Skip to main content
NCBI home page
PLOS One logoLink to PLOS One
. 2023 Aug 3;18(8):e0289640. doi: 10.1371/journal.pone.0289640

Rates and reasons for hospital readmission after acute ischemic stroke in a US population-based cohort

Lily W Zhou 1,*, Maarten G Lansberg 2, Adam de Havenon 3
Editor: Karthik Gangu4
PMCID: PMC10399731  PMID: 37535655

Abstract

Hospital readmissions following stroke are costly and lead to worsened patient outcomes. We examined readmissions rates, diagnoses at readmission, and risk factors associated with readmission following acute ischemic stroke (AIS) in a large United States (US) administrative database. Using the 2019 Nationwide Readmissions Database, we identified adults discharged with AIS (ICD-10-CM I63*) as the principal diagnosis. Survival analysis with Weibull accelerated failure time regression was used to examine variables associated with hospital readmission. In 2019, 273,811 of 285,451 AIS patients survived their initial hospitalization. Of these, 60,831 (22.2%) were readmitted within 2019. Based on Kaplan Meyer analysis, readmission rates were 9.7% within 30 days and 30.5% at 1 year following initial discharge. The most common causes of readmissions were stroke and post stroke sequalae (25.4% of 30-day readmissions, 15.0% of readmissions between 30–364 days), followed by sepsis (10.3% of 30-day readmissions, 9.4% of readmissions between 30–364 days), and acute renal failure (3.2% of 30-day readmissions, 3.0% of readmissions between 30–364 days). After adjusting for multiple patient and hospital-level characteristics, patients at increased risk of readmission were older (71.6 vs. 69.8 years, p<0.001) and had longer initial lengths of stay (7.6 vs. 6.2 day, p<0.001). They more often had modifiable comorbidities, including vascular risk factors (hypertension, diabetes, atrial fibrillation), depression, epilepsy, and drug abuse. Social determinants associated with increased readmission included living in an urban (vs. rural) setting, living in zip-codes with the lowest median income, and having Medicare insurance. All factors were significant at p<0.001. Unplanned hospital readmissions following AIS were high, with the most common reasons for readmission being recurrent stroke and post stroke sequalae, followed by sepsis and acute renal failure. These findings suggest that efforts to reduce readmissions should focus on optimizing secondary stroke and infection prevention, particularly among older socially disadvantaged patients.

Introduction

Acute ischemic stroke (AIS) is the most common stroke subtype and accounts for 87% of all strokes in the United States (US) [1]. Hospital readmission following AIS has previously been estimated to occur in 17.4% of patients at 30 days and 42.5% of patients at 1 year [2]. The total direct medical cost of stroke is among the top 10 contributors to Medicare expenditure [3], and is projected to more than double from $71.55 billion to $184.13 billion between the years of 2012–2035 [4]. Furthermore, each additional day spend in health care facilities (vs. home) in the 3 months following stroke has been associated with decreased quality of life in addition to increased healthcare costs [5]. Measuring and reducing readmissions is a vital component in improving quality of life after stroke and in managing the staggering growth in stroke-related healthcare expenditure.

To better understand the frequency and the causes of readmission after AIS, we report the rates of readmission by patient-level medical comorbidities, stroke severity, discharge status and socioeconomic status from the 2019 Nationwide Readmissions Database (NRD), which contains 18 million discharges representing 60.4% of all U.S. hospitalizations from 30 states and from all types of insurance payers [6]. The NRD does not allow data linkage across calendar years for the same patient resulting in variable length of follow up for patients admitted in different times of the year. Because of this, previous work using the NRD studied 30-day readmissions following stroke [7, 8]. We describe a novel technique using time-to-event (ie the product limit method) which 1) provides readmission rates using Kaplan-Meyer estimates including survival time contributions from right-censored patients at 30-days and 1-year following hospital discharge, 2) allows data from all patients in the sample to contribute to the estimates (including patients with less than 30-days follow-up), and 3) allows examination of risk factors for readmission using survival analysis.”

Methods

Cohort, outcomes, and co-variates

We included individuals with AIS (ICD-10-CM I63*) as the principal discharge diagnosis of their baseline hospital admission and excluded patients under age 18 and elective hospital admissions. The primary outcome was any non-elective hospital readmission during 2019. The secondary outcome was readmission with 1) recurrent AIS (ICD 10 codes I63*) as the principal readmission diagnosis. The tertiary outcome was readmission due to Major Adverse Cardiovascular Events (MACE) defined as: 1) AIS, 2) transient ischemic attack (TIA) (G45*, G46*), 3) Peripheral Vascular Disease (I7*), 4) myocardial infarction (MI) (I21*, I22*) or 5) cardiac arrest (I46*) as the principal readmission diagnosis or 6) a hospital readmission ending in death with a principal readmission diagnosis starting with I (diseases of the circulatory system) to capture cardiovascular death.

The study exposures of individual-level medical comorbidities were pre-defined by NRD, which uses the Elixhauser Comorbidity Measure [9], with the exception of the following comorbidities (defined by ICD-10-CM coding): atrial fibrillation (AF) (I48*), smoking/nicotine dependence (I17*), hyperlipidemia (E78.1-E78.5), and epilepsy (G40*). Because our study used de-identified data, it was exempt from IRB approval. The data is publicly available at https://www.hcup-us.ahrq.gov. We report our findings according to the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guidelines.

Statistical analysis

For the primary analysis, we fit time-to-event models. For individuals with the primary outcome of hospital readmission, the exact number of days between the index admission and the readmission is captured in the NRD. Time to readmission is calculated using this information and length of stay (LOS) of the index admission. Within the NRD, hospital records from patient transfers are combined to avoid counting hospitalization at the second hospital as a readmission. For those without recurrence, NRD only captures the month of discharge. Because the resulting timescale mix of days and months would prevent an accurate time to event model, we imputed the discharge day of index admission for those without the primary outcome. The imputation selected a random discharge day in the discharge month and creates a timescale in days for individuals without a second hospitalization. This novel methodology reduces bias by allowing the right-censoring to assume a standard distribution.

Stratified Cox models were initially used to assess variables associated with readmission but there were significant violations of the proportional hazards assumption. This was confirmed graphically on plots of Schoenfeld residuals and log-log plots of survival (S1 Fig). Because of this, we used Weibull accelerated failure time regression models for survival analysis, which does not assume constant hazard in survival data [10], to examine the association between patient characteristics (demographics, socio-economic variables, comorbidities) and time to hospital readmission.

This was done first in univariate analysis and then after adjusting for patient age, sex, LOS, discharge disposition, hospital bed size using National Inpatient Sample criteria (small/medium/large) [11], location/teaching status (urban teaching, urban non-teaching, rural), and ownership (government, private non-profit, private for-profit), and baseline National Institute of Health Stroke Scale (NIHSS), made available after October of 2016 by the Centers for Medicare & Medicaid Services (using ICD-10-CM code (R29.7*) to be used in conjunction with the diagnosis of AIS (I63*). Admission NIHSS scores are submitted from the admitting hospital’s medical record to the National Readmission Database as a component of administrative data [12]. Adjustment for stroke severity using NIHSS submitted in this form to the Healthcare Cost and Utilization Project has previously been shown to have important implications in analysis of the association between poststroke outcome and patient sex or stroke interventions [13].

Baseline NIHSS and LOS were stratified into categories as there was significant non-linearity on the Box-Tidwell test. The Box-Tidwell test was not significant for age, so it was included in the multivariable model as a continuous variable. The study exposures measured at the index admission included hypertension, diabetes, hyperlipidemia, AF, obesity, drug abuse, smoking or nicotine dependence, depression, dementia, epilepsy, malignancy, rural residence, median income by zip code, primary payer.

Principal diagnoses at readmission were divided into categories using the ICD-10-CM code first letter, which corresponds to organ system. Kaplan-Meyer failure curves divided by principal diagnosis at readmissions were calculated and overlaid on failure curves of all cause readmissions. Missing data is noted where presented and trimmed in the analysis. All statistical analyses were conducted in Stata 17.0 (StataCorp, College Station, TX).

Results

Rates of hospital readmission

Of the 285,451 initial stroke hospitalizations in the 2019 NRD (median LOS 3 days, IQR 2–7 days; median hospitalization charge $48 504 USD, IQR $27 226–91 089 USD), 4.1% resulted in mortality (Fig 1).

Fig 1. Outcomes of patients discharged with ischemic stroke and reasons for readmission.

Fig 1

Open in a new tab

a There is no missing data on mortality at discharge within this cohort.

Of the 273,811 patients who survived their initial hospitalization, 22.2% were readmitted to hospital within 2019 (median LOS 4 days, IQR 2–7; median hospitalization charge $40105, IQR $22 115–76 600 USD). Based on Kaplan-Meier analysis, hospital readmission from any cause occurred in 9.7% at 30 days and in 30.5% at 1 year (Fig 2). Readmissions from MACE occurred in 2.6% of patients at 30 days and in 7.6% of patients at 1 year after discharge, and due to AIS in 2.2% of patients at 30 days and 5.8% at 1 year.

