Brief Summary:
Findings present key risk factors for early readmission among sepsis survivors transitioning from acute care to HHC and indicate the need to affirm the presence of a sepsis diagnosis and risk factors across settings.
Keywords: sepsis, transitional care, hospital discharge, home health care, readmission
Introduction
Hospitals in the United States discharge over 1 million sepsis survivors annually.1 Sepsis survivors are a vulnerable population that experience substantial morbidity and mortality, with readmission rates rivaling or exceeding those for heart failure (HF), pneumonia and myocardial infarction.2 Sepsis survivors are twice as likely to be readmitted by 30 days as non-sepsis patients,3 with 32% of readmissions occurring within seven days.3
After discharge, a substantial proportion of sepsis survivors (32%) receive skilled home health care (HHC) to monitor for reinfection and support continued recovery and rehabilitation.3 However, the characteristics of those who receive HHC and of those who are at risk of readmission are not fully understood.4 A published framework for sepsis evaluation and treatment within 90 days after discharge does not even mention HHC.4 The resulting information gap presents an untapped opportunity to improve sepsis care.
This paper uses national data to profile the characteristics of sepsis survivors receiving HHC by type of sepsis before, during, and after a sepsis hospitalization and to identify which of these characteristics are significantly associated with 7-day readmission. We focus on readmissions within 7 days because little is known about the large proportion readmitted by that timeframe and the outcome is potentially most affected by clinical factors that may be amenable to care transitions and HHC interventions.5,6 Early post-acute care attention – shown to significantly reduce readmission hazard among selected populations,7 including 8–7 percentage point rehospitalization reductions among HHC heart failure and sepsis patients8,9 provides the opportunity for prompt medication reconciliation, early nursing surveillance, vital sign monitoring, antibiotic stewardship, wound care, patient education, care coordination, and early outpatient assessment.
Others have used acute care electronic record information5 or claims data to predict 7-day readmissions. No previous studies have used the detailed acute and post-acute care assessment information that we did to profile patients by type of sepsis during the critical transition from hospital to HHC. The ultimate goal of this paper, therefore, is to produce new knowledge that providers can use to improve community transitions and post-discharge outcomes among the large and growing number of sepsis survivors receiving HHC.
Methods
Study Design and Conceptual Framework
This cross sectional, cohort, descriptive study is a secondary analysis of national data of Medicare beneficiaries who survived sepsis and received skilled HHC services in the United States. The Andersen Behavioral Model of Health Service Utilization, positing the role of “predisposing,” “enabling” and “needs” factors as contributors to service use – guided the analysis.10 The Institutional Review Board of the Visiting Nurse Service of New York approved this study.
Study Sample
The sample is from a national dataset of Medicare beneficiaries hospitalized for sepsis who were discharged to HHC services between July 1, 2013 to June 30, 2014. For this analysis, we chose the first instance of a sepsis hospital stay (index admission) followed immediately by a HHC episode where there was a complete Outcome and Assessment Information Set (OASIS-C) admission assessment, and prior to any other health care facility use (e.g., an observation unit or skilled nursing facility stay).
Data Sources and Measurement of Individual Health Determinants
Pre-hospital and index stay variables include the Medicare administrative and claims files from the Centers for Medicare and Medicaid (CMS) Chronic Condition Data Warehouse (CCW), and the United States Census. HHC episode data are from OASIS C, a comprehensive CMS-mandated assessment tool, that includes nearly 100 structured data items related to a HHC recipient’s functional status, clinical status, and service needs.11 Item reliability ranges from fair to excellent.12–14
Predisposing Characteristics
Socio-demographics.
Age, race and gender (Medicare administrative data); ethnicity and caregiver availability to assist with equipment/medications/medical procedures (OASIS-C).15,16 Enabling Characteristics
Medicaid eligibility and annual median family income in the county where the patient lives (Medicare administrative file, US Census).
Need Illness Characteristics
Reasons for Medicare eligibility [disability, end-stage renal disease or old age (≥65)] (Medicare administrative file).
Prior Health Care Utilization.
Previous hospital stays (0–3 or more) and number of previous hospital days (0–30 or more) in six months prior to index sepsis admission (Medicare claims files).
Sepsis Indicators.
Sepsis is a life-threatening, toxic response to infection that can lead to tissue damage, organ failure, and death.17 Using ICD-9 codes from the Medicare claims, patients were grouped into three sepsis types. We included all short-term general and critical access hospital stays with an explicit or implicit ICD-9 sepsis diagnosis code. Explicit sepsis: ICD-9 diagnosis codes 995.91 (sepsis without acute organ dysfunction), 995.92 (severe sepsis), or 785.52 (septic shock). We included implicit “severe” sepsis, as defined by Angus18 and validated by another group;19 one or more ICD-9 bacterial or fungal infection codes and an acute organ dysfunction code. We include whether sepsis was present on admission and types of infections (e.g. urinary, gastrointestinal).
Index Stay Characteristics.
Main reason for hospital care (medical condition or surgical procedure); length of stay (LOS); days in an intensive care unit (ICU); primary diagnoses and Elixhauser comorbidities (ICD-9) (Medicare claims).20
Home Care Clinical Status.
