (mod- est; consultant/advisory board), Roche (significant; research grant), The Medicine’s Company (modest; consultant/advisory board), and ZS Pharma (modest; research grant). A Pritchett would to like to report an industry relationship with St. Jude (modest; ownership interest). All remaining authors have no disclosures to report.

References

1. Cohen RA, Martinez ME, Zammitti EP. Health insurance coverage: early release of es- timates from the National Health Interview Survey, 2015. National Center for Health Statistics; 2015.
2. The Commonwealth Fund. The problem of underinsurance and how rising deduct- ibles will make it worse. Available from: http://www.commonwealthfund.org/ publications/issue-briefs/2015/may/problem-of-underinsurance; 2015.
3. America’s uninsured crisis: consequences for health and health care. Washington (DC); 2009.
4. Keenan PS, Normand SL, Lin Z, et al. An administrative claims measure suitable for profiling hospital performance on the basis of 30-day all-cause readmission rates among patients with heart failure. Circ Cardiovasc Qual Outcomes 2008;1(1): 29-37. https://doi.org/10.1161/CIRCOUTCOMES.108.802686 [PubMed PMID:

   20031785].

   Krumholz HM, Merrill AR, Schone EM, et al. Patterns of hospital performance in acute myocardial infarction and heart failure 30-day mortality and readmission. Circ Cardiovasc Qual Outcomes 2009;2(5):407-13. https://doi.org/10.1161/ CIRCOUTCOMES.109.883256 [PubMed PMID: 20031870].

5. Ekman I, Andersson B, Ehnfors M, et al. Feasibility of a nurse-monitored, outpatient- care programme for elderly patients with moderate-to-severe, chronic heart failure. Eur Heart J 1998;19(8):1254-60 [PubMed PMID: 9740348].
6. McDonald K, Ledwidge M, Cahill J, et al. Elimination of early rehospitalization in a randomized, controlled trial of multidisciplinary care in a high-risk, elderly heart failure population: the potential contributions of specialist care, clinical stability and optimal angiotensin-converting enzyme inhibitor dose at discharge. Eur J Heart Fail 2001;3(2):209-15 [PubMed PMID: 11246059].
7. Harrison MB, Browne GB, Roberts J, et al. Quality of life of individuals with heart fail- ure: a randomized trial of the effectiveness of two models of hospital-to-home tran- sition. Med Care 2002;40(4):271-82 [PubMed PMID: 12021683].
8. McDonald K, Ledwidge M, Cahill J, et al. Heart failure management: multidisciplinary care has intrinsic benefit above the optimization of medical care. J Card Fail 2002;8

   (3):142-8 [PubMed PMID: 12140806].

   Riegel B, Carlson B, Kopp Z, et al. Effect of a standardized nurse case-management telephone intervention on resource use in patients with chronic heart failure. Arch Intern Med 2002;162(6):705-12 [PubMed PMID: 11911726].

9. Laramee AS, Levinsky SK, Sargent J, et al. Case management in a heterogeneous con- gestive heart failure population: a randomized controlled trial. Arch Intern Med 2003;163(7):809-17. https://doi.org/10.1001/archinte.163.7.809 [PubMed PMID: 12695272].
10. Stewart S, Pearson S, Horowitz JD. Effects of a home-based intervention among pa- tients with congestive heart failure discharged from acute hospital care. Arch Intern Med 1998;158(10):1067-72 [PubMed PMID: 9605777].
11. Stromberg A, Martensson J, Fridlund B, et al. Nurse-led heart failure clinics improve survival and self-care behaviour in patients with heart failure: results from a pro- spective, randomised trial. Eur Heart J 2003;24(11):1014-23 [PubMed PMID: 12788301].
12. Stewart S, Marley JE, Horowitz JD. Effects of a multidisciplinary, home-based inter- vention on Unplanned readmissions and survival among patients with chronic con- gestive heart failure: a randomised controlled study. Lancet 1999;354(9184): 1077-83 [PubMed PMID: 10509499].
13. Sales VL, Ashraf MS, Lella LK, et al. Utilization of trained volunteers decreases 30-day readmissions for heart failure. J Card Fail 2013;19(12):842-50. https://doi.org/10. 1016/j.cardfail.2013.10.008 [PubMed PMID: 24331204].
14. Weinberger M, Oddone EZ, Henderson WG. Does increased access to primary care reduce hospital readmissions? Veterans Affairs Cooperative Study Group on Primary Care and Hospital Readmission. N Engl J Med 1996;334(22):1441-7. https://doi.org/ 10.1056/NEJM199605303342206 [PubMed PMID: 8618584].
15. Yancy CW, Jessup M, Bozkurt B, et al. 2013 ACCF/AHA guideline for the management of heart failure: executive summary: a report of the American College of Cardiology Foundation/American Heart Association Task Force on practice guidelines. Circula- tion 2013;128(16):1810-52. https://doi.org/10.1161/CIR.0b013e31829e8807

    [PubMed PMID: 23741057].

