Improving quality of care for pediatric cardiac surgery in New York State

Progress in Pediatric Cardiology 18 (2003) 13–25

Improving quality of care for pediatric cardiac surgery in New York State Donna R. Dorana, Casey S. Roarkb, Edward L. Hannanc,* a

Cardiac Services Program, New York State Department of Health, New York, USA b Research Foundation of SUNY, New York, USA c Department of Health Policy, Management, and Behavior, University at Albany School of Public Health, One University Place, Rensselaer, NY 12144, USA Accepted 4 January 2003

Abstract In 1989, the New York State Department of Health introduced a computer-based cardiac surgery reporting system that gathered demographic, procedural, clinical, complication and discharge information on every adult and pediatric patient undergoing cardiac surgery in a New York State hospital. In 1991, a separate pediatric cardiac surgery reporting system was developed to deal with the unique clinical factors associated with pediatric cardiac surgery. For the last 10 years, the Department of Health has been working with its Cardiac Advisory Committee to modify this system and to use data from it to assure and improve the quality of pediatric cardiac surgery in the State. These efforts have included the assurance that the data in this database are accurate and complete through the implementation of annual medical record audits; the use of the database to explore volume–outcome relationships for pediatric cardiac surgery and to feed back this information to providers and the public; the identification of significant independent risk factors for in-hospital mortality of pediatric cardiac surgery patients, the adjustment of hospital mortality rates based on differences in the prevalence of these risk factors among hospitals in the State, and the reporting to each hospital of comparative outcomes data for benchmarking; and the use of annual hospital pediatric cardiac surgery volumes and outcomes analyses to trigger focused reviews consisting of alert letters, requirements for hospitals to address low patient volumes, and site visits conducted by the Cardiac Advisory Committee. 䊚 2003 Elsevier Ireland Ltd. All rights reserved. Keywords: Pediatric cardiac surgery; Quality of care; Mortality rates; Quality assurance; Quality assessment

1. History of New York’s Cardiac Advisory Committee and pediatric cardiac surgery data system The New York State Cardiac Advisory Committee (CAC) has a long and distinguished history in implementing innovative quality improvement initiatives. Historically, these initiatives have utilized a combination of objective data and peer reviews to identify factors associated with outcomes and use that information to improve care. Interestingly, pediatric cardiac surgery was the stimulus leading to the committee’s origins. During the early 1950s, cardiac surgery and essential related services were added to the scope of the New York State handicapped children’s program. This change provided finan*Corresponding author. Tel.: q1-518-402-0333; fax: q1-518-4020414. E-mail address: [email protected] (E.L. Hannan).

cial aid for the treatment of pediatric patients to meet the costs of cardiac care with the condition that quality and costs met minimum standards. A two-committee system evolved—one under the auspices of the New York State Health Department, the other under the New York City Department of Health. A formal inspection plan was established that called for a surgeon and a cardiologist to visit a program, review all aspects of the program and provide recommendations to the Department of Health. The two committees were merged and formally designated as a statewide commissioner’s advisory body in 1974. Minimum volumes were codified for pediatric and adult cardiac catheterization and surgery programs, and a statewide data collection system was instituted. The information collected, which included hospital specific volume, mortality and complication data for these services, was tabulated and provided back to hospitals

1058-9813/03/$ - see front matter 䊚 2003 Elsevier Ireland Ltd. All rights reserved. doi:10.1016/S1058-9813(03)00072-9

14

D.R. Doran et al. / Progress in Pediatric Cardiology 18 (2003) 13–25

in an annual report for use in hospital and Cardiac Advisory Committee quality assurance activities. The Committee’s role has expanded substantially over the years. However, a focus on pediatric cardiac services remains and is carried out through a standing pediatric cardiac services subcommittee. The subcommittee, which includes practicing pediatric cardiac surgeons and cardiologists, is chaired by Dr Roberta Williams, Vice President of Pediatrics and Academic Affairs at Los Angeles Children’s Hospital and Professor and Chair of Pediatrics at Keck School of Medicine at the University of Southern California. Our analyses and data collection initiatives are carried out under the guidance of this subcommittee and the full Cardiac Advisory Committee. The process of utilizing data and peer review to ensure and improve quality of care continues today. This process has been enhanced by the implementation of much more comprehensive data systems. Advances in the New York State data collection initiatives for pediatric cardiac surgery have, for the most part, paralleled the advances in the adult cardiac surgery system. Both systems gather patient specific demographic, procedural, clinical and outcomes data for use in analyzing provider specific risk-adjusted mortality rates. However, it soon became apparent that evaluating risk-adjusted outcomes for pediatric cardiac surgery is not as straightforward as it is for coronary artery bypass surgery. The history of our pediatric data collection initiatives is as follows: ● Mid 1970s–1988: Aggregate data were gathered from each hospital on an annual basis. Summaries of volume, mortality and morbidity for congenital heart surgeries and pediatric cardiac catheterization were distributed to hospitals and used by the Cardiac Advisory Committee to focus quality reviews. ● 1989: A computer-based cardiac surgery reporting system was introduced that gathered demographic, procedural, clinical, complication and discharge information on every adult and pediatric patient undergoing cardiac surgery in a New York State hospital. Data entry software was provided to hospitals. (The software was subsequently enhanced to allow hospitals to track outcomes). This system provided some additional specificity for reviewing pediatric cardiac surgery, but the clinical factors gathered in this system (aimed primarily at the adult population) did not yield a mechanism to risk-adjust pediatric cardiac surgery data. ● 1991: A separate pediatric cardiac surgery reporting system was developed to deal with the unique clinical factors associated with pediatric cardiac surgery. The system allowed for one diagnosis code and one procedural code—with some codes representing a combination of procedures (see Appendix 1A and Appendix 1B for a complete list of codes).

