Risk-adjusted Mortality Ratio Research Articles

Assessment of risk-adjusted mortality ratio (RAMR) in bloodstream infections using all-patient refined diagnosis-related groups (APR-DRGs).

To calculate a risk-adjusted mortality ratio (RAMR) for bloodstream infections (BSIs) using all-patient refined diagnosis-related groups (APR-DRGs) and compare it with the crude mortality rate (CMR). Retrospective observational study of prevalent BSI at our institution from January 2019 to December 2022. In-hospital mortality was adjusted with a binary logistic regression model adjusting for sex, age, admission type and mortality risk for the hospitalization episode according to the four severity levels of APR DRGs. The RAMR was calculated as the ratio of observed to expected in-hospital mortality, and the CMR was calculated as the proportion of deaths among all bacteraemia episodes. Of 2939 BSIs, 2541 were included: Escherichia coli (n = 1310), Klebsiella pneumoniae (n = 428), Pseudomonas aeruginosa (n = 209), Staphylococcus aureus (n = 498) and candidaemia (n = 96). A total of 436 (17.2%) patients died during hospitalization and 279 died within the first 14 days after the onset of BSI. Throughout the period, all BSI cases had a mortality rate above the expected adjusted mortality (RAMR value greater than 1), except for Escherichia coli (1.03; 95% CI 0.86-1.21). The highest overall RAMR values were observed for P. aeruginosa, Candida and S. aureus with 2.06 (95% CI 1.57-2.62), 1.99 (95% CI 1.3-2.81) and 1.8 (95% CI 1.47-2.16), respectively. The temporal evolution of CMR may differ from RAMR, especially in E. coli, where it was reversed. RAMR showed higher than expected mortality for all BSIs studied except E. coli and provides complementary to and more clinically comprehensive information than CMR, the currently recommended antibiotic stewardship programme mortality indicator.

¿Qué hay del efecto fin de semana? Impacto del día de ingreso sobre la mortalidad intrahospitalaria, la duración de la estancia y el coste de la hospitalización

IntroductionThere is no agreement on the existence of the weekend effect in healthcare or, if it exists, on its possible causes. The objective of the study was to evaluate the differences in healthcare outcomes between patients admitted on weekdays or weekends in a high-complexity hospital. MethodsObservational and retrospective study of patients admitted between 2016 and 2019 in a public hospital with more than 1300 beds. Hospitalization episodes were classified according to whether admission took place between Friday at 3:00 p.m. and the following Monday at 8:00 a.m. (weekend admission) or not (admission on weekdays). Mortality, length of stay and associated costs were compared, applying their respective risk-adjustment models. ResultsOf the total 169,495 hospitalization episodes analyzed, 48,201 (28.44%) corresponded to the weekend, presenting an older age (54.9 years vs. 53.9; P<.001), a higher crude mortality rate (5.22% vs. 4.59%; P<0.001), and a longer average length of stay (7.42 days vs. 6.74; P<.001), than those admitted on weekdays. The median crude cost of stay was lower (€731.25 vs. €850.88; P<0.001). No significant differences were found when applying the adjustment models, with a risk-adjusted mortality ratio of 1.03 (0.99–1.08) vs. 0.98 (0.95–1.01), risk-adjusted length of stay of 1.002 (0.98–1.005) vs. 0.999 (0.997–1.002) and risk-adjusted cost of stay of 0.928 (0.865–0.994) vs. 0.901 (0.843–0.962). ConclusionThe results of the study reveal that the assistance provided during the weekends does not imply worse health outcomes or increased costs. Comparing the impact between hospitals will require a future homogenization of temporal criteria and risk adjustment models.

Patient-level and hospital-level variation and related time trends in COVID-19 case fatality rates during the first pandemic wave in England: multilevel modelling analysis of routine data

BackgroundA report suggesting large between-hospital variations in mortality after admission for COVID-19 in England attracted much media attention but used crude rates. We aimed to quantify these variations between hospitals...