HHC diagnoses; readmission risk factors; infusion therapy in the home; overall health status, depression, pain, pressure ulcers, dyspnea, incontinence, cognition, anxiety, activities of daily living (ADL), instrumental ADLs (IADL), and fall risk (all from OASIS-C). HHC clinicians are instructed by Medicare to list on the OASIS the ICD codes and diagnosis terms associated with the reason for HHC.
Primary Outcome.
Hospital readmission prior to any other health care facility use (e.g., an observation unit or skilled nursing facility stay) out to 30-days (Medicare claims).
Data Analysis
We used descriptive statistics including frequencies, means, standard deviations, and proportions (%) to summarize patient characteristics six months prior to the index sepsis hospitalization, during the index hospital stay, and upon HHC admission. We described hospital readmission out to 30 days after the index sepsis hospital discharge. Using multivariate logistic regression, we examined the associations between 7-day readmissions and the demographic and clinical characteristics that were highly significant by type of sepsis (p<0.05). Results did not differ by sepsis type through stratification or by including interaction terms in a single model. Therefore, we used multivariate logistic regression to estimate a multivariate model of the average marginal effect21 (i.e. the percentage point change in risk) of the selected predisposing, enabling, need/illness characteristics on 7-day hospital readmission, including sepsis type as one of the covariates. To account for the large sample size, we used a p-value of ≤ 0.001 and a marginal effect of at least 0.5% to highlight the variables with the highest impact. Sensitivity analysis was completed by removing the small number who died before 7 days or readmission (N=623, 0.83%) All descriptive statistics were calculated using RStudio version 3.3.3. Marginal effects of the logistic regression model were produced using STATA version 14.2.
Results
The sample consists of 165,228 sepsis survivors discharged to HHC (e-Figure 1 in the supplement). The majority of patients (N=133,397; 80.7%) had an explicit hospital diagnosis of severe sepsis (ICD-9 code of 995.92 in any diagnosis field) or hospital diagnoses indicating implicit severe sepsis using the Angus method (i.e., ICD-9 combination of an infection with organ dysfunction); 5.7% had a septic shock diagnosis (ICD-9 code 785.52) and the remainder (13.6%) had sepsis without acute organ system dysfunction (ICD-9 code 995.91). For simplicity we refer to this final group as “sepsis.”
Patient Characteristics Prior to and During Sepsis Hospitalization
Table 1 describes and compares predisposing, enabling and need/illness factors of all sepsis survivors and by sepsis type, prior to and during the index sepsis stay. Overall, average age was 75.1 (SD=12.6), 55.3% female, 79.5% non-Hispanic white, 12.4% non-Hispanic black, and 5.6% Hispanic (predisposing factors). The median annual family income in the county where the patient lived was on average $54,100, and 26.5% of patients were eligible for Medicaid-funded services in addition to Medicare (enabling factors).
Table 1.
Patient Characteristics Prior to and During Sepsis Hospitalizationa
| Sepsis Only (ICD-9, 995.91) | Severe Sepsis (ICD-9, 995.92 or Angus | Septic Shock (ICD-9, 785.52) | [ALL] | |
|---|---|---|---|---|
| N=22,382 (13.6%) | N=133,397 (80.7%) | N=9,449 (5.7%) | N=165,228 | |
| Predisposing Factors | ||||
| Age, mean (SD) | 73.5 (14.1) | 75.7 (12.2) | 70.4 (13.3) | 75.1 (12.6) |
| Female gender | 12343 (55.1%) | 74239 (55.7%) | 4751 (50.3%) | 91333 (55.3%) |
| Race | ||||
| Non-Hispanic White | 17717 (79.2%) | 106400 (79.8%) | 7292 (77.2%) | 131409 (79.5%) |
| Non-Hispanic Black | 2458 (11.0%) | 16362 (12.3%) | 1095 (11.6%) | 19915 (12.1%) |
| Hispanic | 1423 (6.4%) | 6935 (5.2%) | 659 (7.0%) | 9017 (5.5%) |
| Other | 784 (3.5%) | 3700 (2.8%) | 403 (4.3%) | 4887 (3.0%) |
| Enabling Factors | ||||
| Eligible for both Medicaid and Medicare | 6356 (28.4%) | 34419 (25.8%) | 2958 (31.3%) | 43733 (26.5%) |
| County level annual median family income in dollars, mean (SD) | 54007 (14799) | 54099 (14796) | 54341 (15020) | 54100 (14809) |
| Need/Illness Factors | ||||
| Three or more inpatient stays in previous 6 months | 1685 (7.5%) | 10020 (7.5%) | 904 (9.6%) | 12609 (7.6%) |
| Sepsis not present on admission | 1282 (5.7%) | 26022 (19.5%) | 938 (9.9%) | 28242 (17.1%) |
| Type of infection (5 most frequent) | ||||
| Kidney, Urinary Tract and Other Genitourinary | 8239 (36.8%) | 54099 (40.6%) | 3573 (37.8%) | 65911 (39.9%) |
