Effect Of Case Mix Research Articles

Automated surveillance of antimicrobial consumption in intensive care, northern Sweden: an observational case study

BackgroundThe digitalization of information systems allows automatic measurement of antimicrobial consumption (AMC), helping address antibiotic resistance from inappropriate drug use without compromising patient safety.ObjectivesDescribe and characterize a new automated AMC surveillance service for intensive care units (ICUs), with data stratified by referral clinic and linked with individual patient risk factors, disease severity, and mortality.MethodsAn automated service collecting data from the electronic medical record was developed, implemented, and validated in a healthcare region in northern Sweden. We performed an observational study from January 1, 2018, to December 31, 2021, encompassing general ICU care for all ≥18-years-olds in a catchment population of 270000 in secondary care and 900000 in tertiary care. We used descriptive analyses to associate ICU population characteristics with AMC outcomes over time, including days of therapy (DOT), length of therapy, defined daily doses, and mortality.ResultsThere were 5608 admissions among 5190 patients with a median age of 65 (IQR 48-75) years, 41.2% females. The 30-day mortality was 18.3%. Total AMC was 1177 DOTs in secondary and 1261 DOTs per 1000 patient days and tertiary care. AMC varied significantly among referral clinics, with the highest total among 810 general surgery admissions in tertiary care at 1486 DOTs per 1000 patient days. Case-mix effects on the AMC were apparent during COVID-19 waves highlighting the need to account for case-mix. Patients exposed to more than three antimicrobial drug classes (N = 242) had a 30-day mortality rate of 40.6%, with significant variability in their expected rates based on admission scores.ConclusionWe introduce a new service and instructions for automating local ICU-AMC data collection. The versatile long-term ICU-AMC metrics presented, covering patient factors, referral clinics and mortality outcomes, are expected to be beneficial in refining antimicrobial drug use.

Validity of Early Outcomes as Indicators for Comparing Hospitals on Quality of Stroke Care.

Background Insight into outcome variation between hospitals could help to improve quality of care. We aimed to assess the validity of early outcomes as quality indicators for acute ischemic stroke care for patients treated with endovascular therapy (EVT). Methods and Results We used data from the MR CLEAN (Multicenter Randomized Controlled Trial of Endovascular Treatment for Acute Ischemic Stroke in the Netherlands)Registry, a large multicenter prospective cohort study including 3279 patients with acute ischemic stroke undergoing EVT. Random effect linear and proportional odds regression were used to analyze the effect of case mix on between-hospital differences in 2 early outcomes: the National Institutes of Health Stroke Scale (NIHSS) score at 24 to 48 hours and the expanded thrombolysis in cerebral infarction score. Between-hospital variation in outcomes was assessed using the variance of random hospital effects (tau2). In addition, we estimated the correlation between hospitals' EVT-patient volume and (case-mix-adjusted) outcomes. Both early outcomes and case-mix characteristics varied significantly across hospitals. Between-hospital variation in the expanded thrombolysis in cerebral infarction score was not influenced by case-mix adjustment (tau 2=0.17 in both models). In contrast, for the NIHSS score at 24 to 48 hours, case-mix adjustment led to a decrease in variation between hospitals (tau 2 decreases from 0.19 to 0.17). Hospitals' EVT-patient volume was strongly correlated with higher expanded thrombolysis in cerebral infarction scores (r=0.48) and weakly with lower NIHSS score at 24 to 48 hours (r=0.15). Conclusions Between-hospital variation in NIHSS score at 24 to 48 hours is significantly influenced by case-mix but not by patient volume. In contrast, between-hospital variation in expanded thrombolysis in cerebral infarction score is strongly influenced by EVT-patient volume but not by case-mix. Both outcomes may be suitable for comparing hospitals on quality of care, provided that adequate adjustment for case-mix is applied for NIHSS score.

Variation in Emergency Department Use of Racemic Epinephrine and Associated Outcomes for Croup.