Inpatient hospital performance is associated with post-discharge sepsis mortality

BackgroundPost-discharge deaths are common in patients hospitalized for sepsis, but the drivers of post-discharge deaths are unclear. The objective of this study was to test the hypothesis that hospitals with high risk-adjusted inpatient sepsis mortality also have high post-discharge mortality, readmissions, and discharge to nursing homes.MethodsRetrospective cohort study of age-qualifying Medicare beneficiaries with sepsis hospitalization between January 2013 and December 2014. Hospital survivors were followed for 180-days post-discharge, and mortality, readmissions, and new admission to skilled nursing facility were measured. Inpatient hospital-specific sepsis risk-adjusted mortality ratio (observed: expected) was the primary exposure.ResultsA total of 830,721 patients in the cohort were hospitalized for sepsis, with inpatient mortality of 20% and 90-day mortality of 48%. Higher hospital-specific sepsis risk-adjusted mortality was associated with increased 90-day post-discharge mortality (aOR 1.03 per each 0.1 increase in hospital inpatient O:E ratio, 95% CI 1.03–1.04). Higher inpatient risk adjusted mortality was also associated with increased probability of being discharged to a nursing facility (aOR 1.03, 95% CI 1.02–1.03) and 90-day readmissions (aOR 1.03, 95% CI 1.02–1.03).ConclusionsHospitals with the highest risk-adjusted sepsis inpatient mortality also have higher post-discharge mortality and increased readmissions, suggesting that post-discharge complications are a modifiable risk that may be affected during inpatient care. Future work will seek to elucidate inpatient and healthcare practices that can reduce sepsis post-discharge complications.

Mortality related to drug-resistant organisms in surgical sepsis-3: an 8-year time trend study using sequential organ failure assessment scores

The difference in sequential organ failure assessment (SOFA) scores from the baseline to sepsis is a known predictor of sepsis-3 outcome, but the prognostic value of drug-resistant organisms for mortality is unexplained. We employed sepsis stewardship and herein report an observational study. Study subjects were patients admitted to the Departments of Surgery/Chest Surgery from 2011 through 2018 with a diagnosis of sepsis and a SOFA score of 2 or more. Our sepsis stewardship methods included antimicrobial and diagnostic stewardship and infection control. We determined the primary endpoint as in-hospital death and the secondary endpoint as the annual trend of the risk-adjusted mortality ratio (RAMR). For mortality, we performed logistic regression analysis based on SOFA score, age, sex, comorbid disease, and the presence of methicillin-resistant Staphylococcus aureus (MRSA) and extended-spectrum beta-lactamase inhibitor–producing bacteria. In a total of 457 patients, two factors were significant predictors for fatality, i.e., SOFA score of 9 or more with an odds ratio (OR) 4.921 and 95% confidence interval [95% CI] 1.968–12.302 (P = 0.001) and presence of MRSA with an OR 1.83 and 95% CI 1.003–3.338 (P = 0.049). RAMR showed a decrease during the study years (P < 0.05). Early detection of MRSA may help patients survive surgical sepsis-3. Thus, MRSA-oriented diagnosis may play a role in expediting treatment with anti-MRSA antimicrobials.

CAUSES, mortality rates and risk factors of death in community-dwelling Europeans aged 50 years and over: Results from the Survey of Health, Ageing and Retirement in Europe 2013–2015

ObjectiveTo determine mortality rates and to rank the causes and predictors of mortality using a wide range of sociodemographic and clinical variables. Materials and methodsIt is a prospective population-based cohort study of adults living in the community, 2013–15 N = 48,691, age ≥50; deceased = 1,944. Clinical and sociodemographic data were obtained from the Survey of Health, Ageing and Retirement in Europe SHARE: Age, Gender, Marital Status, Years of Schooling, Income, Loneliness, Cognition, Self-Rated Health, Diseases, Activities of daily living ADL, and Frailty. Mortality rates were calculated. A Cox proportional hazards model were used to determine risk-adjusted mortality ratios. ResultsThe crude mortality rate was 18.39 (1000 person-years at risk), (99 % CI, 18.37–18.42). The factors most associated with an increased mortality risk were older age, lower self-rated health, lower cognition, male gender, ADL deficits, higher comorbidity, frailty and loneliness. The diseases with a higher mortality risk were: cancer (Hazard ratio, HR = 2.67), dementia (HR = 2.19), depressive symptoms (HR = 2.10), fractures (hip, femur) (HR = 1.57), stroke (HR = 1.55), chronic lung disease (HR = 1.52), diabetes (HR = 1.36) and heart attack (HR = 1.21). ConclusionsThe main mortality risk factors, associated independently in the eight diseases were: older age, poor self-rated health, ADL deficits, male gender, lower cognition, comorbidity and the presence of depressive symptoms, with a different influence in the European regions. The need to evaluate and treat the depressive symptoms that accompanies diseases with higher risk of mortality is stressed.

