The relationship between severity of illness and hospital length of stay and mortality

Social determinants of health predict readmission following COVID-19 hospitalization: a health information exchange-based retrospective cohort study

Introduction

Since February 2020, over 104 million people in the United States have been diagnosed with SARS-CoV-2 infection, or COVID-19, with over 8.5 million reported in the state of Texas. This study analyzed social determinants of health as predictors for readmission among COVID-19 patients in Southeast Texas, United States.

Methods

A retrospective cohort study was conducted investigating demographic and clinical risk factors for 30, 60, and 90-day readmission outcomes among adult patients with a COVID-19-associated inpatient hospitalization encounter within a regional health information exchange between February 1, 2020, to December 1, 2022.

Results and discussion

In this cohort of 91,007 adult patients with a COVID-19-associated hospitalization, over 21% were readmitted to the hospital within 90  days (n = 19,679), and 13% were readmitted within 30  days (n = 11,912). In logistic regression analyses, Hispanic and non-Hispanic Asian patients were less likely to be readmitted within 90  days (adjusted odds ratio [aOR]: 0.8, 95% confidence interval [CI]: 0.7-0.9, and aOR: 0.8, 95% CI: 0.8-0.8), while non-Hispanic Black patients were more likely to be readmitted (aOR: 1.1, 95% CI: 1.0-1.1, p = 0.002), compared to non-Hispanic White patients. Area deprivation index displayed a clear dose-response relationship to readmission: patients living in the most disadvantaged neighborhoods were more likely to be readmitted within 30 (aOR: 1.1, 95% CI: 1.0-1.2), 60 (aOR: 1.1, 95% CI: 1.2-1.2), and 90  days (aOR: 1.2, 95% CI: 1.1-1.2), compared to patients from the least disadvantaged neighborhoods. Our findings demonstrate the lasting impact of COVID-19, especially among members of marginalized communities, and the increasing burden of COVID-19 morbidity on the healthcare system.

Socioeconomic Factors Are Associated With Severity of Hospitalization in Pediatric Lupus: An Analysis of the 2016 Kids' Inpatient Database

OBJECTIVE

Health disparities in adult lupus, including higher disease severity and activity among those in poverty, have been identified. Similar associations in pediatric lupus have not been clearly established. This study was undertaken to investigate the relationship of income level and other socioeconomic factors with length of stay (LOS) in the hospital and severe lupus features using the 2016 Kids' Inpatient Database (KID).

METHODS

Lupus hospitalizations were identified in children ages 2-20 years in the 2016 KID using International Statistical Classification of Diseases and Related Health Problems, Tenth Revision (ICD-10) codes (M32). Univariate and multivariate negative binomial regression analyses were used to analyze the association of income level, race and ethnicity, and insurance status with LOS in the hospital. Univariate and multivariate logistic regression analyses were used to analyze the association of the same predictors with the presence of severe lupus features, defined using ICD-10 codes associated with lupus sequelae (e.g., lupus nephritis).

RESULTS

A total of 3,367 unweighted (4,650 weighted) lupus hospitalizations were identified. Income level was found to be a statistically significant predictor of increased LOS in the hospital for those in the lowest income quartile (adjusted incidence rate ratio 1.12 [95% confidence interval (95% CI) 1.02-1.23]). Black race, "other" race, and public insurance were also associated with severe lupus features (adjusted odds ratio [ORadj ] 1.51 [95% CI 1.11-2.06]; ORadj 1.61 [95% CI 1.01-2.55]; and ORadj 1.51 [95% CI 1.17-2.55], respectively).

CONCLUSION

Using a nationally representative data set, income level was found to be a statistically significant predictor of LOS in the hospital among those with the lowest reported income, highlighting a potential target population for intervention. Additionally, Black race and public insurance were associated with severe lupus features.

Impact of workforce flexibility on quality of care: moderating effects of workload and severity of illness

Purpose

As a way of alleviating nursing workforce shortages, health care managers are employing two types of workforce flexibility: supplemental staffing and floating among units. In this paper, the authors investigate the moderating effects of two critical situational variables – namely, job-level workload and severity of illness (SOI) in a given unit – on the relationship between workforce flexibility and quality of care as assessed by the nurses at the unit-level.

Design/methodology/approach

The authors empirically test the relationship between a unit's floating of nurses and the use of supplemental workforce on the quality of patient care and the moderating role of patient SOI and job-level workload on this relationship using 357 hospital-unit observations.

Findings

The authors find that situational variables play a critical role in flexible staffing strategies and they should be accounted for carefully to obtain the best quality of care outcomes. The authors find that the well-known negative effect of supplemental staffing on quality of care is not universal and appears to be moderated by the situational factors studied in this paper.

Practical implications

For best outcomes, staffing manager who oversee multiple units should use supplemental staff on units that have lower job-level workload and on units that have high severity of illness. The authors also find that managers of units with patients who are less-severely ill should encourage nurses to float out and return to their home unit. This strategy will improve quality of patient care in the home unit.

Originality/value

While some research analyzes the direct link between flexibility and quality performance, how this relationship is affected by varying situational factors within a unit has not been studied so far.

Relationship of patient characteristics and inpatient rehabilitation services to 5-year outcomes following spinal cord injury: A follow up of the SCIRehab project

OBJECTIVE

To examine associations of patient characteristics and treatment quantity delivered during inpatient spinal cord injury (SCI) rehabilitation with outcomes at 5 years post-injury and compare them to the associations found at 1 year post-injury.

DESIGN

Observational study using Practice-Based Evidence research methodology in which clinicians documented treatment details. Regression modeling was used to predict outcomes.

SETTING

Five inpatient SCI rehabilitation centers in the US.

PARTICIPANTS

Participants were 792 SCIRehab participants who were >12 years of age, gave informed consent, and completed both a 1-year and 5-year post-injury interview.

OUTCOME MEASURES

Outcome data were derived from Spinal Cord Injury Model Systems (SCIMS) follow-up interviews at 5 years post-injury and, similar to the 1-year SCIMS outcomes, included measures of physical independence, societal participation, life satisfaction, and depressive symptoms, as well as place of residence, school/work attendance, rehospitalization, and presence of pressure ulcers.

RESULTS

Consistent with 1-year findings, patient characteristics continue to be strong predictors of outcomes 5-years post-injury, although several variables add to the prediction of some of the outcomes. More time in physical therapy and therapeutic recreation were positive predictors of 1-year outcomes, which held less true at 5 years. Greater time spent with psychology and social work/case management predicted greater depressive symptomatology 5-years post-injury. Greater clinician experience was a predictor at both 1- and 5 -years, although the related positive outcomes varied across years.

CONCLUSION

Various outcomes 5-years post-injury were primarily explained by pre-and post-injury characteristics, with little additional variance offered by the quantity of treatment received during inpatient rehabilitation.

Hospital Length of Stay Prediction Methods: A Systematic Review

OBJECTIVE

This systematic review sought to establish a picture of length of stay (LOS) prediction methods based on available hospital data and study protocols designed to measure their performance.

MATERIALS AND METHODS

An English literature search was done relative to hospital LOS prediction from 1972 to September 2019 according to the PRISMA guidelines. Articles were retrieved from PubMed, ScienceDirect, and arXiv databases. Information were extracted from the included papers according to a standardized assessment of population setting and study sample, data sources and input variables, LOS prediction methods, validation study design, and performance evaluation metrics.

RESULTS

Among 74 selected articles, 98.6% (73/74) used patients' data to predict LOS; 27.0% (20/74) used temporal data; and 21.6% (16/74) used the data about hospitals. Overall, regressions were the most popular prediction methods (64.9%, 48/74), followed by machine learning (20.3%, 15/74) and deep learning (17.6%, 13/74). Regarding validation design, 35.1% (26/74) did not use a test set, whereas 47.3% (35/74) used a separate test set, and 17.6% (13/74) used cross-validation. The most used performance metrics were R2 (47.3%, 35/74), mean squared (or absolute) error (24.4%, 18/74), and the accuracy (14.9%, 11/74). Over the last decade, machine learning and deep learning methods became more popular (P=0.016), and test sets and cross-validation got more and more used (P=0.014).