| Pneumonia and Other Respiratory | 7407 (33.1%) | 50485 (37.8%) | 3397 (36.0%) | 61289 (37.1%) |
| Other | 5005 (22.4%) | 38642 (29.0%) | 2062 (21.8%) | 45709 (27.7%) |
| Bone, Joint and Skin/Soft Tissue | 3863 (17.3%) | 19693 (14.8%) | 1032 (10.9%) | 24588 (14.9%) |
| Gastrointestinal | 2069 (9.2%) | 12790 (9.6%) | 1489 (15.8%) | 16348 (9.9%) |
| Index Stay - MS-DRG | ||||
| Medical | 18727 (83.7%) | 107637 (80.7%) | 7194 (76.1%) | 133558 (80.8%) |
| Surgical | 3655 (16.3%) | 25760 (19.3%) | 2255 (23.9%) | 31670 (19.2%) |
| Index hospitalization length of stay, mean (SD) | 7.69 (5.65) | 8.78 (7.38) | 13.0 (14.0) | 8.87 (7.79) |
| ICU days, mean (SD) | 1.84 (3.50) | 3.01 (5.49) | 6.70 (7.9) | 3.06 (5.52) |
| Number of Elixhauser comorbid conditions | 3.81 (1.82) | 4.42 (1.88) | 4.68 (1.89) | 4.35 (1.89) |
| Select Elixhauser comorbid conditions | ||||
| Hypertension | 15660 (70.0%) | 92780 (69.6%) | 6071 (64.3%) | 114511 (69.3%) |
| Fluid and electrolyte disorders | 9810 (43.8%) | 74894 (56.1%) | 6549 (69.3%) | 91253 (55.2%) |
| Renal failure | 4706 (21.0%) | 48503 (36.4%) | 2966 (31.4%) | 56175 (34.0%) |
| Congestive heart failure | 4546 (20.3%) | 33526 (25.1%) | 2780 (29.4%) | 40852 (24.7%) |
| Depression | 3200 (14.3%) | 18187 (13.6%) | 1223 (12.9%) | 22610 (13.7%) |
| Coagulopathy | 125 (0.6%) | 23128 (17.3%) | 1929 (20.4%) | 25182 (15.2%) |
| Weight loss | 2247 (10.0%) | 13630 (10.2%) | 1813 (19.2%) | 17690 (10.7%) |
all variables significant at ≤.001 except: annual median family income in dollars p=0.183, depression p=0.003
The need/illness characteristics describe severe illness burden (not all data shown). Thirty-one percent initially became eligible for Medicare due to disability or end-stage renal disease. A sizeable minority (40.8%) had at least one prior hospital stay in the past 6 months. The most common type of infection was kidney or urinary tract (39.9%), followed closely by pneumonia (37.1%). Average LOS was 8.9 days (SD=7.8). Average number of medical conditions was 4.35 (SD=1.89) out of 15 Elixhauser comorbid conditions.20 Almost 70% of the sample had hypertension and nearly 40%, diabetes mellitus. Medical conditions were the primary reason for 80.8% of hospital stays; surgical procedures for 19.2%.
Home Health Admission Diagnoses.
The primary and five secondary diagnoses listed for the sepsis survivors by HHC clinicians on the OASIS were very diverse. The 5 most common primary diagnoses listed upon admission to HHC were pneumonia (10.2%), other aftercare (9.5%), congestive heart failure (8.4%), urinary tract infection (UTI) (8.4%), and obstructive chronic bronchitis (6.2%) (Table 2). Looking across all six diagnosis categories, and combining all three sepsis ICD-9 codes, the sepsis codes appeared only 4% of the time, an additional 3% noted septicemia or bacteremia.
Table 2.
Top 20 most frequent diagnoses recorded at home care admission
| Frequency | Cumulative Frequency | ||
|---|---|---|---|
| 1 | Pneumonia (except that caused by TB or STD) | 16810 (10.2%) | 16810 |
| 2 | Other aftercare | 15731 (9.5%) | 32541 |
| 3 | Congestive heart failure; non-hypertensive | 13939 (8.4%) | 46480 |
| 4 | Urinary tract infections | 13801 (8.4%) | 60281 |
| 5 | Obstructive chronic bronchitis | 10252 (6.2%) | 70533 |
| 6 | Rehabilitation care; fitting of prostheses; and adjustment of devices | 7622 (4.6%) | 78155 |
| 7 | Cellulitis and abscess | 6821 (4.1%) | 84976 |
| 8 | Septicemia (except in labor) | 4769 (2.9%) | 89745 |
| 9 | Complications of surgical procedures or medical care | 3729 (2.3%) | 93474 |
| 10 | Other connective tissue disease | 3292 (2%) | 96766 |
| 11 | Decubitus ulcer | 3240 (2%) | 100006 |
| 12 | Cardiac dysrhythmias | 2894 (1.8%) | 102900 |
| 13 | Diabetes mellitus without complication | 2887 (1.8%) | 105787 |
| 14 | Hypertension with complications and secondary hypertension | 2728 (1.7%) | 108515 |
| 15 | Complication of device; implant or graft | 2069 (1.3%) | 110584 |
| 16 | Late effects of cerebrovascular disease | 2015 (1.2%) | 112599 |
| 17 | Genitourinary symptoms and ill-defined conditions | 1809 (1.2%) | 114408 |
| 18 | Acute myocardial infarction | 1688 (1%) | 116096 |
| 19 | Diabetes with other manifestations | 1641 (1%) | 117737 |
| 20 | Essential hypertension | 1614 (1%) | 119351 |
| 21 | Remaining Sample | 45877 (27.8%) | 165228 |
For 50% of patients in this category, “other aftercare” includes: aftercare following surgery of the teeth, oral cavity, digestive system, or circulatory system. The remainder are aftercare for a variety of surgical procedures, medication monitoring, or wound care.