Racemic epinephrine (RE) is commonly administered for croup in the emergency department (ED). Our objectives were to examine variation in RE use between EDs, to determine whether ED variation in RE use is associated with hospital or patient factors, and to evaluate the associations between the rates of hospital-specific ED RE use and patient outcomes. We performed a retrospective cohort study using the Pediatric Heath Information System of children aged 3 months to 10 years with croup in the ED. We used mixed-effects regression to calculate risk-standardized proportions of patients receiving RE in each ED and to analyze the relationship between risk-standardized institutional RE use and individual patient odds of hospital admission, ICU admission, and ED revisits. We analyzed 231 683 patient visits from 39 hospitals. ED administration of RE varied from 14% to 48% of visits (median, 24.5%; interquartile range, 20.0%-27.8%). A total of 8.6% of patients were hospitalized and 1% were admitted to the ICU. After standardizing for case mix and site effects, increasing ED use of RE per site was associated with increasing patient odds of hospital admission (odds ratio [OR], 1.39-95%; confidence interval [CI], 1.01-1.91), but not ICU admission (OR, 1.39; 95% CI, 0.99-1.97) or ED revisit (OR, 1.00; 95% CI, 0.92-1.09). In this large, observational study, RE administration varied widely across EDs. Increased RE use by site was associated with increased odds of hospital admission for individual patients when controlling for patient factors. These results suggest further standardization of RE use in children with croup is warranted.

Case-Mix Moderation of the Relationship Between OR Performance Metrics and Utilization.

We investigated the impact of the case-mix ratio of inpatients to outpatients on the relationships between OR utilization and late starts, turnover time, delays, cancellations, and idle time at an academic medical center in the southeastern United States. After extracting 55 months of data from the surgical repository, we used simple and multiple linear regression models to analyze the data and determine the strength and direction of the relationships among the variables. We compared models comprising proportionally more inpatients to models comprising proportionally more outpatients for each metric to ascertain the effects of case mix on OR utilization. Idle time had the greatest effect on OR utilization, followed by late starts and turnover time. Case mix moderated the relationship between OR utilization and the metrics of cancellations and turnover time. Perioperative leaders may enhance OR utilization by monitoring and addressing idle time and late starts and scheduling an appropriate mix of inpatients and outpatients.

P-P35 Textbook outcomes after pancreaticoduodenectomy in high risk patients: results from a high volume UK centre

Abstract Background Textbook Outcome (TO) after pancreaticoduodenectomy (PD) is a quality metric that may be used to compare outcomes between centres, but the effect of casemix on TO is unknown. The aim of this study was to determine if TO after PD is affected by casemix. Methods TO was evaluated in a prospectively maintained database of 830 consecutive patients who underwent PD between 2009-2019 in a high volume centre. TO was defined as an absence of POPF, bile leak, haemorrhage, Clavien III+ complications, readmission and hospital mortality. Frequency of TO was compared between high and low risk cases. High risk was defined as any of the following: age ≥ 75 years, significant comorbidity (Charlson index ≥5), vascular resection or additional procedures. Multivariable analysis using binary logistic regression analysis was performed to assess factors associated with TO. Results Overall, 599/830 patients (72%) had TO after PD. There has been no change during the study period (2009-2013 v 2014-2018: 70% v 75%; p = 0.148). There was no difference in TO in elderly patients (p = 0.774), severe comorbidity (p = 0.483), vascular resection (p = 0.187) or additional procedures (p = 0.189). On multivariable analysis, cardiac disease (OR 0.47, 95%CI 0.28-0.81; p = 0.006), pancreatic adenocarcinoma (OR 1.55 95%CI 1.02-2.35; p = 0.039) and hard gland (OR 3.12, 95%CI 2.06-4.736; p < 0.001) were independently associated with TO. Conclusions Acceptable Textbook Outcomes can be achieved in high risk patients and those undergoing complex surgery, when performed in high volume specialist centres with appropriate patient selection.

Model-Based ROC Curve: Examining the Effect of Case Mix and Model Calibration on the ROC Plot.