● 2000: The reporting system was revised to update and expand diagnosis and procedure codes. The new system allows for the coding of multiple diagnoses and multiple procedure codes (see Appendix 2A and Appendix 2B for a complete list of codes). 2. Current steps taken to improve quality of care Regulations governing the approval of new programs and minimum standards for existing programs provide a foundation for ensuring high quality of care in New York State hospitals. Regulations governing existing programs (NYCRR405.22) set forth minimum volumes as well as organizational and staffing requirements for centers that provide cardiac catheterizations andyor cardiac surgery programs. Services must be provided under the direction of a qualified specialist and all personnel must be prepared for their responsibilities through appropriate training and educational programs. A comprehensive range of services is required and pediatric cardiac surgery programs are expected to meet an annual minimum volume of 50 cases per year. The approval of new services is governed by certificate of need (CON) regulations (10NYCRR–704.14). The goal of these regulations is to ensure access to high quality services while avoiding an unnecessary duplication of services. Pediatric cardiac catheterization and pediatric cardiac surgery programs are viewed as regional services. Facility capabilities and available resources within a region are two major components in the review process. Currently, 13 hospitals perform pediatric cardiac surgery in New York State. In recent years, volume for pediatric cardiac surgery has declined (a trend seen nationally) and the focus in New York State has been on consolidating services rather than adding services. Approximately 1600 pediatric patients undergo pediatric cardiac surgery in New York State each year. This represents an 8.43y100 000 use rate. Recent work done by Chang and Klitzner w3x together with 2000 census data suggests that the use rate for pediatric cardiac surgery in California is lower—on the order of 6.48y 100 000, even though more centers provide the service. They identify 65 hospitals with at least one case between 1995 and 1997, 30 hospitals performed only one case between 1995 and 1997, of the remaining 35 hospitals– 20 performed over 10 cases between 1995 and 1997. In addition to the regulations that provide minimum standards, several steps have been implemented through the Cardiac Advisory Committee to monitor and improve quality of care. These steps include: developing an analytical model to assess the significant risk factors associated with pediatric cardiac surgery and assessing risk-adjusted outcomes, investigating the relationship between provider volume and patient outcomes, providing baseline information to hospitals, and conducting

D.R. Doran et al. / Progress in Pediatric Cardiology 18 (2003) 13–25

focused reviews for programs where volume andyor outcomes are a concern. A description of the current status of these activities is outlined below. 3. Predicting in-hospital patient mortality using complexity categories Adult cardiac surgery procedures, in particular coronary artery bypass graft (CABG) surgery, are among the most studied procedures in the medical and health services research literature. Numerous studies have identified the significant risk factors for short-term mortality and other adverse outcomes for CABG surgery and for various cardiac valve replacements and repairs. The reasons for the large number of studies in these areas include the frequency with which the adult cardiac procedures are performed, the expense of the procedures and the fact that the relative homogeneity of the procedures enables analysts to arrive at accurate predictions. For pediatric cardiac procedures, it is much more difficult to obtain accurate estimates of patient risk because of the relatively low number of total procedures that are performed (7169 in the 1992–1995 time period in all of New York State) and the extremely large number of different pediatric cardiac procedures that are performed (54 combinations of procedures are reported in the New York State system). These impediments make it very difficult to predict mortality for each of the pediatric cardiac procedures. Building on earlier work by Jenkins et al. w1x, an approach was developed for grouping procedures into relatively homogeneous groups with regard to shortterm mortality risk. First, the procedures were arrayed in order of increasing in-hospital mortality and then natural breaks in contiguous categories were identified with an eye toward defining between three and five different groups. Pediatric surgeons and cardiologists then examined the groups and re-arranged them so that similar procedures would be contained in the same complexity group even if their mortality rates were somewhat dissimilar. This situation arose in particular for some procedures with low volumes of cases whose mortality rates were not necessarily representative of what they would have been with a large number of patients. The previous two steps were repeated a few times until the clinicians were comfortable with the resulting categorization, which consisted of four complexity categories with between seven and eighteen procedures in each category, and between 893 and 3453 cases in each category. These categories are presented in Table 1. Once the complexity categories were developed, a statistical model (stepwise logistic regression) was used to determine which patient risk factors available on the data form were significant independent predictors of in-

15

hospital mortality in addition to the complexity categories, and how those factors and the complexity categories should be weighted to obtain the best prediction of the probability of each patient dying in the hospital based on the patient’s risk factors. The model, published in Pediatrics w2x, is presented in Table 2. As indicated, the significant predictors of inhospital mortality were age, four comorbidities (severe cyanosis or hypoxia, arterial pH-7.25 pre-operative, significant extracardiac anomalies, and pulmonary hypertension), and the complexity categories. Patient age of less than 90 days was the highest risk, with odds of dying in the hospital that were 2.317 times the odds for patients who were more than 1 year of age (the reference category). Patients between 90 days and 1-year-old had odds of dying in the hospital that were 1.958 times the odds of patients who were more than 1 year of age. The odds of mortality for patients with each of the four comorbidities mentioned above relative to patients without that specific comorbidity ranged from 1.796 for pulmonary hypertension to 2.107 for arterial pH less than 7.25 pre-operatively. As expected, the risk of mortality rose with each successive complexity category relative to category I, the reference category. Patients in complexity categories II, III and IV had odds of mortality that were 2.898, 5.316 and 9.311 times as high, respectively, as the odds for patients in complexity category I. The statistical model predicted mortality quite well, with a C-statistic of 0.818 and a Hosmer–Lemeshow P-value of 0.37. 4. Investigating the volume–mortality relationship for pediatric cardiac surgery The statistical model just mentioned was also used to determine whether or not there were significant differences in risk-adjusted in-hospital mortality associated with hospital andyor surgeon volume. Different ranges of hospital and surgeon volume were inspected in order to find volume intervals with enough cases and roughly homogenous mortality rates within intervals. Ultimately, hospital volume and surgeon volume were each subdivided into two groups, with respective cut points for 1992–1995 annual surgeon volume and hospital volume of 75 and 100. The next step was to use the statistical model and the observed mortality results to calculate the observed, expected and risk-adjusted mortality rates for high- and low-volume surgeons, high- and low-volume hospitals, and the four intersections of those groups. For each group of interest, the observed mortality rate was simply the number of in-hospital deaths divided by the number of patients in the group. The expected mortality rate was obtained by summing the predicted probabilities of mortality from the statistical model for each of the patients in the group and then dividing by the number