Characteristics of Sepsis Survivors upon Homecare Admission (Table 3).
Table 3.
Characteristics of Sepsis Survivors Upon Homecare Admission
| Sepsis Only (ICD-9, 995.91) | Severe Sepsis (ICD-9, 995.92 or Angus) | Septic Shock (ICD-9, 785.52) | [ALL] | |
|---|---|---|---|---|
| N=22382 | N=133397 | N=9449 | N=165228 | |
| Enabling Factors | ||||
| Caregiver ability and willingness to provide assistance with medical equipment when needed | ||||
| No assistance needed in this area | 11616 (51.9%) | 73513 (55.1%) | 4530 (47.9%) | 89659 (54.3%) |
| Caregiver(s) currently provide assistance | 6278 (28.0%) | 37488 (28.1%) | 2898 (30.7%) | 46664 (28.2%) |
| Poor Caregiver Availability | 4488 (20.1%) | 22396 (16.8%) | 2021 (21.4%) | 28905 (17.5%) |
| Caregiver ability and willingness to provide assistance with Medication Administration | ||||
| No assistance needed in this area | 4612 (20.6%) | 22360 (16.8%) | 1576 (16.7%) | 28548 (17.3%) |
| Caregiver(s) currently provide assistance | 11586 (51.8%) | 72921 (54.7%) | 5171 (54.7%) | 89678 (54.3%) |
| Poor Caregiver Availability | 6184 (27.6%) | 38116 (28.6%) | 2702 (28.6%) | 47002 (28.4%) |
| Caregiver ability and willingness to provide assistance with Medical Procedures | ||||
| No assistance needed in this area | 11704 (52.3%) | 73863 (55.4%) | 4387 (46.4%) | 89954 (54.4%) |
| Caregiver(s) currently provide assistance | 4783 (21.4%) | 27596 (20.7%) | 2295 (24.3%) | 34674 (21.0%) |
| Poor Caregiver Availability | 5895 (26.3%) | 31938 (23.9%) | 2767 (29.3%) | 40600 (24.6%) |
| Need/Illness Factors | ||||
| Signs or symptoms indicating patient at risk for hospitalization | ||||
| Taking 5 or more medications | 19716 (88.1%) | 120228 (90.1%) | 8410 (89.0%) | 148354 (89.8%) |
| Multiple hospitalizations (2 or more) in past 12 months | 11087 (49.5%) | 66558 (49.9%) | 4876 (51.6%) | 82521 (49.9%) |
| Frailty indicators (e.g., weight loss, self-reported exhaustion)a | 9061 (40.5%) | 55637 (41.7%) | 4245 (44.9%) | 68943 (41.7%) |
| History of falls (2 or more falls - or any fall with an injury-in past year) | 5132 (22.9%) | 35354 (26.5%) | 2003 (21.2%) | 42489 (25.7%) |
| Recent decline in mental, emotional, or behavioral statusa | 3060 (13.7%) | 20997 (15.7%) | 1408 (14.9%) | 25465 (15.4%) |
| Severe pain based on a standardized assessment | 4846 (21.7%) | 27390 (20.5%) | 2153 (22.8%) | 34389 (20.8%) |
| IV Infusion Therapy in the Home | 4089 (18.3%) | 11321 (8.5%) | 1381 (14.6%) | 16791 (10.2%) |
| Unhealed Pressure Ulcer at Least Stage II | 1608 (7.2%) | 8358 (6.3%) | 1120 (12.0%) | 11086 (6.7%) |
| When is patient dyspneic or noticeably short of breath | ||||
| Patient not short of breath | 5730 (25.6%) | 26760 (20.1%) | 2121 (22.4%) | 34611 (20.9%) |
| With moderate exertion | 6944 (31.0%) | 44710 (33.5%) | 3128 (33.1%) | 54782 (33.2%) |
| With minimal exertion | 3898 (17.4%) | 26103 (19.6%) | 1780 (18.8%) | 31781 (19.2%) |
| At rest (during day or night) | 1106 (4.9%) | 7945 (5.9%) | 563 (5.9%) | 9614 (5.8%) |
| Multi-factor risk assessment conducted and risk of falls indicated | 18888 (84.4%) | 116888 (87.6%) | 8074 (85.4%) | 143850 (87.1%) |
| Patient requires a urinary catheter | 2062 (9.2%) | 10697 (8.0%) | 1117 (11.8%) | 13876 (8.4%) |
| Alert/oriented | 12223 (54.6%) | 68572 (51.4%) | 5186 (54.9%) | 85981 (52.0%) |
| Sum of Activities of Daily Living (ADL) dependenciesb (mean, SD) | 4.93 (2.7) | 5.09 (2.59) | 5.45 (2.67) | 5.09 (2.62) |
all variables significant at ≤.001 except: Frailty Indicators, p=0.082 and Recent decline in mental, emotional, or behavioral status, p=0.002
ADL dependencies included: grooming, bathing, upper and lower body dressing, toilet transfer, toilet hygiene, feeding, transferring and ambulation
The enabling factor caregiver availability for equipment use, medications, and medical procedures showed poor caregiver availability for 17.5%, 28.4%, and 24.6% of patients respectively for each task.