BackgroundThe performance of risk prediction models is often characterized in terms of discrimination and calibration. The receiver-operating characteristic (ROC) curve is widely used for evaluating model discrimination. However, when comparing ROC curves across different samples, the effect of case mix makes the interpretation of discrepancies difficult. Further, compared with model discrimination, evaluating model calibration has not received the same level of attention. Current methods for examining model calibration require specification of smoothing or grouping factors.MethodsWe introduce the “model-based” ROC curve (mROC) to assess model calibration and the effect of case mix during external validation. The mROC curve is the ROC curve that should be observed if the prediction model is calibrated in the external population. We show that calibration-in-the-large and the equivalence of mROC and ROC curves are together sufficient conditions for the model to be calibrated. Based on this, we propose a novel statistical test for calibration that, unlike current methods, does not require any subjective specification of smoothing or grouping factors.ResultsThrough a stylized example, we demonstrate how mROC separates the effect of case mix and model miscalibration when externally validating a risk prediction model. We present the results of simulation studies that confirm the properties of the new calibration test. A case study on predicting the risk of acute exacerbations of chronic obstructive pulmonary disease puts the developments in a practical context. R code for the implementation of this method is provided.ConclusionmROC can easily be constructed and used to interpret the effect of case mix and calibration on the ROC plot. Given the popularity of ROC curves among applied investigators, this framework can further promote assessment of model calibration.HighlightsCompared with examining model discrimination, examining model calibration has not received the same level of attention among investigators who develop or examine risk prediction models.This article introduces the model-based ROC (mROC) curve as the basis for graphical and statistical examination of model calibration on the ROC plot.This article introduces a formal statistical test based on mROC for examining model calibration that does not require arbitrary smoothing or grouping factors.Investigators who develop or validate risk prediction models can now also use the popular ROC plot for examining model calibration, as a critical but often neglected component in predictive analytics.

Hospital-level variation in follow-up strategies after percutaneous coronary intervention, revealed in health claims data of Korea

This study sought to determine hospital variation in the use of follow-up stress testing (FUST) and invasive coronary angiography (FUCAG) after percutaneous coronary intervention (PCI). The claims records of 150,580 Korean patients who received PCI in 128 hospitals between 2008 and 2015 were analyzed. Patient were considered to have undergone FUST and FUCAG, when these testings were performed within two years after discharge from the index hospitalization. Hierarchical generalized linear and frailty models were used to evaluate binary and time-to-event outcomes. Hospital-level risk-standardized FUCAG and FUST rates were highly variable across the hospitals (median, 0.41; interquartile range [IQR], 0.27–0.59; median, 0.22; IQR, 0.08–0.39, respectively). The performances of various models predicting the likelihood of FUCAG and FUST were compared, and the best performance was observed with the models adjusted for patient case mix and individual hospital effects as random effects (receiver operating characteristic curves, 0.72 for FUCAG; 0.82 for FUST). The intraclass correlation coefficients of the models (0.41 and 0.68, respectively) indicated that a considerable proportion of the observed variation was related to individual institutional effects. Higher hospital-level FUCAG and FUST rates were not preventive of death or myocardial infarction. Increased repeat revascularizations were observed in hospitals with higher FUCAG rates.

A Novel Method for Assessing the Effect of Case-Mix and Model Calibration on the Receiver Operating Characteristic (ROC) Plot, With a Case Study in Chronic Obstructive Pulmonary Disease

A Novel Method for Assessing the Effect of Case-Mix and Model Calibration on the Receiver Operating Characteristic (ROC) Plot, With a Case Study in Chronic Obstructive Pulmonary Disease

Effect of Case-Mix and Random Variation on Breast Cancer Care Quality Indicators and Their Rankability.

Hospital comparisons to improve quality of care require valid and reliable quality indicators. We aimed to test the validity and reliability of 6 breast cancer indicators by quantifying the influence of case-mix and random variation. The nationwide population-based database included 79 690 patients with breast cancer from 91 Dutch hospitals between 2011 and 2016. The indicator-scores calculated were: (1) irradical breast-conserving surgery (BCS) for invasive disease, (2) irradical BCS for ductal carcinoma-in-situ, (3) breast contour-preserving treatment, (4) magnetic resonance imaging (MRI) before neo-adjuvant chemotherapy, (5) radiotherapy for locally advanced disease, and (6) surgery within 5 weeks from diagnosis. Case-mix and random variation adjustments were performed by multivariable fixed and random effect logistic regression models. Rankability quantified the between-hospital variation, representing unexplained differences that might be the result of the level of quality of care, as low (<50%), moderate (50%-75%), or high (>75%). All of the indicators showed between-hospital variation with wide (interquartile) ranges. Case-mix adjustment reduced variation in indicators 1 and 3 to 5. Random variation adjustment (further) reduced the variation for all indicators. Case-mix and random variation adjustments influenced the indicator-scores of individual hospitals and their ranking. Rankability was poor for indicator 1, 2, and 5, and moderate for 3, 4, and6. The 6 indicators lacked validity and/or reliability to a certain extent. Although measuring quality indicators may stimulate quality improvement in general, comparisons and judgments of individual hospital performance should be made with caution if based on indicators that have not been tested or adjusted for validity and reliability, especially in benchmarking.