Fig 2. Kaplan-Meier failure functions for hospital readmission after AIS.

Fig 2

Open in a new tab

* 2 readmitted patients excluded due to missing length of stay data.

Diagnoses at hospital readmission

Twenty three percent of hospital readmissions were due to MACE with AIS accounting for most MACE-related readmissions (77%) (Figs 1, 2 Panel B). Based on the principal ICD 10 code, diseases of the circulatory system were the most common cause for readmission (34.5%%) followed by infectious diseases (10.3%) (Fig 2 Panel A, and S1 Table). Readmission rates by sex, age, stroke severity and length of stay are also shown in Fig 2.

When examining for specific disease diagnoses, the top ten most common diagnosis accounted for 50.2% of all readmissions at 30 days and 44.0% of all readmissions between 30–364 days (Table 1).

Table 1. Top 10 diagnosis codes for readmission within first 30 days of discharge vs. after 30 days.

Readmissions within 30 days (n = 25858) Readmission between 30–364 days (n = 34971)
Rank ICD 10 code n % ICD 10 code n %
1 I63 Cerebral infarction 5594 21.6% I63 Cerebral infarction 5232 15.0%
2 A41 Other sepsis 2651 10.3% A41 Other sepsis 3297 9.4%
3 I69 Sequelae of cerebrovascular disease 963 3.7% N17 Acute renal failure 1032 3.0%
4 N17 Acute renal failure 839 3.2% I13 Hypertensive heart and renal disease 1008 2.9%
5 I13 Hypertensive heart and renal disease 581 2.2% I21 Acute myocardial infarction 892 2.6%
6 J69 Pneumonitis due to solids and liquids 498 1.9% N39a Other disorders of the urinary system 865 2.5%
7 I48 Atrial fibrillation and flutter 477 1.8% I11 Hypertensive heart disease 840 2.4%
8 N39* Other disorders of the urinary system 472 1.8% E11 Type 2 diabetes mellitus 799 2.3%
9 I61 Intracranial hemorrhage 466 1.8% S72 Fracture of femur 737 2.1%
10 I21 Acute myocardial infarction 437 1.7% I48 Atrial fibrillation and flutter 678 1.9%
Total 50.2% Total 44.0%
Open in a new tab

aN39.0: Urinary tract infection, site not specified

The most common causes of readmission were stroke and post stroke sequelae (25.4% of 30-day readmissions, 15% of readmissions between 30–364 days), followed by sepsis (10.3% of 30-day readmissions, 9.4% of readmissions between 30–364 days), and acute renal failure (3.2% of 30-day readmissions, 3.0% of readmissions between 30–364 days).

Of the 10,826 individuals with a recurrent stroke admission, 7,667 (70.8%) had an NIHSS documented on readmission. Amongst those individuals, the initial mean±SD NIHSS from their index admission was 5.9±6.6 and with readmission it was 6.7±7.1 (p<0.001). In-hospital mortality on readmission was 4.6% among all cause readmissions and 3.7% among AIS readmissions.

Patient and hospital factors associated with hospital readmission

Comparisons between patients with and without readmissions are listed in Table 2.

Table 2. Baseline patient and hospital characteristics shown after stratification by recurrent admission from any cause, MACE events and AIS vs. no recurrent admission.

Variable Any Recurrent Admission vs. No Recurrent Admission Recurrent MACE Admission vs. No Recurrent MACE Admission Recurrent Stroke Admission vs. No Recurrent Stroke Admission
N = 60,831 vs. 212,980 N = 14,045 vs. 259,766 N = 10,826 vs. 262,985
Time to readmission in days (median, IQR) 41 (13–101) vs. N/A 33 (8–90) vs. N/A 28 (7–82) vs. N/A
Died during readmission 4.6% vs. N/A 6.9% vs. N/A 3.7% vs. N/A
Mean age (SD), years 71.6 (13.3) vs. 69.8 (13.9), p <0.001 69.3 (13.4) vs. 70.2 (13.8), p <0.001 68.9 (13.3) vs. 70.2 (13.8), p <0.001
Female 50.8% vs. 48.8%, p <0.001 48.7% vs. 49.3%, p = 0.016 48.7% vs. 49.2%, p = 0.28
Mean baseline NIHSS (SD) (missing n = 117,118) a 6.4 (6.7) vs. 5.8 (6.6), p <0.001 5.2 (5.7) vs. 5.9 (6.7), p <0.001 5.3 (5.7) vs. 5.9 (6.6), p <0.001
Mean LOS (SD), days (missing n = 8) a 7.6 (10.0) vs. 6.2 (8.9), p <0.001 5.8 (7.8) vs. 6.5 (9.3), p <0.001 5.7 (7.8) vs. 6.5 (9.3), p <0.001
Insurance Status p<0.001 p<0.001 p<0.001
    Medicare 73.1% vs. 64.3% 66.7% vs. 66.2% 65.5% vs. 66.3%
    Medicaid 10.0% vs. 9.2% 11.4% vs. 9.3% 11.8% vs. 9.3%
    Private 12.0% vs. 19.7% 15.4% vs. 18.1% 15.9% vs. 18.1%
    Self-pay/other 4.8% vs. 6.6% 6.4% vs. 6.2% 6.7% vs. 6.2%
    Missing (n = 382) a 0.1% vs. 0.1% 0.1% vs. 0.1% 0.1% vs. 0.1%
Median income by zip code (USD) p <0.001 p<0.001 p<0.001
    ≤47,999 30.2% vs. 28.1% 30.9% vs. 28.4% 30.6% vs. 28.5%
    48–60,999 25.6% vs. 25.8% 25.6% vs. 25.7% 25.6% vs. 25.7%
    61–81,999 23.7% vs. 24.8% 23.5% vs. 24.6% 23.8% vs. 24.6%
    ≥82,000 19.2% vs. 20.0% 18.7% vs. 19.9% 18.9% vs. 19.9%
    Missing(n = 3,608) a 1.3% vs. 1.3% 1.2% vs. 1.3% 1.1% vs. 1.3%
Hypertension 88.6% vs. 85.8%, p <0.001 89.4% vs. 86.3%, p<0.001 89.3% vs. 86.3%, p <0.001
Diabetes 46.0% vs. 37.3%, p <0.001 46.6% vs. 38.8%, p<0.001 46.6% vs. 38.9%, p <0.001
Hyperlipidemia 62.6% vs. 62.2% 65.7% vs. 62.1%, p<0.001 65.4% vs. 62.2%, p <0.001
p = 0.065
Atrial Fibrillation 29.4% vs. 23.4%, p<0.001 22.5% vs. 24.8%, p<0.001 21.4% vs. 24.9, p<0.001
Obesity 15.1% vs. 15.3% 15.0% vs. 15.3%, p = 0.47 15.3% vs. 15.2%, p = 0.85
p = 0.36
Drug Abuse 3.7% vs. 3.0%, p <0.001 3.6% vs. 3.1%, p = 0.002 4.0% vs. 3.1%, p <0.001
Smoking or nicotine dependence 18.5% vs. 19.8%, p <0.001 22.0% vs. 19.4%, p<0.001 22.8% vs. 19.4%, p <0.001
Alcohol Abuse 4.6% vs. 4.8%, p = 0.055 4.6% vs. 4.8%, p = 0.26 4.9% vs. 4.8%, p = 0.53
Depression 14.1% vs. 11.4%, p <0.001 12.4% vs. 12.0%, p = 0.22 12.4% vs. 12.0%, p = 0.28
Dementia 14.2% vs. 11.2%, p <0.001 10.6% vs. 12.0%, p<0.001 10.3% vs. 12.0%, p <0.001
Epilepsy 5.6% vs. 3.7%, p <0.001 4.1% vs. 4.1%, p = 0.90 4.1% vs. 4.1%, p = 1.00
Malignancy 6.8% vs. 4.3%, p <0.001 5.4% vs. 4.8%, P <0.001 5.6% vs. 4.8%, p <0.001
Rural (missing n = 988) a 13.0% vs. 15.2%, p <0.001 13.7% vs. 14.8%, p<0.001 13.2% vs. 14.8%, p <0.001
Discharge disposition p <0.001 p<0.001 p <0.001
Routine/home 31.3% vs. 45.2% 40.6% vs. 42.2% 41.0% vs. 42.2%
Short-term hospital 1.8% vs. 1.3% 2.9% vs. 1.3% 3.4% vs. 1.3%
Skilled facility/other 42.2% vs. 31.7% 30.9% vs. 34.2% 30.1% vs. 34.2%
Home health care 23.2% vs. 20.7% 23.3% vs. 21.1% 22.9% vs. 21.2%
AMA 1.6% vs. 1.1% 2.4% vs. 1.1% 2.6% vs. 1.1%
Hospital Type p <0.001 p = 0.016 p = 0.009
Urban non-teaching 19.1% vs. 18.8% 19.8% vs. 18.8% 19.9% vs. 18.8%
Urban teaching 75.2% vs. 75.1% 74.5% vs. 75.2% 74.5% vs. 75.2%
Rural 5.6% vs. 6.1% 5.7% vs. 6.0% 5.6% vs. 6.0%
Hospital Bedsize p <0.001 p = 0.037 p = 0.17
Small 15.1% vs. 15.8% 15.9% vs. 15.7% 16.1% vs. 15.7%
Medium 27.4% vs. 27.0% 27.9% vs. 27.1% 27.6% vs. 27.1%
Large 57.5% vs. 57.2% 56.2% vs. 57.3% 56.4% vs. 57.3%
Hospital Control p <0.001 p = 0.79 p = 0.33
Government 10.9% vs. 11.4% 11.2% vs. 11.3% 11.5% vs. 11.3%
Private, non-profit 75.4% vs. 76.2% 75.9% vs. 76.0% 76.2% vs. 76.0%
Private, for profit 13.6% vs. 12.4% 12.9% vs. 12.7% 12.3% vs. 12.7%
Open in a new tab