Epidemiology and prognoses in a medical intermediate care unit.

The purpose of medical intermediate care units is the observation and treatment of patients with incipient or manifest organ failure. We wished to obtain data on which conditions result in admission to these units and the prognosis for these patients. All patients admitted to the medical intermediate care unit at Akershus University Hospital in 2014 were registered prospectively with reason for admission, period of hospitalisation, degree of severity, comorbidity, last place of hospitalisation prior to medical intermediate care and treatment limitations (do-not-resuscitate order and/or do-not-intubate order). Mortality in the hospital and one year after hospitalisation were registered retrospectively. Multiple regression analysis was performed with hospital mortality as the outcome variable. Altogether 1369patient hospitalisations for 1118 unique patients were included. The most frequent reasons for admission were pneumonia, chronic obstructive pulmonary disease, sepsis, poisonings and hyponatraemia. The degree of severity of the condition for which patients were admitted corresponded to that reported by intensive care departments in Norwegian local hospitals. A total of 13% died during their stay in hospital and a further 14% in the course of one year. The highest mortality was for patients with severe infection, cardiac failure and restrictive/neuromuscular respiratory disorder. The degree of severity, age, infection, comorbidity and ward as admitting unit were predictors of mortality during the hospitalisation period. Risk-adjusted mortality ratio of 0.64 satisfied the quality objective for intensive care departments (<0.7). A total of 5.6% of hospitalisations in the medical intermediate care unit entailed transfer to the intensive care ward. The degree of severity of the condition for which patients were admitted was high, and the treatment outcomes judged upon expected mortality were good. Medical intermediate care units can relieve pressure on wards with seriously ill patients without taking up intensive care beds.

When high-volume PCI operators in high-volume hospitals move to lower volume hospitals-Do they still maintain high volume and quality of outcomes?

The aim of this quasi-experimental study was to examine whether high-volume percutaneous coronary intervention (PCI) operators still maintain high volume and quality of outcomes when they moved to lower volume hospitals. Systematic reviews have indicated that high-volume PCI operators and hospitals have higher quality outcomes. However, little is known on whether high PCI volume and high quality outcomes are mainly due to operator characteristics (i.e., skill and experience) and is portable across organizations or whether it is due to hospital characteristics (i.e., equipment, team, and management system) and is less portable. We used Taiwan National Health Insurance claims data 2000-2012 to identify 98 high-volume PCI operators, 10 of whom moved from one hospital to another during the study period. We compared the PCI volume, risk-adjusted mortality ratio, and major adverse cardiovascular event (MACE) ratio before and after moving. Of the 10 high-volume operators who moved, 6 moved from high- to moderate- or low-volume hospitals, with median annual PCI volumes (interquartile range) of 130 (117-165) in prior hospitals and 54 (46-84) in subsequent hospitals (the hospital the operator moved to), and the remaining 4 moved from high to high-volume hospitals, with median annual PCI volumes (interquartile range) of 151 (133-162) in prior hospitals and 193 (178-239) in subsequent hospitals. No significant differences were observed in the risk-adjusted mortality ratios and MACE ratios between high-volume operators and matched controls before and after moving. High-volume operators cannot maintain high volume when they moved from high to moderate or low-volume hospitals; however, the quality of care is maintained. High PCI volume and high-quality outcomes are less portable and more hospital bound.

Outcomes of trauma care at centers treating a higher proportion of older patients: the case for geriatric trauma centers.

