Surgical Outcomes: A synopsis & commentary on the Cardiac Care Quality Indicators Report. May 2018

Transcription

1 Surgical Outcomes: A synopsis & commentary on the Cardiac Care Quality Indicators Report May 2018 Prepared by the Canadian Cardiovascular Society (CCS)/Canadian Society of Cardiac Surgeons (CSCS) Cardiac Surgery Quality Working Group

2 Surgical Outcomes: A synopsis & commentary on the Cardiac Care Quality Indicators Report, May 2018 Background In 2013 the Cardiac Surgery Working Group of the Canadian Cardiovascular Society (CCS) Quality Project defined quality indicators in cardiac surgery for pan-canadian reporting. Mortality indicators were defined as 30-day mortality after the three most common cardiac surgical procedures, coronary artery bypass (CABG), aortic valve replacement (AVR), and combined AVR and CABG, as well as 30-day readmission after CABG. Although CIHI had been reporting pan-canadian cardiac outcomes, substantial consultation and partnership with CIHI occurred, with extensive cohort and methodologic refinement. In 2016 a formal memorandum of understanding between CCS and CIHI was agreed to, with the purpose of facilitating and reporting of quality indicators publicly. The Cardiac Care Quality Indicators (CCQI) public report and CCS commentary were released in October 2017 based on a three year cohort of pooled data for fiscal years to The updated indicator results are based on 3 years of most recent pooled data to the end of fiscal year The purpose of public release of this data is for assessment of quality of cardiac care, for education and enlightenment, to promote transparency in the system of care, and to foster and stimulate quality improvement. Although many provinces and hospitals have high quality clinical databases, very few have systematic regular reviews for the purposes of quality, and until this process, national outcome reporting permitting comparisons have not been possible. Questions regarding administrative data methods and quality have arisen which will lead to improved chart documentation, data consistency and quality. Ironically, with rare exception, the high quality data in clinical

3 cardiac surgical databases is not available for hospital chart abstraction. This is an attainable objective. Data exchange across provincial health care jurisdictions is also challenging, and although CIHI is able to access national in-hospital mortality, facilitated access to vital statistics data for longer term mortality is essential, and should be expedited. The indicators reported reflect the highest volume cardiac surgical procedures, which should be associated with consistent excellent results, and which lend themselves to reproducible systems and processes of care. The first public report has increased the awareness of the outcomes reported in the cardiac community and has increased engagement of institutions and provinces. The risk factor analyses provide important information for clinicians and patients in estimating individual procedural risk. Methodology and Cohort Selection Model specifications (coefficients, odds ratios and p-values) for the indicators were developed using the logistic regression model based on three years of pooled data ( to ). This can be regarded as the model-derivation cohort. The reported (test) cohort includes the most recent three fiscal year data to : , , In-hospital mortality is reported and for hospital stays less than 30 days will underestimate true 30-day mortality. The table of contents contained in these updated indicator results provide important links to this May 2018 public release and to detailed technologic methodology.

4 Select Sample from Table 1 Cardiac Care quality indicator crude and risk-adjusted results, by reporting level (Canada, province, cardiac care centre), to (3 years of pooled data) Reporting level Organization Indicator name Crude rate Predicted rate Riskadjusted rate 95% confidence interval riskadjusted rate: LCL 95% confidence interval riskadjusted rate: UCL Statistically significant Canada Province Canada 1.4 n/a n/a n/a n/a n/a Newfoundland and Labrador 2.7 n/a Yes Province Nova Scotia 2.1 n/a Yes Province New Brunswick 1.3 n/a No Province Quebec 1.7 n/a No Province Ontario 1.4 n/a No Province Manitoba 0.9 n/a Yes Province Saskatchewan 30-Day In- 0.9 n/a No Hospital Province Province Alberta British Columbia Mortality After Isolated CABG n/a n/a No Yes PCI: Percutaneous coronary intervention. AVR: Aortic valve replacement. n/a: Not applicable. Mortality indicators are based on in-hospital mortality only. Isolated means that no other cardiac surgeries, valve procedures or core concomitant procedures were performed during the episode of care. There is no comprehensive capture of PCI data in Quebec, so Quebec could not be included in analyses for 30-day in-hospital mortality after PCI, 30-day readmission after PCI and PCI volume by centre. Model specifications (coefficients, odds ratios and p-values) were developed using the logistic regression model based on 3 years of pooled data ( to ). Discharge Abstract Database, Hospital Morbidity Database and National Ambulatory Care Reporting System, , and , Canadian Institute for Health Information. This table provides the numeric crude and risk-adjusted outcomes for all reported data by province and hospital for the first 11 months of each fiscal year, and the sum total of the fiscal years, totaling 33 months of data to the end of February The 11 month fiscal cohorts allowed 30-day hospital readmission rates to fiscal year end to be reported for patients discharged within the first 11 months of the fiscal year. The Table contains the numeric values for the data displayed in Figures 1-8. The arrows at the top of each column allow filtering of the displayed data by province, hospital, indicator, and year and permit ordering of the rows as well.