CONCLUSIONS

Methods to predict LOS are more and more elaborate and the assessment of their validity is increasingly rigorous. Reducing heterogeneity in how these methods are used and reported is key to transparency on their performance.

Differences in functional improvement based on history of substance abuse and pain severity following spinal cord injury

Background: This study explored the relationship between history of substance abuse and pain severity during inpatient rehabilitation following traumatic spinal cord injury (SCI). Methods: Secondary analysis of a prospective longitudinal study. An adjusted general linear model was used to examine differences in functional improvement based on history of substance abuse and pain severity. Results: Over 50% of the sample had a history of substance abuse, and 94% reported moderate or severe pain. There was a significant interaction between the history of substance abuse and pain severity (p = 0.01, partial η² = 0.012). A difference in functional improvement was found among individuals who reported low pain; those with a history of substance abuse achieved less functional improvement than those without a history of substance abuse, M = 5.32, SE = 1.95, 95% CI 0.64-10.01. Conclusions: A history of substance abuse and post-injury pain are prevalent among individuals with SCI in rehabilitation, and there may be a meaningful relationship between these two patient characteristics and functional improvement. The results provide potential new insights into the characteristics of vulnerable subpopulations during SCI rehabilitation. Furthering our understanding of these results warrants future investigation to prevent and minimize poor outcomes among vulnerable SCI patients.

Association of prurigo nodularis and infectious disease hospitalizations: a national cross-sectional study

BACKGROUND

Prurigo nodularis (PN) is associated with a variety of systemic comorbidities, including infectious diseases such as HIV and viral hepatitis. There are limited data on other infectious disease comorbidities in patients with PN.

AIM

To characterize infectious disease hospitalizations among patients with PN and the associated cost burden.

METHODS

We searched the 2016-2017 National Inpatient Sample, a cross-sectional sample of 20% of all US hospitalizations, for infectious disease hospitalizations among patients with PN. Associations of PN with infections and related costs were determined using multivariable logistic and linear regression, adjusting for age, race, sex and insurance type.

RESULTS

PN was associated with any infection overall (OR = 2.98, 95% CI 2.49-3.56), and with HIV, cutaneous, hepatobiliary, central nervous system, bacterial, viral and fungal/parasitic infections and for sepsis. Patients with PN had a higher mean cost of care (US$11 667 vs. US$8893, P < 0.001) and length of stay (5.5 vs. 4.2 days, P < 0.001) for any infection overall and for 7 of 13 other infections. Adjusting for age, race, sex and insurance coverage, PN was associated with higher cost (+30%, 95% CI +17 to +44%) and higher length of stay (+30%, 95% CI +18 to +44%) for any infection overall, and for several specific infections. These associations remained with alternate regression models adjusting for severity of illness.

CONCLUSION

There is a high infectious disease burden among patients with PN, corresponding to higher healthcare utilization and spending. Clinicians must be aware of these associations when treating these patients with immunomodulatory drugs.

Development and Testing of Compatible Diagnosis Code Lists for the Functional Comorbidity Index: International Classification of Diseases, Ninth Revision, Clinical Modification and International Classification of Diseases, 10th Revision, Clinical Modification

BACKGROUND

The Functional Comorbidity Index (FCI) was developed for community-based adult populations, with function as the outcome. The original FCI was a survey tool, but several International Classification of Diseases (ICD) code lists-for calculating the FCI using administrative data-have been published. However, compatible International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) and ICD-10-CM versions have not been available.

OBJECTIVE

We developed ICD-9-CM and ICD-10-CM diagnosis code lists to optimize FCI concordance across ICD lexicons.

RESEARCH DESIGN

We assessed concordance and frequency distributions across ICD lexicons for the FCI and individual comorbidities. We used length of stay and discharge disposition to assess continuity of FCI criterion validity across lexicons.

SUBJECTS

State Inpatient Databases from Arizona, Colorado, Michigan, New Jersey, New York, Utah, and Washington State (calendar year 2015) were obtained from the Healthcare Cost and Utilization Project. State Inpatient Databases contained ICD-9-CM diagnoses for the first 3 calendar quarters of 2015 and ICD-10-CM diagnoses for the fourth quarter of 2015. Inpatients under 18 years old were excluded.

MEASURES

Length of stay and discharge disposition outcomes were assessed in separate regression models. Covariates included age, sex, state, ICD lexicon, and FCI/lexicon interaction.

RESULTS

The FCI demonstrated stability across lexicons, despite small discrepancies in prevalence for individual comorbidities. Under ICD-9-CM, each additional comorbidity was associated with an 8.9% increase in mean length of stay and an 18.5% decrease in the odds of a routine discharge, compared with an 8.4% increase and 17.4% decrease, respectively, under ICD-10-CM.

CONCLUSION

This study provides compatible ICD-9-CM and ICD-10-CM diagnosis code lists for the FCI.

Multitask learning and benchmarking with clinical time series data

Health care is one of the most exciting frontiers in data mining and machine learning. Successful adoption of electronic health records (EHRs) created an explosion in digital clinical data available for analysis, but progress in machine learning for healthcare research has been difficult to measure because of the absence of publicly available benchmark data sets. To address this problem, we propose four clinical prediction benchmarks using data derived from the publicly available Medical Information Mart for Intensive Care (MIMIC-III) database. These tasks cover a range of clinical problems including modeling risk of mortality, forecasting length of stay, detecting physiologic decline, and phenotype classification. We propose strong linear and neural baselines for all four tasks and evaluate the effect of deep supervision, multitask training and data-specific architectural modifications on the performance of neural models.

Analysing repeated hospital readmissions using data mining techniques

Few studies have examined how to identify future readmission of patients with a large number of repeat emergency department (ED) visits. We explore 30-day readmission risk prediction using Microsoft's AZURE machine learning software and compare five classification methods: Logistic Regression, Boosted Decision Trees (BDTs), Support Vector Machine (SVM), Bayes Point Machine (BPM), and Two-Class Neural Network (TCNN). We predict the last readmission visit of frequent ED patients extracted from the electronic health records of their 8455 penultimate visits. The methods show differential improvement, with the BDT indicating marginally better AUC (area under the ROC curve) than logistic regression and BPM, followed by the TCNN and SVM. A comparison of BDT and Logistic Regression results for correct and incorrect classification highlights the similarities and differences in the significant predictors identified by each method. Future research may incorporate time-varying covariates to identify other longitudinal factors that can lead to readmission risk reduction.

Length of stay overestimates severity of post-ERCP pancreatitis: Is it time to revise the consensus definition?

INTRODUCTION

 Length of stay (LOS) is an important determinant of the severity of post-ERCP pancreatitis (PEP) in the consensus definition. The aim of our study was to evaluate and compare severity of PEP based on the revised Atlanta classification (RAC) and the consensus definition.

PATIENTS AND METHODS

 Between 1/2000 and 12/2011, all adult patients admitted with suspicion of PEP after outpatient ERCP were evaluated. PEP was defined using the RAC, but the severity of PEP was defined using both revised Atlanta and consensus definitions.

RESULTS

 A total of 341 patients (mean age 49 years and 75 % females) were diagnosed with PEP. The consensus definition classified 57 %, 37 %, and 8 % of patients with mild, moderate, and severe PEP, respectively. The RAC diagnosed 94 %, 6 %, and 0 % with mild, moderate, and severe acute pancreatitis, respectively. Of the patients diagnosed with moderate-severe PEP by consensus definition, only 12.5 % had clinical parameters of pancreatitis severity, such as acute fluid collection(s), pancreatic necrosis, transient organ failure and/or required percutaneous or surgical drainage, while 87.5 % were classified only based on a LOS ≥ 4 days. The most common reason for increased LOS was persistent post-procedural abdominal pain in 47 % of patients, followed by other reasons not related to pancreatitis in 17 %.