    Phillips CO, Wright SM, Kern DE, et al. Comprehensive discharge planning with postdischarge support for older patients with congestive heart failure: a meta-anal- ysis. JAMA 2004;291(11):1358-67. https://doi.org/10.1001/jama.291.11.1358 [PubMed PMID: 15026403].

16. Koelling TM, Johnson ML, Cody RJ, et al. Discharge education improves clinical out- comes in patients with chronic heart failure. Circulation 2005;111(2):179-85. https://doi.org/10.1161/01.CIR.0000151811.53450.B8 [PubMed PMID: 15642765].
17. Kosiborod M, Lichtman JH, Heidenreich PA, et al. National trends in outcomes among elderly patients with heart failure. Am J Med 2006;119(7):616 e1-7. https://doi.org/ 10.1016/j.amjmed.2005.11.019 [PubMed PMID: 16828634].
18. Ross JS, Chen J, Lin Z, et al. Recent national trends in readmission rates after heart failure hospitalization. Circ Heart Fail 2010;3(1):97-103. https://doi.org/10.1161/ CIRCHEARTFAILURE.109.885210 [PubMed PMID: 19903931; PMCID: PMC2830811].
19. The University of Texas Health Science Center at Houston School of Public Health. Harris County health care safety net: where we stand; 2010.
20. Kawaguchi A, Koch GG. Sanon: an R package for stratified analysis with nonpara- metric covariable adjustment, 67(9); 2015; 37. https://doi.org/10.18637/jss.v067.

    i09 [Epub 2015-10-07].

    Gilbert EH, Lowenstein SR, Koziol-McLain J, et al. Chart reviews in emergency med- icine research: where are the methods? Ann Emerg Med 1996;27(3):305-8 [PubMed PMID: 8599488].

21. Worster A, Haines T. Advanced statistics: understanding medical record review (MRR) studies. Acad Emerg Med 2004;11(2):187-92 [PubMed PMID: 14759964].
22. McHugh ML. Interrater reliability: the kappa statistic. Biochem Med (Zagreb) 2012; 22(3):276-82 [PubMed PMID: 23092060; PMCID: PMC3900052].
23. Harrison A, Morrison LK, Krishnaswamy P, et al. B-type natriuretic peptide predicts future cardiac events in patients presenting to the emergency department with dys- pnea. Ann Emerg Med 2002;39(2):131-8 [PubMed PMID: 11823766].
24. McCullough PA, Kuncheria J, Mathur VS. Diagnostic and therapeutic utility of B-type natriuretic peptide in patients with renal insufficiency and decompensated heart failure. Rev Cardiovasc Med 2003;4(Suppl. 7):S3-S12.
25. Bland JM, Altman DG. Comparisons against baseline within randomised groups are often used and can be highly misleading. Trials 2011;12:264. https://doi.org/10. 1186/1745-6215-12-264 [PubMed PMID: 22192231; PMCID: PMC3286439].
26. Vickers AJ, Altman DG. statistics notes: analysing controlled trials with baseline and follow up measurements. BMJ 2001;323(7321):1123-4 [PubMed PMID: 11701584; PMCID: PMC1121605].
27. Graham S, Brookey J. Do patients understand? Perm J 2008;12(3):67-9 [PubMed PMID: 21331214; PMCID: PMC3037129].
28. Banerjee P, Gill L, Muir V, et al. Do heart failure patients understand their diagnosis or want to know their prognosis? Heart failure from a patient’s perspective. Clin Med (Lond) 2010;10(4):339-43 [PubMed PMID: 20849006].
29. Bock BC, Becker B, Monteiro R, et al. Physician intervention and patient risk percep- tion among smokers with acute respiratory illness in the emergency department. Prev Med 2001;32(2):175-81. https://doi.org/10.1006/pmed.2000.0799 [PubMed PMID: 11162344].
30. Gorin AA, Phelan S, Hill JO, et al. Medical triggers are associated with better short- and long-term weight loss outcomes. Prev Med 2004;39(3):612-6. https://doi.org/ 10.1016/j.ypmed.2004.02.026 [PubMed PMID: 15313102].
31. Hungerford DW, Pollock DA. Emergency department services for patients with alco- hol problems: research directions. Acad Emerg Med 2003;10(1):79-84 [PubMed PMID: 12511321].
32. Schillinger D, Piette J, Grumbach K, et al. Closing the loop: physician communication with Diabetic patients who have low health literacy. Arch Intern Med 2003;163(1): 83-90 [PubMed PMID: 12523921].
33. Lindenauer PK, Remus D, Roman S, et al. Public reporting and pay for performance in hospital quality improvement. N Engl J Med 2007;356:486-96.
34. The Patient Protection and Affordable Care Act; 2010.
35. Readmissions Reduction Program (HRRP). https://www.cms.gov/medicare/ medicare-fee-for-service-payment/acuteinpatientpps/readmissions-reduction- program.html.
36. Desai AS, Stevenson LW. Rehospitalization for heart failure: predict or prevent? Cir- culation 2012;126:501-6.
37. Bradley EH, Curry L, Horwitz LI, et al. Contemporary evidence about hospital strate- gies for reducing 30-day readmissions: a national study. J Am Coll Cardiol 2012;60: 607-14.