This study assessed the relationship between demand-side factors, such as age, gender, SES and clinical need, and wait times for coronary angiography in a cohort of 74,254 patients. We found that a structured urgency rating system, reflecting clinical need, was effective in appropriately triaging patients waiting for coronary angiography with more urgent patients treated more rapidly. Nonetheless, there remained wide regional variations in wait-times. Although supply-side factors had an important impact on residual wait-times, there was differential role of cath lab, invasive cardiologist and GP supply depending on patient urgency. The most consistent contribution to this geographic variation was cath lab supply, with the impact of invasive cardiologist supply confined to urgent patients and that of GP supply confined to semi-urgent and elective patients. Our findings suggest that regional disparities in coronary angiography wait times may be addressed by initiatives aimed at improving both the supply of procedural facilities and access to both specialty and GP care.
Currently, most jurisdictions use implicit judgment in managing wait lists for coronary angiography. Accordingly, there is the potential for non-clinical factors, such as SES having an inappropriate impact on patient triaging. The urgency score incorporates clinical determinants of patient need and has been validated against physician judgment and adverse events. As such, when used as an explicit structured wait-time management tool to aid the appropriate triaging of patients, the urgency score may reduce previously documented access disparities across age, SES and geographic location.
We determined that the urgency score was a strong predictor of wait-times for angiography. Nonetheless, other variables continued to impact wait-times. For example, the effect of age was most pronounced in urgent patients, with older patients having delays to angiography despite being at the greatest risk and therefore having the potential for the greatest absolute risk reduction with an invasive strategy. Our findings reinforce the risk-treatment paradox that has been identified in multiple areas of medicine, including invasive cardiovascular care[19, 20].
In contrast, SES differences in wait times were minimal, suggesting that a structured wait time management system has been successful in reducing access disparities for coronary angiography. Importantly, our study diverges from findings in previous studies. Prior studies have shown that patients in lower socio-economic strata have a higher prevalence of coronary risk factors, and therefore would potentially be of higher urgency. In our analysis, by adjusting for urgency score prior to SES, we may have mitigated much of the impact of SES. This is reinforced by the finding that SES had the greatest impact in elective patients, in whom clinical risk would be lower. Importantly, our analysis is limited by the absence of patient level data on income and SES, requiring postal codes instead to derive an income quintile. As such, given the risk for ecological fallacy, we cannot exclude the possibility that SES did play a demand-side role on wait times.
Despite accounting for patient urgency, there remained significant regional differences in wait times to coronary angiography. Although absolute differences in median wait times between LHINs were modest, there were substantial differences in the proportion of patients who were treated within RMWT. Previous work from the CCN registry suggests that, irrespective of urgency strata, the majority of all waiting-time deaths for angiography occur after the RMWT; moreover, had these patients been treated within the RMWT, an estimated 18.5 deaths per 10,000 patients may have been averted. In our cohort from 2005-06, this translates to 15 potentially avoidable wait-time deaths across the three urgency groups.
In our analysis, cath lab supply had the most pronounced and consistent impact on regional differences in wait-times. The impact of procedural capacity was greatest in elective patients. This suggests that in regions with limited procedural capacity, there is appropriate prioritization of urgent patients, which in turn will amplify the delays for more elective angiograms. Cath lab supply is downstream from the CCN referral and therefore, predictably would have a direct impact on the delay to angiography.
Both invasive cardiology and GP supply had relatively modest impact on wait-times. Moreover, the impact of invasive cardiologists supply was restricted to urgent patients while that of GP supply was in less sick patients. A potential explanation for this observation is that urgent patients would more likely enter the health care system via a hospital admission and therefore require access to a specialist. In contrast, the primary care physician would more likely serve as the point of entry into the health care system for elective patients. As wait times are calculated from the time of CCN referral to the procedure, the impact of GP supply on time to cath is indirect. This impact may occur through greater patient advocacy, with the GP acting to ensure their patients receive prompt treatment. Therefore, we hypothesize that GP supply corresponds with the overall efficiency of health care delivery within the region.
These differential impacts underscore the inter-relationship between these components of care. As such, our data suggests that if greater equity in access to angiography across regions is to be achieved, initiatives should be directed at improving all supply components. An important caveat to our conclusions is the absence of a direct measure of the appropriateness of coronary angiography. The appropriateness for coronary angiography should be a reflection of the clinical need for invasive cardiac investigation. Although the urgency score captures some elements of clinical need, it is not an ideal proxy for appropriateness, because it fails to incorporate other important factors such as frailty, which may 'appropriately' delay angiography. The inability to fully evaluate procedural appropriateness into our analysis highlights the need for a more direct metric of this critical demand-supply factor.
Several additional limitations of our study merit consideration. First, this study used data from 2005 to 2006. Since that time, coronary angiography facilities in Ontario have increased substantially, and we would anticipate overall wait times have reduced. However, concurrent to this increase in supply, the indications for coronary angiography have expanded, especially for patients hospitalized with acute coronary syndromes[24, 25]. As a sensitivity analysis, we repeated our analysis using data from 2001-2004. As seen in Additional file 1, we found similar results using this earlier cohort, with both cath lab and physician supply playing a larger role. Indeed, as procedural capacity has improved in recent years, physician factors have played a lesser role in residual wait times. Therefore, we anticipate that our conclusions will apply to contemporary practice, although the magnitude may vary.
Second, the use of the urgency score may vary from institution to institution. Although some may use the score itself to triage, others may use it as a monitoring tool to assess if implicit triage is appropriate. Nonetheless, given the allocation of RMWT to urgency scores by the CCN and ongoing surveillance of institutional adherence to these guidelines, the urgency score provides a structured framework for wait time management.
Third, we performed a complete case analysis, assuming that the 12% of patients with missing data were missing completely at random (MCAR). In Additional file 3, we examined the characteristics of patients with complete data and those with missing data elements, and found the cohorts were reasonably matched. However, there were differences in the urgency score and wait times between the complete case cohort and those with missing data (urgency score of 4.1 for missing data versus 3.9 for complete case cohort; wait-time of 12.6 days for the missing data versus 9.7 days in the complete case cohort). In addition, it appeared that the missing data was not evenly distributed between regions. As such, we cannot rule out the possibility that the pattern of missing data is contingent on the other available observed parameters such as regional supply or on any unobserved characteristics of the patients themselves.
Finally, our study only addresses the differences in access to angiography of patients who have already been assessed by a physician and thereby referred for a diagnostic procedure. Arguably, a substantial burden of disease exists in the proportion of the population who has no access to physicians, and therefore never evaluated nor referred. Therefore, it is likely that the impact of supply measures on regional access disparities, especially that of GP supply, is in fact greater than we have described.