ralssBioMed CentJournal of Cardiothoracic SurgeryOpen AcceResearch articleCumulative incidence for wait-list death in relation to length of queue for coronary-artery bypass grafting: a cohort studyBoris G Sobolev*1,2, Lisa Kuramoto2, Adrian R Levy1,3 and Robert Hayden4Address: 1Department of Health Care and Epidemiology, University of British Columbia, Canada, 2Centre for Clinical Epidemiology and Evaluation, Vancouver, Canada, 3Centre for Health Evaluation and Outcome Sciences, Vancouver, Canada and 4British Columbia Cardiac Registries Surgical Research Committee, Vancouver, CanadaEmail: Boris G Sobolev* - bsobolev@shaw.ca; Lisa Kuramoto - Lisa.Kuramoto@vch.ca; Adrian R Levy - alevy@cheos.ubc.ca; Robert Hayden - erh@telus.net* Corresponding author AbstractBackground: In deciding where to undergo coronary-artery bypass grafting, the length of surgicalwait lists is often the only information available to cardiologists and their patients. Our objectivewas to compare the cumulative incidence for death on the wait list according to the length of waitlists at the time of registration for the operation.Methods: The study cohort included 8966 patients who registered to undergo isolated coronary-artery bypass grafting (82.4% men; 71.9% semi-urgent; 22.4% non-urgent). The patients werecategorized according to wait-list clearance time at registration: either "1 month or less" or "morethan 1 month". Cumulative incidence for wait-list death was compared between the groups, andthe significance of difference was tested by means of regression models.Results: Urgent patients never registered on a wait list with a clearance time of more than 1month. Semi-urgent patients registered on shorter wait lists more often than non-urgent patients(79.1% vs. 44.7%). In semi-urgent and non-urgent patients, the observed proportion of wait-listdeaths by 52 weeks was lower in category "1 month or less" than in category "more than 1 month"(0.8% [49 deaths] vs. 1.6% [39 deaths], P < 0.005). After adjustment, the odds of death beforesurgery were 64% higher in patients on longer lists, odds ratio [OR] = 1.64 (95% confidence interval[CI] 1.02–2.63). The observed death rate was higher in category "more than 1 month" than incategory "1 month or less", 0.79 (95%CI 0.54–1.04) vs. 0.58 (95% CI 0.42–0.74) per 1000 patient-weeks, the adjusted OR = 1.60 (95%CI 1.01–2.53). Longer wait times (log-rank test = 266.4, P <0.001) and higher death rates contributed to a higher cumulative incidence for death on the waitlist with a clearance time of more than 1 month.Conclusion: Long wait lists for coronary-artery bypass grafting are associated with increasedprobability that a patient dies before surgery. Physicians who advise patients where to undergocardiac revascularization should consider the risk of pre-surgical death that is associated with thelength of a surgical wait list.Published: 24 August 2006Journal of Cardiothoracic Surgery 2006, 1:21 doi:10.1186/1749-8090-1-21Received: 24 May 2006Accepted: 24 August 2006This article is available from: http://www.cardiothoracicsurgery.org/content/1/1/21© 2006 Sobolev et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.Page 1 of 10(page number not for citation purposes)Journal of Cardiothoracic Surgery 2006, 1:21 http://www.cardiothoracicsurgery.org/content/1/1/21BackgroundIn patients with coronary artery disease (CAD) who are toundergo coronary artery bypass grafting (CABG), delayingthat operation may lead to the deterioration of thepatient's condition, a poor clinical outcome, and anincreased risk of death [1-3]. A patient who presents withthe symptoms of CAD is usually referred to a cardiologist,who evaluates the results of coronary angiography andrecommends treatment. If coronary angioplasty is notindicated, that patient is referred to a cardiac surgeon,who assesses the need for and suitability of CABG surgery.Patients who require immediate care are admitted to ahospital cardiac ward directly from the catheterizationlaboratory. Elective patients are scheduled for outpatientconsultation with the cardiac surgeon. After the consulta-tion in which a CABG is deemed necessary, surgeons reg-ister patients on their wait lists. The detailed pathway tosurgical revascularization has been described elsewhere[4]. Surgical wait lists hold patient names until surgery canbe scheduled. Patients are removed from the wait listwithout having undergone surgery if they die, refuse theoperation, accept surgery from another surgeon, move outof the province, or experience a health-related decline thatcontraindicates surgery.It has been argued that cardiologists and their patientsshould assess the likely extent of treatment delay and asso-ciated risks when they choose a cardiac surgeon [5]. Indeciding