Which patients improve most after total hip arthroplasty? Influence of patient characteristics on patient-reported outcome measures of 22,357 total hip arthroplasties in the Dutch Arthroplasty Register.

Patient-reported outcome measures (PROMs) after total hip arthroplasty (THA), can be influenced by patient characteristics (case-mix factors). We used the Dutch Arthroplasty Register (LROI) to determine the effect of case-mix on improvement of PROMs after primary THA. We included all primary THAs (n = 22,357) performed in the Netherlands between 2014 and 2018. The Hip disability and Osteoarthritis Outcome Score Physical function short form (HOOS-PS), Oxford Hip Score (OHS), EQ-5D index score and thermometer, and Numeric Rating Scales (NRS) measuring pain during activities and at rest, were recorded. The difference between preoperative and 3- and 12-month postoperative scores was calculated (delta-PROM) and used as primary outcome variable. Multivariable linear regression was used to examine the association between patient characteristics (age, sex, ASA score, body mass index (BMI), Charnley class, smoking, and previous operations to the affected hip) and PROMs. Cohens' d was used to measure effect size. Postoperative improvement (delta-PROM) on HOOS-PS, OHS, EQ-5D, and pain relief were significantly higher in patients <60 years, in patients with female gender, a high ASA score (III-IV), a BMI >30 kg/m2, and patients without a previous operation to the hip. Cohen's d indicated clinically small differences (0.2). Patients benefiting most in terms of postoperative improvement of self-reported physical functioning, pain relief and quality of life after primary THA were young, female, with a high ASA or BMI score, and without previous operations to the hip.

The Effect of Case-Mix on Hemolysis Rates Following the Christchurch Earthquake

Understanding the effects of different clinical contexts on key quality indicators will help laboratories account for unforeseen errors when circumstances change. Our aim was to assess hemolysis in relation to changing case-mix before and after a major earthquake.

Patient and Hospital Factors Associated With Unexpected Newborn Complications Among Term Neonates in US Hospitals

IMPORTANCEUnexpected complications in term newborns have been recently adopted by the Joint Commission as a marker of obstetric care quality.OBJECTIVETo understand the variation and patient and hospital factors associated with severe unexpected complications in term neonates among hospitals in the United States.DESIGN, SETTING, AND PARTICIPANTSThis cross-sectional study collected data from all births in US counties with 1 obstetric hospital using county-identified birth certificate data and American Hospital Association annual survey data from January 1, 2015, through December 31, 2017. All live-born, term, singleton infants weighing at least 2500 g were included. The data analysis was performed from December 1, 2018, through June 30, 2019.EXPOSURESSevere unexpected newborn complication, defined as neonatal death, 5-minute Apgar score of 3 or less, seizure, use of assisted ventilation for at least 6 hours, or transfer to another facility.MAIN OUTCOMES AND MEASURESBetween-hospital variation and patient and hospital factors associated with unexpected newborn complications.RESULTSA total of 1 754 852 births from 576 hospitals were included in the analysis. A wide range of hospital complication rates was found (range, 0.6–89.9 per 1000 births; median, 15.3 per 1000 births [interquartile range, 9.6–22.0 per 1000 births]). Hospitals with high newborn complication rates were more likely to care for younger, white, less educated, and publicly insured women with more medical comorbidities compared with hospitals with low complication rates. In the adjusted models, there was little effect of case mix to explain the observed between-county variation (11.3%; 95% CI, 10.0%−12.6%). Neonatal transfer was the primary factor associated with complication rates, especially among hospitals with the highest rates (66.0% of all complications). The risk for unexpected neonatal complication increased by more than 50% for those neonates born at hospitals without a neonatal intensive care unit compared with those with a neonatal intensive care unit (adjusted odds ratio, 1.55; 95% CI, 1.38–1.75).CONCLUSIONS AND RELEVANCEIn this study, severe unexpected complication rates among term newborns varied widely. When included in the metric numerator, neonatal transfer was the primary factor associated with complications, especially among hospitals with the highest rates. Transfers were more likely to be necessary when infants were born in hospitals with lower levels of neonatal care. Thus, if this metric is to be used in its current form, it would appear that accreditors, regulatory bodies, and payers should consider adjusting for or stratifying by a hospital’s level of neonatal care to avoid disincentivizing against appropriate transfers.