The need/illness characteristics describe a population with severe illness burden: 98.1% of patients had at least one risk factor for readmission; 89.8% took five or more medications and 41.7% had frailty indicators; 15.4% had a recent decline in mental, emotional, or behavioral status and more than 10% required home intravenous therapy. Thirty-nine percent were classified as fragile health, 20.8% with severe pain, 79.1% shortness of breath, and 87.1% at risk for falls; barely half were alert and oriented (52%). Nearly half (48.1%) reported urinary incontinence and another 8.4% had a urinary catheter while 6.7% had unhealed stage 2 or greater pressure ulcers. The large majority (88.4%) were dependent on human help in three or more activities of daily living, while 51.5% were unable to self-manage their medications.
Differences by Type of Sepsis.
Septic shock patients presented with a significantly different socioeconomic and medical complexity profile compared to the overall sample, and compared to sepsis and severe sepsis survivors (Tables 1 and 3). Compared to the overall sample, septic shock survivors were younger (25.8% versus 16.0% were <65), less likely to be white (77.2% versus 79.5%), and less likely to be female (50.3% versus 55.3%). A higher proportion of septic shock patients were Medicaid-eligible (31.3%) compared to the overall sample (26.5%). A larger proportion of septic shock patients entered the hospital for surgery (23.9%) than overall (19.2%) and they had the longest stays on average at 13 days versus 8.87 days overall.
Fifty percent of the overall sample spent time in the intensive care unit (ICU) compared to 93.1% of septic shock patients. Almost 30% of the septic shock patients spent 8–30 days in the ICU compared to nearly 12% overall having long ICU stays. Septic shock patients presented with the most severe comorbidity profile with higher proportions having fluid and electrolyte disorders, anemia, congestive heart failure, coagulopathy, and weight loss. Septic shock patients had nearly double the incidence of unhealed stage 2 or greater pressure ulcers (11.9%) than the sepsis (7.2%) or severe sepsis (6.3%) patients.
Readmission Outcomes of Sepsis Survivors Discharged to HHC.
Within 30 days of index discharge, 22.5% of HHC sepsis patients were rehospitalized prior to admission to any other health care facility or a hospice program. Nearly one-third of these readmissions occurred within seven days affecting 6.64% of the overall sample. Septic shock patients had the highest proportion of readmissions across all timeframes (Figure 1).
Figure 1.
Cumulative Readmission Rates Overall and By Sepsis Type Within 30 days of Hospital Discharge
Associations among Predisposing, Enabling and Need Illness Characteristics and 7-day Readmission.
Given the large sample size, and therefore numerous significant variables in the full model (e-table 1), we present a parsimonious selection of high impact variables in Table 4 to show 17 variables where p = ≤.001 with a marginal effect point estimate of at least 0.5%. Several need/illness factors significantly increased the early readmission risk above the baseline of 6.64% to a range of 7.15% for patients with fluid and electrolyte disorder to 10.61 % for those with 3 or more previous inpatient stays in the past 6 months. This translates into a 60% relative increase in the risk of 7-day readmission for the latter group. As seen in the last column of Table 4, our model highlights risk factors that present significant increases in relative risk such as dyspnea at rest (47.8% relative increase), 8–9 ADL/IADL dependencies (38.8% relative increase), index length of stay ≥8 days (29.7% relative increase), intravenous infusion in the home (21.4% relative increase) and a diagnosis of septic shock (17.5% relative increase). Some risk factors are potentially modifiable (LOS, dyspnea, ADL deficits, pain, depression, frailty). Non-modifiable risk factors include diagnosis, and history of previous admissions. There were no differences in results using samples with or without the small number who died prior to readmission or 7 days.
Table 4.
Incremental Effects of Sepsis Patient Risk Factors on Hospital Readmission within 7 Days
| Overall Sample Readmission Risk = 6.64% | Percentage Point Change in Riska | ||||
|---|---|---|---|---|---|
| Point Estimate | 95% Confidence Interval | Adjusted Risk of Readmission with Factorb | Relative Percent Change in Readmission Riskc | ||
| Sepsis Type (Ref = Sepsis without Organ Failure 995.91) | |||||
| Severe Sepsis (995.92 or Angus) | 0.651% | 0.290% | 1.012% | 7.29% | 9.8% |
| Septic Shock (785.52) | 1.163% | 0.558% | 1.767% | 7.80% | 17.5% |
| Number of Inpatient Stays in Past 6 Months (Ref = 0) | |||||
| 1–2 Stays | 1.395% | 1.131% | 1.658% | 8.03% | 21.0% |
| 3+ Stays | 3.975% | 3.449% | 4.501% | 10.61% | 59.9% |
| Sepsis Not Present on Hospital Admission | 0.825% | 0.478% | 1.172% | 7.46% | 12.4% |
| Hospital LOS 8+ Days | 1.973% | 1.704% | 2.242% | 8.61% | 29.7% |
| Elixhauser Comorbidity: Fluid and Electrolyte Disorder | 0.508% | 0.264% | 0.753% | 7.15% | 7.7% |
| Diagnosis of Depression | 0.708% | 0.445% | 0.970% | 7.35% | 10.7% |
| Risk for Hosp- Recent Decline | 0.650% | 0.292% | 1.007% | 7.29% | 9.8% |
| Risk for Hosp- Frailty | 0.779% | 0.524% | 1.035% | 7.42% | 11.7% |
| Pain Assessed, Severe Pain | 0.759% | 0.454% | 1.064% | 7.40% | 11.4% |
| IV Infusion on Admission to Home Health Care | 1.709% | 1.235% | 2.182% | 8.35% | 25.7% |
| When Dyspneic (Ref = Not Short of Breath) | |||||
| With Minimal Exertion | 1.422% | 1.014% | 1.830% | 8.06% | 21.4% |
| At rest, during the day | 3.176% | 2.530% | 3.822% | 9.81% | 47.8% |
| Count of number of ADL dependenciesd (Ref ≤ 2) | |||||
| 6–7 | 1.326% | 0.844% | 1.807% | 7.96% | 20.0% |
| 8–9 | 2.576% | 2.052% | 3.100% | 9.21% | 38.8% |
all variables significant at ≤.001
adjusted risk of readmission with risk factor = overall risk + point estimate risk
Relative percent change in readmission risk = point estimate/overall readmission risk × 100
Activities of Daily Living dependencies include: grooming, bathing, upper and lower body dressing, toilet transfer, toilet hygiene, feeding, transferring and ambulation
Discussion.