where to undergo treatment, wait-list size isoften the only information available to cardiologists andtheir patients because the length of the wait list for surgeryis a common correlate of the expected wait for hospitaladmission. Indeed, all patients on a wait list must betreated before a patient who has just registered for surgerycan be scheduled for treatment. We previously performedan empirical analysis of a population-based registry andfound that the length of queue at registration affected thetime to elective surgery [4]. Surprisingly, few studies havecorrelated the health effects of the pre-surgical wait withwait-list size at the time of registration for an electiveCABG. The common concern is whether the decision torefer a patient to a specific cardiac surgeon can be madewithout considering the length of the current wait list.We performed a prospective study of all patients who reg-istered to undergo isolated CABG surgery from 1991through 2000 in British Columbia, Canada. We estimatedthe time-dependent probability for death during or beforea certain wait-list week in a patient who could be removedfrom a surgical waiting list for surgery, death, or other rea-sons. The objective of this study was to compare thecumulative incidence of wait-list death between twogroups of patients classified according to the length oftest for significant differences in the risk of death resultingfrom registration on a longer wait list.Patients and methodsData sourcesThe data were taken from the British Columbia CardiacRegistries [6]. That prospectively collected database con-tains information about registration, procedure, or with-drawal dates, and about disease severity and other riskfactors for all patients registered for surgical coronaryrevascularization in 1 of the 4 tertiary-care hospitals thatprovide cardiac care to adult residents of the Canadianprovince of British Columbia since 1991 [4]. To identifythe date and underlying cause of death of registeredpatients who died before they could undergo CABG, welinked the registry to British Columbia Linked HealthDatabase Deaths File by patients' Provincial HealthNumber [7]. Underlying causes of death were codedaccording to the International Classification of Diseases,9th revision (ICD-9). To identify coexisting medical con-ditions in the study cohort we linked the registry to the BCLinked Health Database Hospital Separations File [8] forthe period of 1990 through 2001 and retrieved diagnosesreported in discharge abstracts within 1 year before regis-tration for CABG [9]. The University of British ColumbiaEthics Board approved the protocol for this study.PatientsBetween January 1991 and December 2000, 9366 recordsof patients who registered for isolated CABG were addedto the registry. We excluded 30 records of patients whowere coded as emergency cases, 99 who had the same datefor registration and removal, 4 whose operating roomreport was missing, and 267 who underwent surgerywithin one to three days after having been registered on await list. The remaining 8966 records had either the sur-gery date or the date and reason of removal from the waitlist without surgery. Because patients whose angiographicfindings indicated the need for immediate surgery werenot added to a wait list, they were not included in theanalysis of wait-list mortality but instead contributed todemand for service figures.Urgency groupsWhen accepting patients on wait lists for CABG in BritishColumbia, all cardiac surgeons use a common guidelineto indicate the priority for booking the operating roomaccording to the patient's anginal symptoms, coronaryanatomy, and left ventricular function so that surgery canbe performed within a clinically appropriate time [10]. Inthis analysis, patients are classified as "urgent" if the sug-gested time to surgery was 3 days after the treatment deci-sion had been made, "semi-urgent" if that time was 6Page 2 of 10(page number not for citation purposes)wait lists at the time of their registration for CABG and to weeks, or "non-urgent" if that time was 12 weeks.Journal of Cardiothoracic Surgery 2006, 1:21 http://www.cardiothoracicsurgery.org/content/1/1/21Demand for surgeryFor each calendar week during the study period, thedemand for surgery was characterized by the size of exist-ing wait lists and by the number of direct admissions, i.e.,patients admitted to a hospital ward immediately afterangiography. For each patient, the wait-list size at registra-tion was a count of patients with higher or equal urgencyto undergo CABG in the same hospital. Each patient con-tributed 1 count to the list size for each week that he or sheremained on the wait list, except for the week of registra-tion. Because CABG surgeries are confirmed 1 week inadvance, patients