Patient Characteristics Influence Revision Rate of Total Hip Arthroplasty: American Society of Anesthesiologists Score and Body Mass Index Were the Strongest Predictors for Short-Term Revision After Primary Total Hip Arthroplasty

BackgroundOutcome and survival after primary total hip arthroplasty (THA) can be affected by patient characteristics. We examined the effect of case-mix on revision after primary THA using the Dutch Arthroplasty Register. MethodsOur cohort included all primary THAs (n = 218,214) performed in patients with osteoarthritis in the Netherlands between 2007 and 2018. Multivariable logistic regression analysis was used to calculate the difference in survivorship in patients with different patient characteristics (age, gender, American Society of Anesthesiologists [ASA] score, body mass index [BMI], Charnley score, smoking, and previous operations to the hip). ResultsCase-mix factors associated with an increased risk for revision 1 year after THA were the following: a high ASA score (II and III-IV) (odds ratio [OR] 1.5, 95% confidence interval [CI] 1.1-2.0 and OR 3.0, 95% CI 1.7-5.3), a higher BMI (30-40 and >40) (OR 1.4, 95% CI 1.2-1.5 and OR 2.0, 95% CI 1.4-1.7), age ≥75 years (OR 1.5, 95% CI 1.1-2.0), and male gender (OR 1.3, 95% CI 1.2-1.4). A similar model for 3-year revision showed comparable results. High BMI (OR 1.9, 95% CI 1.3-2.9), a previous hip operation (OR 1.8, 95% CI 1.3-2.5), ASA III-IV (OR 1.2, 95% CI 1-1.6), and Charnley score C (OR 1.5, 95% CI 1.1-2.2) were associated with increased risk for revision. Main reasons for revision in obese and ASA II-IV patients were infection, dislocation, and periprosthetic fracture. Patients with femoral neck fracture and late post-traumatic pathology were more likely to be revised within 3 years, compared to osteoarthritis patients (OR 1.5, 95% CI 1.3-1.7 and OR 1.5, 95% CI 1.2-1.7). ConclusionThe short-term risk for revision after primary THA is influenced by case-mix factors. ASA score and BMI (especially >40) were the strongest predictors for 1-year revision after primary THA. After 3 years, BMI and previous hip surgery were independent risk factors for revision. This will help surgeons to identify and counsel high-risk patients and take appropriate preventive measures.

Smooth Bayesian network model for the prediction of future high-cost patients with COPD

IntroductionThe clinical course of chronic obstructive pulmonary disease (COPD) is marked by acute exacerbation events that increase hospitalization rates and healthcare spending. The early identification of future high-cost patients with COPD may decrease healthcare spending by informing individualized interventions that prevent exacerbation events and decelerate disease progression. Existing studies of cost prediction of other chronic diseases have applied regression and machine-learning methods that cannot capture the complex causal relationships between COPD factors. Thus, the exploration of these factors through nonlinear, high-dimensional but explainable modeling is greatly needed. ObjectivesWe aimed to develop a machine-learning model to identify future high-cost patients with COPD. Such a model should incorporate expert knowledge about causal relationships, and the method for estimating the model could provide more accurate predictions than other machine learning methods. MethodsWe used the 2011–2013 medical insurance data of patients with COPD in a large city. The data set included demographic information and admission records. Leveraging on developments in graphical modeling methods, we proposed a smooth Bayesian network (SBN) model for the prediction of high-cost individuals using medical insurance data. The modeling method incorporated some expert knowledge about causal relationships (i.e., about the Bayesian network structure). We employed a smoothing kernel based on the weighted nearest neighborhood method in the SBN model to address overfitting, case-mix effect, and data sparsity (i.e., using data about “similar patients”). ResultsThe proposed SBN achieved the area under curve (AUC) of 0.80 and showed considerable improvement over the baseline machine-learning methods. Besides confirming the known factors from the literature, we found “region” (i.e., a suburban or urban area) to be a significant factor, and that in a 3-tier system with primary, secondary and tertiary hospitals, COPD patients who had been admitted to primary hospitals were more likely to develop into future high-cost patients than patients who had been admitted to tertiary hospitals. ConclusionThe proposed SBN model not only obtained higher prediction accuracy and stronger generalizability than a number of benchmark machine-learning methods, but also used the Bayesian network to capture the complex causal relationships between different predictors by incorporating expert knowledge. Furthermore, a framework was developed to establish the relationships between exposure to historical trajectory and future outcome, which can also be applied to other temporal data to model different trajectory information and predict other outcomes.