a Total missing data for 273,811 patients who survived their initial hospitalization. Data are complete for all variables where no missing data are reported

Patients with readmissions from any cause were more likely to be female, older, and had longer initial LOS. Individuals who were readmitted were also more likely to have traditional vascular risk factors (hypertension, diabetes, and AF) as well as other comorbidities like depression, dementia, epilepsy and a history of drug abuse and malignancy (Table 2). Readmission was more common amongst individuals in the lowest quartile of income, those with Medicaid insurance, those discharged from urban hospitals and those who had a discharge disposition other than home at the end of their index admission (e.g. left against medical advice, discharged to a skill nursing facility). All-cause readmissions were less common among those with a history of smoking or nicotine dependency. Variables associated with readmissions due to AIS or MACE were similar to those associated with all-cause readmissions, with the exception of age, stroke severity, history of atrial fibrillation and dementia, and initial length of stay (Table 2).

Variables associated with time to readmission from all causes are listed in Table 3.

Table 3. Association between variables and time to recurrent admission after AIS hospitalization (using a Weibull accelerated failure time model) in univariate analysis (Model 1) and after controlling for patient age, sex, NIHSS category, LOS, discharge disposition, hospital bed size, location/teaching status, and ownership (Model 2).

Variable Model 1 Coefficient exponentiated 95% CI Lower 95% CI Upper p-value Model 2 adjusted coefficient exponentiated 95% CI Lower 95% CI Upper p-value
Age 0.986 0.985 0.987 <0.001 0.990 a 0.989 0.992 <0.001
Female 0.883 0.861 0.907 <0.001 0.965 a 0.931 1.001 0.055
LOS quartile      
0–2 Ref - - - Ref - - -
3 0.614 0.588 0.640 <0.001 0.727 a 0.686 0.770 <0.001
4–7 0.447 0.431 0.463 <0.001 0.607 a 0.577 0.640 <0.001
>7 0.371 0.358 0.384 <0.001 0.516 a 0.489 0.545 <0.001
NIHSS        
<6 Ref - - - Ref - - -
6–10 0.675 0.644 0.708 <0.001 0.920 a 0.877 0.966 0.001
11–15 0.567 0.532 0.603 <0.001 0.868 a 0.814 0.925 <0.001
16–20 0.670 0.621 0.724 <0.001 1.105 a 1.022 1.196 0.013
>20 0.765 0.707 0.828 <0.001 1.315 a 1.212 1.427 <0.001
Hospital Type      
Urban non-teaching Ref - - - Ref - - -
Urban teaching 1.015 0.981 1.049 0.398 1.034 a 0.988 1.083 0.154
Rural 1.154 1.084 1.228 <0.001 1.083 a 0.981 1.196 0.116
Hospital Bedsize         - - - -
Small Ref - - - Ref - - -
Medium 0.914 0.877 0.953 <0.001 0.948 a 0.893 1.006 0.078
Large 0.928 0.893 0.963 <0.001 0.983 a 0.930 1.039 0.545
Hospital Control         - - - -
Government Ref - - - Ref - - -
Private, non-profit 0.952 0.913 0.993 0.024 0.990 a 0.933 1.050 0.728
Private, for profit 0.810 0.768 0.854 <0.001 0.827 a 0.769 0.888 <0.001
Discharge disposition                
Routine/home Ref - - - Ref - - -
Short-term hospital 0.293 0.265 0.324 <0.001 0.335 a 0.291 0.385 <0.001
Skilled facility/other 0.370 0.359 0.382 <0.001 0.496 a 0.472 0.521 <0.001
Home health care 0.481 0.464 0.498 <0.001 0.592 a 0.562 0.624 <0.001
AMA 0.309 0.278 0.344 <0.001 0.272 a 0.234 0.316 <0.001
Hypertension 0.679 0.651 0.707 <0.001 0.785 0.741 0.831 <0.001
Diabetes 0.588 0.573 0.604 <0.001 0.623 0.601 0.646 <0.001
Hyperlipidemia 0.967 0.941 0.993 0.014 0.960 0.925 0.997 0.033
Atrial fibrillation 0.627 0.609 0.645 <0.001 0.758 0.727 0.790 <0.001
Obesity 1.004 0.968 1.041 0.851 0.957 0.910 1.006 0.083
Drug abuse 0.722 0.674 0.774 <0.001 0.683 0.619 0.754 <0.001
Smoking or nicotine dependence 1.135 1.098 1.174 <0.001 0.966 0.921 1.013 0.154
Alcohol abuse 1.073 1.009 1.142 0.026 1.125 1.032 1.225 0.007
Depression 0.699 0.674 0.726 <0.001 0.761 0.723 0.801 <0.001
Dementia 0.672 0.647 0.698 <0.001 0.918 0.869 0.971 0.003
Epilepsy 0.542 0.512 0.574 <0.001 0.595 0.548 0.645 <0.001
Malignancy 0.464 0.441 0.489 <0.001 0.525 0.487 0.565 <0.001
Urban residence 0.764 0.735 -0.230 <0.001 0.782 0.739 0.827 <0.001
Median income                
≤47,999 0.848 0.815 0.881 <0.001 0.832 0.790 0.877 <0.001
48–60,999 0.947 0.911 0.986 0.007 0.919 0.870 0.969 0.002
61–81,999 0.996 0.957 1.038 0.863 0.993 0.941 1.048 0.794
≥82,000 Ref - - - Ref - - -
Insurance status                
Medicare 0.395 0.379 0.412 <0.001 0.470 0.442 0.502 <0.001
Medicaid 0.421 0.398 0.445 <0.001 0.483 0.448 0.522 <0.001
Private Ref - - - Ref - - -
Self-pay/other 0.772 0.720 0.829 <0.001 0.780 0.710 0.856 <0.001
Open in a new tab

a Coefficients in base multi-variable model with patient age, sex, NIHSS category, LOS, discharge disposition, hospital bed size, location/teaching status, and ownership without other co-variates

Stroke severity had a non-monotonic effect on risk of all-cause readmission (Table 3 and Fig 2, panel E). Time to readmission was longest in those with NIHSS <6 and shortest in those with NIHSS 11–15. The comorbidity with the strongest association with shortened time to readmission was malignancy, where time to readmission was 0.46 (95% CI 0.44–0.49) times as long as patients without malignancy. Other co-morbidities associated with markedly shorter times to readmission include hypertension (0.68, 95%CI 0.65–0.71), diabetes (0.59, 95%CI 0.57–0.60), AF (0.63, 95%CI 0.61–0.64), drug abuse (0.72, 95%CI 0.67–0.77), depression (0.70, 95%CI 0.67–0.73), dementia (0.67, 95%CI 0.65–0.70), and epilepsy (0.54, 95%CI 0.51–0.57).

Social determinants associated with shorter time to readmission included having Medicare (0.40, 95%CI 0.38–0.41) or Medicaid (0.42, 95%CI 0.40–0.45) insurance compared to private insurance (reference), living in urban settings (0.764, 95%CI 0.74–0.80) and for those living in zip codes with the lowest (0.84, 95% CI 0.82–0.88) and second lowest quartile (0.95, 95%CI0.91–0.99) of median income.

A history of smoking or nicotine dependence was associated with increased time to readmission on univariate analysis but not in multivariable analysis (Model 2). All other associations with medical co-morbidities remained significant in multivariable analysis.