CONCLUSION

 The consensus definition overestimates the rates of severe PEP when compared to the RAC. The majority of PEP patients classified as moderate-severe PEP have extended LOS, due to post-procedural abdominal pain rather than complications of PEP.

Impact of computerized provider order entry (CPOE) on length of stay and mortality

Objective: To examine changes in patient outcome variables, length of stay (LOS), and mortality after implementation of computerized provider order entry (CPOE).

Materials and Methods: A 5-year retrospective pre-post study evaluated 66 186 patients and 104 153 admissions (49 683 pre-CPOE, 54 470 post-CPOE) at an academic medical center. Generalized linear mixed statistical tests controlled for 17 potential confounders with 2 models per outcome.

Results: After controlling for covariates, CPOE remained a significant statistical predictor of decreased LOS and mortality. LOS decreased by 0.90 days, P < .0001. Mortality decrease varied by model: 1 death per 1000 admissions (pre = 0.006, post = 0.0005, P < .001) or 3 deaths (pre = 0.008, post = 0.005, P < .01). Mortality and LOS decreased in medical and surgical units but increased in intensive care units.

Discussion: This study examined CPOE at multiple levels. Given the inability to randomize CPOE assignment, these results may only be applicable to the local setting. Temporal trends found in this study suggest that hospital-wide implementations may have impacted nursing staff and new residents. Differences in the results were noted at the patient care unit and room levels. These differences may partly explain the mixed results from previous studies.

Conclusion: Controlling for confounders, CPOE implementation remained a statistically significant predictor of LOS and mortality at this site. Mortality appears to be a sensitive outcome indicator with regard to hospital-wide implementations and should be further studied.

Analysing repeated hospital readmissions using data mining techniques

Few studies have examined how to identify future readmission of patients with a large number of repeat emergency department (ED) visits. We explore 30-day readmission risk prediction using Microsoft's AZURE machine learning software and compare five classification methods: Logistic Regression, Boosted Decision Trees (BDTs), Support Vector Machine (SVM), Bayes Point Machine (BPM), and Two-Class Neural Network (TCNN). We predict the last readmission visit of frequent ED patients extracted from the electronic health records of their 8455 penultimate visits. The methods show differential improvement, with the BDT indicating marginally better AUC (area under the ROC curve) than logistic regression and BPM, followed by the TCNN and SVM. A comparison of BDT and Logistic Regression results for correct and incorrect classification highlights the similarities and differences in the significant predictors identified by each method. Future research may incorporate time-varying covariates to identify other longitudinal factors that can lead to readmission risk reduction.

Combining adverse pregnancy and perinatal outcomes for women exposed to antiepileptic drugs during pregnancy, using a latent trait model

Background

Application of latent variable models in medical research are becoming increasingly popular. A latent trait model is developed to combine rare birth defect outcomes in an index of infant morbidity.

Methods

This study employed four statewide, retrospective 10-year data sources (1999 to 2009). The study cohort consisted of all female Florida Medicaid enrollees who delivered a live singleton infant during study period. Drug exposure was defined as any exposure to Antiepileptic drugs (AEDs) during pregnancy. Mothers with no AED exposure served as the AED unexposed group for comparison. Four adverse outcomes, birth defect (BD), abnormal condition of new born (ACNB), low birth weight (LBW), and pregnancy and obstetrical complication (PCOC), were examined and combined using a latent trait model to generate an overall severity index. Unidimentionality, local independence, internal homogeneity, and construct validity were evaluated for the combined outcome.

Results

The study cohort consisted of 3183 mother-infant pairs in total AED group, 226 in the valproate only subgroup, and 43,956 in the AED unexposed group. Compared to AED unexposed group, the rate of BD was higher in both the total AED group (12.8% vs. 10.5%, P < .0001), and the valproate only subgroup (19.6% vs. 10.5%, P < .0001). The combined outcome was significantly correlated with the length of hospital stay during delivery in both the total AED group (Rho = 0.24, P < .0001) and the valproate only subgroup (Rho = 0.16, P = .01). The mean score for the combined outcome in the total AED group was significantly higher (2.04 ± 0.02 vs. 1.88 ± 0.01, P < .0001) than AED unexposed group, whereas the valproate only subgroup was not.

Conclusions

Latent trait modeling can be an effective tool for combining adverse pregnancy and perinatal outcomes to assess prenatal exposure to AED, but evaluation of the selected components is essential to ensure the validity of the combined outcome.

Electronic supplementary material

The online version of this article (doi:10.1186/s12884-016-1190-7) contains supplementary material, which is available to authorized users.

Day of Surgery Affects Length of Stay and Charges in Primary Total Hip and Knee Arthroplasty

BACKGROUND

Length of hospital stay (LOS) is a large driver of cost after primary total joint arthroplasty (TJA). Strategies to decrease LOS may help reduce the economic burden of TJA. This study's aim was to investigate the effect of day of the week of surgery on mean LOS and total charges following primary total knee arthroplasty (TKA) and total hip arthroplasty (THA).

METHODS

An administrative clinical database at a large US health care system was reviewed for all primary THA and TKA admissions performed between 2010 and 2012 (n = 15,237). Of these, 14,800 cases met our inclusion criteria and were analyzed. Furthermore, the cohort was divided into early (Monday/Tuesday) and late week (Thursday/Friday) surgeries, excluding Wednesday surgeries (n = 2835). Univariate and multiple regression analyses examined the effect of each variable on LOS.

RESULTS

Mean LOS for THA and TKA on Monday was 3.54 and 3.35 days and increased to 4.12 and 3.66 days on Friday (P < .0001), respectively. Late vs early week admissions had 0.358 (95% confidence interval: 0.29-0.425, P < .001) additional hospital days. Increased age (0.003 days per unit increase in age, P = .02) and severity of illness score (0.781 days per level increase, P < .001) were associated with increased LOS. Late week surgery had a greater effect on LOS for TKA than for THA. TKAs were associated with higher charges for late week surgery vs early week surgery (P < .001).

CONCLUSION

Late week TJA cases, older age, and increasing severity of illness score were associated with increased LOS. Furthermore, late week TKA was associated with increased total charges.

Using geographic variation in unplanned ambulatory care sensitive condition admission rates to identify commissioning priorities: an analysis of routine data from England

Objectives To use geographic variation in unplanned ambulatory care sensitive condition admission rates to identify the clinical areas and patient subgroups where there is greatest potential to prevent admissions and improve the quality and efficiency of care. Methods We used English Hospital Episode Statistics data from 2011/2012 to describe the characteristics of patients admitted for ambulatory care sensitive condition care and estimated geographic variation in unplanned admission rates. We contrasted geographic variation across admissions with different lengths of stay which we used as a proxy for clinical severity. We estimated the number of bed days that could be saved under several scenarios. Results There were 1.8 million ambulatory care sensitive condition admissions during 2011/2012. Substantial geographic variation in ambulatory care sensitive condition admission rates was commonplace but mental health care and short-stay (<2 days) admissions were particularly variable. Reducing rates in the highest use areas could lead to savings of between 0.4 and 2.8 million bed days annually. Conclusions Widespread geographic variations in admission rates for conditions where admission is potentially avoidable should concern commissioners and could be symptomatic of inefficient care. Further work to explore the causes of these differences is required and should focus on mental health and short-stay admissions.

The number of inpatient consultations is negatively correlated with patient satisfaction in patients with prolonged hospital stays

BACKGROUND

Patient satisfaction is often measured using the Hospital Consumer Assessment of Healthcare Providers and Systems survey. Our aim was to examine the structural and clinical determinants of satisfaction among inpatients with prolonged lengths of stays (LOS).

