Patient Factors Associated With Strut Fracture in Björk-Shiley 60° Convexo-Concave Heart Valves
Background Previously established predictors of outlet strut fracture in Björk-Shiley convexo-concave (CC) valves include larger valve size, larger opening angle (70° versus 60°), younger age at implant, and date of manufacture. We sought to identify patient characteristics that might be predictive of strut fracture and to refine the estimates associated with previously identified predictors.
Methods and Results We conducted a case-control study of CC60° valves implanted in the United States and Canada and manufactured between January 1, 1979, and March 31, 1984. Cases included all valves with verified outlet strut fractures reported to the manufacturer from January 1979 through January 1992. Up to 10 controls were selected for each case. Control valves were matched according to implanting surgeon and were required to have been functioning at least as long as their matched case valves. Case and control medical records were reviewed for information on patient medical history before the valve implant. There were 96 case and 634 control valves for which clinical data were available. Patient age and valve size and implant position were confirmed as important determinants of fracture. There was a strong inverse gradient of risk with age. The risk of fracture was 42% lower for each 10-year increment of patient age at time of implant. Large mitral valves were at greatest risk of strut fracture, with the largest mitral valves (33 mm) estimated to be 33 times more likely to fracture than the smallest (21 to 25 mm) aortic valves. Date of manufacture was also associated with risk; valves welded from mid-1981 through March 1984 were more likely to fracture than those manufactured in 1979 and 1980. Body surface area <1.5 m2 was associated with 1/16 the risk of body surface area ≥2.0 m2. No other patient factor was strongly associated with the risk of strut fracture.
Conclusions Few patient features identifiable in the implant record are predictive of strut fracture. Our analysis supports previous work in identifying valve size, patient age, and date of manufacture as predictors of fracture and adds body surface area. A number of these associations suggest that conditions associated with higher cardiac output may also place patients at increased risk.
The development of the Björk-Shiley 60° convexo-concave valve (CC60°) (see the Figure⇓) began in the mid-1970s as an improvement over earlier tilting-disk valve designs. Clinical trials were initiated in 1976, with major production starting in 1979. The manufacturer, Shiley Inc, estimates that more than 81 000 valves have been implanted worldwide. The outlet strut of some of these valves has fractured, allowing the disk to escape and leading to acute valve failure, often resulting in death. The first reported fracture occurred in 1978, and 360 fractures had been reported worldwide by February 1992, when this study began. The CC60° valve was voluntarily withdrawn from the market in 1986, but at present ≈44 000 persons still have this valve.
The annual incidence of fracture reports varies from 0.02% for small valves (21 to 27 mm) to 0.62% for large mitral valves (29 to 33 mm).1 With allowances for presumed rates of underreporting of valve failure, estimates of the annual fracture incidence range from 0.02% to 2.52%, depending on valve size, position, and period of manufacture (R. Brookmeyer, unpublished data, 1993).2
Identification of additional risk factors and confirmation of existing factors associated with increased risk of fracture could enhance the ability of clinicians and valve recipients to determine the need for intensified clinical surveillance, evaluation with high-definition radiography,3 4 acoustic characterization of strut status (R. Brookmeyer, N. Weiss, Y. Yasui, E. Blackstone, unpublished data, 1993), or prophylactic explant. To date, information on risk factors has come predominantly from relatively small cohort studies of patients with CC60° valves and from studies in which the numbers of fracture reports were compared with at-risk populations whose sizes were estimated by actuarial techniques.
Large valve size (29 to 33 mm) has been consistently identified as a major risk factor (R. Brookmeyer, unpublished data, 1992).5 6 7 Two studies have found position to have a supplementary effect, with mitral valves in general carrying a greater risk than aortic valves of the same diameter (R. Brookmeyer, unpublished data, 1992, and Reference 5). Valves implanted in patients <50 years old have been reported to fail more often than valves implanted at later ages.6 7 Valves manufactured in 1981 and the first half of 1982 have been found to be at greater risk than valves manufactured either before or after (R. Brookmeyer, unpublished data, 1993).1 No valve manufactured after March 1984 has been reported to have failed.
We present the results of a case-control study in which we refine previous estimates of the relation between the risk of fracture and age, size, position, and date of manufacture and evaluate the role of new factors obtained from a detailed review of patient data taken from surgeons’ records created around the time of valve implant. Additional information on the manufacturing process was also collected and will be discussed elsewhere.
The source population for this study consisted of all CC60° valves that were implanted in the mitral or aortic position in patients in the United States and Canada, were manufactured between January 1, 1979, and March 31, 1984, and for which the implanting surgeon or hospital submitted implant cards to the manufacturer before the fracture date (if any). This manufacturing period was selected because very few CC60° valves were manufactured before 1979 and no fractures have been reported in valves manufactured after March 1984. The implant card contains information on valve serial number, patient name, surgeon name, date of implant, and hospital. These cards were packaged with the valves and were completed by hospital personnel at the time of the implantation and returned to the manufacturer. Approximately 33 000 CC60° valves are believed to have been implanted in the United States and Canada, and approximately 28 000 implant cards were returned.