who are to undergo surgery are consid-ered removed from the wait list during the week beforetheir admission date. We defined the number of directadmissions as the weekly count of CABG surgeries per-formed without wait-list registration.Statistical analysisPrimary outcomeIn this study, the primary outcome was the death ofpatients awaiting CABG on a wait list referred to as wait-list deaths. The time on a wait list was computed as thenumber of calendar weeks from registration to surgery,death, or wait-list removal. The date of surgeon's requestfor booking the operating room serves as the date of reg-istration on a wait list. The probability of remaining onthe list after a certain time was estimated by the product-limit method [11]; wait-list times were treated as prospec-tive observations that were monitored from registration tothe patient's last week on the list. The log-rank test wasused to compare the time to removal across the studygroups [12]. The average weekly rate of wait-list deathswas determined by dividing the number of deaths by thesum of observed wait-list times.Study variablesThe wait-list size was categorized by clearance time; i.e., ahypothetical time within which the list could be cleared atthe maximum weekly service capacity if there were no newarrivals [13]. We categorized wait-list size as either "1month or less" or "more than 1 month" of clearance time.We chose 1 month as a cut-off, reasoning that registrationon a wait list with a clearance time of 1 month or less per-mits undergoing surgery within the planned access time of6 weeks for semi-urgent patients. In 3 of the 4 participat-ing hospitals, which had a service capacity of performing15 operations per week, a wait list of 59 or fewer patientscorresponded to a clearance time of 1 month or less, anda wait list of 60 or more patients corresponded to a clear-ance time of more than 1 month. In the fourth hospital,which had a service capacity of performing 25 operationsper week, a wait list of 99 or fewer patients correspondedto a clearance time of 1 month or less, and a list of 100 orthan 1 month. The weekly number of direct admissionswas treated as a continuous variable.Cumulative incidence for wait-list deathWe used the cumulative incidence function (CIF) to char-acterize the time-dependent, marginal probability thatpre-operative death occurs on or before a certain wait-listweek. We interpreted the cumulative incidence for wait-list death as the proportion of patients who were toundergo CABG but died before surgery; a number thatincreased over wait-list time. The CIF for wait-list death isdefined as the integration over time of the product of theweekly death rate and the probability of remaining on thelist [14]. The CIF of wait-list death and its standard errorswere estimated using non-parametric methods [15]. Weused a 2-sample test to compare the CIFs between catego-ries of wait-list clearance time [16].Regression modelsThe effect of wait-list clearance time on the weekly deathrate was estimated by means of discrete-time survivalregressions that yield the odds ratio (OR) as a measure ofthe effect size [17]. We used discrete-time survival analysisbecause wait-list time is inherently discrete and is bestmeasured by the number of weekly operating room sched-ules [13]. To test for differences in the CIF between list-size categories, we used competing-risk regression modelsbased on pseudo-values of the CIF [18]. The clearance-time category was added as an indicator variable, with 1denoting a clearance time of more than 1 month. Theexponential of the regression coefficient for that variablegives the odds ratio of pre-operative deaths for category"more than 1 month" relative to category "1 month orless". Pseudo-values for the CIF for wait-list death werecomputed in the presence of surgery and other competingevents at all distinct, observed event times. For eachpatient, the CIF pseudo-values corresponded to a series ofbinary variables equal to zero before and 1 at or afterdeath in the absence of censoring. The CIF models wereadjusted for subject-level correlation between pseudo-val-ues using the generalized estimation equations. The work-ing weight matrix was fixed and estimated as a product-moment correlation matrix among the pseudo-values. Forthe direct admissions, we interpret odds ratios as a changein the weekly odds of wait-list death associated with 1additional surgery performed immediately after angiogra-phy.ConfoundersMultivariate analyses controlled for differences inpatients' characteristics and significant confounders sum-marized in Table 1. Existing literature suggests that elderlypatients are more likely to undergo revascularization as anPage 3 of 10(page number not for citation purposes)more patients