Readmissions in the Year After Percutaneous Coronary Intervention

Research Article | DOI: https://doi.org/10.31579/2641-0419/253

Readmissions in the Year After Percutaneous Coronary Intervention

  • Imran Baig 1
  • Amir Eslami DO 1
  • Andrea Berger MA 2
  • Cara Nordberg MPH 2
  • James Blankenship 3*

1 Geisinger Heart Institute, Danville PA
2 Biostatistics Core Center for Health Research, Geisinger Health System, Danville PA
3 University of New Mexico, Albuquerque NM

*Corresponding Author: James C Blankenship, Division of Cardiology 1 University of New Mexico MC 10 5550 Albuquerque New Mexico.

Citation: Imran Baig, Amir Eslami DO, Andrea Berger MA, Cara Nordberg MPH, James Blankenship. (2022). Readmissions in the Year After Percutaneous Coronary Intervention. J. Clinical Cardiology and Cardiovascular Interventions, 5(4); Doi:10.31579/2641-0419/253

Copyright: © 2022 James C Blankenship, This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.

Received: 04 March 2022 | Accepted: 21 March 2022 | Published: 29 March 2022

Keywords: coronary artery disease; percutaneous coronary intervention; re-admission

Abstract

Prior studies of readmission have evaluated correlates of 30-day readmission but have not evaluated correlates of readmission after 30 days. The study sought to evaluate factors associated with re-hospitalization within one year of undergoing percutaneous coronary intervention (PCI). We analyzed 6265 patients treated at Geisinger hospitals with PCI between January 2010 and December 2015. Correlates of readmission within 1 year were identified.  Sub-groups were compared based on reason for readmission (related to versus not related to ischemic heart disease) and timing of readmission (1-30 days versus 31-365 days).  Mean age was 64.3 years, 70% were male, and 98.8% were Caucasian. In the first year after PCI, 2,767 patients (44.2%) were re-admitted. Within 30 days 931 patients (14.9%) were readmitted; 1836 (29.3%) were readmitted between 31 and 365 days. Nine hundred fifty (15.2%).  In summary, in an unselected patient cohort treated with PCI, approximately 44% of the patients were readmitted within one year.  Two-thirds of these were admitted after the first month.  Efforts to prevent the need for readmission should continue beyond the first month post-discharge and center on risk factor modification.

Introduction

Ischemic heart disease related healthcare costs are amongst the highest for any disease entity in the United States. Approximately one in every six US health care dollars is spent on cardiovascular disease [1]. Hospital readmission rates are variable and add to the health care costs. With the high cost of ischemic heart disease and the additional cost of revascularization for this subset of patients, there is limited data on risk factors that correlate with readmission after percutaneous coronary intervention (PCI) [2-4]. Previous studies have reviewed 30-day readmissions post PCI. However, readmissions from 30 days to one-year post-PCI have received little attention.  The purpose of this analysis was to identify patient and characteristics associated with hospital readmission in the year after PCI [5].

Methods

We studied 7,228 patients undergoing PCI at our medical center between January 1, 2010 and December 31, 2015. For patients with multiple PCI encounters during the study period only the first was included. Patients were excluded if they died during their hospital stay (n = 159), if the encounter was missing the admission and discharge date/time information due to PCI database and electronic health record (EHR) matching issues (n = 24), if they died within 365 days of their procedure without a readmission prior to death (n = 110), or if they were without follow-up in 

the Geisinger Health System for 730 days after the original PCI discharge (n = 670). The remaining 6265 patients composed the study cohort.

Hospital readmissions (including emergency department visits and inpatient admissions) within 365 days of the PCI encounter discharge date were identified. Observation unit stays were excluded. 

Descriptive statistics are provided for all 6265 patients, and for patients readmitted within 30 days, readmitted within 31-365 days, and not readmitted. Categorical variables are characterized using frequency counts and percentages. Continuous variables are characterized using means and standard deviations (S.D.) or medians and interquartile ranges (IQR). Characteristics of patients readmitted within 365 days were compared to those who were not readmitted using two-sample t-tests, Wilcoxon rank-sum tests, and Pearson’s chi-square, or Fisher’s exact tests, as appropriate. Multinomial logistic regression modeling was used to compare patients readmitted within 30 days, those readmitted within 31-365 days, and those that were not readmitted. 

Due to the large number of study variables that were significantly associated with the readmission outcomes, bootstrap resampling was used for variable selection. Starting with 64 variables (Table 1), fast step-down selection keeping factors with significance levels < 0>

 

Readmit vs Not Readmitted Model

30 Day/31 to 365/Not Readmitted Model

Age (Quadratic)

x

x

Sex

 

x

Body Mass Index (Piecewise with knot at 30)

x

x

Insurance Payers

x

x

Current/Recent Smoker

 

x

Hypertension

x

x

Dyslipidemia

  

Family History of Premature coronary artery disease

  

Prior myocardial infarction

 

x

Prior Heart Failure

x

x

Prior Valve Surgery/Procedure

 

x

Prior percutaneous coronary intervention

  

Prior Coronary artery bypass grafting

 

x

Currently on Dialysis

 

x

Cerebrovascular Disease

x

x

Chronic Lung Disease

x

x

Diabetes Mellitus

x

x

Coronary artery disease Presentation

x

x

Angina Classification within 2 Weeks

 

x

Heart Failure Within 2 Weeks

x

x

Cardiomyopathy or LV Systolic Dysfunction

  