The burden of injury among older patients continues to grow and accounts for a disproportionate number of trauma deaths. We wished to determine if older trauma patients have better outcomes at centers that manage a higher proportion of older trauma patients. The National Trauma Data Bank years 2007 to 2011 was used. All high-volume Level 1 and Level 2 trauma centers were included. Trauma centers were categorized by the proportion of older patients seen. Adult trauma patients were categorized as older (≥65 years) and younger adults (16-64 years). Coarsened exact matching was used to determine differences in mortality and length of stay between older and younger adults. Risk-adjusted mortality ratios by proportion of older trauma patients seen were analyzed using multivariate logistic regression models and observed-expected ratios. A total of 1.9 million patients from 295 centers were included. Older patients accounted for one fourth of trauma visits. Matched analysis revealed that older trauma patients were 4.2 times (95% confidence interval, 3.99-4.50) more likely to die than younger patients. Older patients were 34% less likely to die if they presented at centers treating a high versus low proportion of older trauma (odds ratio, 0.66; 95% confidence interval, 0.54-0.81). These differences were independent of trauma center performance. Geriatric trauma patients treated at centers that manage a higher proportion of older patients have improved outcomes. This evidence supports the potential advantage of treating older trauma patients at centers specializing in geriatric trauma. Prognostic and epidemiologic study, level III.

Reducing Mortality and Infections After Congenital Heart Surgery in the Developing World

There is little information about congenital heart surgery outcomes in developing countries. The International Quality Improvement Collaborative for Congenital Heart Surgery in Developing World Countries uses a registry and quality improvement strategies with nongovernmental organization reinforcement to reduce mortality. Registry data were used to evaluate impact. Twenty-eight sites in 17 developing world countries submitted congenital heart surgery data to a registry, received annual benchmarking reports, and created quality improvement teams. Webinars targeted 3 key drivers: safe perioperative practice, infection reduction, and team-based practice. Registry data were audited annually; only verified data were included in analyses. Risk-adjusted standardized mortality ratios (SMRs) and standardized infection ratios among participating sites were calculated. Twenty-seven sites had verified data in at least 1 year, and 1 site withdrew. Among 15,049 cases of pediatric congenital heart surgery, unadjusted mortality was 6.3% and any major infection was 7.0%. SMRs for the overall International Quality Improvement Collaborative for Congenital Heart Surgery in Developing World Countries were 0.71 (95% confidence interval [CI] 0.62-0.81) in 2011 and 0.76 (95% CI 0.69-0.83) in 2012, compared with 2010 baseline. SMRs among 7 sites participating in all 3 years were 0.85 (95% CI 0.71-1.00) in 2011 and 0.80 (95% CI 0.66-0.96) in 2012; among 14 sites participating in 2011 and 2012, the SMR was 0.80 (95% CI 0.70-0.91) in 2012. Standardized infection ratios were similarly reduced. Congenital heart surgery risk-adjusted mortality and infections were reduced in developing world programs participating in the collaborative quality improvement project and registry. Similar strategies might allow rapid reduction in global health care disparities.

Validation of the revised injury severity classification score in patients with moderate-to-severe traumatic brain injury

IntroductionBy analysing risk-adjusted mortality ratios, weaknesses in the process of care might be identified. Traumatic brain injury (TBI) is the main cause of death in trauma, and thus it is crucial that trauma prediction models are valid for TBI patients. Accordingly, we assessed the validity of the RISC score in TBI patients by internal and external validation analyses. MethodsPatients with moderate-to-severe TBI admitted to the TraumaRegister DGU® (TR-DGU) and the trauma registry of Helsinki University Hospital (TR-THEL) in 2006–2011 were included in this retrospective open cohort study. Definition of moderate-to-severe TBI was head abbreviated injury scale of 3 or higher. Subgroup analysis for patients with isolated and polytrauma TBI was performed. The performance of the RISC score was evaluated by assessing its discrimination (area under the curve, AUC) and calibration (Hosmer–Lemeshow [H–L] test). ResultsAmong the 9106 and 809 patients with moderate-to-severe TBI admitted to TR-DGU and TR-THEL, unadjusted mortality was 26% and 23%, respectively. Internal and external validation of the RISC score showed good discrimination (TR-DGU AUC 0.89, 95% confidence interval [CI] 0.88–0.90 and TR-THEL AUC 0.84, 95% CI 0.81–0.87), but poor calibration (p<0.001) in patients with moderate-to-severe TBI. Subgroup analysis found the discrimination only to be modest in isolated TBI (AUC 0.76) and calibration to be particularly poor in polytrauma TBI (TR-DGU H–L=4356, p<0.001; TR-THEL H–L 112, p<0.001). ConclusionThe RISC score was found to be of limited predictive value in patients with moderate-to-severe TBI. A new general trauma scoring system that includes TBI specific prognostic factors is warranted.