The Effect of Transplant Volume and Patient Case Mix on Center Variation in Kidney Transplantation Outcomes.

Background:Kidney transplantation is the optimal treatment for patients with end-stage renal disease; however, long-term outcomes remain suboptimal.Objective:The objectives of our study were to examine the variation in survival rates and determine whether center volume and case mix are associated with transplant outcomes and explain the variation across kidney transplant centers in Ontario, Canada.Design:This was a population-based cohort study using health care administrative databases.Setting:A total of 5 transplant centers across Ontario, Canada.Patients:We included adults (≥18 years) undergoing primary, solitary kidney transplantation between January 1, 2000 to December 31, 2013.Measurements:The co-primary outcomes were death-censored graft loss and total mortality.Methods:Multivariable Cox proportional hazards regression was used to assess potential associations and describe variation, using hazard ratios (HRs) with 95% confidence intervals (CIs) for each center relative to the average across all centers.Results:The study cohort included 5037 patients followed for a median of 5.3 years, interquartile range (2.7-8.6). In multivariable models, recipient age, body mass index, Charlson Index, time on dialysis, donor type, and age were found to be significantly associated with death-censored graft loss, and recipient age and sex, Charlson Index, time on dialysis, donor age, and time era of transplant were associated with total mortality. There was statistically significant variation across centers observed for death-censored graft loss (P = .04) with HRs ranging from 0.72 to 1.22. However, neither adjusting for case mix nor center volume meaningfully changed the HRs reflecting each center-specific effect. There was a tendency toward reduced risk of graft loss (HR, per additional 25 patients, 0.90 [95% CI, 0.78-1.04]) in centers with higher volumes. For total mortality, there was statistically significant variation across centers with HRs ranging from 0.82 to 1.13 (P = .04); however, neither adjusting for case mix or center volume meaningfully changed the HRs. Center volume was not significantly associated with total mortality (HR, per additional 25 patients, 1.04 [95% CI, 0.90-1.20]).Limitations:This study was limited by the small number of centers included.Conclusions:Outcomes differ across the 5 transplant centers in Ontario. We did not find any strong support for our hypotheses that case mix or center volume is responsible for these differences.

Variation in geographical treatment intensity affects survival of non-small cell lung cancer patients in England

ObjectivesWe aimed to determine the geographical variation in the proportion of non-small cell lung cancer (NSCLC) patients undergoing curative treatment and assess the relationship between treatment access rates and survival outcomes. MethodsWe extracted cancer registration data on 144,357 lung cancer (excluding small cell tumours) patients diagnosed between 2009 and 2013. Surgical and radiotherapy treatment intensity quintiles were based on patients’ Clinical Commissioning Group (CCG) of residence. We used logistic regression to assess the effect of travel time and case-mix on treatment use and Cox regression to analyse survival in relation to treatment intensity. ResultsThere was wide variation in the use of curative treatment across CCGs, with the proportion undergoing surgery ranging from 8.9% to 20.2%, and 0.4% to 16.4% for radical radiotherapy. The odds of undergoing surgery decreased with socioeconomic deprivation (OR 0.91, 95% CI 0.85-0.97), whereas the opposite was observed for radiotherapy (OR 1.16, 95% CI 1.08–1.25). There was an overall effect of travel time to thoracic surgery centre on the odds of undergoing surgery (OR 0.81, 95% CI 0.76-0.87 for travel time >55 min vs ≤15 min) which was amplified by the effect of deprivation. No clear association was observed for radiotherapy. Higher mortality rates were observed for the lower resection and radiotherapy quintiles (HR 1.08, 95% CI 1.04–1.12 and HR 1.06, 95% CI 1.02–1.10 for lowest vs. highest resection and radiotherapy quintile). ConclusionThere was wide geographical variation in the use of curative treatment and a higher frequency of treatment was associated with better survival.

President's Pen: What is Impact?

President's Pen: What is Impact?