corresponded to a clearance time of more urgent procedure [19]; smaller coronary vessel diametersmay account for higher risk of adverse events in womenJournal of Cardiothoracic Surgery 2006, 1:21 http://www.cardiothoracicsurgery.org/content/1/1/21[20]; co-existing medical conditions may delay open heartsurgery [21]; and changes in practice or supplementaryfunds may reduce time to surgery [10]. We entered twoindicator variables for three comorbidity categories, refer-ent, no co-existing conditions, and 2 comparison catego-ries: presenting with congestive heart failure, diabetes,chronic obstructive pulmonary disease, cancer or rheuma-toid arthritis as suggested by Naylor and colleagues [22];or presenting with other co-existing chronic conditions asdefined in Romano and colleagues [23].ResultsPatientsTable 1 shows the distribution of wait-listed patients anddirect admissions according to age, sex, calendar period,urgency for surgery, comorbid conditions, and wait-listclearance time at registration. In the group of patients whounderwent surgery without registration on wait lists, theage distribution was similar to the listed patients, with themajority (68%) undergoing surgery between 60 and 79years. Compared with the listed patients, the proportionditions indicate that sicker patients were more likely toundergo operation without delay. For example, less than6% of wait-listed patients were in urgent category com-pared with 51% for directly admitted. Similarly, almost53% of wait-listed patients had no co-existing conditions,compared with only 11% in the other group. Wait listswith 1 month or less of clearance time were observed inall urgent patients and were more prevalent in semi-urgent than non-urgent patients (79.1% vs 44.7%, respec-tively).Outcomes of registration for CABGBy 52 weeks on the list, 7724 (86.1%) patients had under-gone surgery, and 767 (8.6%) had been removed withoutsurgery for various reasons such as having died whileawaiting surgery (92 patients), continuing medical treat-ment (176), refusal of surgery (188), having beenaccepted for surgery by another surgeon or hospital (99),having undergone another type of surgery (23), or otherreasons (189). Death certificates were available for 87 ofthe 92 patients who died while awaiting operation, and 5Table 1: Characteristics of 8,966 wait-listed patients and 10,467 directly admitted patients isolated coronary artery bypass surgery in British Columbia 1991–2001Characteristic Wait-Listed patients N(%) Direct admissions N(%)Age group (yr)<50 yr 717 (8.0) 808 (7.7)50–59 yr 1966 (21.9) 2082 (19.9)60–69 yr 3425 (38.2) 3689 (35.2)70–79 yr 2676 (29.8) 3509 (33.5)≥80 yr 182 (2.0) 379 (3.6)SexWomen 1581 (17.6) 2313 (22.1)Men 7385 (82.4) 8154 (77.9)Period of registration/surgery1991–1992 1675 (18.7) 1770 (16.9)1993–1994 1859 (20.7) 1526 (14.6)1995–1996 1867 (20.8) 1686 (16.1)1997–1998 1853 (20.7) 1997 (19.1)1999–2000 1712 (19.1) 2454 (23.4)2001 1034 (9.9)Urgency at registration/surgeryUrgent 515 (5.7) 5353 (51.1)Semi-urgent 6444 (71.9) 4536 (43.3)Non-urgent 2007 (22.4) 523 (5.0)Not provided 55 (0.5)Comorbidity at registration/surgeryMajor conditions† 1930 (21.5) 4040 (38.6)Other conditions‡ 2304 (25.7) 5268 (50.3)None 4732 (52.8) 1159 (11.1)Wait-list clearance time1 month or less 6512 (72.6)more than 1 month 2454 (27.4)†congestive heart failure, diabetes, chronic obstructive pulmonary disease, rheumatoid arthritis, cancer‡peripheral vascular disease, cerebrovascular disease, dementia, peptic ulcer disease, hemiplegia, renal disease, or liver diseasePage 4 of 10(page number not for citation purposes)of women (22%) was slightly higher. Differences betweenthese two groups by urgency and coexisting medical con-sudden deaths were reported by the participating hospi-tals. Of the 515 urgent patients, 98 (19.0%) were down-Journal of Cardiothoracic Surgery 2006, 1:21 http://www.cardiothoracicsurgery.org/content/1/1/21graded to the semi-urgent or non-urgent category at thetime of surgery.More than 10% (258) of non-urgent patients and about5% (212) of semi-urgent patients were still on the waitlists at 52 weeks. Five patients in the urgent group had cal-culated wait times of more than 52 weeks. One of thosepatients was eventually removed by request, the urgencyfor surgery was downgraded in 2 patients, and the reasonfor the delay in surgery was unknown in 2 patients. Intotal, 254 (2.8%) patients were removed from the waitlists for CABG after being deemed unfit for surgery.Figure 1 shows the estimated probability of remaining onthe list by week since registration and wait-list clearancetime. Lists with longer clearance times were associatedwith longer wait times (the log-rank test = 266.4, df = 1, p< 0.001). When a clearance time was 1 month or less,75%, 