16

D.R. Doran et al. / Progress in Pediatric Cardiology 18 (2003) 13–25

Table 1 Complexity categories for pediatric cardiac surgery Procedure Category I Patent ductus arteriosis closureywt.)1500 g Coarctation of aorta repair Blalock-hanlon septectomy Aortic valvotomy Vascular ring repair Atrial septal defect, secundum ASD(primumysinuous Ven.yPAPVCyw. MV repair) Cor triatrialum or supravalvular mitral stenosis Single VSD VSD and aortic Inc.y VSD wyTri. Valve Tetralogy of fallot without transannular patch Pulmonary valvotomy Coronary fistula closure Aort. Val. Repl. (Mech.yHet.y homografty auto.) Aortic stenosis (subvalvulary supravalvular) Cardiac arrhythmia surgery Category II Multiple VSD Tetralogy of fallot, with transannular patch Bidirectional glenn anastomosisy Bidir. Glenn Aortopulmonary window repair Aortic root repair Mitral valve repair Rastelli repairyintraventricular tunnel repair Category III Blalock-taussigy other shunts Complete atrioventricular canal defect Tet.of fallot (wyright Vent. to Pul. Cond.yPul. Atr.) Tetralogy of fallot, with other intracardiac Proc. Reconstruction of RV outflow tract, with shunt Reconstruction of RV outflow tract, wyo shunt Fontan operationy total cavo-pulm.derivation Ebstein’s malformation repair Aortoventriculoplasty Aortic valvotomy, open Mech. mitral valve replacementy Het. Repl. Arterial switchy art switch and other cardiac Proc. Other Op. for CHD with extra. Corp. Category IV Patent Ductus Arteriosis Closureywt.-1500 g Aortic arch anom repairy Int. Aortic arch repair Banding of pulmonary artery Watersonycentral shunts Pulmonary valvotomy, closed Other Ops. for CH disease wyo extra. Corp. Total anomolous pulmonary venous connection Truncus arteriosus repair Anomalous left coronary from pulmonary artery Aortic valve replacement (other) Other Op. for left ventricular outlet obstruction Hypoplastic LH of aortic atresia (norwood, other) Mitral valve Repl., Creat. or enlargement of ASD Mustardysenning repair-trans. of great arteries LV-PA conduit with or without other card Proc. Other procedure for TGA or DORV Septation(primary or staged) single vent Proc.

MR (%)

No. of cases

Vol. (%)

1.88 1.93 0.00 0.00 0.00 0.63 0.37 0.00 1.84 2.17 2.04 1.79 0.00 2.38 0.71 0.00 1.39

585 466 11 3 51 789 273 19 817 46 147 56 7 42 140 1 3453

8.16 6.50 0.15 0.04 0.71 11.01 3.81 0.27 11.40 0.64 2.04 0.78 0.10 0.59 1.95 0.00 48.15

5.26 5.29 3.33 4.76 0.00 2.7 5.71 4.48

19 435 210 21 29 74 105 893

0.27 6.07 2.93 0.29 0.04 1.03 1.46 12.09

9.11 10.10 13.00 8.16 12.9 11.25 13.51 11.11 7.14 7.50 13.51 11.72 12.46 10.97

428 287 100 98 31 80 185 9 14 40 37 256 313 1878

5.97 4.00 1.39 1.37 0.43 1.12 2.58 0.13 0.20 0.56 0.52 3.57 4.37 26.21

15.56 16.07 19.05 22.41 16.67 17.39 17.86 22.03 14.81 20.00 20.00 34.21 16.67 23.53 40.00 15.38 16.67

257 56 84 58 18 46 112 59 27 5 10 114 6 17 5 52 6

3.58 0.78 1.17 0.81 0.25 0.64 1.56 0.82 0.38 0.07 0.14 1.59 0.08 0.24 0.07 0.73 0.08

D.R. Doran et al. / Progress in Pediatric Cardiology 18 (2003) 13–25

17

Table 1 (Continued) Procedure

MR (%)

Other procedure for single ventricle

46.15 20.11

No. of cases 13 945

Vol. (%) 0.18 13.17

ASDsatrial septal defect. DORVsdouble outlet right ventricle. MVsmitral valve. VSDsventricular septal defect. TGAstransposition of great arteries.

of patients. The risk-adjusted mortality rate for each group was calculated by dividing its observed mortality rate by its expected mortality rate, and then multiplying this quotient by the overall mortality rate for the 1992– 1995 time period (6.75%). Findings from the study published in Pediatrics were that the respective risk-adjusted mortality rates for patients undergoing surgery in hospitals with 1992– 1995 annual volumes below 100 and hospitals with annual volumes of 100 or more were 8.26 and 5.95%, and these percentages were significantly higher and significantly lower, respectively, than the statewide rate of 6.75% w2x. Furthermore, the respective risk-adjusted mortality rates for patients undergoing surgery performed by surgeons with 1992–1995 annual volumes below 75 and with volumes of 75 or more were 8.77 and 5.90%, and these percentages were significantly higher and significantly lower, respectively, than the statewide rate. The worst results (8.94% mortality rate) occurred in patients who underwent surgery performed by low-volume surgeons (-75) in low-volume hospitals (-100). The best results (5.45% mortality rate) occurred in patients who underwent surgery performed Table 2 Multivariable risk factor equation for pediatric cardiac surgery inhospital deaths in New York State 1992–1995 Patient risk factor