538 (PB-171) - Quantifying the effect of case-mix and random effect on quality indicators for breast cancer care and their rankability

538 (PB-171) - Quantifying the effect of case-mix and random effect on quality indicators for breast cancer care and their rankability

928: Case mix and the utility of postpartum readmission rates as a marker of quality in obstetrics

Postpartum readmission rates have been discussed as a potential quality metric in obstetrics. The objective of this project was to determine the relative effects of case mix on a hospital's postpartum readmission rate. This retrospective cohort study was conducted using the State Inpatient Databases from CA, FL, and NY between 2004 and 2013. Women who delivered at hospitals with >100 deliveries/year were included. We compared patient and hospital characteristics among hospitals with low and high readmission rates using chi-squared tests. Risk-adjusted 30-day readmission rates were calculated for patient, delivery, and hospital characteristics to understand how case mix and hospital factors affect a patient’s risk for readmission using fixed and random effects models. The average 30-day readmission rates were 1.15%, 1.57%, and 1.42% in California, Florida, and New York from the 3,164,429, 1,890,802, and 2,257,360 deliveries included from each state, respectively. There were notable differences in the distributions of patients cared for at high readmission hospitals compared to low readmission hospitals. Patients were more likely to identify as white, have private insurance and higher incomes, and have fewer comorbidities at low readmission hospitals in all states. These characteristics were associated with a lower risk of readmission, as shown by the adjusted rates in Table 1. The patient characteristics with the highest risk-adjusted readmission rates varied by state, but included hypertension (range 2.14-3.04%), obesity (1.78-2.94%), preterm labor/delivery (2.50-2.60%), and seizure disorder (1.78-3.35%). Delivery complications were also associated with increased risk-adjusted readmission rates (Table 2), but overall occurred rarely. In comparison, hospital characteristics did not have a profound impact on readmission risk compared to the patient characteristics (Table 2). Postpartum readmissions are an overall uncommon event that are affected by unmodifiable factors, such as demographics, socioeconomic status, and underlying comorbidities. The use of unadjusted readmission rates as a quality metric in obstetrics would disproportionately affect hospitals caring for higher risk, sicker patients.View Large Image Figure ViewerDownload Hi-res image Download (PPT)

Association of an Enhanced Recovery Pilot With Length of Stay in the National Surgical Quality Improvement Program

Enhanced recovery protocols (ERPs) are standardized care plans of best practices that can decrease morbidity and length of stay (LOS). However, many hospitals need help with implementation. The Enhanced Recovery in National Surgical Quality Improvement Program (ERIN) pilot was designed to support ERP implementation. To evaluate the association of the ERIN pilot with LOS after colectomy. Using a difference-in-differences design, pilot LOS before and after ERP implementation was compared with matched controls in a hierarchical model, adjusting for case mix and random effects of hospitals and matched pairs. The setting was 15 hospitals of varied size and academic status from the National Surgical Quality Improvement Program. Preimplementation and postimplementation colectomy cases (July 1, 2013, to December 31, 2015) were collected using novel ERIN variables. Emergency and septic cases were excluded. A propensity score match identified a 2:1 control cohort of patients undergoing colectomy at non-ERIN hospitals. Pilot hospitals developed and implemented ERPs that included expert guidance, multidisciplinary teams, data audits, and opportunities for collaboration. The primary outcome was LOS, and the secondary outcome was serious morbidity or mortality composite. There were 4975 colectomies performed by 15 ERIN pilot hospitals (3437 before implementation and 1538 after implementation) compared with a control cohort of 9950 colectomies (4726 before implementation and 5224 after implementation). The mean LOS decreased by 1.7 days in the pilot (6.9 [interquartile range (IQR), 4-8] days before implementation vs 5.2 [IQR, 3-6] days after implementation, P < .001) compared with 0.4 day in controls (6.4 [IQR, 4-7] days before implementation vs 6.0 [IQR, 3-7] days after implementation, P < .001). Readmission did not differ pre-post for the pilot or controls. Serious morbidity or mortality decreased for pilot participants (485 [14.1%] before implementation vs 162 [10.5%] after implementation, P < .001), with no difference in controls, and remained significant after risk adjustment (adjusted odds ratio, 0.76; 95% CI, 0.60-0.96). After adjusting for differences in case mix and for clustering in hospitals and matched pairs, the adjusted difference-in-differences model demonstrated a decrease in LOS by 1.1 days in the pilot over controls (P < .001). Participating ERIN pilot hospitals achieved shorter LOS and decreased complications after elective colectomy, without increasing readmissions. The ability to implement ERPs across hospitals of varied size and resources is essential. Lessons from the ERIN pilot may inform efforts to scale this effective and evidence-based intervention.