Bariatric surgery in the elderly: 2009–2013

BackgroundAmple evidence supports the safety and effectiveness of bariatric surgery in the general adult population but more information is needed in patients age 60 years and older (elderly). We previously examined the outcome of bariatric surgery performed in the elderly between 1999 and 2005 using the University HealthSystem Consortium (UHC) Clinical Database. The aim of this study was to analyze contemporary outcomes of bariatric surgery in the elderly and to compare them to previous data from 1999–2005. MethodsUsing International Classification of Diseases, 9th Revision diagnosis and procedure codes, we obtained data from the UHC database for all elderly (age >60 yr) and adult nonelderly (age 19–60 yr) patients who underwent bariatric surgery for the treatment of morbid obesity between 2009 and 2013. Outcome measures, such as patient characteristics, LOS, morbidity, and observed-to-expected (risk-adjusted) mortality ratio were compared between elderly and nonelderly patients. ResultsBariatric surgery in the elderly made up 2.7% of all bariatric operations in 1999–2005. This represents an increase to 10.1% of all bariatric operations in 2009–2013. In-hospital mortality was .30% for the nonelderly and .70% for the elderly in 1999–2005, whereas contemporary in-hospital mortality has decreased to .11% for the nonelderly and .05% for the elderly. ConclusionOur results show that the number of bariatric procedures performed in the elderly is increasing and now represents 10% of all bariatric operations performed at academic centers. In-hospital mortality in bariatric surgery in the elderly has improved so much that it is now even better than in-hospital mortality in the nonelderly in 1999–2005.

EComment. Risk-adjusted mortality ratio for EuroSCORE II in everyday clinical practice

I have read with great interest the article by Nishida et al. [1]. The article includes up-to-date information on EuroSCORE II, and the results shown by the authors validate the superiority of the EuroSCORE II over the additive and logistic EuroSCOREs for predicting risk mortality in thoracic aortic surgery. EuroSCORE II is highly recommended for assessing risk in adult cardiac surgery. However, one must not forget that no risk model is perfect, nor do all risk factors appear in all models. EuroSCORE II now offer us the capacity to virtually create a reality. However, this issue must be addressed much more consistently. The variation in outcomes between centres and surgeons should be taken into account [2]. Once a given risk model is in use, all units and surgeons should calculate the risk-adjusted mortality ratio (RAMR) by dividing the actual mortality (observed) by the expected mortality. Then, the predicted mortality calculated by EuroSCORE II is multiplied by the unit's or the individual surgeon's RAMR. This reflects in a more faithful manner the current mortality for a given patient in a given surgical unit. If, for example, the unit's RAMR for thoracic aortic surgery is 2, and the predicted mortality by EuroSCORE II is 7.4%, the corrected mortality rate for thoracic aortic surgery for this hypothetic unit would be 14.8%. In consequence, this is a more effective way to adapt the EuroSCORE II to the ‘real world’. Conflict of Interest: none declared

Performance Analysis of EuroSCORE II Compared to the Original Logistic EuroSCORE and STS Scores for Predicting 30-Day Mortality After Transcatheter Aortic Valve Replacement

The original European System for Cardiac Operative Risk Evaluation (EuroSCORE) has been recently updated as EuroSCORE II to optimize its efficacy in cardiac surgery, but its performance has been poorly evaluated for predicting 30-day mortality in patients who undergo transcatheter aortic valve replacement (TAVR). Consecutive patients (n = 250) treated with TAVR were included in this analysis. Transapical access was used in 60 patients, while 190 procedures were performed using a transfemoral approach. Calibration (risk-adjusted mortality ratio) and discrimination (C-statistic and U-statistic) were calculated for the logistic EuroSCORE, EuroSCORE II, and Society of Thoracic Surgeons (STS) scores for predicting 30-day mortality. Observed mortality was 7.6% in the overall population (6.3% and 11.7% for the transfemoral and transapical cohorts, respectively). Predicted mortality was 22.6 ± 12.8% by logistic EuroSCORE, 7.7 ± 5.8% by EuroSCORE II, and 7.3 ± 4.1% by STS score. The risk-adjusted mortality ratio was 0.34 (95% confidence interval [CI] 0.10 to 0.58) for logistic EuroSCORE, 0.99 (95% CI 0.29 to 1.69) for EuroSCORE II, and 1.05 (95% CI 0.30 to 1.79) for STS score. Moderate discrimination was observed with EuroSCORE II (C-index 0.66, 95% CI 0.52 to 0.79, p = 0.02) compared to the logistic EuroSCORE (C-index 0.63, 95% CI 0.51 to 0.76, p = 0.06) and STS (C-index 0.58, 95% CI 0.43 to 0.73, p = 0.23) score, without a significant difference among the 3 risk scores. Discrimination was slightly better in the transfemoral cohort compared to the transapical cohort with the 3 risk scores. In conclusion, EuroSCORE II and the STS score are better calibrated than the logistic EuroSCORE but have moderate discrimination for predicting 30-day mortality after TAVR.