Logistic regression Odds ratio

P-value

Demographic Age -90 days 90 days–1 year

2.317 1.958

-0.0001 -0.0001

Comorbidities Severe cyanosis or hypoxia Arterial pH-7.25 pre-op Sign. extracardiac anomalies Pulmonary hypertension

1.973 2.107 2.040 1.796

-0.0001 0.0009 -0.0001 -0.0001

Complexity category II III IV

2.898 5.316 9.311

-0.0001 -0.0001 -0.0001

by high-volume surgeons (G75) in high-volume hospitals (G100) w2x. Subsequent analyses were aimed at examining the hospital volume–mortality relationship and surgeon volume–mortality relationship for each of the complexity groups, as well as the tendency for high- and lowvolume providers (hospitals and surgeons) to have highrisk patients. Some of the conclusions of this part of the study were: 1. there was a slight tendency for low-volume hospitals (4-year volume -100) to perform less complex procedures, with the share for these hospitals dropping from 42% in the lowest complexity group to 36% in the highest complexity group, 2. there was a slighter tendency for low-volume surgeons (4-year volume -75) to perform less complex procedures, with the share for these surgeons dropping from 34% in the lowest complexity group to 26% in the second highest complexity group;, however, their share in the highest complexity group was 34%, 3. high-volume hospitals experienced lower risk-adjusted mortality rates than low-volume hospitals for cases in all four complexity groups. However, perhaps surprisingly, the differential in performance between high-volume hospitals and low-volume hospitals was largest in the lowest complexity group, 4. similarly, high-volume surgeons experienced lower risk-adjusted mortality rates than low-volume surgeons in all four complexity groups, but the differential in performance between high-volume surgeons and low-volume surgeons was largest in the lowest complexity group w2x. 5. Distributing benchmarking information to centers

Interceptsy4.8396. C statistics0.818. Hosmer-lemeshow statistics8.62 (Ps0.376).

In order to provide pediatric cardiac surgery programs with benchmarking and quality improvement tools, tables summarizing the characteristics of pediatric patient reports have been distributed to hospitals. These reports compare the facility’s frequency and mortality rate to the statewide experience for categories of patient characteristics, risk factors, procedural information, and complications in the system. In addition, to this information, rates for index procedures identified by the

18

D.R. Doran et al. / Progress in Pediatric Cardiology 18 (2003) 13–25

Pediatric Subcommittee have been tabulated in recent years and distributed. This subset of index procedures represents a variety of complexity groups and allows hospitals to review a ‘snapshot’ of how well they performed for relatively common procedures (see Appendix 3 for a complete list of index procedures). 6. Focused reviews As noted above, analyses of the data provided strong evidence of differences in outcomes among providers. When crude mortality rates, preliminary risk-adjusted analyses using complexity categories andyor extended periods of low volumes were identified, focused reviews were undertaken. Focused reviews are tailored to individual circumstances, but fall into three general categories: ● Alert Letters: When crude mortality rates in initial data submissions are substantially higher than the statewide average, a letter is sent to the program director. The director is asked to review the situation and provide feed back to the Department of Health. If ongoing monitoring indicates that the issue is resolved, no further action is required. ● Low Volume Reviews: Programs with volume below acceptable levels are required to submit an explanation and a plan to address both volume and quality of care issues. ● Site Visits: Blinded data reviews are conducted by an ad hoc group of CAC members and consultants. When risk-adjusted mortality rates are substantially higher than the statewide average, a site survey is typically recommended. Site visits are conducted by a team of practicing clinicians (surgeon, cardiologist, and nurse) with extensive experience in the provision of pediatric cardiac care and Department of Health staff. Data are reviewed by team members in preparation for the review together with background about the program. Surveys are designed by the team – but involve interviews with staff involved in the provision of care, review of hospital quality improvement activities, medical recordy interactive case reviews, a tour of the facility, and interviews with hospital leaders. A great deal of information is exchanged between surveyors and hospital staff during these reviews. Focused reviews conducted to date have indicated that, in some cases, changes occurring in medical personnel—particularly changes or departures of the surgeon performing most of the procedures—had resulted in a substantial, but temporary impact on a program. In other cases, more systematic issues were identified. In some cases, steps taken to deal with the problems were identified and implemented through a hospital’s internal review. In other cases strategies were recommended

through the site visit andyor the CAC review process. Examples of steps taken in recent years include: ● Use of an outside consultant to provide ongoing monitoring and input with regard to clinical issues. ● Use of an experienced surgeon from another center to provide services on a regular, but part time basis. ● Medical record reviews by outside expert consultants. ● Specific facility changes as recommended by the survey team. ● Suspension of services with corresponding arrangements made to ensure that access to a high quality of care for patients in the service area is maintained. 7. Future directions The review of clinically relevant, reliable data combined with focused program reviews has provided an important mechanism for improving quality of care in pediatric cardiac surgery. The extensive use of the data by facilities and the Cardiac Advisory Committee has also identified opportunities to improve our reporting system and consider our analytical approach for evaluating outcomes. The following is a description of the strategies currently underway to further enhance the effectiveness of our activities. 7.1. Revised data system Limitations identified through preliminary analyses of data have led to major revisions to the system. The need to update procedure and diagnosis codes to reflect new procedures led to major revisions in our coding system effective 1, January 2000. In addition to adding codes, the coding scheme was reorganized based on recommendations from the CAC’s pediatric cardiac services subcommittee. The new scheme follows the anatomical structure of the heart and great vessels to aid clinicians in using the most appropriate codes. In addition, it became clear that the combination codes for multiple procedures included in the existing system did not fully reflect the complexity of the disease process or procedural status in many pediatric patients. Therefore, the new system ‘unbundled’ the combined codes, and expanded coding to allow five cardiac diagnosis codes and four procedure codes for each visit to the operating room (see Appendix 2A and Appendix 2B). 7.2. Revisiting analytical approaches As mentioned earlier, the reason for defining complexity groups for the purpose of predicting mortality and testing volume–mortality relationships was that the extremely large number of pediatric cardiac procedures that are performed (54 combinations) along with the