Our study highlights new and important information about risk factors that, if collected, may lead to better identification and earlier intervention for at-risk sepsis patients. We found that although our sample population were all sepsis survivors, the diagnosis codes in HHC noted sepsis only 4% of the time. This information gap puts HHC providers at a disadvantage for recognizing that they are caring for a patient with one of the highest risks for hospital readmission. The CMS instructs HHC clinicians to list each diagnosis for which the patient is receiving HHC in their order of seriousness. We do not know how often sepsis was a diagnoses listed on the referral documents received by the agencies. However, a focus on pneumonia, urinary tract infection, or the ambiguous “other aftercare” may not equate to the same scrutiny and intensity of care afforded to the sepsis survivors who have a high illness burden even when compared to heart failure patients.22 Moreover, the potential of not being aware of the septic shock survivors in particular is of concern as 96% experienced an ICU stay, they demonstrate more frequent and severe illness/need characteristics, and have higher readmission rates than other sepsis patients. In our model, septic shock increased 7-day readmission risk by 17.5%. HHC clinicians could specifically request the additional information such as ICU stay, organ failure, or shock when receiving a hospital discharge referral after an infection. This may help to identify the septic shock survivors, the youngest and by many measures, the sickest of the sepsis subpopulations.
Our findings may inform improved identification of severe sepsis, the majority of sepsis survivors in our sample (80.7%) through systematic documentation and information sharing about infection and organ failure as an indication of the severe sepsis diagnoses.18 Our findings indicate that HHC clinicians note the infections (pneumonia, UTI), but not the life-threatening response to that infection, notably the diagnosis codes for organ failure associated with severe sepsis or septic shock are largely absent. Collecting this information as well as ICU stays, LOS, and number of prior hospitalizations may be helpful. Although non-modifiable by HHC, these “red flag” risk factors can raise the awareness that this is a sepsis survivor and guide the timing and frequency of HHC visits and other interventions.
Most of the risk characteristics associated with early readmission in this study are collected during the HHC admission OASIS. Several modifiable risk factors identified in our study are also reported as risk factors for 7-day readmission in other disease conditions (e.g., depression23, pain5,24–26, dyspnea27, number of functional dependencies28,29). However, because of lags between hospital discharge and HHC referral, and because agencies have up to five days after initial assessment to finalize the OASIS, complete OASIS information may not be available early enough in the HHC stay to maximize its usefulness. There is a need to assess/collect this information earlier, preferably at the time of referral.
Sepsis type, index LOS, and history of previous hospital admissions (also found to be risk factors for 30-day Sepsis readmissions3,30–33) were all risk factors for 7-day readmission in this study, but are largely information unknown by HHC. For other types of patients, lack of information transferred during transitions is associated with increased readmission risk,6,34 and could present a serious problem in sepsis. The gaps we have described represent a lost opportunity to intervene early. Sepsis survivors who received a HHC nursing visit within 48 hours of hospital discharge, at least one more visit the first week, and outpatient provider follow-up by 7 days have demonstrated 30-day readmission rates 7 percentage points lower than those without timely nursing and provider follow-up.9 However, only 28.1% of sepsis survivors nationally received this intervention.9 This early intervention, potentially effective in reducing 7-day readmission as well, makes it critical to identify early on who is at risk to enable receipt of this level of prompt attention.
When interpreting these results, it is important to note that the sample is limited to sepsis survivors who were referred for HHC and received at least one nursing visit within the first week, had complete OASIS-C admission data, and no admissions to other inpatient settings. Sepsis was identified using ICD-9 codes; crosswalks to ICD-10 are available. The OASIS-C was used for the HHC assessments, and the newest OASIS-D contains all risk factors identified by our work.