Editorial Comment: EuroSCORE II and the art and science of risk modelling

Chalmers et al. [4] applied the model to a 5500-patient cohort, concluded that EuroSCORE II is globally better calibrated and found better overall discrimination with a C-statistic of 0.79 (old model 0.77), with best performance in mitral (0.87) and coronary (0.79) surgery and weakest in isolated aortic valve replacement (0.69), marginally better than the old model (0.67). Pooling contemporaneous multi-institutional data provides optimal model validation. A single institution performing exceptionally in one type of surgery may perceive a lack of fit in the entire cohort, as Chalmers indeed found (Hosmer–Lemeshow P-value <0.05) for EuroSCORE II overall, but not in any subdivision. Considering this and other limitations of a single-institution study, the model has achieved its objective of better calibration and discrimination in global cardiac surgery. EuroSCORE II riskstratifies using factors including operation type. Lower discrimination when these are neutralized is therefore unsurprising. We advise caution in applying the model to narrow patient subsets. In the editorials [2, 3], many comments reiterate issues already addressed in the discussion section of the original paper [1]. As Sergeant states, EuroSCORE is widely used and has surgeons’ confidence. There has been misuse, particularly in evaluating patients for transcatheter aortic valve implantation (TAVI), and in predicting non-mortality events. The former is avoidable by our recommendation for using risk-adjusted mortality ratios (RAMR) [1], but that is not enough. Conventional surgery risk factors differ from those of TAVI, and we shall explore our TAVI data to illuminate this. Predicting non-mortality events using EuroSCORE has often succeeded, but its purpose remains to evaluate expeditiously the risk of death. SELECTING AND HANDLING RISK VARIABLES

Calibration of the EuroSCORE II risk stratification model: is the Hosmer-Lemeshow test acceptable any more?

We read with great interest the article by Nashef et al .[ 1] regarding the construction and validation of the EuroSCORE II risk model. This manuscript is an outstanding contribution in the creation of a risk stratification model that is robust enough for use in cardiac surgery worldwide. However, we would like to know which statistical procedure should be used to determine calibration of the EuroSCORE II? The Hosmer–Lemeshow goodness-of-fit test has been the most popular test to validate calibration [2, 3], measuring the differences between expected and observed outcomes (mortality) over deciles (test results are acceptable with cohort divided in at least terciles) of risk. A well-calibrated model gives a corresponding P value >0.05. Recently, it has been claimed that a nonsignificant Hosmer–Lemeshow test meant that there was no evidence of bad calibration, but that this result did not mean that there was good calibration [4]. In our opinion, statistical results are either significant or nonsignificant, black or white; there are no grey results in statistics. However, if that is true, how has it come to pass that we needed more than 10 years (during which Hosmer–Lemeshow statistics was used in more than 95% of manuscripts to test the calibration of additive and logistic EuroSCORE in cardiac surgery) to find out that Hosmer– Lemeshow test is no longer valid to determine calibration, and that it should be replaced with the risk-adjusted mortality ratio, [RAMR = observed/predicted (expected) – O/E mortality], as it is now suggested by Nashef et al [1]. An O/E ratio of 1.0 means that the score predicts mortality perfectly. An O/E ratio > 1.0 means that the model underpredicts mortality [in EuroSCORE II, for a validation data set of 5553 patients, the O/E ratio was 1.058 (4.18%/3.95%)], while an O/E ratio < 1.0 means that the model overpredicts mortality. However, how are we going to check the statistical significance of the RAMR value? Conceptually, if the observed number of deaths is equal to the expected number of deaths (as predicted by the scoring system), the RAMR would have a value of 1.0. Thus, the statistical test for the significance of the RAMR is whether it is different from 1.0. To gauge the statistical significance of the RAMR, we must first calculate the 95% confidence interval for the RAMR. If the 95% confidence interval excludes the value ‘1.0’, it may be considered statistically significant (no matter whether it overpredicts or underpredicts mortality). On the contrary, Bhatti et al .[ 5] suggested χ 2 statistics

Reply to Nezic et al.