Discussion

Our analysis of readmissions following ischemic stroke in the US using the 2019 NRD revealed that unplanned hospital readmissions were frequent, with the most common reasons for readmission being recurrent stroke and post stroke sequalae, followed by sepsis and acute renal failure. We found multiple patient-level factors that are associated with increased hospital readmission including vascular risk factors (hypertension, diabetes, AF), drug abuse, epilepsy, depression, dementia and malignancy as well social determinants (insurance status, residing in urban areas and zip codes with low median income).

Hospital readmission within 30 days has been defined by the centers of Medicare and Medicaid services as an indicator of poor inpatient care and has been linked to payment determination via the Hospital Readmissions Reduction Program (HRRP) [14]. Our results show that in 2019, rates of all cause 30-day readmissions for patients with AIS was 9.7%. Previously, authors studying the 2010–2015 NRD [7] reported an all-cause 30-day readmission rate of 12% with a decreasing trend during the study period consistent with our lower point estimate. Other US estimates vary widely between 6.4% [15] from a large single center cohort up to 21% using 2008 national Medicare claims data [16]. Differences in rates of readmissions by insurance status (with higher readmissions among those insured with Medicare) and by hospital characteristics, as shown in our analyses, likely explain some of the discrepancies between studies.

Causes for hospital readmissions are complex and may be due to failure to plan for post stroke needs and to implement appropriate stroke prevention, the development of a post-stroke sequelae, or new issues separate from the patient’s initial stroke. In our study, more than 50% of all 30-day readmissions could be attributed to ten causes. Ischemic stroke or post stroke sequelae accounted for 25% of readmissions with a further 1.8% of readmissions from hemorrhagic stroke, a proportion of which may represent hemorrhagic transformation. The next most common cause of readmission was from infections, which accounted for 10.3% of readmissions at 30 days, followed by acute renal failure, MI and complications of vascular risk factors such as hypertension, diabetes, and AF. In a 2016 meta-analysis of 7 US, 2 Chinese and 1 Norwegian studies of hospital readmission within 30 days after stroke, the most common causes for readmission were infection (19.9%), coronary artery disease (17.8%) and recurrent stroke (16.0%) respectively [2]. Our findings and other prior research suggest that the highest yield interventions in transitional care after stroke hospitalization would likely include improved infection prevention strategies at discharge and careful transition of responsibility for vascular risk factor management from the inpatient to outpatient settings [17].

Causes for more adverse outcomes following stroke in women compared to men have previously been described and include older age and more severe strokes [18–21]. We found that the higher rate of all cause readmission in women was no longer significant after controlling for stroke severity, age, and hospital characteristics, suggesting that a sex-based difference is likely confounded.

All cause readmission as well as readmission for AIS and MACE was higher amongst patients living in zip codes with the lowest quartile of median income and for patients treated in urban hospitals. Higher 30-day readmissions at “safety net hospitals” located in poor and underserved communities has previously been noted when examining multiple HRRP targeted conditions within the US [22]. These findings highlight persistent health disparities in stroke care and aftercare, which are challenging to address on a policy level. Safety-net hospitals treating vulnerable patients have suffered disproportionally higher penalties under HRRP [23]. However, adjusting penalties based on the income of treated patients may unintentionally entrench lower standards for care provided to disadvantaged populations [24].

When examining only readmissions due to AIS or MACE, readmitted patients appear paradoxically healthier (younger age, less severe baseline strokes, shorter initial LOS and lower rates of AF and dementia) compared to those not readmitted. This effect was not seen when looking at all cause readmissions and may be explained by the fact that patients readmitted for AIS have fewer co-morbidities that could take coding preference during the readmission compared to patients readmitted for other causes [25]. These findings should not be taken to suggest higher stroke recurrence in younger patients or those without AF but rather that younger and healthier patients have less complicated readmissions where stroke recurrence or stroke complications are more likely to be the principal diagnosis.

Our study has limitations. Firstly, the NRD does not allow for linkage of admissions across multiple years or admissions of the same patient occurring across multiple states. We used a novel approach to modelling recurrent ischemic stroke admission in the US using time-to-event analysis, which accurately models the time to right-censoring allowing us to make use of all available data unlike previous works using NRD data that focused solely on risk factors related to 30 days readmissions. However, the data structure still limits the duration of follow up and makes it challenging to study preceding admissions. Our current study examines patients following their first admission with AIS in 2019 but we were unable to differentiate between first ever stroke vs. recurrent stroke admissions in patients with stroke admissions before 2019.

Secondly, for readmissions due to AIS (ICD code I63*), which accounts for more than 20% of readmissions before 30 days, there is no reliable way of differentiating between incident new AIS vs. readmission due to the index event for psycho-social reasons or recrudescence of symptoms. Among AIS readmissions with a documented NIHSS, the NIHSS on readmission is slightly higher compared to the index admission (6.7 vs. 5.9) suggesting worsening of neurological function and the validity of a recurrent stroke diagnosis. Similarly, there has been no validated methods of examining for hemorrhagic transformation during AIS admissions so the effect of this important prognostic factor could not be analyzed.

We found hospital readmission from MACE occurred in 2.6% of patients at 30 days and in 7.6% of patients at 1 year after discharge. Within MACE, readmissions due to AIS occurred in 2.2% and 5.8% of patients at 30 days and 1 year respectively. Rates of recurrent stroke vary significantly based on stroke etiology. Within the POINT study, of patients with minor stroke and high-risk TIA, the recurrence of ischemic stroke, MI, or ischemic vascular death occurred in 4.9% of patients at 30 days [26]. In THALES, the rate of ischemic stroke at 30 days was 5.6%. Annual stroke recurrence rates reported from clinical trials vary from 2.4%, as seen in the SPS3 trial [27] of patients with lacunar infarction, up to 15%, seen in the WASID [28] and SAMMPRIS [29] trials of patients with intracranial atherosclerotic disease. Rates of recurrent stroke within 5 years after first stroke also varies significantly by age, sex and ethnicity ranging from 5% in black men under 65 years old to 22% in black women aged 65–74 [1]. Unfortunately, the NRD does not contain data on race/ethnicity to examine for differences within these subgroups.

Additionally, there is no way of linking anonymized NRD data to other data sources to obtain information on out-of-hospital mortality. In our study, patients with very high NIHSS at their baseline admissions had lower rehospitalization rates which was likely due to increased rates of out of hospital death following discharge. Out of hospital deaths can account for up to 78% of deaths after stroke at 30 days and 49% of deaths at 1 year [30]. Out of hospital death information would have also enabled competing risk analyses [31] and for a more accurate description of cardiovascular death in the MACE composite. Finally, the cohort being studied was identified using ICD diagnosis codes which is subject to human error during the coding process. However, previous work shows that codes for both AIS and AIS risk factors have acceptable accuracy in administrative data for hospitalized patients [32].

Conclusion

Unplanned readmission to US hospitals following AIS was high in 2019, with 9.7% of patients being readmitted within 30 days of discharge and 30.5% readmitted by 1 year. The most common causes of readmission were recurrent stroke and post stroke sequelae, followed by sepsis and acute renal failure. Our findings also highlight significant continued health disparities by social determinants of health with higher readmissions among those living in zip-codes with the lowest median income, those living in urban settings and those with Medicare insurance.

Supporting information

S1 Fig. Example diagnostic plots showing violations of assumption of proportional hazards for hypertension with top panel showing non-parallel lines on log-log plot of survival and bottom panel showing increasing scaled Schoenfeld residuals over time (more prominent in early period).

(PDF)

Click here for additional data file. (245.5KB, pdf)
S1 Table. Principle diagnosis at readmission after discharge with ischemic stroke by ICD category.

(PDF)

Click here for additional data file. (178.9KB, pdf)

Acknowledgments

We would like to acknowledge all of the Healthcare Cost and Utilization Project Data Partners that contributed to the data made available for analysis. More information can be found at www.hcup-us.ahrq.gov/hcupdatapartners.jsp

Data Availability

The data used is publicly available administrative data. It can be purchased for use by researchers after appropriate safety and confidentiality training from the Healthcare Cost and Utilization Project Agency for Healthcare Research and Quality. If a journal or publication is interested in access to data or analytic files, that information including data elements and data structure can be found here: https://hcup-us.ahrq.gov/nrdoverview.jsp.