Cardiogenic Shock Within 24 Hours

x

x

Cardiac Arrest Within 24 Hours

  

Fluoroscopy Time (Piecewise with knot at 19)

x

x

Contrast Volume

  

Procedure Diagnostic

x

x

Intra-aortic balloon pump

  

Arterial Access Site

x

x

Left Main 

  

Proximal left anterior descending artery

  

Distal left anterior descending artery

  

Circumflex

  

Right coronary artery 

x

 

Ramus

  

PCI Scheduling Status

x

x

Cardiogenic Shock at Start of PCI

  

Pre Creatinine (Piecewise with knot at 0.8)

x

x

Pre Hemoglobin (Piecewise with knot at 15)

x

x

Low Molecular Weight Heparin (any)

 

x

Unfractionated Heparin (any) at Procedure

  

Aspirin at Procedure

  

Bivalirudin at Procedure

 

x

Glycoprotein IIb/IIIa Inhibitors at Procedure

  

Clopidogrel at Procedure

x

x

Prasugrel at Procedure

  

Ticagrelor at Procedure

x

x

Angiotensin converting enzyme inhibitors/angiotensin receptor blockers s at Discharge

  

Aspirin at Discharge

  

Beta Blockers at Discharge

x

x

Lipid-Lowering Agents (statins and non-statins) at Discharge

  

Clopidogrel at Discharge

  

Prasugrel at Discharge

x

x

Ticagrelor at Discharge

x

x

Culprit Lesion Identification

x

x

Pre-Stenosis Percent

x

x

Lesion Complexity

 

x

Length (mm)

  

Thrombus

  

Bifurcation

  

Systolic Blood Pressure (Piecewise with knot at 120)

x

x

Diastolic Blood Pressure (Piecewise with knot at 70)

 

x

Year of Intervention

x

x

Charlson Comorbidity Index on procedure Date

x

x

Duration of Index Hospitalization

x

x

Table 1: Variables Used in Bootstrap Resampling for Variable Selection

not readmitted, and readmitted within 31-365 days vs. not readmitted. Variables selected in 50% or more of the 1000 repetitions for readmitted versus not readmitted were retained for a final multivariable logistic regression model. Variables retained in 50% or more of either the 30-day readmission versus not readmitted or 31-365-day readmission versus not readmitted model were retained for a final multivariable multinomial logistic regression model. Odds ratio estimates, 95% confidence intervals, and p-values are reported for the results of the multivariable models. 

Restricted cubic splines were used to assess non-linear relationships between continuous variables and outcomes. Non-linear relationships were included in the multivariable models as quadratic terms or piecewise linear terms, as appropriate. Missingness in variables used in the variable selection process did not exceed 1.3% for categorical variables and 5.8% for continuous variables. Missing values for categorical variables were imputed by random assignment to a category proportional to the frequencies in the non-missing observations.  Missing values for continuous variables were imputed using non-missing median values by sex. Analysis was performed using R version 3.5.0, and SAS 9.4. 

Results

The study cohort included 6265 patients.  Mean age was 64.3 years, 70% were male, and 98.8% were Caucasian. In the first year after PCI, 2,767 patients (44.2%) were re-admitted. Nine hundred thirty-one patients (14.9%) were readmitted within 30 days and 1836 (29.3%) were readmitted between 31 and 365 days. 

Correlates of Readmission within 365 days:  Factors associated with readmission within a year are listed in Table 2 and results of multivariable logistic regression modeling are in Table 3.  These included co-morbidities (Charlston comorbidity score, age, body mass index, history of hypertension, heart failure, cerebrovascular disease, chronic lung disease, diabetes mellitus), status at time of percutaneous coronary intervention (i.e., clinical presentation), cardiogenic shock within 24 hours, pre-procedure creatinine, pre-procedure hemoglobin, systolic blood pressure), procedural factors (culprit lesion, pre-stenosis percentage, arterial access site), and post-procedural factors (beta-blockers at discharge, prasugrel at discharge, ticagrelor at discharge, and length of stay).  Patients with Medicare were more likely to be readmitted than patients with Blue Cross/Blue Shield, Geisinger Health Plan, or other insurance. 

 

 

All Patients (n = 6265)

Readmitted within 1 Year
(n = 2767)

Not Readmitted (n = 3498)

Readmitted vs Not Readmitted P-Value

 

n

%

n

%

n

%

 

Age, mean (SD)

64.3 (12.2)

65.3 (12.8)

63.5 (11.7)

< 0>

Male

4401

70.2%

1830

66.1%

2571

73.5%

< 0>

Body mass index, mean (SD)

30.7 (6.8)

30.3 (7.2)

30.9 (6.4)

0.0013

Insurance Payors

 

 

 

 

 

 

< 0>

Blue cross/Blue Shield

744

11.9%

259

9.4%

485

13.9%

 

Geisinger Health

2345

37.4%

982

35.5%

1363

39.0%

 

Medicaid

37

0.6%

17

0.6%

20

0.6%

 

Medicare

2215

35.4%

1108

40.0%

1107

31.6%

 