We are grateful to Nezic et al. [1] for their interest in, and comments about our work [2], but we do not fully agree with their statements regarding significance. Statistics are rarely absolute and merely quantify the agreement between observed data and a hypothesized model. The type I error, also referred to as the ‘P-value’ or ‘significance level’, represents the probability of observing the current data (or a more extreme dataset) if the proposed model is ‘true’. If the P-value is small, then the data occur only rarely under the proposed model. If it is large, then the data are entirely consistent with the proposed model. Individual researchers may describe a particular P-value as significant and, traditionally, a value of 0.05 (or sometimes 0.01) has been used as a threshold, but the true nature of the P-value is continuous. For example, there is almost no difference between P-values of 0.04 and 0.06, but according to Nezic, the former would represent a poor model and the latter a good model. There is a range of statistics that represent, in one way or another, some summary of the agreement between the observed data and a proposed model. Each method has its advantages and disadvantages. For example, the Hosmer– Lemeshow statistic can specifically show whether the observed and expected number of deaths agree when split into deciles, and an examination of the table shows where any disagreement occurs, such as for high-risk or low-risk patients. However, the groupings are essentially arbitrary, and some groups may contain small numbers of deaths resulting in unstable estimates. Another method is the use of the risk-adjusted mortality ratio, which provides an overall estimate of the average fit without reference to risk subgroups, but may be more robust since it is based on the total number of deaths. There are many other model assessment techniques based on either how well the observed data and the proposed model agree, or on how well the proposed model predicts future patient outcomes. Each gives a different, but often related, test for model fit. Statisticians bear the responsibility for ensuring that models are properly assessed and that, in this difficult area, findings are presented in a clear way to clinical researchers, in order to prevent the misunderstandings described above. On the final point in the letter, we agree fully that it would have been useful to show the Hosmer–Lemeshow statistic for the validation set in addition to the overall dataset. In fact, this and many other findings could also have been included in the manuscript, but we were constrained by limitations of space in which to report these data and analyses of this very large project. For the record, the Hosmer–Lemeshow statistic for the validation set is 14.92 with a P-value of 0.09316, a more than satisfactory result by Nezic’s own rigorous standards.

Abstract 1: Hospital PCI Appropriateness and In-Hospital Procedural Outcomes: Insights from the NCDR ®

Background. Measurement of PCI quality has previously focused on post-procedure outcomes and processes of care. PCI appropriateness provides an assessment of quality as it relates to patient selection and anticipated procedural benefit relative to potential harm. The relationship, if any, between case selection and in-hospital procedural outcomes is unknown. Methods. We identified 203,531 consecutive patients from 779 hospitals undergoing non-acute (elective) PCI within the CathPCI Registry ® between July, 2009 and April, 2011. We examined the association between a hospital’s proportion of inappropriate PCI in non-acute settings (as assessed by the Appropriate Use Criteria (AUC) for Coronary Revascularization) and (1) risk-adjusted in-hospital mortality; (2) risk-adjusted bleeding complications; and (3) use of optimal guideline directed discharge medications (i.e. aspirin, thienopyridine, and statin on discharge). Results. Of the 24,532 patients who received PCI for non-acute indications, 453 (0.2%) died in-hospital and 3699 (1.8%) suffered bleeding events. The hospital proportion of inappropriate PCI in non-acute settings ranged from 0.0% to 58.6%. Categorized as hospital tertiles, the median proportion of inappropriate PCI was 5.3% in hospitals of the lowest-tertile, 10.9% in the middle-tertile, and 20.0% in the highest-tertile. The hospital proportion of inappropriate PCI was not associated with mortality for PCI in non-acute settings (Risk-adjusted mortality ratio; 0.7 vs. 0.7 vs. 0.8 from lowest to highest-tertile, P=0.75 for trend), risk-adjusted bleeding complications (Risk-adjusted bleeding ratio; 0.6 vs. 0.6 vs. 0.6, P=0.81), or use of optimal PCI discharge medications (85.3% vs. 85.7% vs. 85.2%, P=0.58). Conclusions. In a national cohort of non-acute PCIs from 779 hospitals, hospital proportion of inappropriate PCI indications was not associated with in-hospital mortality, bleeding, or process of care measures of PCI quality. These findings suggest PCI appropriateness and post-procedural outcomes measure different aspects of hospital PCI quality.