50%, and 25% of patients remained on wait listsafter 4, 9, and 18 weeks, respectively. For a clearance timeEstimated probability of remaining on a coronary-artery bypass grafting wait list by the number of weeks since registration and wait-list clearance times in sem -urgent and non-urgent group combinedFigure 1Estimated probability of remaining on a coronary-artery bypass grafting wait list by the number of weeks since registration and 1 4 7 10 13 16 19 22 25 28 31 34 37 40 43 46 49 5200.20.40.60.81WAIT−LIST CLEARANCE TIME:less than one monthmore than one monthWEEK SINCE REGISTRATIONPROBABILITY OF REMAINING ON THE LISTPage 5 of 10(page number not for citation purposes)wait-list clearance times in semi-urgent and non-urgent groups combined.Journal of Cardiothoracic Surgery 2006, 1:21 http://www.cardiothoracicsurgery.org/content/1/1/21of more than 1 month, 75%, 50%, and 25% of patientsremained on the list after 6, 14, and 29 weeks, respec-tively.Death rates by clearance timeThe effect of wait-list clearance time was studied in semi-urgent and non-urgent patients because all urgent patientsfell in one clearance time category. There were 49 wait-listweeks in category "more than 1 month". The observedaverage death rate was higher in category "more than 1month" than in category "1 month or less", 0.79 (95%confidence interval [CI] 0.54–1.04) vs. 0.58 (95% CI0.42–0.74) per 1000 patient-weeks. After adjustment forage, sex, urgency for surgery, calendar period, co-existingconditions, and weeks on the list, the weekly odds of wait-list death were 1.6 higher greater for a longer clearanceEstimated cumulative incidence for death on the wait list by the number of weeks since registration and urgency group, thin lines represent standard errors the cumulative incidence estimate for each week.Figu 2Estimated cumulative incidence for death on the wait list by the number of weeks since registration and urgency group, thin lines represent standard errors for the cumulative incidence estimate for each week1 4 7 10 13 16 19 22 25 28 31 34 37 40 43 46 49 5200.0050.010.0150.02semi−urgentnon−urgentURGENCY GROUP:two−sample test: p = 0.002WEEK SINCE REGISTRATIONCUMULATIVE INCIDENCE OF DEATHPage 6 of 10(page number not for citation purposes)deaths over 84,710 patient-weeks of follow-up in category"1 month or less", and 39 deaths over 49,219 patient-time, the adjusted OR = 1.60 (95% CI 1.01–2.53). Insemi-urgent and non-urgent groups, the product of theJournal of Cardiothoracic Surgery 2006, 1:21 http://www.cardiothoracicsurgery.org/content/1/1/21average death rates and weeks on the wait list served as agood approximation for the cumulative hazards, suggest-ing that the hazard functions for wait-list death were con-stant over wait-list time.Cumulative incidence for wait-list deathFigure 2 shows the estimated cumulative incidence forwait-list death by clearance-time categories in semi-urgentand non-urgent patients combined. The observed (unad-justed) proportion of wait-list deaths by 52 weeks waslower in category "1 month or less" than in category"more than 1 month" (0.8% [49 deaths] vs. 1.6% [39deaths], Gray's 2-sample test = 10.1, df = 1, P < 0.005).Higher weekly death rates and longer waits in the groupwith a clearance time of more than 1 month contributedto the differences in the cumulative incidence of wait-listdeath between the groups studied. After adjustment forage, sex, urgency for surgery, calendar period, co-existingconditions, and weeks on the list, the effect of wait-listsize at registration remained significant. The odds of wait-list death were 64% higher in patients on a list with aclearance time of more than 1 month than in those on alist with a clearance time of 1 month or less, the adjustedOR = 1.64 (95%CI, 1.02–2.63), Table 2. As expected, theurgency for surgery had a major influence on the cumula-tive incidence of wait-list death as well. Non-urgentpatients had a higher cumulative incidence of pre-opera-tive death than did semi-urgent patients for almost allweeks on the list (Gray's 2-sample test = 9.3, df = 1, P<0.001). After controlling for confounders, the differencebetween urgency groups remained significant and inde-pendent from the list-size effect, the adjusted OR = 1.69(95%CI, 1.05–2.74). Direct admissions did not alter theodds of death for semi-urgent and non-urgent patients.DiscussionWe examined the relationship between the length of thewait list at the time of registration for CABG and the riskof death before surgery in patients awaiting that operationon any of multiple wait lists in a health system in whichall medically necessary services are publicly funded. Usingrecords from the provincial population-based registry ofpatients identified as needing surgical revascularization,we