D.R. Doran et al. / Progress in Pediatric Cardiology 18 (2003) 13–25

relatively low volumes of these procedures (7169 in the 1992–1995 time period) makes it very difficult to look at each procedure separately. However, grouping the procedures into complexity categories is problematic for a couple of reasons. An obvious drawback is that every procedure in a given complexity group is treated as having exactly the same risk in the prediction of mortality using the statistical model in Table 1. To the extent that this is untrue, the model is inaccurate, and this can provide either an unfair benefit or an unfair disadvantage to individual providers. For example, in complexity category IV based on 1992–1995 data, the mortality rates for individual procedures range from 14.81% for ‘anomalous left coronary from pulmonary artery’ to 46.15% for ‘other procedure for single ventricle.’ Even in the lowest risk category (complexity category I), the range of mortality rates is higher than it would ideally be from 0.00% for six different procedures to 2.38% for aortic valve replacement. To gain a better understanding of how these procedures are distributed among individual hospitals, the complexity category IV cases were divided into three ‘risk’ groups based on the overall mortality rate for the individual procedures. The ‘low risk’ group has a mortality rate less than 16.00%, the ‘middle risk’ group has mortality rates between 16.00 and 22.99%, and the ‘high risk’ group has mortality rates over 23.00%. Of the 16 hospitals that performed pediatric cardiac surgery between 1992 and 1995, seven of these centers had over 50% (55.56–83.33%) of their complexity category IV cases classified as ‘low risk’. Since these procedures are not well-distributed among individual hospitals, a large bias is introduced in the process of evaluating hospital performance. Another drawback of the complexity category approach is that if it is revisited from year to year, or even in three year periods, the process for identifying complexity categories would yield quite different categories. Although this is not troubling if changing mortality rates are a reflection of advances in pediatric cardiac surgery that result in truly lower rates as a result of those advances, it appears that the variations in mortality rates for several procedures, many of which experience a rise in mortality, is more a reflection of random variability and low volumes for individual procedures. Thus, the assignment of complexity categories seems to be a very arbitrary process for many procedures. 7.3. Alternatives to the complexity group approach It is clear that pediatric cardiac surgery patients have considerably different mortality rates as a function of the procedures they undergo and the diagnoses they have related to the need for surgery. For example, the procedural mortality rates in the 1992–1995 time period

19

ranged from 0.00% for seven different procedures to 46.15% for ‘other procedure for single ventricle.’ Consequently, it would not suffice for purposes of predicting in-hospital mortality to simply ignore what procedure the patient underwent and risk-adjust on the basis of patient demographics and comorbidities as is largely the case in the risk-adjustment process for CABG surgery in adults. Essentially the only other procedure-based options are to treat each of the procedures uniquely by either developing a separate statistical model for each procedure, or by allowing the individual procedures to be variables in a single model that predicts mortality on the basis of the individual procedures along with patient demographics and comorbidities. The first option is out of the question because there are extremely low volumes for some procedures, not nearly enough to support a statistical model, which requires roughly twice the number of adverse events as the number of variables in the model when there are binary outcomes. This leaves the possibility of using each of the procedures as binary (performed, not performed) variables in the model along with patient demographics and comorbidities. Another similar approach is to use individual patient diagnoses that are related to the need for surgery as binary variables in the model. When individual procedures or diagnoses are used as binary variables, a subset of them is aggregated to create a reference group. The reference group is created so that the relative risk of the individual procedures or diagnoses can determined. The selection criterion for the reference group has been found to be relatively stable and includes all diagnoses or procedures that have a mortality rate less than 2.00%. Using either the individual procedure approach or individual diagnosis approach to risk-adjusted outcomes yields quite similar results when using the New York database that was in place prior to the year 2000. This is to be expected because, in that coding system, there was essentially a one-to-one relationship between the diagnoses relating to the need for surgery and the procedure performed. However, the expanded 2000 reporting system allows multiple diagnoses and procedure codes to be reported. This enables coding of different procedures for a given diagnosis. As such, we do not anticipate that the results using a procedurebased approach and the results using a diagnosis-based approach will be as close in future analyses. The Cardiac Advisory Committee has reviewed the alternate approaches and discussed the relative strengths of each approach. They have determined that both the diagnosis and the procedure-based approaches should continue to be analyzed and compared. They have also recommended that the diagnosis-based method be used for the analysis and release of hospital specific riskadjusted mortality rates because it is reflective of the patient’s condition rather than a choice on the part of

20

D.R. Doran et al. / Progress in Pediatric Cardiology 18 (2003) 13–25

the surgeon as to how the patient should be treated. It is expected that the diagnosis-based methodology will soon be released as part of a public report along with assessments of hospital performance.