Others

924

14.7%

401

14.5%

523

15.0%

 

Hypertension

4908

78.3%

2273

82.1%

2635

75.3%

< 0>

Dyslipidemia

4746

75.8%

2137

77.2%

2609

74.6%

0.0152

Family History of Premature Coronary Artery Disease

2219

35.4%

930

33.6%

1289

36.8%

0.0078

Prior Myocardial Infarction

1410

22.5%

698

25.2%

712

20.4%

< 0>

Prior Heart Failure

653

10.4%

389

14.1%

264

7.5%

< 0>

Prior Valve Surgery/Procedure

109

1.7%

68

2.5%

41

1.2%

0.0001

Prior Percutaneous Coronary Intervention

1444

23.0%

684

24.7%

760

21.7%

0.0052

Prior Coronary Artery Bypass Surgery

871

13.9%

462

16.7%

409

11.7%

< 0>

Currently on Dialysis

112

1.8%

86

3.1%

26

0.7%

< 0>

Cerebrovascular Disease

645

10.3%

387

14.0%

258

7.4%

< 0>

Chronic Lung Disease

701

11.2%

387

14.0%

314

9.0%

< 0>

Diabetes Mellitus

2123

33.9%

1088

39.3%

1035

29.6%

< 0>

CAD Presentation

 

 

 

 

 

 

< 0>

No Symptoms, No Angina

289

4.6%

152

5.5%

137

3.9%

 

Non-STEMI

1455

23.2%

664

24.0%

791

22.6%

 

STEMI or Equivalent

1542

24.6%

660

23.9%

882

25.2%

 

Stable Angina

917

14.7%

332

12.0%

585

16.7%

 

Symptom Unlikely to be Ischemic

58

0.9%

21

0.8%

37

1.1%

 

Unstable Angina

1998

31.9%

933

33.8%

1065

30.5%

 

Missing

6

0.1%

5

0.2%

1

0.0%

 

Heart Failure Within 2 Weeks
(n missing = 16)

539

8.6%

333

12.1%

206

5.9%

< 0>

Cardiomyopathy or LV Systolic Dysfunction
(n missing = 16)

626

10.0%

331

12.0%

295

8.4%

< 0>

Cardiogenic Shock Within 24 Hours
(n missing = 16)

121

1.9%

75

2.7%

46

1.3%

< 0>

Contrast Volume, mean (SD)
(n missing = 320)

177.3 (77.8)

173.3 (77.8)

180.4 (77.7)

0.0005

Intra-aortic Balloon Pump

160

2.6%

96

3.5%

64

1.8%

< 0>

Arterial Access Site

 

 

 

 

 

 

< 0>

Femoral

2938

46.9%

1413

51.1%

1525

43.6%

 

Radial

3307

52.8%

1345

48.6%

1962

56.1%

 

Brachial/Others

18

0.3%

8

0.3%

10

0.3%

 

Missing

2

0.0%

1

0.0%

1

0.0%

 

PCI Status

 

 

 

 

 

 

< 0>

Elective

1762

28.4%

682

25.0%

1080

31.1%

 

Emergency

1609

25.9%

691

25.3%

918

26.4%

 

Salvage

58

0.9%

28

1.0%

30

0.9%

 

Urgent

2777

44.7%

1332

48.7%

1445

41.6%

 

Missing

59

0.9%

34

1.2%

25

0.7%

 

Cardiogenic Shock at Start of PCI
(n missing = 79)

126

2.0%

69

2.5%

57

1.6%

0.0138

Pre PCI Creatinine, median (IQR) (n missing = 297)

0.9 (0.8, 1.1)

1.0 (0.8, 1.2)

0.9 (0.8, 1.1)

< 0>

Pre PCI Hemoglobin, mean (STD)
(n missing = 362)

13.7 (1.9)

13.3 (2.0)

14.0 (1.7)

< 0>

Discharge Medication

 

 

 

 

 

 

 

Aspirin

6071

96.9%

2665

96.3%

3406

97.4%

0.0165

Clopidogrel

5191

82.9%

2228

80.5%

2963

84.7%

< 0>

Ticagrelor

486

7.8%

247

8.9%

239

6.8%

0.0021

Lesions Characteristic

 

 

 

 

 

 

 

Thrombus

1916

30.6%

792

28.6%

1124

32.1%

0.0028

Previous Analysis

 

 

 

 

 

 

 

Year of Intervention

 

 

 

 

 

 

< 0>

2010

1156

18.5%

450

16.3%

706

20.2%

 

2011

1029

16.4%

425

15.4%

604

17.3%

 

2012

933

14.9%

370

13.4%

563

16.1%

 

2013

1061

16.9%

523

18.9%

538

15.4%

 

2014

1098

17.5%

524

18.9%

574

16.4%

 

2015

988

15.8%

475

17.2%

513

14.7%

 

Charlson Comorbidity Index on Procedure Date

 

 

 

 

 

 

< 0>

0

368

5.9%

144

5.2%

224

6.4%

 

1 to 2

2159

34.5%

802

29.0%

1357

38.8%

 

≥ 3

3738

59.7%

1821

65.8%

1917

54.8%

 

Duration of Index Hospitalization

 

 

 

 

 

 

< 0>

0 to 3 Days

4472

71.4%

1780

64.3%

2692

77.0%

 

4 to 7 Days

1387

22.1%

720

26.0%

667

19.1%

 

> 7 Days

406

6.5%

267

9.6%

139

4.0%

 

Table 2: Summary of patient and encounter characteristics overall and by readmitted and not readmitted, excluding those not used in bootstrap variable selection an excluding those without significant differences between those readmitted and not readmitted.