Funding Statement

LWZ received salary support from a project grant from the Canadian Institute of Health Research (RN387091 - 420683). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. ML has no relevant funding to disclose. AdH is funded by NIH-NINDS K23NS105924. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

References

  • 1.Virani SS, Alonso A, Benjamin EJ, Bittencourt MS, Callaway CW, Carson AP, et al. Heart Disease and Stroke Statistics—2020 Update: A Report From the American Heart Association. Circulation. 2020;141: e139–e596. doi: 10.1161/CIR.0000000000000757 [DOI] [PubMed] [Google Scholar]
  • 2.Zhong W, Geng N, Wang P, Li Z, Cao L. Prevalence, causes and risk factors of hospital readmissions after acute stroke and transient ischemic attack: a systematic review and meta-analysis. Neurol Sci. 2016;37: 1195–1202. doi: 10.1007/s10072-016-2570-5 [DOI] [PubMed] [Google Scholar]
  • 3.Andrews RM, Elixhauser A. The National Hospital Bill: Growth Trends and 2005 Update on the Most Expensive Conditions by Payer: Statistical Brief #42. Healthcare Cost and Utilization Project (HCUP) Statistical Briefs. Rockville (MD): Agency for Healthcare Research and Quality (US); 2006. Available: http://www.ncbi.nlm.nih.gov/books/NBK56314/ [PubMed] [Google Scholar]
  • 4.Ovbiagele B, Goldstein LB, Higashida RT, Howard VJ, Johnston SC, Khavjou OA, et al. Forecasting the Future of Stroke in the United States. Stroke. 2013;44: 2361–2375. doi: 10.1161/STR.0b013e31829734f2 [DOI] [PubMed] [Google Scholar]
  • 5.Dewilde S, Annemans L, Peeters A, Hemelsoet D, Vandermeeren Y, Desfontaines P, et al. The relationship between Home-time, quality of life and costs after ischemic stroke: the impact of the need for mobility aids, home and car modifications on Home-time. Disabil Rehabil. 2020;42: 419–425. doi: 10.1080/09638288.2018.1501438 [DOI] [PubMed] [Google Scholar]
  • 6.NRD Overview. [cited 27 Jun 2022]. Available: https://www.hcup-us.ahrq.gov/nrdoverview.jsp
  • 7.Bambhroliya AB, Donnelly JP, Thomas EJ, Tyson JE, Miller CC, McCullough LD, et al. Estimates and Temporal Trend for US Nationwide 30-Day Hospital Readmission Among Patients With Ischemic and Hemorrhagic Stroke. JAMA Netw Open. 2018;1: e181190. doi: 10.1001/jamanetworkopen.2018.1190 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 8.Vahidy FS, Donnelly JP, McCullough LD, Tyson JE, Miller CC, Boehme AK, et al. Nationwide Estimates of 30-Day Readmission in Patients With Ischemic Stroke. Stroke. 2017;48: 1386–1388. doi: 10.1161/STROKEAHA.116.016085 [DOI] [PubMed] [Google Scholar]
  • 9.Healthcare Cost and Utilization Project (HCUP) NRD Notes. [cited 14 Mar 2022]. Available: https://www.hcup-us.ahrq.gov/db/vars/cmr_aaa/nrdnote.jsp
  • 10.Patel K, Kay R, Rowell L. Comparing proportional hazards and accelerated failure time models: an application in influenza. Pharm Stat. 2006;5: 213–224. doi: 10.1002/pst.213 [DOI] [PubMed] [Google Scholar]
  • 11.Healthcare Cost and Utilization Project (HCUP) NIS Notes. [cited 26 Jan 2021]. Available: https://www.hcup-us.ahrq.gov/db/vars/hosp_bedsize/nisnote.jsp
  • 12.ICD-10-CM Official Guidelines for Coding and Reporting. 2019 [cited 25 Jan 2021]. Available: https://www.cms.gov/Medicare/Coding/ICD10/Downloads/2019-ICD10-Coding-Guidelines-.pdf
  • 13.de Havenon A, Sheth KN, Johnston KC, Anadani M, Yaghi S, Tirschwell D, et al. Effect of Adjusting for Baseline Stroke Severity in the National Inpatient Sample. Stroke. 2021;52: e739–e741. doi: 10.1161/STROKEAHA.121.035112 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 14.Hospital Readmissions Reduction Program (HRRP) | CMS. [cited 12 Jun 2022]. Available: https://www.cms.gov/medicare/medicare-fee-for-service-payment/acuteinpatientpps/readmissions-reduction-program
  • 15.Nahab F, Takesaka J, Mailyan E, Judd L, Culler S, Webb A, et al. Avoidable 30-Day Readmissions Among Patients With Stroke and Other Cerebrovascular Disease. The Neurohospitalist. 2012;2: 7–11. doi: 10.1177/1941874411427733 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 16.Fehnel CR, Lee Y, Wendell LC, Thompson BB, Potter NS, Mor V. Post–Acute Care Data for Predicting Readmission After Ischemic Stroke: A Nationwide Cohort Analysis Using the Minimum Data Set. J Am Heart Assoc Cardiovasc Cerebrovasc Dis. 2015;4: e002145. doi: 10.1161/JAHA.115.002145 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 17.Puhr MI, Thompson HJ. The Use of Transitional Care Models in Patients With Stroke. J Neurosci Nurs J Am Assoc Neurosci Nurses. 2015;47: 223–234. doi: 10.1097/JNN.0000000000000143 [DOI] [PubMed] [Google Scholar]
  • 18.Kilkenny MF, Dalli LL, Kim J, Sundararajan V, Andrew NE, Dewey HM, et al. Factors Associated With 90-Day Readmission After Stroke or Transient Ischemic Attack. Stroke. 2020. [cited 8 Jun 2022]. doi: 10.1161/STROKEAHA.119.026133 [DOI] [PubMed] [Google Scholar]
  • 19.Ramchand P, Thibault DP, Crispo JA, Levine J, Hurst R, Mullen MT, et al. Readmissions After Mechanical Thrombectomy for Acute Ischemic Stroke in the United States: A Nationwide Analysis. J Stroke Cerebrovasc Dis Off J Natl Stroke Assoc. 2018;27: 2632–2640. doi: 10.1016/j.jstrokecerebrovasdis.2018.05.035 [DOI] [PubMed] [Google Scholar]
  • 20.Eriksson M, Åsberg S, Sunnerhagen KS, von Euler M, null null. Sex Differences in Stroke Care and Outcome 2005–2018. Stroke. 2021;52: 3233–3242. doi: 10.1161/STROKEAHA.120.033893 [DOI] [PubMed] [Google Scholar]
  • 21.Gall SL, Tran PL, Martin K, Blizzard L, Srikanth V. Sex Differences in Long-Term Outcomes After Stroke. Stroke. 2012;43: 1982–1987. doi: 10.1161/STROKEAHA.111.632547 [DOI] [PubMed] [Google Scholar]
  • 22.Murray F, Allen M, Clark CM, Daly CJ, Jacobs DM. Socio-demographic and -economic factors associated with 30-day readmission for conditions targeted by the hospital readmissions reduction program: a population-based study. BMC Public Health. 2021;21: 1922. doi: 10.1186/s12889-021-11987-z [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 23.Banerjee S, Paasche-Orlow MK, McCormick D, Lin M-Y, Hanchate AD. Readmissions performance and penalty experience of safety-net hospitals under Medicare’s Hospital Readmissions Reduction Program. BMC Health Serv Res. 2022;22: 338. doi: 10.1186/s12913-022-07741-9 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 24.Krumholz HM, Bernheim SM. Considering the Role of Socioeconomic Status in Hospital Outcomes Measures. Ann Intern Med. 2014;161: 833–834. doi: 10.7326/M14-2308 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 25.Nouh AM, McCormick L, Modak J, Fortunato G, Staff I. High Mortality among 30-Day Readmission after Stroke: Predictors and Etiologies of Readmission. Front Neurol. 2017;8: 632. doi: 10.3389/fneur.2017.00632 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 26.Johnston SC, Easton JD, Farrant M, Barsan W, Conwit RA, Elm JJ, et al. Clopidogrel and Aspirin in Acute Ischemic Stroke and High-Risk TIA. N Engl J Med. 2018;379: 215–225. doi: 10.1056/NEJMoa1800410 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 27.SPS3 Investigators, Benavente OR, Hart RG, McClure LA, Szychowski JM, Coffey CS, et al. Effects of clopidogrel added to aspirin in patients with recent lacunar stroke. N Engl J Med. 2012;367: 817–825. doi: 10.1056/NEJMoa1204133 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 28.Chimowitz MI, Lynn MJ, Howlett-Smith H, Stern BJ, Hertzberg VS, Frankel MR, et al. Comparison of Warfarin and Aspirin for Symptomatic Intracranial Arterial Stenosis. N Engl J Med. 2005;352: 1305–1316. doi: 10.1056/NEJMoa043033 [DOI] [PubMed] [Google Scholar]
  • 29.Derdeyn CP, Chimowitz MI, Lynn MJ, Fiorella D, Turan TN, Janis LS, et al. Aggressive medical treatment with or without stenting in high-risk patients with intracranial artery stenosis (SAMMPRIS): the final results of a randomised trial. Lancet Lond Engl. 2014;383: 333–341. doi: 10.1016/S0140-6736(13)62038-3 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 30.Bravata DM, Ho S-Y, Meehan TP, Brass LM, Concato J. Readmission and Death After Hospitalization for Acute Ischemic Stroke. Stroke. 2007;38: 1899–1904. doi: 10.1161/STROKEAHA.106.481465 [DOI] [PubMed] [Google Scholar]
  • 31.Prentice RL, Kalbfleisch JD, Peterson AV, Flournoy N, Farewell VT, Breslow NE. The Analysis of Failure Times in the Presence of Competing Risks. Biometrics. 1978;34: 541–554. doi: 10.2307/2530374 [DOI] [PubMed] [Google Scholar]
  • 32.Kokotailo RA, Hill MD. Coding of Stroke and Stroke Risk Factors Using International Classification of Diseases, Revisions 9 and 10. Stroke. 2005;36: 1776–1781. doi: 10.1161/01.STR.0000174293.17959.a1 [DOI] [PubMed] [Google Scholar]

Decision Letter 0

Karthik Gangu

19 May 2023

PONE-D-23-02024Rates and Reasons for Hospital Readmission after Acute Ischemic Stroke in a US Population-Based CohortPLOS ONE

Dear Dr. Zhou,

Thank you for submitting your manuscript to PLOS ONE. After careful consideration, we feel that it has merit but does not fully meet PLOS ONE’s publication criteria as it currently stands. Therefore, we invite you to submit a revised version of the manuscript that addresses the points raised during the review process.