METHODS

Adult patients who were admitted between 2009 and 2012, had a LOS of 21 days or more, and completed the Hospital Consumer Assessment of Healthcare Providers and Systems survey, were included. Univariate analyses assessed the relationship between satisfaction and patient/system variables. Recursive partitioning was used to examine the relative importance of the identified variables.

RESULTS

One hundred one patients met inclusion criteria. The average LOS was 35 days and 58% were admitted to a surgical service. Satisfaction with physician communication was significantly associated with fewer consultations (P < .01), nonoperative admission (P < .001), no intensive care unit stay (P < .01), nonsurgical service (P < .01), and non-emergency room admissions (P = .03). Among these, having fewer consultations had the highest relative importance.

CONCLUSIONS

In long stay patients, having fewer inpatient consultations was the strongest predictor of patient satisfaction with physician communication. This suggests that examination of patient-level data in clinically relevant subgroups may be a useful way to identify targets for quality improvement.

Hospital Clostridium difficile Infection Rates and Prediction of Length of Stay in Patients Without C. difficile Infection

BACKGROUND Inpatient length of stay (LOS) has been used as a measure of hospital quality and efficiency. Patients with Clostridium difficile infections (CDI) have longer LOS. OBJECTIVE To describe the relationship between hospital CDI incidence and the LOS of patients without CDI. DESIGN Retrospective cohort analysis. METHODS We predicted average LOS for patients without CDI at both the hospital and patient level using hospital CDI incidence. We also controlled for hospital characteristics (eg, bed size) and patient characteristics (eg, comorbidities, age). SETTING Healthcare Cost and Utilization Project Nationwide Inpatient Sample, 2009-2011. PATIENTS The Nationwide Inpatient Sample includes patients from a 20% sample of all nonfederal US hospitals. RESULTS Inpatient LOS was significantly longer (P<.001) at hospitals with greater CDI incidence at both the hospital and individual level. At a hospital level, a percentage point increase in the CDI incidence rate was associated with more than an additional day's stay (between 1.19 and 1.61 days). At the individual level, controlling for all observable variables, a percentage point increase in the CDI incidence rate at their hospital was also associated with longer LOS (between 0.6 and 1.05 additional days). Hospital CDI incidence had a larger impact on LOS than many other commonly used predictors of LOS. CONCLUSION CDI rates are a predictor of LOS in patients without CDI at an individual and institutional level. CDI rates are easy to measure and report and thus may provide an important marker for hospital efficiency and/or quality.

Outcomes associated with acute exacerbations of chronic obstructive pulmonary disorder requiring hospitalization

BACKGROUND

Acute exacerbations of chronic obstructive pulmonary disorder (AECOPD) are known to be associated with increased morbidity and mortality and have a significant socioeconomic impact. The factors that determine frequent hospital readmissions for AECOPD are poorly understood. The present study was done to ascertain failures rates following AECOPD and to evaluate factors associated with frequent readmissions.

MATERIALS AND METHODS

We conducted a prospective study among 186 patients with COPD with one or more admissions for acute exacerbations in a tertiary care hospital. Frequency of previous re-admissions for AECOPD in the past year, and clinical characteristics, including spirometry were ascertained in the stable state both before discharge and at 6-month post-discharge. Failure rates following treatment were ascertained during the follow-up period. All the patients were followed up for a period of 2 years after discharge to evaluate re-admissions for the AECOPD.

RESULTS

Of 186 COPD patients admitted for AECOPD, 54% had one or more readmission, and another 45% had two or more readmissions over a period of 2 years. There was a high prevalence of current or ex-heavy smokers, associated co-morbidity, underweight patients, low vaccination prevalence and use of domiciliary oxygen therapy among COPD patients. A total of 12% mortality was observed in the present study. Immediate failure rates after first exacerbation was observed to be 34.8%. Multivariate analysis showed that duration >20 years (OR = 0.37; 95% CI: 0.10-0.86), use of Tiotropium (OR = 2.29; 95% CI: 1.12-4.69) and use of co-amoxiclav during first admission (OR = 2.41; 95% CI: 1.21-4.79) were significantly associated with higher immediate failure rates. The multivariate analysis for repeated admissions revealed that disease duration >10 years (OR = 0.50; 95% CI: 0.27-0.93), low usage of inhaled ICS + LABA (OR = 2.21; 95% CI: 1.08-4.54), and MRC dyspnea grade >3 (OR = 2.51; 95% CI: 1.08-5.82) were independently associated with frequent re-admissions for AECOPD.

CONCLUSIONS

The outcomes of patients admitted for an acute exacerbation of COPD were poor. The major factors influencing frequency of repeated COPD exacerbations were disease duration, low usage of inhaled ICS + LABA, and MRC dyspnea grade >3.

Trends in Endovascular Aortic Aneurysm Repair Length of Stay over a Decade at a Tertiary Academic Institution

BACKGROUND

Length of stay (LOS) is used as a quality metric to reduce cost and improve value of delivery of care. We sought to analyze trends in endovascular aortic aneurysm repair (EVAR) LOS at a tertiary academic institution over the last decade.

METHODS

A retrospective review of prospectively collected data was performed. Infrarenal EVARs from 2001 to 2013 were divided into 3 groups: group I (2001-2004), EVARs were performed as part of clinical trials; group II (2005-2008), EVARs were referred to a tertiary referral center with the most experience with EVAR; group III (2009-2013), EVARs were referred to academic institutions in the presence of severe patient comorbidities. Trends in LOS and correlation with severity of illness (SOI) as based on All Patient Refined Diagnosis Related Groups and admission and/or disposition status were analyzed. LOS index (LOSI) at our institution was then compared with University HealthSystem Consortium (UHC) Hospitals over the past 3 years.

RESULTS

A total of 1,265 EVARs were performed during this time period: 325 in group I, 547 in group II, and 393 in group III. The median LOS was 4 days (inter quartile range [IQR], 2-6) vs. 3 days (IQR, 2-5) ± 0.28 vs. 4 days (IQR, 3-7), respectively (P < 0.01). Although moderate SOI was fairly constant over time (P = 0.66), major and/or extreme SOI constituted a greater proportion of patients in group I, was reduced in group II, and was again increased in group III, P < 0.01. The complication rate paralleled this pattern (group I, 15.2%; group II, 8.6%; group III, 10.4%; P = 0.02). The percentage of patients discharged to nursing home and/or rehab was 5.7% in group I, 8.2% in group II, 11.5% in group III (P = 0.03). Cases that were performed urgently and/or emergently increased over time: 11.6% in group I, 14.9% in group II, 21.6% in group III (P = 0.01). The risk-adjusted LOSI at our institution was significantly greater (1.25) when compared with UHC hospitals (0.75).

CONCLUSIONS

Our study suggests a relationship between time period of EVAR, SOI, complications, admission status, and LOS. Attention to these trends could be used to decrease LOS in an increasingly complex patient population.

Variation in Rehabilitation Treatment Patterns for Hip Fracture Treated With Arthroplasty

BACKGROUND

Recommendations for health care redesign often advocate for comparative effectiveness research that is patient-centered. For patients who require rehabilitation services, a first step in this research process is to understand current practices for specific patient groups.

OBJECTIVE

To document in detail the physical and occupational therapy treatment activities for inpatient hip fracture rehabilitation among 3 patient subgroups distinguished by their early rate of functional recovery between time of surgery to rehabilitation admission.

DESIGN

Multicenter prospective observational cohort, practice-based evidence, study.

SETTING

Seven skilled nursing facilities and 11 inpatient rehabilitation facilities across the United States.

PARTICIPANTS

A total of 226 patients with hip fractures treated with hip arthroplasty.

METHODS

Comparisons of physical and occupational therapy treatment activities among 3 groups with different initial recovery trajectory (IRT) rates (slower, moderate, faster).

MAIN OUTCOME MEASURE(S)

Percent of patients in each IRT group exposed to each physical and occupational therapy activity (exposure), and mean minutes per week for each activity (intensity).