compared the cumulative incidence for wait-list deathbetween the two categories of wait-list size according to aclearance time. The list size was a simple count of patientswith higher or equal surgical priority who were on a waitlist at the time of registration of a new patient. Out of 88wait-list deaths that occurred in the two less urgentgroups, 44 deaths in semi-urgent and 15 deaths in non-urgent groups were related to cardiovascular disease. Wereport on all-cause mortality because the accuracy ofdeath certificate codes is a concern in this analysis; usingall-cause mortality could not have induced bias in theresults [24].Our results show that wait-list size is associated with theprobability that a semi-urgent or non-urgent patientwould die before surgery by a certain wait-list week. Thepatients registered on a list with a clearance time of morethan 1 month had 60% higher weekly death rate afteradjustment than those on a list with a clearance time of 1month or less. Longer wait times (p < 0.001) and a higherdeath rate contributed to a higher cumulative incidencefor wait-list death in the patients registered on a list witha clearance time of more than 1 month, the adjusted OR= 1.64 (95% CI 1.02–2.63). The number of patients whounderwent CABG without having been registered on await list in the same hospital exerted no independenteffect.Other investigators concerned with delay in treatment forpatients who require a CABG have reported on the impactof patient prioritization [25,26], risks of delayed treat-ment [1,2,27], and the worsening symptoms and morbid-ity associated with a long wait for surgery [3,28]. Inquantifying the risk of adverse events on wait lists forCABG surgery, the Kaplan-Meier method is often used toestimate the cumulative probability of the occurrence ofan event by certain time after registration for surgery[3,28,29]. It has been found, however, that the comple-ment of Kaplan-Meier estimator overestimates the pro-portion of the event in the competing risks setting [30].Because patients on a wait list are subject to competingevents such as surgery, death, or removal from the wait listfor other reasons, the Kaplan-Meier method producesprobability estimates that are only valid in a hypotheticalsituation in which all competing risks are removed beforeTable 2: Association between urgency, wait-list clearance times and cumulative incidence for death on the wait list as measured by odds ratios derived from discrete-time survival regression modelsEffect Unadjusted OR(95% CI) Adjusted OR*(95% CI)non-urgent vs semi-urgent 1.61(1.00, 2.59) 1.69(1.05, 2.74)clearance time of 1 month or less 1.00 1.00clearance time of more than 1 month 1.67(1.05, 2.66) 1.64(1.02, 2.63)direct admission† -- 1.00(1.00, 1.00)Page 7 of 10(page number not for citation purposes)*Adjusted for age, sex, comorbidity, calendar period, and week on the list†Associated with one additional surgery performed without wait-list registrationJournal of Cardiothoracic Surgery 2006, 1:21 http://www.cardiothoracicsurgery.org/content/1/1/21the patient's death without altering the risk of death [31].Without the assumption of independent competingevents, the Kaplan-Meier method is not valid and shouldnot be used [32]. However, the independence of wait out-comes cannot be verified from data and may not be real-istic, because the low proportion of wait-list deaths mayindicate either a low risk of death or a high rate of surgery.for competing events. The CIF describes the time-depend-ent marginal probability that pre-operative death occurson or before a certain time of registration on a wait listafter the probability of surviving multiple competingevents has been considered [14,33,34]. Pepe and Moriargued that the CIF is a more accurate and comprehensivesummary of the risk of death in a competing-risks settingEstimated cumulative incidence for death on the wait list by the number of weeks since registration and wait-list clearance times in semi-urgent a d no -urgent groups combined, hin lin s repr sent standard errors for he cumulative in idence sti-mate for ach we kFigure 3Estimated cumulative incidence for death on the wait list by the number of weeks since registration and wait-list clearance times in semi-urgent and non-urgent groups combined, thin lines represent standard errors for the cumulative incidence esti-mate for each week1 4 7 10 13 16 19 22 25 28 31 34 37 40 43 46 49 5200.0050.010.0150.02less than one monthgreater than one monthWAIT−LIST CLEARANCE TIME:two−sample test: p = 0.002WEEK SINCE REGISTRATIONCUMULATIVE INCIDENCE OF DEATHPage 8 of 10(page number not for citation purposes)Appropriate statistical instruments include the CIF thatcan be estimated without the independence assumptionthan are death