STEMI = ST elevation myocardial infarction, PCI = Percutaneous Coronary Intervention, SD = standard deviation, IQR = Inter-quartile range, CK-MB = Creatine Kinase-Myocardial Band

Readmission within 365 Days (n = 6265)

OR

95% CI Upper, Lower

P-Value

Age Quadratic- Age 55 (Q1) one year increase

0.974

0.965

0.983

< 0>

Age Quadratic- Age 64 (Median)one year increase

0.991

0.984

0.998

Age Quadratic- Age 73 (Q3) one year increase

1.008

0.999

1.016

Body Mass Index (1 unit increase for BMI < 30>

0.958

0.939

0.978

0.0001

Body Mass Index (1 unit increase for BMI ≥ 30)

1.007

0.995

1.020

Insurance: Geisinger vs Blue Cross/Blue Shield

1.181

0.980

1.422

0.0014

Insurance: Medicaid vs Blue Cross/Blue Shield

1.397

0.689

2.832

Insurance: Medicare vs Blue Cross/Blue Shield

1.447

1.191

1.758

Insurance: Others vs Blue Cross/Blue Shield

1.198

0.969

1.480

Hypertension

1.350

1.174

1.552

< 0>

Prior Heart Failure

1.328

1.086

1.624

0.0057

Cerebrovascular Disease

1.663

1.390

1.989

< 0>

Chronic Lung Disease

1.258

1.060

1.493

0.0085

Diabetes Mellitus

1.268

1.120

1.434

0.0002

Coronary Disease Presentation: No Symptoms/No Angina vs Unlikely Ischemic

1.754

0.943

3.262

0.0023

Coronary Disease Presentation: Non-STEMI vs Unlikely Ischemic

1.477

0.825

2.645

Coronary Disease Presentation: STEMI vs Unlikely Ischemic

1.428

0.746

2.732

Coronary Disease Presentation: Stable Angina vs Unlikely Ischemic

1.327

0.738

2.386

Coronary Disease Presentation: Unstable Angina vs Unlikely Ischemic

1.815

1.022

3.223

Cardiogenic Shock within 24 Hours

1.593

1.030

2.464

0.0363

Fluoroscopy Time (1 minute increase for time ≥ 19)

1.004

0.997

1.011

 

Procedure Diagnostic

1.288

1.048

1.583

0.0161

Arterial Access: Radial vs Femoral

0.815

0.725

0.916

0.0024

Arterial Access: Brachial/Others vs Femoral

1.095

0.417

2.874

PCI Scheduling Status: Emergency vs Elective

1.464

1.050

2.042

0.0062

PCI Scheduling Status: Salvage vs Elective

0.933

0.482

1.805

PCI Scheduling Status: Urgent vs Elective

1.299

1.099

1.535

Creatinine Pre-Procedure (0.1 increase for < 0>

0.896

0.828

0.969

0.0005

Creatinine Pre-Procedure (0.1 increase for ≥ 0.8)

1.017

1.007

1.027

Hemoglobin Pre-Procedure (1 unit increase for < 15>

0.906

0.867

0.946

< 0>

Hemoglobin Pre-Procedure (1 unit increase for ≥ 15)

0.976

0.874

1.090

Beta Blockers at Discharge

0.822

0.688

0.983

0.0316

Prasugrel at Discharge

1.317

1.033

1.679

0.0263

Ticagrelor at Discharge

1.519

1.052

2.193

0.0257

Culprit Lesion Identification vs Unknown

0.760

0.641

0.901

0.0016

Pre-Stenosis Percent (1 unit increase)

0.993

0.987

0.999

0.0152

Systolic Blood Pressure (1 unit increase < 120>

0.989

0.981

0.997

0.0269

Systolic Blood Pressure (1 unit increase ≥ 120)

1.003

0.999

1.007

Procedure Year: 2011 vs 2010

1.135

0.946

1.361

< 0>

Procedure Year: 2012 vs 2010

1.130

0.934

1.367

Procedure Year: 2013 vs 2010

1.796

1.489

2.166

Procedure Year: 2014 vs 2010

1.653

1.367

1.999

Procedure Year: 2015 vs 2010

1.773

1.447

2.172

Charlson Comorbidity Index on Cath Date (3 vs 0) 

1.359

1.000

1.848

Length of Stay: (4-7 Days vs 0 to 3 Days)

1.258

1.094

1.446

< 0>

Length of Stay: (> 7 Days vs 0 to 3 Days)

1.712

1.331

2.201

Table 3: Multivariable Binary Logistic Regression Results for Outcome of Readmitted within 365 Days and Not Readmitted 

STEMI = ST elevation myocardial infarction, PCI = Percutaneous Coronary Intervention, SD = standard deviation, BMI = Body Mass Index

Acute coronary syndrome patients were more likely to be readmitted compared to patients with stable or unlikely ischemic presentations. Radial access patients were less likely to be readmitted compared to femoral access patients. Emergency and urgent procedure patients had higher odds of readmission compared to elective procedure patients. A non-linear association between systolic blood pressure and readmission was observed - as systolic pressure increased for patients with systolic pressure < 120>

Correlates of readmission Days 1-30 days versus Days 31 -334:  Most characteristics associated with 1 to 30-day readmission by multivariable nominal logistic regression were also associated with admission from 31-334 days, but some differences were observed (Tables 4,5). 