RESULTS

The number of patients exposed to different physical or occupational therapy activities varied within the entire sample. More specifically, among the 3 IRT groups, significant differences in exposure occurred for 44% of physical therapy activities and 39% of occupational therapy activities. More patients in the slower recovery group, IRT 1, received basic activities of daily living treatments and more patients in the faster recovery group, IRT 3, received advanced activities. The moderate recovery group, IRT 2, had some treatments similar to IRT 1 group and others similar to IRT 3 group.

CONCLUSIONS

Analyses of practice-based evidence on inpatient rehabilitation of hip fracture patients treated with arthroplasty identified differences in therapy activities among three patient groups classified by IRT rates. These results may enhance physiatrists', other physicians', and rehabilitation teams' understanding of inpatient rehabilitation for these patients and help design future comparative effectiveness research.

OrderRex: clinical order decision support and outcome predictions by data-mining electronic medical records

OBJECTIVE

To answer a "grand challenge" in clinical decision support, the authors produced a recommender system that automatically data-mines inpatient decision support from electronic medical records (EMR), analogous to Netflix or Amazon.com's product recommender.

MATERIALS AND METHODS

EMR data were extracted from 1 year of hospitalizations (>18K patients with >5.4M structured items including clinical orders, lab results, and diagnosis codes). Association statistics were counted for the ∼1.5K most common items to drive an order recommender. The authors assessed the recommender's ability to predict hospital admission orders and outcomes based on initial encounter data from separate validation patients.

RESULTS

Compared to a reference benchmark of using the overall most common orders, the recommender using temporal relationships improves precision at 10 recommendations from 33% to 38% (P < 10(-10)) for hospital admission orders. Relative risk-based association methods improve inverse frequency weighted recall from 4% to 16% (P < 10(-16)). The framework yields a prediction receiver operating characteristic area under curve (c-statistic) of 0.84 for 30 day mortality, 0.84 for 1 week need for ICU life support, 0.80 for 1 week hospital discharge, and 0.68 for 30-day readmission.

DISCUSSION

Recommender results quantitatively improve on reference benchmarks and qualitatively appear clinically reasonable. The method assumes that aggregate decision making converges appropriately, but ongoing evaluation is necessary to discern common behaviors from "correct" ones.

CONCLUSIONS

Collaborative filtering recommender algorithms generate clinical decision support that is predictive of real practice patterns and clinical outcomes. Incorporating temporal relationships improves accuracy. Different evaluation metrics satisfy different goals (predicting likely events vs. "interesting" suggestions).

Factors associated with length of stay following an emergency medical admission

BACKGROUND

Hospitals are under pressure to use resources in the most efficient manner. We have examined the factors predicting Length of Stay (LOS) in one institution, using a database of all episodes of emergency medical admissions prospectively collected over 12 years.

AIM

To examine the ability to predict hospital LOS following an emergency medical hospital admission.

METHODS

All emergency admissions (66,933 episodes; 36,271 patients) to St. James's Hospital, Dublin, Ireland over a 12-year period (2002-2013) were evaluated in relation to LOS. Predictor variables (identified univariately) were entered into a multiple logistic regression model to predict a longer or shorter LOS (bivariate at the median). The data was also modelled as count data (absolute LOS), using zero truncated Poisson regression methodology. Appropriate post-estimation techniques for model fit were then applied to assess the resulting model.

RESULTS

The major predictors of LOS included Acute Illness Severity (biochemical laboratory score at admission), Charlson co-morbidity, Manchester Triage Category at admission, Diagnosis Related Group, sepsis status (based on blood culture result), and Chronic Disease Score Indicator. The full model to predict a LOS above or below the median had an Area Under Receiver Operating Characteristic (AUROC) of 0.71 (95% CI: 0.70, 0.71). The truncated Poisson model appeared to achieve a good model fit (R(2) statistic=0.76).

CONCLUSION

Predictor variables strongly correlated with LOS; there were linear increases within categories and summation between variables. More predictor variables may improve model reliability but predicting LOS ranges or quantiles may be more realistic, based on these results.

Basics of Stratifying for Severity of Illness

Conventional wisdom suggests that those who assess healthcare processes and outcomes always should stratify cases by severity of illness; however, infection control personnel should analyze each quality assessment tool with and without severity adjustment and determine whether such adjustment is necessary. This article briefly reviews severity adjustments for diseases or procedures involving specific organ systems, as well as those applicable to all diseases, including the commercially available systems. Also discussed is whether and how these various systems for severity adjustment can be compared. Finally, the article will provide selected references for individuals who will use these scoring systems and need more information.

Managing for efficiency in health care: the case of Greek public hospitals

This paper evaluates the efficiency of public hospitals with two alternative conceptual models. One model targets resource usage directly to assess production efficiency, while the other model incorporates financial results to assess economic efficiency. Performance analysis of these models was conducted in two stages. In stage one, we utilized data envelopment analysis to obtain the efficiency score of each hospital, while in stage two we took into account the influence of the operational environment on efficiency by regressing those scores on explanatory variables that concern the performance of hospital services. We applied these methods to evaluate 96 general hospitals in the Greek national health system. The results indicate that, although the average efficiency scores in both models have remained relatively stable compared to past assessments, internal changes in hospital performances do exist. This study provides a clear framework for policy implications to increase the overall efficiency of general hospitals.

Relation between inpatient and postdischarge services and outcomes 1 year postinjury in people with traumatic spinal cord injury

OBJECTIVE

To examine the association between inpatient and postdischarge rehabilitation services and function, life satisfaction, and community participation 1 year after spinal cord injury (SCI).

DESIGN

Prospective, observational.

SETTING

Six rehabilitation facilities.

PARTICIPANTS

Patients with SCI (N=1376).

INTERVENTIONS

None.

MAIN OUTCOME MEASURES

Satisfaction with Life Scale (SWLS), Craig Handicap Assessment and Reporting Technique (CHART), motor FIM (mFIM), and return to work/school at 1 year post-SCI.

RESULTS

Demographic and injury characteristics explained 49% of the variance in mFIM and 9% to 25% of the variance in SWLS and CHART social integration, mobility, and occupation scores. Inpatient rehabilitation services explained an additional 2% of the variance for mFIM and 1% to 3% of the variance for SWLS and CHART scores. More time in inpatient physical therapy (PT) was associated with higher mFIM scores; more time in inpatient therapeutic recreation (TR) and social work and more postdischarge nursing (NSG) were associated with lower mFIM scores. More inpatient PT and TR and more postdischarge PT were associated with higher mobility scores; more inpatient psychology (PSY) was associated with lower mobility scores. More postdischarge TR was associated with higher SWLS; more postdischarge PSY services was associated with lower SWLS. Inpatient TR was positively associated with social integration scores; postdischarge PSY was negatively associated with social integration scores. More postdischarge vocational counseling was associated with higher occupation scores. Differences between centers did not explain additional variability in the outcomes studied.

CONCLUSIONS

Inpatient and postdischarge rehabilitation services are weakly associated with life satisfaction and societal participation 1 year after SCI. Further study of the type and intensity of postdischarge services, and the association with outcomes, is needed to ascertain the most effective use of therapy services after SCI.

Rehospitalization in the first year of traumatic spinal cord injury after discharge from medical rehabilitation

OBJECTIVE

To determine rates of rehospitalization among discharged rehabilitation patients with traumatic spinal cord injury (SCI) in the first 12 months postinjury, and to identify factors associated with rehospitalization.

DESIGN

Prospective observational cohort study.

SETTING

Six geographically dispersed rehabilitation centers in the U.S.

PARTICIPANTS

Consecutively enrolled individuals with new traumatic SCI (N=951), who were discharged from participating rehabilitation centers and participated in a 1-year follow-up survey.

INTERVENTIONS

Not applicable.