rates or cumulative hazards, which cannotbe translated to the probability of death [15].Journal of Cardiothoracic Surgery 2006, 1:21 http://www.cardiothoracicsurgery.org/content/1/1/21Misclassification of the recorded urgency for treatment isa concern in this analysis. Retrieved from the registry, theurgency category is a composite variable that is based ona variety of clinical factors. No audit was performed toevaluate the quality of those records. The observation thathigher priority patients were more likely to undergoCABG via direct admission indicates that the degree ofmisclassification of priority was likely small. Another con-cern is that in some patients, the urgency for surgery wasreclassified at the time of surgery. However, the timing ofchanges in urgency was not recorded.ConclusionThe contribution of this paper is two-fold. First, the cumu-lative incidence for wait-list death in relation to wait-listsize at the time of registration for CABG, to our knowl-edge, has not been reported previously. We found thatlong wait lists are associated with increased probabilitythat a patient dies before surgery after accounting for thesurgery rate in semi-urgent and non-urgent patients. Sec-ond, physicians who advise patients to undergo revascu-larization with a cardiac surgeon can use our results toconsider the risk of pre-surgical death that is associatedwith the current length of wait list of the surgeon.Authors' contributionsBS conceived the study concept and design, participatedin analysis and interpretation, and drafted the manu-script. LK performed statistical analysis and drafted themanuscript. AL participated in data acquisition and criti-cally revised the manuscript. RH participated in dataacquisition and critically revised the manuscript. Allauthors read and approved the final manuscript.AcknowledgementsThis study received financial support from the Canada Research Chairs Program (BS), the Canada Foundation for Innovation (BS, AL), the Michael Smith Foundation for Health Research (AL), the Vancouver Coastal Health Research Institute (BS, LK), and the St. Paul's Hospital Foundation (AL). None of the sponsors had a role in the study design; in the collection, anal-ysis, and interpretation of data; in the writing of the report; or in the deci-sion to submit the paper for publication.The following cardiac surgeons are members of the Surgical Research Committee: James Abel, Richard Brownlee, Larry Burr, Anson Cheung, James Dutton, Guy Fradet, Virginia Gudas, Robert Hayden, Eric Jamieson, Michael Janusz, Shahzad Karim, Tim Latham, Jacques LeBlanc, Sam Lichten-stein, Hilton Ling, John Ofiesh, Michael Perchinsky, Peter Skarsgard and Frank Tyers. The authors gratefully acknowledge the contributions of Mark FitzGerald, Martin Schechter, and Rita Sobolyeva.References1. Bernstein SJ, Rigter H, Brorsson B, Hilborne LH, Leape LL, Meijler AP,Scholma JK, Nord AS: Waiting for coronary revascularization:a comparison between New York State, The Netherlandsand Sweden. Health Policy 1997, 42:15-27.3. Ray AA, Buth KJ, Sullivan JA, Johnstone DE, Hirsch GM: Waiting forcardiac surgery: results of a risk-stratified queuing process.Circulation 2001, 104:I92-I98.4. Sobolev B, Levy AL, Hayden RH, Kuramoto L: Does wait-list sizeat registration influence time to surgery? Analysis of a popu-lation-based cardiac surgery registry. Health Serv Res 2006,41:23-49.5. Naylor CD: A different view of queues in Ontario. Health Aff1991, 10:110-128.6. Volk T, Hahn L, Hayden R, Abel J, Puterman ML, Tyers GF: Reliabil-ity audit of a regional cardiac surgery registry. J Thorac Cardi-ovasc Surg 1997, 114:903-910.7. Sobolev BG, Levy AR, Kuramoto L, Hayden R, FitzGerald M: Dolonger delays for coronary bypass surgery contribute to pre-operative mortality in less urgent patients? Med Care 2006,44:.8. Chamberlayne R, Green B, Barer ML, Hertzman C, Lawrence WJ,Sheps SB: Creating a population-based linked health database:a new resource for health services research. Can J Public Health1998, 89:270-273.9. Humphries KH, Rankin JM, Carere RG, Buller CE, Kiely FM, SpinelliJJ: Co-morbidity data in outcomes research: are clinical dataderived from administrative databases a reliable alternativeto chart review? J Clin Epidemiol 2000, 53:343-349.10. Levy A, Sobolev B, Hayden R, Kiely M, FitzGerald M, Schechter M:Time on wait lists for coronary bypass surgery in BritishColumbia, Canada, 1991 - 2000. BMC Health Services Research2005, 5:22 [http://www.biomedcentral.com/1472-6963/5/22].11. Bland JM, Altman DG: Survival probabilities (the Kaplan-Meiermethod). BMJ 1998, 317:1572.12. Bland JM, Altman DG: The logrank test. BMJ 2004, 328:1073.13. Sobolev B, Brown P, Zelt D, Kuramoto L: Waiting time in relationto wait-list size at registration: statistical analysis of a wait-ing-list registry. Clin Invest Med 2004, 27:298-305.14. Bryant J, Dignam JJ: Semiparametric models for cumulativeincidence functions. Biometrics 2004, 60:182-190.15. Pepe MS, Mori M: Kaplan-Meier, marginal or conditional-prob-ability curves in summarizing competing risks failure timedata. Stat Med 1993, 12:737-751.16. Gray RJ: A class of K-sample tests for comparing the cumula-tive incidence of a competing risk. The Annals of Statistics 1988,16:1141-1154.17. Cox DR, Oakes D: Analysis of survival data London, Chapman Hall;1984. 