Building on the work of Graham et al5, where 7-day readmissions in a general hospital population were largely preventable by acute care interventions, our study indicates a role for both acute and post-acute providers in preventing early sepsis readmissions. Early knowledge of risk factors may trigger delayed discharge, referral to higher levels of care (eg. skilled nursing facility), preventive interventions such as early home nursing visits, early medical provider follow-up, more intense monitoring, physical therapy, patient education, and having equipment such as oxygen or intravenous supplies and antibiotics ready in the home. In sum, the key information we have identified through this study should facilitate a plan of care and associated interventions that better meet the needs of this vulnerable population to enhance sepsis recovery with reduced need for rehospitalization. Future work will aim to create a validated risk screening tool for use during discharge planning and referral.
Supplementary Material
E-Figure 1. Person-Level Analytic Sample Selection (N=165,228) I
E-Table 1: Full Model Marginal Effects of Readmission within 7 days (Overall Risk = 6.64%)
Funding/Support:
Research reported in this publication was supported by the National Institute of Nursing Research of the National Institutes of Health under Award Number R01NR016014 and National Institute of Nursing Research (T32NR009356). The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.
Footnotes
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Conflict of Interest Statement
The Authors have no conflict of interests to disclose
Contributor Information
Kathryn H. Bowles, 418 Curie Boulevard, University of Pennsylvania School of Nursing, Philadelphia, PA 19146.
Christopher M. Murtaugh, Center for Home Care Policy & Research, Visiting Nurse Service of New York, 5 Pennsylvania Plaza, 12th Floor, New York, NY 10001
Lizeyka Jordan, Center for Home Care Policy & Research, Visiting Nurse Service of New York, 5 Pennsylvania Plaza, 12th Floor, New York, NY 10001.
Yolanda Barrón, Center for Home Care Policy & Research, Visiting Nurse Service of New York, 5 Pennsylvania Plaza, 12th Floor, New York, NY 10001
Mark E. Mikkelsen, Perelman School of Medicine of the University of Pennsylvania, 3400 Spruce Street Philadelphia, PA 19104
Christina R. Whitehouse, M. Louise Fitzpatrick College of Nursing, Driscoll Hall, Room 312, 800 E. Lancaster Avenue, Villanova, PA 19085.
Jo-Ana D. Chase, University of Missouri, S343 Sinclair School of Nursing, Columbia, MO 65211.
Miriam Ryvicker, Center for Home Care Policy & Research, Visiting Nurse Service of New York, 5 Pennsylvania Plaza, 12th Floor, New York, NY 10001.
Penny Hollander Feldman, Center for Home Care Policy & Research, Visiting Nurse Service of New York, 107 East 70th Street, New York, NY 10021.
References
- 1.De Backer D, Dorman T. Surviving Sepsis Guidelines: A Continuous Move Toward Better Care of Patients With Sepsis. Jama. 2017;317(8):807–808. [DOI] [PubMed] [Google Scholar]
- 2.Mayr FB, Talisa VB, Balakumar V, Chang CH, Fine M, Yende S. Proportion and Cost of Unplanned 30-Day Readmissions After Sepsis Compared With Other Medical Conditions. Jama. 2017;317(5):530–531. [DOI] [PubMed] [Google Scholar]
- 3.Jones TK, Fuchs BD, Small DS, et al. Post-Acute Care Use and Hospital Readmission after Sepsis. Annals of the American Thoracic Society. 2015;12(6):904–913. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Prescott HC, Angus DC. Enhancing Recovery From Sepsis: A Review. Jama. 2018;319(1):62–75. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Graham KL, Auerbach AD, Schnipper JL, et al. Preventability of Early Versus Late Hospital Readmissions in a National Cohort of General Medicine Patients. Annals of internal medicine. 2018;168(11):766–774. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Jones CD, Jones J, Richard A, et al. “Connecting the Dots”: A Qualitative Study of Home Health Nurse Perspectives on Coordinating Care for Recently Discharged Patients. J Gen Intern Med. 2017;32(10):1114–1121. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Carnahan JL, Slaven JE, Callahan CM, Tu W, Torke AM. Transitions From Skilled Nursing Facility to Home: The Relationship of Early Outpatient Care to Hospital Readmission. Journal of the American Medical Directors Association. 2017;18(10):853–859. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Murtaugh CM, Deb P, Zhu C, et al. Reducing Readmissions among Heart Failure Patients Discharged to Home Health Care: Effectiveness of Early and Intensive Nursing Services and Early Physician Follow-Up. Health services research. 2017;52(4):1445–1472. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Deb P, Murtaugh C, Bowles KH, Mikkelsen ME, Khajavi HN, Moore S, Barron Y, Feldman PH Does Early Follow-up Improve the Outcomes of Sepsis Survivors Discharged to Home Health Care. Medical Care. 2019, 57 (8), 633–640. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Andersen RM. Revisiting the behavioral model and access to medical care: does it matter? Journal of health and social behavior. 1995;36(1): 1–10. [PubMed] [Google Scholar]
- 11.Siegler EL, Murtaugh CM, Rosati RJ, et al. Improving the transition to home healthcare by rethinking the purpose and structure of the CMS 485: first steps. Home health care services quarterly. 2006;25(3–4):27–38. [DOI] [PubMed] [Google Scholar]
- 12.Hittle DF, Shaughnessy PW, Crisler KS, et al. A study of reliability and burden of home health assessment using OASIS. Home health care services quarterly. 2003;22(4):43–63. [PubMed] [Google Scholar]
- 13.Kinatukara S, Rosati RJ, Huang L. Assessment of OASIS reliability and validity using several methodological approaches. Home health care services quarterly. 2005;24(3):23–38. [DOI] [PubMed] [Google Scholar]
- 14.O’Connor M, Davitt JK. The Outcome and Assessment Information Set (OASIS): a review of validity and reliability. Home health care services quarterly. 2012;31(4):267–301. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Waldo DR. Accuracy and Bias of Race/Ethnicity Codes in the Medicare Enrollment Database. Health care financing review. 2004;26(2):61–72. [PMC free article] [PubMed] [Google Scholar]
- 16.Eicheldinger C, Bonito A. More accurate racial and ethnic codes for Medicare administrative data. Health care financing review. 2008;29(3):27–42. [PMC free article] [PubMed] [Google Scholar]
- 17.Definition of Sepsis https://www.sepsis.org/sepsis-basics/what-is-sepsis/. Accessed October 24, 2019.