MAIN OUTCOME MEASURES

Occurrence of postrehabilitation rehospitalization within 1 year of injury, length of rehospitalization stays, and causes of rehospitalizations.

RESULTS

More than one third (36.2%) of participants were rehospitalized at least once in the 12-month follow-up period; 12.5% were rehospitalized at least twice. The average number of rehospitalizations among those rehospitalized at least once was 1.37 times, with an average length of stay (LOS) of 15.5 days across all rehospitalization episodes. The 3 most common health conditions associated with rehospitalization were those related to the genitourinary system (eg, urinary tract infection), respiratory system (eg, pneumonia), and skin and subcutaneous tissue (eg, pressure ulcer). Being a woman (95% confidence interval [CI], 1.034-2.279), having Medicaid as the main payer (95% CI, 1.303-2.936), and more severe case mix were associated with increased odds of rehospitalization. Those who had more intensive physical therapy (95% CI, .960-.981) had lower odds of rehospitalization. Some center-to-center variation in rehospitalization rates remained unexplained after case mix and practice differences were considered. The 6 SCI rehabilitation centers varied nearly 2-fold in rates at which their former SCI patients were rehospitalized--from 27.8% to 50%. Center-to-center variation diminished when patient case mix was considered.

CONCLUSIONS

Compared with earlier studies, rehospitalization rates among individuals with SCI in the first postinjury year remain high and vary by level and completeness of injury. Rehospitalization risk was associated with younger age, being a woman, unemployment and retirement, and Medicaid coverage. Those who had more intensive physical therapy had lower odds of rehospitalization. Future studies should examine center-to-center variations in rehospitalization rates and availability of patient education and community resources.

Association of various comorbidity measures with spinal cord injury rehabilitation outcomes

OBJECTIVE

To investigate the amount of variation in short- and medium-term spinal cord injury (SCI) rehabilitation outcomes explained by various comorbidity measures, over and above patient preinjury characteristics and neurologic and functional status.

DESIGN

Prospective observational cohort study of traumatic SCI patients receiving inpatient rehabilitation and followed up at 1 year postinjury.

SETTING

Inpatient rehabilitation and community follow-up at 6 SCI treatment centers.

PARTICIPANTS

Participants (N=1376) included 1032 patients randomly selected for model development and 344 patients selected for cross-validation.

INTERVENTIONS

Not applicable.

MAIN OUTCOME MEASURES

Rehabilitation length of stay (LOS), return to acute care during rehabilitation, discharge motor FIM, discharge home, rehospitalization after discharge, 1-year return to work/school and 1-year depression symptomatology, motor FIM, and residence. Comorbidity measures used were case-mix groups tier weights, Charlson Comorbidity Index (CCI), and the Comprehensive Severity Index (CSI).

RESULTS

Multivariable regression analyses, controlling for patient preinjury and injury characteristics, found that the maximum Comprehensive Severity Index (MCSI) was a significant and stronger predictor of LOS, return to acute care during rehabilitation, and 1-year motor FIM compared with the case-mix groups tier weight or the CCI. The admission CSI was a strong predictor of LOS. For rehospitalization after discharge, only the case-mix groups tier weight was significant. No comorbidity measure was significant beyond patient preinjury and injury characteristics for discharge home, discharge motor FIM, living at home, depression symptomatology, major depressive syndrome, and return to work/school.

CONCLUSIONS

Patient preinjury and injury characteristics are sufficient to predict most SCI outcomes. For rehabilitation LOS and return to acute care during rehabilitation, one achieves substantially better explanation when taking clinical comorbidity based on the MCSI into account.

Relationship of patient characteristics and rehabilitation services to outcomes following spinal cord injury: the SCIRehab project

BACKGROUND/OBJECTIVE

To examine associations of patient characteristics along with treatment quantity delivered by seven clinical disciplines during inpatient spinal cord injury (SCI) rehabilitation with outcomes at rehabilitation discharge and 1-year post-injury.

METHODS

Six inpatient SCI rehabilitation centers enrolled 1376 patients during the 5-year SCIRehab study. Clinicians delivering standard care documented details of treatment. Outcome data were derived from SCI Model Systems Form I and II and a project-specific interview conducted at approximately 1-year post-injury. Regression modeling was used to predict outcomes; models were cross-validated by examining relative shrinkage of the original model R(2) using 75% of the dataset to the R(2) for the same outcome using a validation subsample.

RESULTS

Patient characteristics are strong predictors of outcome; treatment duration adds slightly more predictive power. More time in physical therapy was associated positively with motor Functional Independence Measure at discharge and the 1-year anniversary, CHART Physical Independence, Social Integration, and Mobility dimensions, and smaller likelihood of rehospitalization after discharge and reporting of pressure ulcer at the interview. More time in therapeutic recreation also had multiple similar positive associations. Time spent in other disciplines had fewer and mixed relationships. Seven models validated well, two validated moderately well, and four validated poorly.

CONCLUSION

Patient characteristics explain a large proportion of variation in multiple outcomes after inpatient rehabilitation. The total amount of treatment received during rehabilitation from each of seven disciplines explains little additional variance. Reasons for this and the phenomenon that sometimes more hours of service predict poorer outcome, need additional study. Note: This is the first of nine articles in the SCIRehab series.

Outcomes of social work and case management services during inpatient spinal cord injury rehabilitation: the SCIRehab project

OBJECTIVE

To investigate associations of social work/case management (SW/CM) services during inpatient rehabilitation following spinal cord injury (SCI) and patient characteristics with outcomes.

DESIGN

Prospective observational cohort of individuals with SCI receiving inpatient rehabilitation.

SETTING

Six inpatient rehabilitation centers.

PARTICIPANTS

1032 individuals with traumatic SCI.

INTERVENTIONS

Not applicable.

MAIN OUTCOME MEASURE(S)

Type of residence at the time of rehabilitation discharge. Employment/school status, presence of a pressure ulcer, Patient History Questionnaire, Satisfaction with Life Scale, Craig Handicap Assessment and Reporting Technique (CHART) subscales, and rehospitalization at 1-year post-injury.

RESULTS

The intensity of specific SW/CM services is associated with multiple outcomes examined. More sessions dedicated to discharge planning for a home discharge and financial planning were associated positively with more discharge to home, while more sessions focused on planning for discharge to a location other than home, e.g. nursing home or long-term acute care facilities, have negative associations with societal participation outcomes (CHART Social Integration, Occupation, and Mobility scores) as well as with residing at home at the time of the 1-year injury anniversary.

CONCLUSION(S)

The intensity and type of SW/CM services are associated with outcomes at rehabilitation discharge and at 1-year post-injury. Discharge to home may require assistance from SW/CM in the area of discharge planning and financial planning, while discharge to non-home residence demands directed SW/CM services for such placement. Note: This is the eighth of nine articles of this SCIRehab series.

Relationship of occupational therapy inpatient rehabilitation interventions and patient characteristics to outcomes following spinal cord injury: the SCIRehab project

BACKGROUND/OBJECTIVE

Describe associations of occupational therapy (OT) interventions delivered during inpatient spinal cord injury (SCI) rehabilitation and patient characteristics with outcomes at the time of discharge and 1-year post-injury.

METHODS

Occupational therapists at six inpatient rehabilitation centers documented detailed information about treatment provided. Least squares regression modeling was used to predict outcomes at discharge and 1-year injury anniversary for a 75% subset; models were validated with the remaining 25%. Functional outcomes for injury subgroups (motor complete low tetraplegia and motor complete paraplegia) also were examined.

RESULTS

OT treatment variables explain a small amount of variation in Functional Independence Measure (FIM) outcomes for the full sample and significantly more in two functionally homogeneous subgroups. For patients with motor complete paraplegia, more time spent in clothing management and hygiene related to toileting was a strong predictor of higher scores on the lower body items of the self-care component of the discharge motor FIM. Among patients with motor complete low tetraplegia, higher scores for the FIM lower body self-care items were associated with more time spent on lower body dressing, manual wheelchair mobility training, and bathing training. Active patient participation during OT treatment sessions also was predictive of FIM and other outcomes.