Please submit your revised manuscript by June 2nd 2023. If you will need more time than this to complete your revisions, please reply to this message or contact the journal office at plosone@plos.org. When you're ready to submit your revision, log on to https://www.editorialmanager.com/pone/ and select the 'Submissions Needing Revision' folder to locate your manuscript file.

Please include the following items when submitting your revised manuscript:

  • A rebuttal letter that responds to each point raised by the academic editor and reviewer(s). You should upload this letter as a separate file labeled 'Response to Reviewers'.

  • A marked-up copy of your manuscript that highlights changes made to the original version. You should upload this as a separate file labeled 'Revised Manuscript with Track Changes'.

  • An unmarked version of your revised paper without tracked changes. You should upload this as a separate file labeled 'Manuscript'.

If you would like to make changes to your financial disclosure, please include your updated statement in your cover letter. Guidelines for resubmitting your figure files are available below the reviewer comments at the end of this letter.

If applicable, we recommend that you deposit your laboratory protocols in protocols.io to enhance the reproducibility of your results. Protocols.io assigns your protocol its own identifier (DOI) so that it can be cited independently in the future. For instructions see: https://journals.plos.org/plosone/s/submission-guidelines#loc-laboratory-protocols. Additionally, PLOS ONE offers an option for publishing peer-reviewed Lab Protocol articles, which describe protocols hosted on protocols.io. Read more information on sharing protocols at https://plos.org/protocols?utm_medium=editorial-email&utm_source=authorletters&utm_campaign=protocols.

We look forward to receiving your revised manuscript.

Kind regards,

Karthik Gangu

Academic Editor

PLOS ONE

Journal requirements:

When submitting your revision, we need you to address these additional requirements.

1. Please ensure that your manuscript meets PLOS ONE's style requirements, including those for file naming. The PLOS ONE style templates can be found at

https://journals.plos.org/plosone/s/file?id=wjVg/PLOSOne_formatting_sample_main_body.pdf and

https://journals.plos.org/plosone/s/file?id=ba62/PLOSOne_formatting_sample_title_authors_affiliations.pdf.

2. We note that the grant information you provided in the ‘Funding Information’ and ‘Financial Disclosure’ sections do not match. When you resubmit, please ensure that you provide the correct grant numbers for the awards you received for your study in the ‘Funding Information’ section.

3. Thank you for stating the following in the Acknowledgments/ Funding Section of your manuscript:

“LWZ received salary support from a fellowship grant from StrokeNet (NINDS U24NS107220) and a project grant from the Canadian Institute of Health Research (RN387091 - 420683). Dr. de Havenon is funded by NIH-NINDS K23NS105924.”

We note that you have provided funding information that is not currently declared in your Funding Statement. However, funding information should not appear in the Acknowledgments section or other areas of your manuscript. We will only publish funding information present in the Funding Statement section of the online submission form.

Please remove any funding-related text from the manuscript and let us know how you would like to update your Funding Statement. Currently, your Funding Statement reads as follows:

 “LWZ received salary support from a fellowship grant from StrokeNet (NINDS U24NS107220) and a project grant from the Canadian Institute of Health Research (RN387091 - 420683).  The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

ML has no relevant funding to disclose.

AdH is funded by NIH-NINDS K23NS105924.  The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.”

Please include your amended statements within your cover letter; we will change the online submission form on your behalf.

4. Thank you for stating the following in the Competing Interests section:

“Dr. de Havenon has received investigator initiated clinical research funding from Regeneron, AMGEN, and AMAG pharmaceuticals, has received consultant fees from Integra and Novo Nordisk, has equity in TitinKM and Certus, and receives author fees from UpToDate.”

Please confirm that this does not alter your adherence to all PLOS ONE policies on sharing data and materials, by including the following statement: ""This does not alter our adherence to  PLOS ONE policies on sharing data and materials.” (as detailed online in our guide for authors http://journals.plos.org/plosone/s/competing-interests).  If there are restrictions on sharing of data and/or materials, please state these. Please note that we cannot proceed with consideration of your article until this information has been declared.

Please include your updated Competing Interests statement in your cover letter; we will change the online submission form on your behalf.

5. In your Data Availability statement, you have not specified where the minimal data set underlying the results described in your manuscript can be found. PLOS defines a study's minimal data set as the underlying data used to reach the conclusions drawn in the manuscript and any additional data required to replicate the reported study findings in their entirety. All PLOS journals require that the minimal data set be made fully available. For more information about our data policy, please see http://journals.plos.org/plosone/s/data-availability.

Upon re-submitting your revised manuscript, please upload your study’s minimal underlying data set as either Supporting Information files or to a stable, public repository and include the relevant URLs, DOIs, or accession numbers within your revised cover letter. For a list of acceptable repositories, please see http://journals.plos.org/plosone/s/data-availability#loc-recommended-repositories. Any potentially identifying patient information must be fully anonymized.

Important: If there are ethical or legal restrictions to sharing your data publicly, please explain these restrictions in detail. Please see our guidelines for more information on what we consider unacceptable restrictions to publicly sharing data: http://journals.plos.org/plosone/s/data-availability#loc-unacceptable-data-access-restrictions. Note that it is not acceptable for the authors to be the sole named individuals responsible for ensuring data access.

We will update your Data Availability statement to reflect the information you provide in your cover letter.

6. We note that you have indicated that data from this study are available upon request. PLOS only allows data to be available upon request if there are legal or ethical restrictions on sharing data publicly. For more information on unacceptable data access restrictions, please see http://journals.plos.org/plosone/s/data-availability#loc-unacceptable-data-access-restrictions.

In your revised cover letter, please address the following prompts:

a) If there are ethical or legal restrictions on sharing a de-identified data set, please explain them in detail (e.g., data contain potentially sensitive information, data are owned by a third-party organization, etc.) and who has imposed them (e.g., an ethics committee). Please also provide contact information for a data access committee, ethics committee, or other institutional body to which data requests may be sent.

b) If there are no restrictions, please upload the minimal anonymized data set necessary to replicate your study findings as either Supporting Information files or to a stable, public repository and provide us with the relevant URLs, DOIs, or accession numbers. For a list of acceptable repositories, please see http://journals.plos.org/plosone/s/data-availability#loc-recommended-repositories.

We will update your Data Availability statement on your behalf to reflect the information you provide.

7. Please note that in order to use the direct billing option the corresponding author must be affiliated with the chosen institute. Please either amend your manuscript to change the affiliation or corresponding author, or email us at plosone@plos.org with a request to remove this option.

[Note: HTML markup is below. Please do not edit.]

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #1: Partly

Reviewer #2: Yes

Reviewer #3: Partly

**********

2. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: Yes

Reviewer #2: I Don't Know

Reviewer #3: Yes

**********

3. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #1: Yes

Reviewer #2: Yes

Reviewer #3: No

**********

4. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #1: Yes

Reviewer #2: Yes

Reviewer #3: Yes

**********

5. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #1: 1. In methods section, this sentence needs to be corrected: "Secondary outcomes included readmission with 1) recurrent AIS (ICD 10 codes I63*) as the principal discharge diagnosis, and 2) readmissions due

to Major Adverse Cardiovascular Events (MACE) defined as: 1) AIS, 2) transient

ischemic attack (TIA) (G45*, G46*), 3) Peripheral Vascular Disease (I7*), 4) myocardial

infarction (MI) (I21*, I22*) or 5) cardiac arrest (I46*) as the principal discharge diagnosis

or 6) a hospital readmission ending in death with a principal discharge diagnosis starting

with I (diseases of the circulatory system) to capture cardiovascular death". It is not readmission outcome with discharge diagnosis, rather it is readmission outcome with principal readmission diagnosis. Also correct in abstract and in other parts of manuscript.