- 18.Angus DC, Linde-Zwirble WT, Lidicker J, Clermont G, Carcillo J, Pinsky MR. Epidemiology of severe sepsis in the United States: analysis of incidence, outcome, and associated costs of care. Critical care medicine. 2001;29(7):1303–1310. [DOI] [PubMed] [Google Scholar]
- 19.Iwashyna TJ, Odden A, Rohde J, et al. Identifying patients with severe sepsis using administrative claims: patient-level validation of the angus implementation of the international consensus conference definition of severe sepsis. Med Care. 2014;52(6):e39–43. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Elixhauser A, Steiner C. Readmissions to U.S. Hospitals by Diagnosis, 2010: Statistical Brief #153 Healthcare Cost and Utilization Project (HCUP) Statistical Briefs. Rockville (MD): Agency for Healthcare Research and Quality (US); 2006. [PubMed] [Google Scholar]
- 21.Norton EC, Dowd BE, Maciejewski ML. Marginal Effects-Quantifying the Effect of Changes in Risk Factors in Logistic Regression Models. Jama. 2019. [DOI] [PubMed] [Google Scholar]
- 22.Scharpf TP, Madigan EA. Functional status outcome measures in home health care patients with heart failure. Home health care services quarterly. 2010;29(4):155–170. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Sharif R, Parekh TM, Pierson KS, Kuo YF, Sharma G. Predictors of early readmission among patients 40 to 64 years of age hospitalized for chronic obstructive pulmonary disease. Annals of the American Thoracic Society. 2014;11(5):685–694. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Kwok CS, Shah B, Al-Suwaidi J, et al. Timing and Causes of Unplanned Readmissions After Percutaneous Coronary Intervention: Insights From the Nationwide Readmission Database. JACC Cardiovascular interventions. 2019;12(8):734–748. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Fingar KR, Barrett ML, Jiang HJ. A Comparison of All-Cause 7-Day and 30-Day Readmissions, 2014: Statistical Brief #230 Healthcare Cost and Utilization Project (HCUP) Statistical Briefs. Rockville (MD): Agency for Healthcare Research and Quality (US); 2006. [PubMed] [Google Scholar]
- 26.Brown EG, Burgess D, Li CS, Canter RJ, Bold RJ. Hospital readmissions: necessary evil or preventable target for quality improvement. Annals of surgery. 2014;260(4):583–589; discussion 589–591. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Eastwood CA, Quan H, Howlett JG, King-Shier KM. Factors Associated With 7-Day Rehospitalization After Heart Failure Admission. The Journal of cardiovascular nursing. 2017;32(4):339–347. [DOI] [PubMed] [Google Scholar]
- 28.Shih SL, Gerrard P, Goldstein R, et al. Functional Status Outperforms Comorbidities in Predicting Acute Care Readmissions in Medically Complex Patients. J Gen Intern Med. 2015;30(11):1688–1695. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Slocum C, Gerrard P, Black-Schaffer R, et al. Functional Status Predicts Acute Care Readmissions from Inpatient Rehabilitation in the Stroke Population. PloS one. 2015;10(11):e0142180. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Sun A, Netzer G, Small DS, et al. Association Between Index Hospitalization and Hospital Readmission in Sepsis Survivors. Critical care medicine. 2016;44(3):478–487. [DOI] [PubMed] [Google Scholar]
- 31.Donnelly JP, Hohmann SF, Wang HE. Unplanned Readmissions After Hospitalization for Severe Sepsis at Academic Medical Center-Affiliated Hospitals. Critical care medicine. 2015;43(9):1916–1927. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Gadre SK, Shah M, Mireles-Cabodevila E, Patel B, Duggal A. Epidemiology and Predictors of 30-Day Readmission in Patients With Sepsis. Chest. 2019;155(3):483–490. [DOI] [PubMed] [Google Scholar]
- 33.Goodwin AJ, Rice DA, Simpson KN, Ford DW. Frequency, cost, and risk factors of readmissions among severe sepsis survivors. Critical care medicine. 2015;43(4):738–746. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Auerbach AD, Kripalani S, Vasilevskis EE, et al. Preventability and Causes of Readmissions in a National Cohort of General Medicine Patients. JAMA internal medicine. 2016;176(4):484–493. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
E-Figure 1. Person-Level Analytic Sample Selection (N=165,228) I
E-Table 1: Full Model Marginal Effects of Readmission within 7 days (Overall Risk = 6.64%)