CONCLUSION

OT treatments add to explained variance (in addition to patient characteristics) for multiple outcomes. The impact of OT treatment on functional outcomes is more evident when examining more homogeneous patient groupings and outcomes specific to the groupings. Note: This is the third of nine articles in the SCIRehab series.

Relationship of physical therapy inpatient rehabilitation interventions and patient characteristics to outcomes following spinal cord injury: the SCIRehab project

BACKGROUND/OBJECTIVE

Examine associations of type and quantity of physical therapy (PT) interventions delivered during inpatient spinal cord injury (SCI) rehabilitation and patient characteristics with outcomes at the time of discharge and at 1 year post-injury.

METHODS

Physical therapists delivering routine care documented details of PT interventions provided. Regression modeling was used to predict outcomes at discharge and 1 year post-injury for a 75% subset; models were validated with the remaining 25%. Injury subgroups also were examined: motor complete low tetraplegia, motor complete paraplegia, and American Spinal Injury Association (ASIA) Impairment Scale (AIS) D motor incomplete tetra-/paraplegia.

RESULTS

PT treatment variables explain more variation in three functionally homogeneous subgroups than in the total sample. Among patients with motor complete low tetraplegia, higher scores for the transfer component of the discharge motor Functional Independence Measure () are strongly associated with more time spent working on manual wheelchair skills. Being male is the most predictive variable for the motor FIM score at discharge for patients with motor complete paraplegia. Admission ASIA lower extremity motor score (LEMS) and change in LEMS were the factors most predictive for having the primary locomotion mode of "walk" or "both (walk and wheelchair)" on the discharge motor FIM for patients with AIS D injuries.

CONCLUSION

Injury classification influences type and quantity of PT interventions during inpatient SCI rehabilitation and is a strong predictor of outcomes at discharge and 1 year post-injury. The impact of PT treatment increases when patient groupings become more homogeneous and outcomes become specific to the groupings. Note: This is the second of nine articles in the SCIRehab series.

Risk-adjusted mortality: problems and possibilities

The ratio of observed-to-expected deaths is considered a measure of hospital quality and for this reason will soon become a basis for payment. However, there are drivers of that metric more potent than quality: most important are medical documentation and patient acuity. If hositals underdocument and therefore do not capture the full “expected mortality” they may be tempted to lower their observed/expected ratio by reducing “observed mortality” through limiting access to the very ill. Underdocumentation occurs because hospitals do not recognize, and therefore cannot seek to confirm, specific comorbidities conferring high mortality risk. To help hospitals identify these comorbidities, this paper describes an easily implemented spread-sheet for evaluating comorbid conditions associated, in any particular hospital, with each discharge. This method identifies comorbidities that increase in frequency as mortality risk increases within each diagnostic grouping. The method is inductive and therefore independent of any particular risk-adjustment technique.

A comparison of administrative and physiologic predictive models in determining risk adjusted mortality rates in critically ill patients

Background

Hospitals are increasingly compared based on clinical outcomes adjusted for severity of illness. Multiple methods exist to adjust for differences between patients. The challenge for consumers of this information, both the public and healthcare providers, is interpreting differences in risk adjustment models particularly when models differ in their use of administrative and physiologic data. We set to examine how administrative and physiologic models compare to each when applied to critically ill patients.

Methods

We prospectively abstracted variables for a physiologic and administrative model of mortality from two intensive care units in the United States. Predicted mortality was compared through the Pearsons Product coefficient and Bland-Altman analysis. A subgroup of patients admitted directly from the emergency department was analyzed to remove potential confounding changes in condition prior to ICU admission.

Results

We included 556 patients from two academic medical centers in this analysis. The administrative model and physiologic models predicted mortalities for the combined cohort were 15.3% (95% CI 13.7%, 16.8%) and 24.6% (95% CI 22.7%, 26.5%) (t-test p-value<0.001). The r² for these models was 0.297. The Bland-Atlman plot suggests that at low predicted mortality there was good agreement; however, as mortality increased the models diverged. Similar results were found when analyzing a subgroup of patients admitted directly from the emergency department. When comparing the two hospitals, there was a statistical difference when using the administrative model but not the physiologic model. Unexplained mortality, defined as those patients who died who had a predicted mortality less than 10%, was a rare event by either model.

Conclusions

In conclusion, while it has been shown that administrative models provide estimates of mortality that are similar to physiologic models in non-critically ill patients with pneumonia, our results suggest this finding can not be applied globally to patients admitted to intensive care units. As patients and providers increasingly use publicly reported information in making health care decisions and referrals, it is critical that the provided information be understood. Our results suggest that severity of illness may influence the mortality index in administrative models. We suggest that when interpreting “report cards” or metrics, health care providers determine how the risk adjustment was made and compares to other risk adjustment models.

Group physical therapy during inpatient rehabilitation for acute spinal cord injury: findings from the SCIRehab Study

BACKGROUND

Inpatient rehabilitation for spinal cord injury (SCI) includes the use of both individual and group physical therapy sessions. A greater understanding of group physical therapy use will help in the evaluation of the appropriateness of its use and contribute to the development of standards of practice.

OBJECTIVE

This report describes the extent to which group physical therapy is being used in inpatient rehabilitation for SCI, identifies group physical therapy interventions being delivered, and examines patterns in the types of activities being used for people with different levels and completeness of injury (ie, injury groups).

DESIGN

The SCIRehab Study is a 5-year, multicenter investigation that uses practice-based evidence research methodology.

METHODS

Data on characteristics of participants and treatments provided were collected through detailed chart review and customized research documentation completed by clinicians at the point of care. The analyses described here included data from 600 participants enrolled during the first year of the project.

RESULTS

Most of the participants (549/600) spent time in group physical therapy, and 23% of all documented physical therapy time was spent in group sessions. The most common group physical therapy activities were strengthening, manual wheelchair mobility, gait training, endurance activities, and range of motion/stretching. Time spent in group physical therapy and the nature of activities performed varied among the injury groups.

LIMITATIONS

Physical therapy use patterns observed in the 6 participating centers may not represent all facilities providing inpatient rehabilitation for SCI. Research documentation did not include all factors that may affect group physical therapy use, and some sessions were not documented.

CONCLUSIONS

The majority of physical therapy was provided in individual sessions, but group physical therapy contributed significantly to total physical therapy time. Group physical therapy time and activities differed among the injury groups in patterns consistent with clinical goals.

Physical therapy activities in stroke, knee arthroplasty, and traumatic brain injury rehabilitation: their variation, similarities, and association with functional outcomes

BACKGROUND

The mix of physical therapy services is thought to be different with different impairment groups. However, it is not clear how much variation there is across impairment groups. Furthermore, the extent to which the same physical therapy activities are associated with functional outcomes across different types of patients is unknown.

OBJECTIVE

The purposes of this study were: (1) to examine similarities and differences in the mix of physical therapy activities used in rehabilitation among patients from different impairment groups and (2) to examine whether the same physical therapy activities are associated with functional improvement across impairment groups.

DESIGN

This was a prospective observational cohort study.

METHODS

The study was conducted in inpatient rehabilitation facilities. The participants were 433 patients with stroke, 429 patients with total knee arthroplasty (TKA), and 207 patients with traumatic brain injury (TBI). Measures used in this study included: (1) the Comprehensive Severity Index to measure the severity of each patient's medical condition, (2) the Functional Independence Measure (FIM) to measure function, and (3) point-of-care instruments to measure time spent in specific physical therapy activities.