Predicting prognosis in cardiac surgery: a prophecy?

Physicians need to make predictions on the prognosis of a treatment that helps them in the choice of therapy. Medicine used to be much more subjective than in the current evidence-based era. Shared decision-making, where physicians and patients both participate in deciding on choices for therapy, is also more common [1]. Clinical prediction models like EuroSCORE may provide the evidence-based input for shared decision-making by providing an estimate of the operative risk of patients undergoing cardiac surgery. An ideal clinical model would be something simple dividing the patients into ‘good’ and ‘bad’ without further specification of the survival chances. The original EuroSCORE was a compromise between the ‘statistical ideal’ and the ‘clinical ideal’. It was developed from a large multinational European population and was a model predicting mortality based on 17 variables, either from a logistic regression equation or from an additive model. Numerous institutions throughout the world have tested and validated EuroSCORE. Loss of calibration with the additive and logistic EuroSCORE has been observed by many investigators and an update of the EuroSCORE was warranted. One possible reason for the poor calibration of the original EuroSCORE score is that the score was developed from patients undergoing surgery almost 20 years ago. As surgical and perioperative care evolves and the impact of clinical variables change, prediction models therefore require revision. These factors may also vary between institutions and it is well known that the quality of care and comorbidities of patients differs between countries. The original EuroSCORE already identified major differences in the risk profile of national samples [2]. This is therefore one of the major concerns with EuroSCORE II: 154 hospitals from 43 countries participated, of which many were outside Europe [3]. One may, therefore, question whether the term EuroSCORE is still valid or another name should be used that reflects the fact that so many countries outside Europe participated. With this in mind, it becomes even more important that, as indicated by the authors, units and surgeons calculate their own risk-adjusted mortality ratio. The model is probably more reliable in the prediction of death over a wide range of risk groups rather than the prediction of the vital status of an individual patient. Another reason for the poor calibration in the original EuroSCORE might be that a large number of risk factors in the model are highly correlated. It is important to recognize correlation between predicting variables, as the additional risk contribution of certain variables can in some part be explained by the effect of other variables. Some predicting variables may also be more important for some types of operations then for others. The large number of risk factors with potential interaction may overestimate risk in certain categories of patients (e.g. intermediate risk or extreme risk). It is therefore a pity that the authors have not explored possible interaction terms in the new

Is there an association between the quality of hospitals’ research and their quality of care?

It is often claimed that hospitals that are leaders in biomedical research provide higher health care quality, or vice versa. Although several studies have shown a relationship between teaching status and quality of care, none has analysed the association between research output and hospital outcomes. Our aim was to determine whether there is a relationship between bibliometric measures of research output in acute hospitals and hospital mortality for two common cardiac conditions. A cross-sectional analysis of secondary data of in-hospital risk-adjusted mortality for congestive heart failure and acute myocardial infarction (2002-2004) and several bibliometric measures of publications (1996-2004) in cardiovascular disease. The setting was 50 acute Spanish public hospitals, voluntarily participating in an external quality initiative, with more than 30 medical cases of congestive heart failure and acute myocardial infarction per year, and more than five citable papers in the field of heart disease. Spearman's rho non-parametric correlation coefficient was used to assess association. There was a low-to-moderate negative correlation between the risk-adjusted mortality ratio and the weighted citations ratio: -0.43 (95% CI -0.17 to -0.63) for congestive heart failure and -0.37 (-0.10 to -0.59) for acute myocardial infarction. Teaching status and the technological level of the hospital had a stronger correlation with hospital mortality. Measures of research output could be considered for incorporation into comparisons of the quality of hospitals. A weighted citations ratio is the most suitable measure of research output, but more research is needed on the interplay between research and practice as complementary ways of developing medical knowledge.