Comparing performance between coronary intervention centres requires detailed case-mix adjusted analysis

Abstract

This study compares 12 month clinical outcomes and procedural costs at two interventional centres with significant differences in crude mortality and revascularization outcomes between 1997 and 1998. Percutaneous coronary intervention (PCI) registry data on 1046 consecutive patients treated contemporaneously at two university centres were linked to hospital discharge and death data to provide 12 month follow-up information on survival and repeat revascularization. Costs were determined by detailed analysis of equipment use, length of stay and staff from 100 contemporary cases at each centre to derive a procedural cost model. This model was then applied retrospectively to estimate cost per procedure. Stents were used more frequently at one centre (56 versus 26 per cent, chi(2) test, p < 0.001) resulting in greater procedural cost [mean (SE), pounds sterling 1970 (34) versus pounds sterling 1521 (39), t-test, p < 0.001). One year repeat target vessel PCI was significantly greater at the centre using more stents (10.3 versus 5.6 per cent, chi(2) test, p = 0.005) and the need for any repeat revascularization (PCI or coronary artery by-pass surgery) was also significantly greater at this centre (18.4 versus 10.8 per cent, chi(2) test, p < 0.001). Cox regression revealed that after correction for case-mix the difference in the need for repeat target vessel PCI between the two centres was no longer significant (p = 0.15). In the two centres studied, crude differences in cost per case, mortality and the need for revascularization were largely accounted for by significant differences in case-mix. Comparison of outcomes and costs between centres should not be published without careful adjustment for differences in case-mix.

Cost-effectiveness of bivalirudin versus heparin plus glycoprotein IIb/IIIa inhibitor in the treatment of non–ST-segment elevation acute coronary syndromes

“…Double-counting of ward costs during the initial hospitalization period was avoided by estimating PCI and CABG procedure costs from the HRG costs on a pro-rata basis. We used a nonward cost proportion of 55% for PCI [20] and a cost proportion of 30% incurred in theater in cardiac surgery for CABG [21]. Non-Q-wave MIs occurring after initial hospitalization discharge were assumed to cost 31% of a Q-wave MI [22].…”

“…The proportion of upstream as opposed to catheter laboratory-induced GPI use was switched from 60% to 20% in the ACUITYbased analysis and from 20% to 60% in the GRACE-based analysis, under the simplifying assumption of no changes in clinical effectiveness [29,30]. To capture change over time, the Angiography procedure cost (IQR) [19] 268 (254-474) PCI cost (IQR) [18] 2614 (2087-2955) PCI procedure cost (IQR) ʈ [18,20] 1438 (1148-1625) CABG cost (IQR) [18] 8631 (7456-10,212) CABG procedure cost (IQR) ʈ [18,21] 2624 (2267-3104) Q-wave MI (IQR) [18] 1934 (882-2376) Non-Q-wave MI (IQR) [18,22] 600 (274-737) Major bleed examination and procedure cost ʈ [18,24,25] 1,078 Minor bleed examination and procedure cost ʈ [18,24,25] 67 Ward costs Normal ward day (IQR) [27] 257 (251-275) CCU/ICU day (IQR) [18] 938 (777-1,060) Long-term annual cardiovascular treatment cost of 1-year survivors [28] …”

The year of transparency: measuring quality of cardiac care

“…A common remark heard by doctors is that they worry about case-mix correction and that a negative outcome compared with others can be explained by the more difficult patient population served. A good case-mix correction applied in crude data could change the compared clinical outcome and is important to avoid unintended consequences [ 35 , 41 , 42 ]. Furthermore, it is important to remember that quality indicators are just an indication of the real quality of care.…”

Effects of changing clinical practice on costs and outcomes of percutaneous coronary intervention between 1998 and 2002

“…Angioplasty balloons, bed-days and adjuvant drug treatment also contribute appreciably to procedural costs. 11 Thirdly, it is not known whether the net effect of multiple new interventions is achieved in ''real-world'' clinical practice. ''Real-world'' patients tend to be older, have more comorbidity and are generally more heterogeneous than those in randomised trials.…”