Appendix 1: A. Primary cardiac diagnosis codes, 1991–1999

Patent ductus arteriosus. 8. Discussionyconclusions The New York State Department of Health has made some significant strides in assuring that the pediatric cardiac surgery performed in New York is of the highest possible quality. A critical role in this effort has been played by the Department’s Cardiac Advisory Committee, which has provided a forum in which providers, regulators and researchers can collaborate to explore and implement innovative quality improvement strategies. Some of the most important initiatives in this ongoing effort include: ● The development and refinement of a database that contains important information about each patient undergoing pediatric cardiac surgery in the State, including demographics, procedures performed, cardiac diagnoses, and comorbidities. ● The assurance that the data in this database is accurate and complete through the implementation of annual medical record audits. ● The use of this clinical database to explore volumeoutcome relationships for pediatric cardiac surgery and to feed back this information to providers and the public. ● The use of this database to identify significant independent risk factors for in-hospital mortality of pediatric cardiac surgery patients, the adjustment of hospital mortality rates based on differences in the prevalence of these risk factors among hospitals in the State, and the reporting to each hospital of its baseline information. ● The use of annual hospital pediatric cardiac surgery volumes and results from outcomes analyses to trigger focused reviews consisting of alert letters, requirements for hospitals to address low patient volumes, andyor site visits conducted by the CAC. ● The continued development and improvement of methods to predict adverse outcomes and subsequently assess the relative quality of care of hospitals in the. ● State in which pediatric cardiac surgery is performed. ● The preparation of the first pediatric cardiac surgery report containing hospital risk-adjusted mortality rates to be released to public (in 2003). The Department of Health and its Cardiac Advisory Committee are committed to exploring and implementing initiatives aimed at assuring the best possible pediatric cardiac surgical care in New York State hospitals.

Atrial septal defect, secundum. Atrial septal defect, primum. Sinus venosus defect. Other partial anomalous pulmonary venous connection (PAPVC). Atrial septal defect and PAPVC. Atrial septal defect and mitral valve anomaly. Total anomalous pulmonary venous connection. To the left innominate vein. To the superior vena cava. To the coronary sinus. To the right atrium. To the infradiaphragmatic vein. Mixed Cor triatrialum. Other supravalvular mitral stenosis. Complete atrioventricular canal defect. Paramembranous ventricular septal defect. Subpulmonary ventricular septal defect. Other single ventricular septal defect. Multiple ventricular septal defects. Ventricular septal defect with aortic incompetence. Ventricular septal defect with straddling or overriding tricuspid valve. Aneurysm of sinus valsalva. Aortic-ventricular tunnel. Tetralogy Tetralogy Tetralogy Tetralogy

of of of of

fallot. fallot with pulmonary atresia. fallot with absent pulmonary valve. fallot with other major cardiac defect.

Pulmonary valve stenosis. Pulmonary atresia with intact ventricular septum. Tricuspid atresia. Ebstein’s malformation. Truncus arteriosus. Origin of left or right pulmonary artery from aorta. Aortopulmonary window. Coronary fistula. Anomalous origin of left coronary from pulmonary artery. Aortic stenosis. Valvular. Discrete subvalvular, localized. Discrete subvalvular, tunnel type. Supravalvular.

D.R. Doran et al. / Progress in Pediatric Cardiology 18 (2003) 13–25

Coarctation of aorta. Coarctation of aorta with ventricular septal defect. Coarctation of aorta with other cardiac defect. Interrupted aortic arch. Vascular rings and slings. Hypoplastic left heart or aortic atresia. Congenital mitral valve disease. Simple transposition of the great arteries. Transposition of the great arteries and ventricular septal defect. Transposition of the great arteries, ventricular septal defect, and pulmonary stenosis. Other transposition of the great arteries. Double outlet right ventricle (DORV). With subaortic ventricular septal defect. With subpulmonary ventricular septal defect. Other DORV. Congenitally corrected transposition of the great arteries. Single ventricle. Other anomalies of atrial situs. Cardiac arrhythmia. Other congenital heart defect, including all cases with abnormal situs of the atria. B. Primary cardiac procedure codes, 1991–1999 Patent ductus arteriosus closure. Coarctation of aorta repair. Aortic arch anomalies repair. Banding of pulmonary artery. Blalock-hanlon septectomy. Blalock-taussig shunt (Classical or Modified). Waterson shunt. Central shunt. Other shunt. Pulmonary valvotomy (with Inflow Occlusion). Aortic valvotomy (with Inflow Occlusion). Vascular ring repair. Other operations for congenital heart disease, performed without extracorporeal circulation. Atrial septal defect, secundum. Atrial septal defect, primum. Atrial septal defect, sinosus venosus. Partial anomalous pulmonary venous connection (PAPVC). Atrial septal defect and PAPVC. Atrial septal defect and mitral valve repair.

21

Total anomalous pulmonary venous connection. To the left innominate vein. To the superior vena cava. To the coronary sinus. To the right atrium. To the infradiaphragmatic vein. Mixed. Cor triatrialum or supravalvular mitral stenosis. Complete atrioventricular canal defect. Single ventricular septal defect. Multiple ventricular septal defect. Ventricular septal defect closure and aortic incompetence repair. Ventricular septal defect closure with straddling or overriding tricuspid valve. Tetralogy of fallot, with transannular patch. Tetralogy of fallot, without transannular patch. Tetralogy of fallot, with right ventricle to pulmonary artery conduit. Tetralogy of fallot, with pulmonary atresia. Any of the above, with other intracardiac procedure. Pulmonary Valvotomy. Reconstruction of RV outflow tract, with aortopulmonary shunt. Reconstruction of RV outflow tract, without aortopulmonary shunt. Fontan (or modified) operation. Bidirectional glenn anastomosis. Total cavo-pulmonary derivation. Ebstein’s malformation repair. Truncus arteriosus repair. Aortopulmonary window repair. Coronary fistula closure. Anomalous left coronary from pulmonary artery. Aortic valve replacement. Mechanical. Heterograft. Homograft. Autograft. Other. Aortic root replacement. Mechanical. Homograft. Autograft. Other. Other operation for left ventricular outlet obstruction. Apical aortic conduit.