18. Klein JP, Andersen PK: Regression modeling of competing risksdata based on pseudovalues of the cumulative incidencefunction. Biometrics 2005, 61:223-229.19. Christenson JT, Simonet F, Schmuziger M: The influence of age onthe outcome of primary coronary artery bypass grafting. JCardiovasc Surg (Torino) 1999, 40:333-338.20. O'Connor NJ, Morton JR, Birkmeyer JD, Olmstead EM, O'ConnorGT: Effect of coronary artery diameter in patients undergo-ing coronary bypass surgery. Northern New England Cardi-ovascular Disease Study Group. Circulation 1996, 93:652-655.21. Naylor CD, Baigrie RS, Goldman BS, Basinski A: Assessment of pri-ority for coronary revascularisation procedures. Lancet 1990,335:1070-1073.22. Naylor CD, Levinton CM, Baigrie RS: Adapting to waiting lists forcoronary revascularization. Do Canadian specialists agreeon which patients come first? Chest 1992, 101:715-722.23. Romano PS, Roos LL, Jollis JG: Adapting a clinical comorbidityindex for use with ICD-9-CM administrative data: differingperspectives. J Clin Epidemiol 1993, 46:1075-1079 [http://www.sciencedirect.com/science?_ob=IssueURL&_tockey=%23TOC%235076%231993%23999539989%23480974%23FLP%23Volume_46,_Issue_10,_Pages_1075-1218_(October_1993)&_auth=y&view=c&_acct=C000050484&_version=1&_urlVersion=0&_userid=1022551&md5=fce23529a85f5d276af1c8f8b231cc85].24. Kircher T, Anderson RE: Cause of death. Proper completion ofthe death certificate. JAMA 1987, 258:349-352.25. Cox JL, Petrie JF, Pollak PT, Johnstone DE: Managed delay for cor-onary artery bypass graft surgery: the experience at oneCanadian center. J Am Coll Cardiol 1996, 27:1365-1373.Page 9 of 10(page number not for citation purposes)2. Morgan CD, Sykora K, Naylor CD: Analysis of deaths while wait-ing for cardiac surgery among 29,293 consecutive patients inOntario, Canada. Heart 1998, 79:345-349.Publish with BioMed Central and every scientist can read your work free of charge"BioMed Central will be the most significant development for disseminating the results of biomedical research in our lifetime."Sir Paul Nurse, Cancer Research UKYour research papers will be:available free of charge to the entire biomedical communitypeer reviewed and published immediately upon acceptancecited in PubMed and archived on PubMed Central Journal of Cardiothoracic Surgery 2006, 1:21 http://www.cardiothoracicsurgery.org/content/1/1/2126. Seddon ME, French JK, Amos DJ, Ramanathan K, McLaughlin SC,White HD: Waiting times and prioritization for coronaryartery bypass surgery in New Zealand. Heart 1999, 81:586-592.27. Rexius H, Brandrup-Wognsen G, Oden A, Jeppsson A: Mortality onthe waiting list for coronary artery bypass grafting: incidenceand risk factors. Ann Thorac Surg 2004, 77:769-774.28. Koomen EM, Hutten BA, Kelder JC, Redekop WK, Tijssen JG, KingmaJH: Morbidity and mortality in patients waiting for coronaryartery bypass surgery. Eur J Cardiothorac Surg 2001, 19:260-265.29. Jackson NW, Doogue MP, Elliott JM: Priority points and cardiacevents while waiting for coronary bypass surgery. Heart 1999,81:367-373.30. Gooley TA, Leisenring W, Crowley J, Storer BE: Estimation of fail-ure probabilities in the presence of competing risks: Newrepresentations of old estimators. Stat Med 1999, 18:695-706.31. Gaynor JJ, Feuer EJ, Tan CC, Wu DH, Little CR, Straus DJ, ClarksonBD, Brennan MF: On the use of cause-specific failure and con-ditional failure probabilities - examples from clinical oncol-ogy data. Journal of the American Statistical Association 1993,88:400-409.32. Alberti C, Metivier F, Landais P, Thervet E, Legendre C, Chevret S:Improving estimates of event incidence over time in popula-tions exposed to other events - Application to three largedatabases. J Clin Epidemiol 2003, 56:536-545.33. Benichou J, Gail MH: Estimates of Absolute Cause-Specific Riskin Cohort Studies. Biometrics 1990, 46:813-826.34. Caplan RJ, Pajak TF, Cox JD: Analysis of the Probability and Riskof Cause-Specific Failure. International Journal of Radiation OncologyBiology Physics 1994, 29:1183-1186.yours — you keep the copyrightSubmit your manuscript here:http://www.biomedcentral.com/info/publishing_adv.aspBioMedcentralPage 10 of 10(page number not for citation purposes)