RESULTS

All 3 groups had similar admission motor FIM scores but varying cognitive FIM scores. Patients with TKA spent more time on exercise than the other 2 groups (average=31.7 versus 6.2 minutes per day). Patients with TKA received the most physical therapy (average=65.3 minutes per day), whereas the TBI group received the least physical therapy (average=38.3 minutes per day). Multivariate analysis showed that only 2 physical therapy activities (gait training and community mobility) were both positively associated with discharge motor FIM outcomes across all 3 groups. Three physical therapy activities (assessment time, bed mobility, and transfers) were negatively associated with discharge motor FIM outcome.

LIMITATIONS

The study focused primarily on physical therapy without concurrently considering other therapies such as occupational therapy, speech-language pathology, nursing care, and case management or the potential interaction of these inputs. This analysis did not consider the interventions that physical therapists used when patients participated in discrete physical therapy activities.

CONCLUSIONS

All 3 patient groups spent a considerable portion of their physical therapy time in gait training relative to other activities. Both gait training and community mobility are higher-level activities that were positively associated with outcomes, although all 3 groups spent little time in community mobility activities. Further research studies, such as randomized clinical trials and predictive validity studies, are needed to investigate whether higher-level or more-integrated therapy activities are associated with better patient outcomes.

Efficient rehabilitation care for joint replacement patients: skilled nursing facility or inpatient rehabilitation facility?

OBJECTIVE

There has been lengthy debate as to which setting, skilled nursing facility (SNF) or inpatient rehabilitation facility (IRF), is more efficient in treating joint replacement patients. This study aims to determine the efficiency of rehabilitation care provided by SNF and IRF to joint replacement patients with respect to both payment and length of stay (LOS).

METHODS

This study used a prospective multisite observational cohort design. Tobit models were used to examine the association between setting of care and efficiency. The study enrolled 948 knee replacement patients and 618 hip replacement patients from 11 IRFs and 7 SNFs between February 2006 and February 2007. Output was measured by motor functional independence measure (FIM) score at discharge. Efficiency was measured in 3 ways: payment efficiency, LOS efficiency, and stochastic frontier analysis efficiency.

RESULTS

IRF patients incurred higher expenditures per case but also achieved larger motor FIM gains in shorter LOS than did SNF patients. Setting of care was not a strong predictor of overall efficiency of rehabilitation care. Great variation in characteristics existed within IRFs or SNFs and severity groups. Medium-volume facilities among both SNFs and IRFs were most efficient. Early rehabilitation was consistently predictive of efficient treatment.

CONCLUSIONS

The advantage of either setting is not clear-cut. Definition of efficiency depends in part on preference between cost and time. SNFs are more payment efficient; IRFs are more LOS efficient. Variation within SNFs and IRFs blurred setting differences; a simple comparison between SNF and IRF may not be appropriate.

The SCIRehab project: treatment time spent in SCI rehabilitation. Speech-language pathology treatment time during inpatient spinal cord injury rehabilitation: the SCIRehab project

BACKGROUND/OBJECTIVE

Following spinal cord injury (SCI), speech-language pathologists (SLPs) perform assessments and provide treatment for swallowing, motor speech, voice, and cognitive-communication disorders that result from the SCI and/or co-occurring brain injuries. This paper describes the nature and distribution of speech-language pathology (SLP) activities delivered during inpatient SCI rehabilitation and discusses predictors (patient and injury characteristics) of the amount of time spent in specific SLP treatment activities.

METHODS

Six rehabilitation centers enrolled 600 patients with traumatic SCI for an observational study of acute inpatient rehabilitation treatment (SCIRehab). SLPs documented the details of assessment and treatment and time spent on each of a set of specific SLP activities during each patient encounter. Patterns of time use are described for all patients by neurological injury category. Ordinary least squares stepwise regression models are used to identify patient and injury characteristics predictive of treatment time in the specific SLP activities identified.

RESULTS

SLP consults were requested for 40% of SCIRehab patients. Fifty-seven percent of these patients received intense therapy (defined as more than five sessions during the rehabilitation stay); the remainder received primarily evaluation or less intense services (one to five sessions). The patients who participated in intense treatment received a mean total of 16.1 hours (range 2.5-105.2 hours, standard deviation (SD) 16.5, median 9.7 hours) of SLP; significant differences were seen in the amount of time spent in each activity among neurological injury groups. Cognitive-communication and swallowing therapy were the most common SLP activities. Patients with motor levels of injury at C1-C4 spent the highest percentage of their therapy time working on swallowing therapy while patients with low tetraplegia and paraplegia, and those classified as AIS D (regardless of motor level of injury) focused the greatest percentage of time on cognitive-communication work. Patient and injury characteristics explained a portion of the variation in time spent on cognitive-communication therapy but did not explain the variation in time spent on swallowing and other SLP treatment activities.

CONCLUSION

The need for swallowing and cognitive treatment by SLP is common during inpatient rehabilitation due to dysfunction resulting from use of artificial airways and feeding approaches, as well as secondary brain injuries. The large amount of variability seen in SLP treatment time, which is not explained well by patient and injury characteristics, sets the stage for future analyses to associate treatments with outcomes.

The SCIRehab project: treatment time spent in SCI rehabilitation. Physical therapy treatment time during inpatient spinal cord injury rehabilitation

BACKGROUND/OBJECTIVE

To describe the nature and distribution of activities during physical therapy (PT) delivered in inpatient spinal cord injury (SCI) rehabilitation and discuss predictors (patient and injury characteristics) of the amount of time spent in PT for specific treatment activities.

METHODS

Six hundred patients from six inpatient SCI centers were enrolled in the SCIRehab study. Physical therapists documented details, including time spent, of treatment provided during 37 306 PT sessions that occurred during inpatient SCI rehabilitation. Ordinary least squares regression models associated patient and injury characteristics with time spent in specific PT activities.

RESULTS

SCIRehab patients received a mean total of 55.3 hours of PT over the course of their rehabilitation stay. Significant differences among four neurologic groups were seen in the amount of time spent on most activities, including the most common PT activities of strengthening exercises, stretching, transfer training, wheelchair mobility training, and gait training. Most PT work (77%) was provided in individual therapy sessions; the remaining 23% was done in group settings. Patient and injury characteristics explained only some of the variations seen in time spent on wheelchair mobility, transfer and bed mobility training, and range of motion/ stretching.

CONCLUSION

Analysis yielded both expected and unexpected trends in SCI rehabilitation. Significant variation was seen in time spent on PT activities within and among injury groups. Providing therapeutic strengthening treatments consumed the greatest proportion of PT time. About one-quarter of all PT services were provided in group settings. Details about services provided, including time spent, will serve as a starting point in detailing the optimal treatment delivery for maximal outcomes.

The SCIRehab project: treatment time spent in SCI rehabilitation. Therapeutic recreation treatment time during inpatient rehabilitation

OBJECTIVE

Following spinal cord injury (SCI), certified therapeutic recreation specialists (CTRSs) work with patients during rehabilitation to re-create leisure lifestyles. Although there is much literature available to describe the benefits of recreation, little has been written about the process of inpatient or outpatient rehabilitation therapeutic recreation (TR) programs or the effectiveness of such programs. To delineate how TR time is used during inpatient rehabilitation for SCI.

METHODS

Six rehabilitation centers enrolled 600 patients with traumatic SCI for an observational study. CTRSs documented time spent on each of a set of specific TR activities during each patient encounter. Patterns of time use are described, for all patients and by neurologic category. Ordinary least-squares stepwise regression models are used to identify patient and injury characteristics predictive of total treatment time (overall and average per week) and time spent in TR activities.

RESULTS

Ninety-four percent of patients enrolled in the SCIRehab study participated in TR. Patients received a mean total of 17.5 hours of TR; significant differences were seen in the amount of time spent in each activity among and within neurologic groups. The majority (76%) of patients participated in at least one structured therapeutic outing. Patient and injury characteristics explained little of the variation in time spent within activities.

CONCLUSION

The large amount of variability seen in TR treatment time within and among injury group categories, which is not explained well by patient and injury characteristics, sets the stage for future analyses to associate treatments with outcomes.