22

D.R. Doran et al. / Progress in Pediatric Cardiology 18 (2003) 13–25

Aortoventriculoplasty (konno, pulmonary autograft). Aortic valvotomy. Aortic stenosis, subvalvular; resection or enucleation. Aortic stenosis, supravalvular. Interrupted aortic arch repair. Hypoplastic left heart or aortic atresia. Norwood. Other. Mitral valve repair.

B.3. Anomalies of pulmonary veins Partial anomalous return. Total anomalous return. Supracardiac. Cardiac. Infracardiac. Mixed. Pulmonary vein stenosis. Cor triatrialum. B.4. Anomalies of atrial septum

Mitral valve replacement. Mechanical. Heterograft. Other. Creation or enlargement of atrial septal defect.

Secundum atrial septum defect. Single atrium. Unroofed coronary sinus. Sinus venosus atrial septum defect. Patent foramen ovale. B.5. Anomalies of atrioventricular (AV) valve(s)

Transposition of great arteries (TGA) or double outlet right ventricle (DORV). Mustard repair. Senning repair. Arterial switch. Arterial switch and VSD closure. Arterial switch and other cardiac procedure. Rastelli repair. Intraventricular tunnel repair. LV-PA conduit, with or without other cardiac procedure. Other procedure for TGA or DORV. Single ventricle procedures. Fontan (or modified). Bidirectional glenn. Septation (primary or staged). Other procedure for single ventricle. Cardiac arrhythmia surgery. Other operation for congenital heart disease, performed with extracorporeal circulation. Appendix 2: A. Primary cardiac diagnosis codes, 2000–Present B.1. Atrial situs anomalies Situs Inversus Situs ambiguousyheterotaxy Syndrome B.2. Cardiac position anomalies Dextrocardia. Mesocardia. Ectopia cordis.

Tricuspid valve. Ebstein’s anomaly. Tricuspid Stenosis. Tricuspid regurgitation. Straddling tricuspid valve. Mitral valve. Supravalvular mitral stenosis. Valvular mitral stenosis. Subvalvular mitral stenosis. Mitral regurgitation Straddling mitral valve. Papillary muscle abnormality. Common AV valve abnormality. Stenosis. Regurgitation. Malaligned. B.6. Anomalies of ventricular septum Perimembranous ventricular septal defect. Doubly committed ventricular septal defect (subarterial). Inlet ventricular septal defect. Muscular ventricular septal defect. Multiple ventricular septal defects. B.7. Atrioventricular septal defects (AVSD) Partial AVSD (primum atrial septal defect). Complete AVSD. Balanced. Unbalanced. B.8. Univentricular heart (single ventricle) Doubleycommon inlet left ventricle. Doubleycommon inlet right ventricle. Tricuspid atresia. With inlet ventricular septum.

D.R. Doran et al. / Progress in Pediatric Cardiology 18 (2003) 13–25

With ventricular septum defect. With transposition of great arteries. Mitral atresia. Indeterminate ventricle. Hypoplastic right ventricle. Pulmonary atresia with IVS. Other type of hypoplastic right ventricle. Hypoplastic left ventricle. Classical hypoplastic left heart syndrome. (Aortic atresia wy hypoplastic left ventricle). Any other hypoplastic left ventricle. B.9. Anomalies of ventricular outflow tracts Pulmonary ventricular outflow tract. Pulmonary valve stenosis. Subvalvularyinfundibular pulmonary stenosis. Double chamber right ventricle. Branch pulmonary artery stenosis. Hypoplastic pulmonary arteries. Pulmonary valve regurgitation. Main pulmonary artery atresia. Branch pulmonary artery atresia. Aortic ventricle outflow tract. Valvular aortic stenosis. Subvalvular aortic stenosis. Discrete. Long segmentytunnel. Supravalvular aortic stenosis. Aortic valve atresia. Aortic valve regurgitation. Aorto-ventricular tunnel. B.10. Tetralogy of fallot (TOF) Right ventricle-pulmonary artery continuity. TOF with pulmonary valve atresia. Absent pulmonary valve syndrome. B.11. Truncus arteriosus Type I. Type II. Type III. B.12. Transposition of the great arteries (TGA)

B.14. Great vessel anomalies Aortopulmonary window. Patent ductus arteriosus. Origin of leftyright PA from aorta. Sinus of valsalva aneurysmyfistula. Aortic coarctation. Aortic interruption. Aortic aneurysm. Ascending. Descending. Transverse. Vascular ring. Origin of LPA from RPA (PA sling). Discontinuous pas. Bronchial PA blood flow (MAPCA). Isolated LSVC. Bilateral SVCs. Azygousyhemiazygous continuous IVC. Other great vessel anomalies. B.15. Coronary artery anomalies Coronary artery fistula. Coronary artery sinusoids. Coronary artery stenosis. Coronary artery aneurysm. Anomalous origin coronary artery. Atresia left main coronary artery. Atresia right main coronary artery. B.16. Cardiac rhythm anomalies Supraventricular tachycardia. Ventricular tachycardia. Sinus bradycardia. Heart block. B.17. Cardiomyopathies Hypertrophic. Left ventricle. Right ventricle. Dilated. B.18. Acquired disease

D-TGA. Congenitally corrected transposition.

Kawasaki’s disease. Endocarditis. Myocarditis. Traumatic.

B.13. Double outlet right ventricle (DORV)

B.19. Organ failure

Subaortic ventricular septal defect. Subpulmonic ventricular septal defect. Uncommitted ventricular septal defect. Doubly committed ventricular septal defect. Restrictive ventricular septal defect.