 

All Patients
(n = 6265)

Readmission within 30 Days
(n = 931)

Readmission 31-365 Days
(n = 1836)

Not Readmitted
(n = 3498)

30 Day Readmission vs Not Readmitted 
P-Value

31-365 Day Readmission vs Not Readmitted P-Value

 

n

%

n

%

n

%

n

%

Age, mean (SD)

64.3 (12.2)

65.2 (13.1)

65.4 (12.6)

63.5 (11.7)

0.0002

< 0>

Male

4401

70.2%

578

62.1%

1252

68.2%

2571

73.5%

< 0>

< 0>

Body Mass Index, mean (SD)

30.7 (6.8)

30.0 (7.2)

30.5 (7.1)

30.9 (6.4)

0.0001

0.0676

Insurance Payors

 

 

 

 

 

 

 

 

< 0>

< 0>

Blue cross/Blue Shield

744

11.9%

92

9.9%

167

9.1%

485

13.9%

 

 

Geisinger Health

2345

37.4%

314

33.7%

668

36.4%

1363

39.0%

 

 

Medicaid

37

0.6%

9

1.0%

8

0.4%

20

0.6%

 

 

Medicare

2215

35.4%

361

38.8%

747

40.7%

1107

31.6%

 

 

Others

924

14.7%

155

16.6%

246

13.4%

523

15.0%

 

 

Hypertension

4908

78.3%

764

82.1%

1509

82.2%

2635

75.3%

< 0>

< 0>

Dyslipidemia

4746

75.8%

697

74.9%

1440

78.4%

2609

74.6%

0.8613

0.0018

Family History of Premature Coronary Artery Disease

2219

35.4%

307

33.0%

623

33.9%

1289

36.8%

0.0288

0.0348

Prior Myocardial Infarction

1410

22.5%

199

21.4%

499

27.2%

712

20.4%

0.4935

< 0>

Prior Heart Failure

653

10.4%

130

14.0%

259

14.1%

264

7.5%

< 0>

< 0>

Prior Valve Surgery/Procedure

109

1.7%

24

2.6%

44

2.4%

41

1.2%

0.0020

0.0009

Prior PCI

1444

23.0%

196

21.1%

488

26.6%

760

21.7%

0.6568

< 0>

Prior Coronary Artery Bypass Surgery

871

13.9%

123

13.2%

339

18.5%

409

11.7%

0.2047

< 0>

Currently on Dialysis

112

1.8%

30

3.2%

56

3.1%

26

0.7%

< 0>

< 0>

Cerebrovascular Disease

645

10.3%

124

13.3%

263

14.3%

258

7.4%

< 0>

< 0>

Chronic Lung Disease

701

11.2%

131

14.1%

256

13.9%

314

9.0%

< 0>

< 0>

Diabetes Mellitus

2123

33.9%

354

38.0%

734

40.0%

1035

29.6%

< 0>

< 0>

Coronary Artery Disease Presentation

 

 

 

 

 

 

 

 

< 0>

< 0>

No Symptoms, No Angina

289

4.6%

44

4.7%

108

5.9%

137

3.9%

 

 

Non-STEMI

1455

23.2%

233

25.0%

431

23.5%

791

22.6%

 

 

STEMI or Equivalent

1542

24.6%

266

28.6%

394

21.5%

882

25.2%

 

 

Stable Angina

917

14.7%

91

9.8%

241

13.2%

585

16.7%

 

 

Symptom Unlikely to be Ischemic

58

0.9%

12

1.3%

9

0.5%

37

1.1%

 

 

Unstable Angina

1998

31.9%

285

30.6%

648

35.4%

1065

30.5%

 

 

Missing

6

0.1%

0

0.0%

5

0.3%

1

0.0%

 

 

Heart Failure Within 2 Weeks
(n missing = 16)

539

8.6%

118

12.7%

215

11.8%

206

5.9%

< 0>

< 0>

Cardiomyopathy or LV Systolic Dysfunction
(n missing = 16)

626

10.0%

120

12.9%

211

11.6%

295

8.4%

< 0>

0.0002

Cardiogenic Shock Within 24 Hours
(n missing = 16)

121

1.9%

38

4.1%

37

2.0%

46

1.3%

< 0>

0.0492

Cardiac Arrest Within 24 Hours
(n missing = 16)

142

2.3%

28

3.0%

39

2.1%

75

2.1%

0.1234

0.9772

Contrast Volume, mean (SD)
(n missing = 320)

177.3 (77.8)

171.6 (76.1)

174.2 (78.7)

180.4 (77.7)

0.0028

0.0074

IABP

160

2.6%

42

4.5%

54

2.9%

64

1.8%

< 0>

0.0094

Arterial Access Site

 

 

 