5 Canadian average rate (per 100) Coronary Artery Bypass (CABG) The 33 month cohort having this isolated procedure comprised 43,782 patients from 32 cardiac surgical centres nationally. Figure 1 Canadian average mortality rates by indicator and fiscal year, to Day In-Hospital Mortality After PCI 30-Day In-Hospital Mortality After 30-Day In-Hospital Mortality After Isolated CABG Isolated AVR Mortality indicators 30-Day In-Hospital Mortality After CABG and AVR * Mortality indicators are based on in-hospital mortality only. Isolated means that no other cardiac surgeries, valve procedures or core concomitant procedures were performed during the hospitalization episode of care. PCI: Percutaneous coronary intervention. AVR: Aortic valve replacement. Discharge Abstract Database, National Ambulatory Care Reporting System and Hospital Morbidity Database, , and , Canadian Institute for Health Information. Average annual in-hospital mortality after isolated CABG is low at 1.4% and has been constant year to year.

6 Figure 4 Cardiac care centre risk-adjusted results for 30-Day In-Hospital Mortality After Isolated CABG, to (3 years of pooled date) Canadian average mortality rates by indicator and fiscal year, to * Risk-adjusted rate is statistically significantly different from the Canadian average. 3 years of pooled data: to Model specifications (coefficients, odds ratios and p-values) were developed using the logistic regression model based on 3 years of pooled data ( to ). Discharge Abstract Database, National Ambulatory Care Reporting System and Hospital Morbidity Database, , and , Canadian Institute for Health Information. Figure 4 shows average in-hospital risk-adjusted mortality over the 33 month time frame with 95% confidence limits. Risk-adjusted mortality ranged from 0.4% to 3.5%. Hospital mortality rates with 95% confidence limits not overlapping the national mean of 1.4% or each other are significantly different from the comparator. The ability of the model used for risk-adjustment to discriminate mortality was 0.81 (C-statistic 0.81). Risk factors used in the model and the percentage of the cohort with each factor appear in Table 4:

7 Table 4 Risk factors for 30-Day In-Hospital Mortality After CABG, to (3 years of pooled data) Risk factor Number Percentage Coefficient Odds ratio Significance Intercept n/a n/a -5.6 n/a < Age (vs ) 2, Age (vs ) 13, < Age 80+ (vs ) 3, < Male (vs. female) 35, < Urgent/emergent admission 22, Shock < NSTEMI AMI/unspecified AMI (vs. stable 13, CAD) STEMI AMI (vs. stable CAD) 3, < Unstable angina (vs. stable CAD) 6, Previous AMI 3, Previous cardiac surgery Cardiac dysrhythmias (non-quebec 2, Cardiac dysrhythmias (Quebec 1, Multiple cardiac interventions in same 1, < episode of care (PCI, CABG) Peripheral vascular disease (non-quebec < Peripheral vascular disease (Quebec 1, < Acute renal failure (non-quebec Acute renal failure (Quebec Charlson group 1 (vs. 0) 17, < Charlson group 2 (vs. 0) 3, < PCI: Percutaneous coronary intervention. CAD: Coronary artery disease. AMI: Acute myocardial infarction. STEMI: ST elevation myocardial infarction. NSTEMI: Non ST elevation myocardial infarction. n/a: Not applicable. Model specifications (coefficients, odds ratios and p-values) were developed using the logistic regression model based on 3 years of pooled data ( to ). Discharge Abstract Database, Hospital Morbidity Database and National Ambulatory Care Reporting System, , and , Canadian Institute for Health Information.

8 Canadian average rate (per 100) Age greater than 70, female gender, urgent hospital admission, shock, acute myocardial infarction with or without ST-segment elevation, percutaneous coronary intervention within the same hospital episode of care, peripheral vascular disease, and a Charlson Comorbidity Index greater than 0 were all associated with increased mortality with the probability indicated by the odds ratio. The Charlson Comorbidity Index incorporates a number of preoperative comorbidities related to mortality such as congestive heart failure, dementia, chronic pulmonary disease, rheumatologic disease, liver disease, diabetes with organ failure, hemiplegia or paraplegia, renal disease, and HIV infection. Aortic Valve Replacement (AVR) The 33 month cohort of patients analyzed, was 7,523. Average national in-hospital mortality after isolated AVR over the three years was low at 1.2 to 1.3%. Figure 1 Canadian average mortality rates by indicator and fiscal year, to Day In-Hospital Mortality After PCI 30-Day In-Hospital Mortality After 30-Day In-Hospital Mortality After Isolated CABG Isolated AVR Mortality indicators Day In-Hospital Mortality After CABG and AVR * Mortality indicators are based on in-hospital mortality only. Isolated means that no other cardiac surgeries, valve procedures or core concomitant procedures were performed during the hospitalization episode of care. PCI: Percutaneous coronary intervention. AVR: Aortic valve replacement. Discharge Abstract Database, National Ambulatory Care Reporting System and Hospital Morbidity Database, , and , Canadian Institute for Health Information.