Cardiac. Pulmonary. B.20. Cardiac neoplasms Atrial.

23

24

D.R. Doran et al. / Progress in Pediatric Cardiology 18 (2003) 13–25

Ventricular. Valvular. Great vessel. B. Primary cardiac procedure codes, 2000–present

Repair of complete AV canal. Repair of partial AV canal.

Prosthetic. Non-valved. Transannular patch. With monocusp valve. Without monocusp valve. Repair branch PS. Aortic ventricular outflow tract. Aortic valvuloplasty. Aortic valvotomy. Repair supravalvular AS. Resection of discrete subvalvular AS. Aortoventriculoplasty (Konno Procedure). Aortic valve replacement. Autograft. Homograft. Prosthetic. Aortic root replacement. Autograft. Homograft. Prosthetic. Left ventricular apex to aorta conduit.

B.24. Anomalies of ventricular septum

B.27. Tetralogy of fallot

B.21. Anomalies of pulmonary veins Repair of anomalous pulmonary venous return. Repair of pulmonary vein stenosis. Repair of partial anomalous pulmonary venous return. B.22. Anomalies of atrial septum Atrial septal defect closure. Creation of atrial septal defect. Repair of cor triatrialum. Patent foramen ovale closure. B.23. Atrioventricular septal defect (AVSD)

Repair of ventricular septal defect. Creationyenlargement of ventricular septal defect. Fenestration of ventricular septal defect patch. B.25. Anomalies of atrioventricular (AV) valves Tricuspid valve. Repair (Non-Ebstein’s Valve). Replacement. Homograft. Prosthetic. Tricuspid valve closure. Repair ebstein’s anomaly. Mitral valve. Resect supramitral ring. Repair (including annuloplasty). Replacement. Homograft. Prosthetic. Common AV valve repair. B.26. Anomalies of ventricular outflow tract(s) Pulmonary ventricular outflow tract. Pulmonary valvotomyyvalvectomy. Resection of subvalvular PS. Repair of supravalvular PS. Pulmonary Valve Replacement. Homograft. Prosthetic. Pulmonary outflow conduit. Valved. Homograft.

Repair with pulmonary valvotomy. Repair with transannular patch. Repair with non-valved conduit. Repair with valved conduit. Homograft. Prosthetic. Repair with reductionyplasty of Pas. Repair with pulmonary valve replacement. Homograft. Prosthetic. B.28. Truncus arteriosus Repair with non-valved conduit. Repair with valved conduit. Homograft. Prosthetic. B.29. Univentricular heart (single ventricle) Fontan Operations Direct RA-PA connection. Total cavopulmonary connection. Lateral tunnel – non-fenestrated. Lateral tunnel – fenestrated. Extracardiac – non-fenestrated. Extracardiac – fenestrated. Septation of single ventricle. Hypoplastic right ventricle. Valved. Homograft. Prosthetic. Non-valved.

D.R. Doran et al. / Progress in Pediatric Cardiology 18 (2003) 13–25

Transannular Patch.

B.33. Cardiomyopathies

With monocusp valve. Without monocusp valve. Hypoplastic left ventricle.

Left ventricular reduction (Batista). Radical myomectomy.

Norwood. Damus Kaye Stansel (DSK). B.30. Transposition of great arteries or double outlet RV Arterial switch. Senning procedure. Mustard procedure. Intraventricular repair of DORV. Rastelli procedure. RV-PA conduit. Valved. Homograft. Prosthetic. Non-valved. REV operation (Modified Rastelli). LV-PA conduit. Valved. Homograft. Prosthetic. Non-valved.

B.34. Interval procedures Pulmonary artery band. Unifocalization of pulmonary vessels. Shunts Central aortopulmonary shunt. Blalock taussig shunts. Classical. Modified. Glenn shunts. Unidirectional (Classical). Bidirectional. Bilateral bidirectional. Cardiac arrhythmia surgery. Other operations for congenital heart disease. Appendix 3: Index procedures for pediatric cardiac surgery, NYS

Index procedure

Complexity category

Single ventricular septal defect (G90 days) Bidirectional glenn anastomosis (G90 days) Tetralogy of fallot, wytransannular patch (G90 days) Complete atrioventricular canal defects (G90 days) Fontan operationsytotal cavopulmonary derivation (G90 days) Isolated arterial switch (-90 days) Norwood (-90 days) Total anomalous venous connection (-90 days)

1 2 2 3

B.31. Great vessel anomalies Patent ductus arteriosus ligation. Repair aortopulmonary window. Reimplantation of left or right pulmonary artery. Repair sinus of valsalva aneurysm. Aortic repair (Coarctation or Interruption). End to end anastomosis. Subclavian flap angioplasty. Onlay patch. Interposition graft. Vascular ring division. Repair of PA sling. Reimplantation of innominate artery. Aortoplexy. B.32. Coronary artery anomalies Translocation of left coronary artery to aorta. Direct. Transpulmonary tunnel (Takeuchi). Coronary artery ligation. Coronary fistula ligation.

25

3 3 4 4

References w1x Jenkins KJ, Newburger JW, Lock JE, Davis RB, Coffman GA, Iezzoni LI. In-hospital mortality for surgical repair of congenital heart defects: preliminary observations of variation by hospital caseload. Pediatrics 1995;95:323 –330. w2x Hannan EL, Racz M, Kavey R-E, Quaegebeur JM, Williams R. Pediatric cardiac surgery: the effect of hospital and surgeon volume on in-hospital mortality. Pediatrics 1998;101(6):963 – 969. w3x Chang R-KR, Klitzner TS. Can regionalization decrease the number of deaths for children who undergo cardiac surgery? a theoretical analysis. Pediatrics 2002;109(2):173 –181.