 

 

 

 

 

0.0005

< 0>

Femoral

2938

46.9%

471

50.6%

942

51.3%

1525

43.6%

 

 

Radial

3307

52.8%

456

49.0%

889

48.4%

1962

56.1%

 

 

Brachial/Others

18

0.3%

4

0.4%

4

0.2%

10

0.3%

 

 

Missing

2

0.0%

0

0.0%

1

0.1%

1

0.0%

 

 

PCI Status

 

 

 

 

 

 

 

 

< 0>

0.0001

Elective

1762

28.4%

174

18.9%

508

28.0%

1080

31.1%

 

 

Emergency

1609

25.9%

270

29.3%

421

23.2%

918

26.4%

 

 

Salvage

58

0.9%

16

1.7%

12

0.7%

30

0.9%

 

 

Urgent

2777

44.7%

460

50.0%

872

48.1%

1445

41.6%

 

 

Missing

59

0.9%

11

1.2%

23

1.3%

25

0.7%

 

 

Cardiogenic Shock at Start of PCI
(missing = 79)

126

2.0%

37

4.0%

32

1.8%

57

1.6%

< 0>

0.7307

Pre Creatinine, median (IQR)
(n missing = 297)

0.9 (0.8, 1.1)

0.9 (0.8, 1.2)

1.0 (0.8, 1.2)

0.9 (0.8, 1.1)

< 0>

< 0>

Pre Hemoglobin, mean (STD)
(n missing = 362)

13.7 (1.9)

13.2 (2.0)

13.4 (2.0)

14.0 (1.7)

< 0>

< 0>

Discharge Medication

 

 

 

 

 

 

 

 

 

 

Aspirin

6071

96.9%

901

96.8%

1764

96.1%

3406

97.4%

0.3270

0.0099

Lipid Lowering Agents (Statins and Non-Statins)

6028

96.2%

893

95.9%

1746

95.1%

3389

96.9%

0.1448

0.0012

Thienopyridines

 

 

 

 

 

 

 

 

 

 

Clopidogrel

5191

82.9%

735

78.9%

1493

81.3%

2963

84.7%

< 0>

0.0015

Ticagrelor

486

7.8%

95

10.2%

152

8.3%

239

6.8%

0.0006

0.0545

Lesions and Devices

 

 

 

 

 

 

 

 

 

 

Pre-Stenosis Percent, median (IQR)
(n missing = 29)

95 (90, 100)

95 (90, 100)

95 (90, 100)

95 (90, 100)

0.8073

0.0326

Thrombus

1916

30.6%

302

32.4%

490

26.7%

1124

32.1%

0.8592

< 0>

Previous Analysis

 

 

 

 

 

 

 

 

 

 

Year of Intervention

 

 

 

 

 

 

 

 

< 0>

< 0>

2010

1156

18.5%

146

15.7%

304

16.6%

706

20.2%

 

 

2011

1029

16.4%

139

14.9%

286

15.6%

604

17.3%

 

 

2012

933

14.9%

115

12.4%

255

13.9%

563

16.1%

 

 

2013

1061

16.9%

171

18.4%

352

19.2%

538

15.4%

 

 

2014

1098

17.5%

184

19.8%

340

18.5%

574

16.4%

 

 

2015

988

15.8%

176

18.9%

299

16.3%

513

14.7%

 

 

Charlson Comorbidity Index on Procedure Date

 

 

 

 

 

 

 

 

< 0>

< 0>

0

368

5.9%

40

4.3%

104

5.7%

224

6.4%

 

 

1 to 2

2159

34.5%

285

30.6%

517

28.2%

1357

38.8%

 

 

≥ 3

3738

59.7%

606

65.1%

1215

66.2%

1917

54.8%

 

 

Duration of Index Hospitalization

 

 

 

 

 

 

 

 

< 0>

< 0>

0 to 3 Days

4472

71.4%

536

57.6%

1244

67.8%

2692

77.0%

 

 

4 to 7 Days

1387

22.1%

277

29.8%

443

24.1%

667

19.1%

 

 

> 7 Days

406

6.5%

118

12.7%

149

8.1%

139

4.0%

 

 

Table 4: Summary of Patient and Encounter Characteristics Overall and by Readmitted Days 1-30, Readmitted Days 31 - 365, and Not Readmitted, Excluding those not use in bootstrap variable selection 

STEMI = ST elevation myocardial infarction, PCI = Percutaneous Coronary Intervention, SD = standard deviation, IQR = Inter-quartile range, CK-MB = Creatine Kinase-Myocardial Band, CCS= Canadian Cardiovascular Society Class

Readmission within 30 Days, Readmssion Between 31 and 365 Days, and Not Readmitted (n = 6265)

Readmission within 30 Days vs Not Readmitted

Readmission 31-365 Days vs Not Readmitted

OR

95% CI
Upper, Lower

P-Value

OR

95% CI
Upper, Lower

P-Value

Age Quadratic- Age 55 (Q1) one year increase

0.966

0.953

0.978

< 0>

0.981

0.971

0.992

< 0>

Age Quadratic- Age 64 (Median) one year increase

0.986

0.976

0.995

0.995

0.987

1.003

Age Quadratic- Age 73 (Q2) one year increase

1.006

0.994

1.018

1.010

1.000

1.019

BMI (1 unit increase for BMI < 30>

0.949

0.924

0.976

0.0001

0.964

0.943

0.986

0.0056

BMI (1 unit increase for BMI ≥ 30)