2. Authors should explain in details how they got NIHSS scale in methodology.

3. In Abstract, please mention the cut off for age (older) and LOS.

4. In Abstract, please mention the comparison groups of social determinants of health.

5. ICD coding errors are major limitations of National databases and it should be mentioned in the limitations section.

6. Authors should mention the missing data in Figure-1.

Reviewer #2: The manuscript is well written and highlights one major reasons for hospital readmission. A few ideas to include in the outcomes includes, the cost of readmission. Previous studies have been done on national readmission database have compared the trends over time from 2010-2025. It would be interesting to see the trends thereafter from 2016-2020, whether there was any impact of COVID-19 infection on readmission post stroke. Lastly, the tables needs to be accurately labelled and missing values needs to be updated. They can cut short the tables to include the variables relevant to the study.

Reviewer #3: Overall, this peer-reviewed article provides valuable insights into the rates, causes, and risk factors associated with hospital readmissions following acute ischemic stroke in the United States. The reviewer has following concerns:

1.Multiple studies have been published on the 30-day readmission rate following acute ischemic stroke using the same database. The authors should address how their manuscript differs from the already published literature. Additionally, they should consider trend analysis in the 30-day readmission rate, as it decreased from 12% in 2013 to 9.7% in 2019.

https://jamanetwork.com/journals/jamanetworkopen/fullarticle/2696869

https://www.ahajournals.org/doi/full/10.1161/STROKEAHA.116.016085

2.Can the authors include the number of patients who had hemorrhagic conversion following AIS during the index stay?

3.The NRD does not track patients beyond the calendar year. For example, patients admitted in December 2019 cannot be tracked for 30-day readmission as there is no link between the year 2019 and 2020. Therefore, December month admissions should be excluded because the 30-day readmission rate cannot be calculated. Many studies have been published using the same methodology. The authors should clarify if they have excluded December month admissions and add it to the limitations.

4.The NRD provides information regarding inter-hospital transfers. If a patient was admitted to a small hospital and transferred to a large hospital, there is a small chance that the patient might be included twice. Therefore, inter-hospital transfers must be excluded. If the authors have excluded them, please clarify in the methods section. If they have not excluded them, please provide a reason for not doing so.

**********

6. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: No

Reviewer #2: No

Reviewer #3: No

**********

[NOTE: If reviewer comments were submitted as an attachment file, they will be attached to this email and accessible via the submission site. Please log into your account, locate the manuscript record, and check for the action link "View Attachments". If this link does not appear, there are no attachment files.]

While revising your submission, please upload your figure files to the Preflight Analysis and Conversion Engine (PACE) digital diagnostic tool, https://pacev2.apexcovantage.com/. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Registration is free. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email PLOS at figures@plos.org. Please note that Supporting Information files do not need this step.

PLoS One. 2023 Aug 3;18(8):e0289640. doi: 10.1371/journal.pone.0289640.r002

Author response to Decision Letter 0

5 Jul 2023

The authors of the present study does not have any special access privileges in accessing this publicly available dataset which other interested researchers would not have. Similarly, we do not have additional information on methodology beyond what is publicly available on the HCUP website: https://hcup-us.ahrq.gov/nrdoverview.jsp. Please see below:

"The Nationwide Readmissions Database (NRD) is part of a family of databases and software tools developed for the Healthcare Cost and Utilization Project (HCUP). The NRD is a unique and powerful database designed to support various types of analyses of national readmissions for all patients, regardless of the expected payer for the hospital stay. This database addresses a large gap in healthcare data - the lack of nationally representative information on hospital readmissions for all ages. Unweighted, the 2020 NRD contains data from approximately 17 million discharges. Weighted, it estimates roughly 32 million discharges. Developed through a Federal-State-Industry partnership sponsored by the Agency for Healthcare Research and Quality (AHRQ), HCUP data inform decision making at the national, State, and community levels."

With regards to data sharing, the data used is publicly available administrative data. It can be purchased for use by researchers after appropriate safety and confidentiality training from the Healthcare Cost and Utilization Project Agency for Healthcare Research and Quality. We are unable to upload the minimal anonymized data set necessary to replicate the study findings. As a condition of access within the data use agreement, authors agreed that “I will not redistribute HCUP data by posting on any website or publishing in any other publicly accessible online repository. If a journal or publication requests access to data or analytic files, I will cite restrictions on data sharing in this Data Use Agreement and direct them to AHRQ HCUP (www.hcup-us.ahrq.gov) for more information on accessing HCUP data.”

Many thanks to the editor and reviewers for comments to strengthen our paper. We’ve responded to the Reviewer’s Comments below and revised the manuscript with changes below.

Reviewer #1: 1. In methods section, this sentence needs to be corrected: "Secondary outcomes included readmission with 1) recurrent AIS (ICD 10 codes I63*) as the principal discharge diagnosis, and 2) readmissions due to Major Adverse Cardiovascular Events (MACE) defined as: 1) AIS, 2) transient

ischemic attack (TIA) (G45*, G46*), 3) Peripheral Vascular Disease (I7*), 4) myocardial infarction (MI) (I21*, I22*) or 5) cardiac arrest (I46*) as the principal discharge diagnosis or 6) a hospital readmission ending in death with a principal discharge diagnosis starting with I (diseases of the circulatory system) to capture cardiovascular death". It is not readmission outcome with discharge diagnosis, rather it is readmission outcome with principal readmission diagnosis. Also correct in abstract and in other parts of manuscript.

Thank you for this recommendation for clarification. The manuscript now reads:

The secondary outcome was readmission due to recurrent AIS (ICD 10 codes I63*) as the principal readmission diagnosis. The tertiary outcome was readmission due to Major Adverse Cardiovascular Events (MACE) defined as: 1) AIS, 2) transient ischemic attack (TIA) (G45*, G46*), 3) Peripheral Vascular Disease (I7*), 4) myocardial infarction (MI) (I21*, I22*) or 5) cardiac arrest (I46*) as the principal readmission diagnosis or 6) a hospital readmission ending in death with a principal readmission diagnosis starting with I (diseases of the circulatory system) to capture cardiovascular death.

Similar terminology does not appear in the abstract or in other locations of the manuscript.

2. Authors should explain in detail how they got NIHSS scale in methodology.

The NIHSS scores are submitted from admitting hospitals to the National Readmission Database as a component of administrative data and are extracted from patient records. The following has been added to the methods for clarification.

“…. baseline National Institute of Health Stroke Scale (NIHSS), made available after October of 2016 by the Centers for Medicare & Medicaid Services (using ICD-10-CM code (R29.7*) to be used in conjunction with the diagnosis of AIS (I63*). Admission NIHSS scores are submitted from the admitting hospital’s medical record to the National Readmission Database as a component of administrative data.[12] Adjustment for stroke severity using NIHSS submitted in this form to the Healthcare Cost and Utilization Project has previously been shown to have important implications in analysis of the association between poststroke outcome and patient sex or stroke interventions.[13]"

3. In Abstract, please mention the cut off for age (older) and LOS.

Based on this comment we realized that our writing was unclear and suggested that we used cutoffs, which is not the case. Instead, we meant to report differences in means. We have tried to clarify this in the revised abstract which now reads: “….patients at increased risk of readmission were older (71.6 vs. 69.8 years, p<0.001) and had longer initial lengths of stay (7.6 vs. 6.2 days, p<0.001).”

4. In Abstract, please mention the comparison groups of social determinants of health.

This abstract now reads: “Social determinants associated with increased readmission included living in an urban (vs. rural) setting, living in zip-codes with the lowest median income, and having Medicare insurance.” These are the main three social determinants. Other insurance status groups (Medicaid, private, self-pay or other) are listed in the main text but not included in the abstract due to word count limitations (300 words max).

The abstract has been revised in other locations to make room within the word limitations for these additions.

5. ICD coding errors are major limitations of National databases, and it should be mentioned in the limitations section.

This has been acknowledged with the following in the discussion: “Finally, the cohort being studied was identified using ICD diagnosis codes which is subject to human error during the coding process. However, previous work shows that codes for both AIS and AIS risk factors have acceptable accuracy in administrative data for hospitalized patients.[32]"

6. Authors should mention the missing data in Figure-1.

There is no missing data for Figure 1 as all patients had complete information on death at discharge and all other diagnostic information is coded only when present. This has been clarified at the bottom of the figure.

“a There is no missing data on mortality at discharge within this cohort”

Reviewer #2: The manuscript is well written and highlights one major reasons for hospital readmission. A few ideas to include in the outcomes includes, the cost of readmission.