9 Overall 3-year risk-adjusted rate (per 100) Figure 5 Cardiac care centre risk-adjusted results for 30-Day In-Hospital Mortality After Isolated AVR, to (3 years of pooled data) Cardiac care centre Canadian average (1.3) * Risk-adjusted rate is statistically significantly different from the Canadian average. AVR: Aortic valve replacement. 3 years of pooled data: to Model specifications (coefficients, odds ratios and p-values) were developed using the logistic regression model based on 3 years of pooled data ( to ). Discharge Abstract Database, National Ambulatory Care Reporting System and Hospital Morbidity Database, , and , Canadian Institute for Health Information. Figure 5 shows average in-hospital risk-adjusted mortality over the 33 month time frame with 95% confidence limits. Risk-adjusted mortality ranged from 0 to 6.3%. Hospital mortality rates with 95% confidence limits not overlapping the national mean of 1.3% or each other are significantly different from the comparator. The ability of the model used for risk-adjustment to discriminate mortality was 0.82 (C-statistic 0.82). Risk factors used in the model and the percentage of the cohort with each factor appear in Table 5:

10 Table 5 Risk factors for 30-Day In-Hospital Mortality After Isolated AVR, to (3 years of pooled data) Risk factor Number Percentage Coefficient Odds ratio Significance Intercept n/a n/a -5.8 n/a < Age (vs ) Age (vs ) 2, Age 80+ (vs ) 1, < Male (vs. female) 4, Urgent/emergent admission 1, Shock Previous cardiac surgery Cerebrovascular disease (non-quebec Cerebrovascular disease (Quebec Cardiac dysrhythmias (non Quebec Cardiac dysrhythmias (Quebec Acute renal failure (non < Quebec Acute renal failure (Quebec Endocarditis Charlson group 1 (vs. 0) 1, < Charlson group 2 (vs. 0) < AVR: Aortic valve replacement. n/a: Not applicable. Model specifications (coefficients, odds ratios and p-values) were developed using the logistic regression model based on 3 years of pooled data ( to ). Discharge Abstract Database, Hospital Morbidity Database and National Ambulatory Care Reporting System, , and , Canadian Institute for Health Information. Age greater than 70, shock, previous cardiac surgery, acute renal failure (all jurisdictions except Quebec), cardiac dysrhythmias (Quebec only) and increased comorbidities as indicated by Charlson group were all predictive of increased mortality, with the odds ratios indicated in Table 5. Note that there are some risk factors with two coefficients one for the data submitted by Quebec and one for all other jurisdictions. Quebecspecific coefficients were included in the model to address differences in data collection.

11 Canadian average rate (per 100) Aortic Valve Replacement and Coronary Artery Bypass The 33 month cohort of patients analyzed was 5,934. Average national in-hospital mortality after combined AVR and CABG is higher than each procedure in isolation at 3.1%, and more variable, ranging from a high of 3.7% in to 2.6% in , and 3.1% in Figure 1 Canadian average mortality rates by indicator and fiscal year, to Day In-Hospital Mortality After PCI 30-Day In-Hospital Mortality After 30-Day In-Hospital Mortality After Isolated CABG Isolated AVR Mortality indicators 30-Day In-Hospital Mortality After CABG and AVR * Mortality indicators are based on in-hospital mortality only. Isolated means that no other cardiac surgeries, valve procedures or core concomitant procedures were performed during the hospitalization episode of care. PCI: Percutaneous coronary intervention. AVR: Aortic valve replacement. Discharge Abstract Database, National Ambulatory Care Reporting System and Hospital Morbidity Database, , and , Canadian Institute for Health Information.