0.995

0.978

1.014

1.012

0.998

1.025

Insurance: Geisinger vs Blue Cross/Blue Shield

1.105

0.840

1.453

0.0838

1.213

0.980

1.502

0.0026

Insurance: Medicaid vs Blue Cross/Blue Shield

2.014

0.836

4.850

1.034

0.432

2.479

Insurance: Medicare vs Blue Cross/Blue Shield

1.333

1.003

1.771

1.482

1.187

1.851

Insurance: Others vs Blue Cross/Blue Shield

1.301

0.960

1.762

1.139

0.892

1.454

Recent Smoker

0.949

0.786

1.146

0.5874

1.158

1.001

1.340

0.0485

Hypertension

1.550

1.259

1.907

< 0>

1.249

1.065

1.464

0.0062

Prior Myocardial Infarction

0.943

0.773

1.150

0.5628

1.174

1.012

1.362

0.0344

Previous Coronary Bypass Surgery

0.951

0.735

1.230

0.6993

1.222

1.012

1.474

0.0369

Cerebrovascular Disease

1.576

1.232

2.016

0.0003

1.665

1.370

2.024

< 0>

Diabetes Mellitus

1.237

1.036

1.478

0.0188

1.271

1.107

1.459

0.0007

Coronary disease Presentation: No Symptoms/No Angina vs Unlikely Ischemic

0.950

0.416

2.168

0.0725

2.916

1.277

6.659

0.0012

Coronary disease Presentation: Non-STEMI vs Unlikely Ischemic

0.792

0.374

1.676

2.189

0.998

4.800

Coronary disease Presentation: STEMI vs Unlikely Ischemic

0.936

0.401

2.182

1.934

0.827

4.520

Coronary disease Presentation: Stable Angina vs Unlikely Ischemic

0.632

0.296

1.349

1.998

0.912

4.378

Coronary disease Presentation: Unstable Angina vs Unlikely Ischemic

0.988

0.472

2.068

2.767

1.271

6.024

Cardiogenic Shock within 24 Hours

1.818

1.076

3.070

0.0254

1.414

0.855

2.338

0.1769

Fluoroscopy Time (1 minute increase for time < 19>

0.977

0.961

0.994

0.0258

0.987

0.974

1.000

0.1376

Fluoroscopy Time (1 minute increase for time ≥ 19)

1.007

0.997

1.017

1.002

0.994

1.009

Procedure Diagnostic

1.461

1.094

1.950

0.0101

1.298

1.042

1.616

0.0201

Arterial Access: Radial vs Femoral

0.886

0.745

1.054

0.1919

0.812

0.710

0.930

0.0104

Arterial Access: Brachial/Others vs Femoral

1.947

0.580

6.540

0.770

0.233

2.543

PCI Scheduling Status: Emergency vs Elective

1.690

1.049

2.723

0.0039

1.437

0.986

2.093

0.0352

PCI Scheduling Status: Salvage vs Elective

1.361

0.597

3.103

0.681

0.303

1.534

PCI Scheduling Status: Urgent vs Elective

1.609

1.241

2.085

1.213

1.004

1.465

Creatinine Pre-Procedure (0.1 increase for < 0>

0.965

0.858

1.084

0.3804

0.890

0.812

0.976

0.0156

Creatinine Pre-Procedure (0.1 increase for ≥ 0.8)

1.011

0.995

1.027

1.013

0.999

1.027

Hemoglobin Pre-Procedure (1 unit increase for < 15>

0.898

0.844

0.954

0.0015

0.926

0.880

0.973

0.0036

Hemoglobin Pre-Procedure (1 unit increase for ≥ 15)

1.001

0.850

1.179

0.970

0.854

1.100

Prasugrel at Discharge

1.358

0.959

1.924

0.0849

1.331

1.014

1.745

0.0392

Ticagrelor at Discharge

1.636

0.989

2.704

0.0550

1.519

1.001

2.306

0.0496

Culprit Lesion Identification vs Unknown

0.743

0.580

0.952

0.0188

0.765

0.634

0.922

0.0050

Pre-Stenosis Percent (1 unit increase)

0.992

0.983

1.000

0.0604

0.992

0.986

0.999

0.0219

Diastolic Blood Pressure (1 unit increase ≥ 70)

1.005

0.992

1.018

 

1.006

0.996

1.016

 

Procedure Year: 2011 vs 2010

1.140

0.870

1.493

< 0>

1.155

0.941

1.417

< 0>

Procedure Year: 2012 vs 2010

1.085

0.816

1.444

1.160

0.936

1.437

Procedure Year: 2013 vs 2010

1.812

1.379

2.382

1.761

1.426

2.175

Procedure Year: 2014 vs 2010

1.703

1.282

2.263

1.507

1.207

1.882

Procedure Year: 2015 vs 2010

1.861

1.361

2.545

1.515

1.182

1.941

Charlson Comorbidity Index on Cath Date (1 to 2 vs 0) 

1.631

1.086

2.448

0.0180

0.945

0.703

1.270

0.2158

Charlson Comorbidity Index on Cath Date (3 vs 0) 