Thank you for taking the time to improve this manuscript. The following has been added to our manuscript:

Of the 285,451 initial stroke hospitalizations in the 2019 NRD (median LOS 3 days, IQR 3-7 days; median hospitalization charge $48 504 USD, IQR $27 226-91 089 USD), 4.1% resulted in mortality (Fig 1). Of the 273,811 patients who survived their initial hospitalization, 22.2% were readmitted to hospital within 2019 (median LOS 4 days, IQR 2-7; median hospitalization charge $40 105, IQR $22 115- 76 600 USD).

Previous studies have been done on national readmission database have compared the trends over time from 2010-2025. It would be interesting to see the trends thereafter from 2016-2020, whether there was any impact of COVID-19 infection on readmission post stroke.

We agree these are important elements to consider in our analysis. However, we exclusively analyzed the 2019 National Readmission Data (NRD) for this manuscript so are unable to conduct a temporal trend analysis. Additionally, information from COVID cannot be captured with 2019 NRD data (as the code for COVID, U071, was introduced in HCUP data starting in 2020).

Lastly, the tables needs to be accurately labelled and missing values needs to be updated. They can cut short the tables to include the variables relevant to the study.

Thank you for this suggestion. The table labels have been checked for accuracy and missing data. Within table 2, the number of missing values has been added for all variables where there were missing values. We have provided a statement in the table’s footnote that ‘Data are complete for all variables where no missing data are reported.’

Reviewer #3: Overall, this peer-reviewed article provides valuable insights into the rates, causes, and risk factors associated with hospital readmissions following acute ischemic stroke in the United States. The reviewer has following concerns:

1.Multiple studies have been published on the 30-day readmission rate following acute ischemic stroke using the same database. The authors should address how their manuscript differs from the already published literature. Additionally, they should consider trend analysis in the 30-day readmission rate, as it decreased from 12% in 2013 to 9.7% in 2019.

https://jamanetwork.com/journals/jamanetworkopen/fullarticle/2696869

https://www.ahajournals.org/doi/full/10.1161/STROKEAHA.116.016085

Thank you for these additional resources. They have been added to our introduction as references 7 and 8. The main difference with the prior studies is that we were able to look at 1-year readmission rates. We have further clarified this difference within our introduction: “The NRD does not allow data linkage across calendar years for the same patient resulting in variable length of follow up for patients admitted in different times of the year. Because of this, previous work using the NRD studied 30-day readmissions following stroke. [7-8] We describe a novel technique using time-to-event (ie the product limit method) which 1) provides readmission rates using Kaplan-Meyer estimates including survival time contributions from right-censored patients at 30-days and 1-year following hospital discharge, 2) allows data from all patients in the sample to contribute to the estimates (including patients with less than 30-days follow-up), and 3) allows examination of risk factors for readmission using survival analysis.”

We agree that temporal trends in readmission would be of interest, but this manuscript analyzed the 2019 National Readmission Database exclusively and does not have sufficient information for temporal trend analysis.

2.Can the authors include the number of patients who had hemorrhagic conversion following AIS during the index stay?

We agree with the reviewer that hemorrhagic transformation is an important prognostic predictor in stroke and would be of interest with regards to readmission.

However, there has been no validated methods for capturing hemorrhagic transformation following acute ischemic stroke within administrative data. One approach is the use of ICD 10 codes for intra-cranial hemorrhage codes and DRG codes related to complications of thrombolysis (61-63). From our experience working with NRD data, this approach leads to rates of hemorrhagic transformation which are not clinically intuitive (such as a hemorrhagic transformation rate of ~10% in AIS following thrombolysis).

This might, in part, be because patients who had both a primary intra-cranial hemorrhage and an ischemic stroke within the same admission would be miscoded as ‘hemorrhagic transformation’ using this strategy. Because of these concerns, information on hemorrhagic transformation was not included. We have included a statement to the discussion to describe this: “Similarly, there has been no validated methods of examining for hemorrhagic transformation during AIS admissions so the effect of this important prognostic factor could not be analyzed.”

3.The NRD does not track patients beyond the calendar year. For example, patients admitted in December 2019 cannot be tracked for 30-day readmission as there is no link between the year 2019 and 2020. Therefore, December month admissions should be excluded because the 30-day readmission rate cannot be calculated. Many studies have been published using the same methodology. The authors should clarify if they have excluded December month admissions and add it to the limitations.

Thank you for this suggestion for clarification. This is an important limitation of using the NRD using a conventional approach, and why we opted for a survival analysis approach which allows for better use of NRD data compared to previous analysis for stroke readmissions. In our analyses, using Kaplan-Meyer estimates (i.e. the product limit method), patients from December contributed to the 30 days (and 1-year) readmission rate as censored data. In our response to your first comment above, we have detailed how we have modified the introduction to better describe the benefits of using a Kaplan Meyer approach.

4.The NRD provides information regarding inter-hospital transfers. If a patient was admitted to a small hospital and transferred to a large hospital, there is a small chance that the patient might be included twice. Therefore, inter-hospital transfers must be excluded. If the authors have excluded them, please clarify in the methods section. If they have not excluded them, please provide a reason for not doing so.

This is a valid concern addressed in the NRD data structure by HCUP. “Readmission analyses do not usually allow the hospitalization at the receiving hospital to be counted as a readmission. To eliminate this possibility, pairs of records representing a transfer are collapsed into a single "combined" record in the NRD.” Further information is available from the HCUP website: https://hcup-us.ahrq.gov/db/vars/samedayevent/nrdnote.jsp

We have added a brief statement to the methods to describe this: “Within the NRD, hospital records from patient transfers are combined to avoid counting hospitalization at the second hospital as a readmission.”

Attachment

Submitted filename: 16June FINAL Responses to reviewer comments.docx

Click here for additional data file. (41.3KB, docx)

Decision Letter 1

Karthik Gangu

24 Jul 2023

Rates and Reasons for Hospital Readmission after Acute Ischemic Stroke in a US Population-Based Cohort

PONE-D-23-02024R1

Dear Dr. Zhou,

We’re pleased to inform you that your manuscript has been judged scientifically suitable for publication and will be formally accepted for publication once it meets all outstanding technical requirements.

Within one week, you’ll receive an e-mail detailing the required amendments. When these have been addressed, you’ll receive a formal acceptance letter and your manuscript will be scheduled for publication.

An invoice for payment will follow shortly after the formal acceptance. To ensure an efficient process, please log into Editorial Manager at http://www.editorialmanager.com/pone/, click the 'Update My Information' link at the top of the page, and double check that your user information is up-to-date. If you have any billing related questions, please contact our Author Billing department directly at authorbilling@plos.org.

If your institution or institutions have a press office, please notify them about your upcoming paper to help maximize its impact. If they’ll be preparing press materials, please inform our press team as soon as possible -- no later than 48 hours after receiving the formal acceptance. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information, please contact onepress@plos.org.

Kind regards,

Karthik Gangu

Academic Editor

PLOS ONE

Acceptance letter

Karthik Gangu

27 Jul 2023

PONE-D-23-02024R1

Rates and reasons for hospital readmission after acute ischemic stroke in a US population-based cohort

Dear Dr. Zhou:

I'm pleased to inform you that your manuscript has been deemed suitable for publication in PLOS ONE. Congratulations! Your manuscript is now with our production department.

If your institution or institutions have a press office, please let them know about your upcoming paper now to help maximize its impact. If they'll be preparing press materials, please inform our press team within the next 48 hours. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information please contact onepress@plos.org.

If we can help with anything else, please email us at plosone@plos.org.

Thank you for submitting your work to PLOS ONE and supporting open access.

Kind regards,

PLOS ONE Editorial Office Staff

on behalf of

Dr. Karthik Gangu

Academic Editor

PLOS ONE

Associated Data

    This section collects any data citations, data availability statements, or supplementary materials included in this article.

    Supplementary Materials

    S1 Fig. Example diagnostic plots showing violations of assumption of proportional hazards for hypertension with top panel showing non-parallel lines on log-log plot of survival and bottom panel showing increasing scaled Schoenfeld residuals over time (more prominent in early period).

    (PDF)

    Click here for additional data file. (245.5KB, pdf)
    S1 Table. Principle diagnosis at readmission after discharge with ischemic stroke by ICD category.

    (PDF)

    Click here for additional data file. (178.9KB, pdf)
    Attachment

    Submitted filename: 16June FINAL Responses to reviewer comments.docx

    Click here for additional data file. (41.3KB, docx)

    Data Availability Statement

    The data used is publicly available administrative data. It can be purchased for use by researchers after appropriate safety and confidentiality training from the Healthcare Cost and Utilization Project Agency for Healthcare Research and Quality. If a journal or publication is interested in access to data or analytic files, that information including data elements and data structure can be found here: https://hcup-us.ahrq.gov/nrdoverview.jsp.

    Articles from PLOS ONE are provided here courtesy of PLOS

    RESOURCES