12 Overall 3-year risk-adjusted rate (per 100) Figure 6 Cardiac care centre risk-adjusted results for 30-Day In-Hospital Mortality After CABG and AVR, to (3 years of pooled data) Cardiac care centre Canadian average (3.1) * Risk-adjusted rate is statistically significantly different from the Canadian average. AVR: Aortic valve replacement. 3 years of pooled data: to Model specifications (coefficients, odds ratios and p-values) were developed using the logistic regression model based on 3 years of pooled data ( to ). Discharge Abstract Database, National Ambulatory Care Reporting System and Hospital Morbidity Database, , and , Canadian Institute for Health Information. Figure 6 shows average in-hospital risk-adjusted mortality over the 33 month time frame with 95% confidence limits. Risk-adjusted mortality ranged from 0 to 6.6%. Hospital mortality rates with 95% confidence limits not overlapping the national mean of 3.1% or each other are significantly different from the comparator. The ability of the model used for risk-adjustment to discriminate mortality was 0.72 (C-statistic 0.72). Risk factors used in the model and the percentage of the cohort with each factor appear in Table 6:

13 Table 6 Risk factors for 30-Day In-Hospital Mortality After CABG and AVR, to (3 years of pooled data) Risk factor Number Percentage Coefficient Odds ratio Significance Intercept n/a n/a -3.9 n/a < Age (vs ) Age (vs ) 2, Age 80+ (vs ) 1, Male (vs. female) 4, Urgent/emergent admission 1, Shock < NSTEMI AMI Previous AMI Previous cardiac surgery Cardiac dysrhythmias (non Quebec Cardiac dysrhythmias (Quebec Multiple cardiac interventions in same episode of care (PCI, CABG) Acute renal failure (non Quebec Acute renal failure (Quebec Charlson group 1 (vs. 0) 2, Charlson group 2 (vs. 0) PCI: Percutaneous coronary intervention. AVR: Aortic valve replacement. AMI: Acute myocardial infarction. NSTEMI: Non ST elevation myocardial infarction. n/a: Not applicable. Model specifications (coefficients, odds ratios and p-values) were developed using the logistic regression model based on 3 years of pooled data ( to ). Discharge Abstract Database, Hospital Morbidity Database and National Ambulatory Care Reporting System, , and , Canadian Institute for Health Information.

14 Age above 70, female gender, urgent/emergent hospital admission, shock, acute myocardial infarction without ST-segment elevation, and previous acute myocardial infarction and PCI in the same episode of care, acute renal failure (all jurisdictions except Quebec), and higher composite comorbidity as indicated by Charlson group 2, increased the probability of in-hospital mortality with odds ratios as indicated in Table Day Readmission after Isolated Coronary Artery Bypass There were 3,932 readmissions within 30 days of discharge in 42,005 patients surviving to discharge. The incremental mortality from a variable discharge date to 30 days post discharge is not known, hence the true denominator of patients alive to be readmitted is smaller, and calculated readmission rates are underestimated. Average national 30- day readmission after isolated CABG over the three years was 9.4% and almost identical each year (Figure 1). Hospital readmission rates ranged from 6.2 to 12.8%.

15 Figure 8 Cardiac care centre risk-adjusted results for 30-Day All-Cause Readmission Rate After Isolated CABG, to (3 years of pooled data) * Risk-adjusted rate is statistically significantly different from the Canadian average. 3 years of pooled data: to Model specifications (coefficients, odds ratios and p-values) were developed using the logistic regression model based on 3 years of pooled data ( to ). Discharge Abstract Database, National Ambulatory Care Reporting System and Hospital Morbidity Database, , and , Canadian Institute for Health Information. Figure 8 shows average hospital 30-day readmission rates over the 33 month time frame with 95% confidence limits. Risk-adjusted readmission ranged from 6.8 to 12.8%. Hospital mortality 95% confidence limits not overlapping the national mean of 9.4% or each other are significantly different. Readmission rates with 95% confidence limits not overlapping the national mean of 1.4% or each other are significantly different from the comparator.

16 Risk factors predicting 30-day readmission were age greater than 70, female gender, urgent/emergent admission, acute myocardial infarction, cardiac dysrhythmias, PCI within the same episode of care, peripheral vascular disease, and increased Charlson Comorbidity Index. Table 8 Risk factors for 30-Day All-Cause Readmission Rate After Isolated CABG, to (3 years of pooled data) Risk factor Number Percentage Coefficient Odds ratio Significance Intercept n/a n/a -2.5 n/a < Age (vs ) 1, Age (vs ) 12, < Age 80+ (vs ) 2, < Male (vs. female) 34, < Urgent/emergent admission 21, Previous AMI 3, < Cardiac dysrhythmias (non-quebec 2, < Cardiac dysrhythmias (Quebec 1, Hypertension (non-quebec 5, Hypertension (Quebec 8, Multiple cardiac interventions in 1, same episode of care (PCI, CABG) Peripheral vascular disease (non < Quebec Peripheral vascular disease (Quebec 1, Acute renal failure (non-quebec Acute renal failure (Quebec Charlson group 1 (vs. 0) 16, < Charlson group 2 (vs. 0) 2, < PCI: Percutaneous coronary intervention. n/a: Not applicable. Model specifications (coefficients, odds ratios and p-values) were developed using the logistic regression model based on 3 years of pooled data ( to ). Discharge Abstract Database, Hospital Morbidity Database and National Ambulatory Care Reporting System, , and , Canadian Institute for Health Information.