1.945

1.226

3.085

1.102

0.782

1.552

Length of Stay: (4-7 Days vs 0 to 3 Days)

1.467

1.210

1.779

< 0>

1.157

0.988

1.355

0.0186

Length of Stay: (> 7 Days vs 0 to 3 Days)

2.180

1.590

2.989

1.449

1.091

1.923

Table 5: Multivariable Multinomial Logistic Regression Results for Outcome of Readmitted within 30 Days, Readmitted Between 31 and 365 Days, and Not Readmitted 

STEMI = ST elevation myocardial infarction, PCI = Percutaneous Coronary Intervention, SD = standard deviation, BMI = Body Mass Index

Factors correlating with later readmission, but NOT early readmission included smoking, prior MI, prior coronary artery bypass graft surgery, radial access, and creatinine.  Factors associated with early readmission, but not later readmission included male gender, insurance status, pre-procedural angina classification, acute shock, radial access, beta blocker therapy at discharge, systolic blood pressure, and Charlson Co-morbidity Index.

Figure: Flow-chart of PCI patients included in the study

Discussion

The most important finding of this study is that the two-fifths of patients undergoing PCI were readmitted within one year of the procedure. Of those readmitted, one-third were readmitted within the first 30 days and two-thirds were readmitted over the next 11 months. Most studies of readmission after PCI focus only on 30-day re-admissions.  Our study shows that a 30-day assessment is limited and under-estimates the burden of morbidity in post-PCI patients.  Care of post-PCI patients beyond 30 days should be of interest to the interventionalist. 

Only two factors over which clinicians have control were associated with likelihood of readmission.  First, radial access (versus femoral access) had an odds ratio of .81 (p = 0.002) of re-admission.  Multiple studies have demonstrated that radial access compared to femoral access decreases vascular complications and bleeding which might explain this correlation [6].  However, use of radial access correlated with reduced rates of late but not early re-admission which is inconsistent with the hypothesis that reduced rates of procedure-related vascular access complications or bleeding are responsible for the lower re-admission rates.  The association with reduced late re-admissions may be due to unknown confounders related to co-morbidities, since femoral access was often reserved for older and sicker patients.

Second, beta-blockers at discharge were associated with an odds ratio of 0.82 (p = 0.03) of any readmission. The correlation was limited to readmissions within the first 30 days (odds ratio 0.74, p = 0.02) but not after 30 days (odds ratio 0.85, p = .11).  This observation has been previously reported [7], but again we cannot exclude unknown confounders since the benefit of beta-blockers is unclear in patients without MI.

Several additional findings are of interest. Urgent and emergent PCI (linked to acute coronary syndromes) status was associated with a 30-46% excess rate of re-admission compared to elective (stable) PCI status. Patients discharged on ticagrelor or prasugrel had 31-52% excess re-admission rates compared to patients discharged on clopidogrel, perhaps because these drugs were used preferentially for acute coronary syndrome patients and because they increase bleeding risk compared to clopidogrel. Finally, length of stay of 4-7 days and > 7 days correlated with 26% and 71% increased risks of readmission, respectively, compared to shorter lengths of stay, presumably because length of stay is a marker for severity of the initial hospitalization and overall clinical status.

We are aware of only 1 other study that evaluated re-admissions up to 1 year after PCI [5]. That study was conducted in Denmark, where the National Health Service guarantees free hospital access.  In the Danish cohort 50% of patients were re-admitted within 1 year compared to 44% in our study.  Similar to our findings, the Denmark investigators concluded that readmission is common in the year following PCI, that many of the re-admissions are due to ischemic heart disease, that co-morbidities correlate most closely with readmission, and that we could identify few modifiable correlates of readmission.  Both studies point to the importance of secondary risk factor modification and careful follow-up after PCI. Differences may be due in part to our inability to identify re-admissions to hospitals outside of our 13-hospital system, although we excluded patients lost-to follow-up from our initial cohort.  Half of the Denmark patients were readmitted due to angina or myocardial infarction whereas two-thirds of our patients were readmitted with ischemia-related diagnoses.  Both studies identified age, Charlson Comorbidity Index </=3, and diabetes as correlates of readmission. The Denmark study identified female gender as a correlate of readmission although our study did not, perhaps because of closer association with other demographic variables that were not available to the Denmark investigators.  However, our study showed a 25% increased risk of readmission for women in the first 30 days that equalized over the next 11 months. 

Readmission rates after PCI within 30 days have been extensively studied, perhaps because 30 days is the time window considered by the Center for Medicare and Medicaid Services Hospital Re-admission Reduction Program (HRRP).  However, this focus may result in lack of focus on care of post-PCI patients over the next 11 months. Attendance at cardiac rehabilitation sessions, adherence to medications, and control of risk factors have all been shown to decrease morbidity and/or mortality after PCI [8-9]. Close follow-up by cardiovascular specialists is important to ensure these goals are met.

Limitations

Our study has several limitations.  We were unable to identify re-admissions to a non-Geisinger hospital, although Geisinger operates 13 hospitals in central and northeastern Pennsylvania where the population is very stable and where most patients treated at Geisinger return to Geisinger.  The population included in this study was 98

References

a