The cases were limited to the 186 valves in the source population (see above) in which a strut fracture was reported to the manufacturer before February 6, 1992. These reports came through communications from physicians, patients, patient families, regulatory agencies, and in the course of product liability litigation. A total of 160 potential cases were identified from the 186 reports; the other 26 were ineligible for inclusion as cases because no implantation data card had been returned. For all case valves, we sought documentation of the occurrence of strut fracture, including at least one of the following: examination of the explanted valve by the manufacturer, visual confirmation of the fracture by the explanting surgeon or pathologist, or confirmation based on an imaging technique. As a result of the verification procedure, four potential case valves were dropped because they either were misclassified or had insufficient documentation.
During the subsequent data collection, additional information became available on date of manufacture, valve type, and other factors related to eligibility criteria. This resulted in the elimination of six additional potential case valves. The final condition for study inclusion was that the implanting surgeon provide access to the medical and surgical records for the implant surgery for each valve. A further 54 cases were eliminated because the surgeons who implanted those valves did not participate or could not make office records available.
An independent review of Food and Drug Administration (FDA) reports of strut fractures in CC valves suggests that a very high percentage of the known cases were identified by our case ascertainment procedures. Because of incomplete valve identifiers in the data reported to the FDA, it is not possible to perfectly match company and FDA records or to eliminate all duplicate reports. A maximum of 27 fractured valves were reported to the FDA but not to Shiley; essentially all fractures that the FDA list indicated were reported to Shiley were included in our initial case list.
Up to 10 potential control valves were randomly selected for each case valve from the source population. Potential control valves were implanted by the same surgeon who implanted the case valve and were required to have survived intact in the recipient for at least as long as the case valve functioned in its recipient. For 183 controls, the available follow-up data proved to be ambiguous as to the survival of the control for the requisite interval. These patients were included, but with appropriate statistical adjustment for their uncertain survival status (see “Statistical Procedures”).
Patient characteristics at the time of implant were obtained from the records of the implanting surgeons. The surgeons were identified from the implant cards and were asked to participate in the study either by submitting photocopies of specified documents (cardiologist’s referral letter, admission note, operative report, discharge note or letter to referring cardiologist, catheterization reports, discharge ECGs, or radiograph reports) or by permitting record abstraction in their offices. When the implanting surgeon had retired, died, or left the institution at which the implant was performed for other reasons, we attempted to trace the surgeon’s records through colleagues, chiefs of service, forwarding addresses, and specialty directories. Data were collected on patient demographics, medical history, indication for implant, implant position, preimplant and postimplant hemodynamics, and surgical/recovery complications. Although postimplant values for blood pressure and heart rate were of particular interest, we did not think that the data recorded in the hospital record gave satisfactory information on these factors, and we omitted them. Abstracted data were regularly reabstracted during review, and an interactive data entry program was designed that included range and logic checks.
The study valves (cases plus controls) implanted by each surgeon were assigned sequential numbers according to implant date. This temporal ranking was used to evaluate the risks associated with the surgeons’ implant experience.
We obtained the weld date for each valve and confirmed the valve size from the manufacturer.
All patient characteristics were initially examined by review of the number of cases and controls present within strata of each recorded characteristic. Factors were discarded from further analysis at this level if they had very large amounts of missing information (eg, cardiac catheterization data), if there were fewer than five cases possessing the key characteristic even after collapsing of adjacent categories, or if they had neither an a priori expectation of an influence on risk nor any hint of an empirical association in case-control tabulations. Remaining factors were examined in univariate and multivariate conditional logistic regression analyses, taking the matching factors into account. Terms that could be expressed as continuous measurements were entered into the model both as categorical variables and as simple linear functions. When the categorical variable was preferable, the test statistic for the linear term was used as a test of trend. We have adjusted for all terms in the final model in the presentation of results for characteristics that did not enter into that model.
For the 183 controls with uncertain vital status on the index date, we calculated a survival probability, which was incorporated into the regression analysis as an “offset” term. The survival probability at the index date was calculated conditionally on known dates of surviving and being dead (which in these cases bracketed the index date), using published survival statistics for persons with implanted mechanical heart valves.6 We also performed all the final analyses presented here with simple exclusion of these controls; in no case did the results differ by an amount that would change the qualitative inferences from the study.
Valve size and position and age at implant were the major determinants of valve fracture (Table 1⇓). With the 29-mm mitral valves taken as the reference group, the valve-specific fracture rate ratio estimates ranged from 0.10 (corresponding to a 10-fold lower risk) in small aortic valves to a 3.3-fold elevation in risk for 33-mm mitral valves. For age, the estimated decline in the incidence of fracture was 42% for each incremental decade of age at implant (expressed in Table 1⇓ as a rate ratio of 0.58). This change corresponds to a 9-fold higher risk (1/0.584) in a patient who received an implant at age 25 years compared with a person who received one at age 65 years. Valves manufactured in 1979 and 1980, as indicated by the weld date, were at lower risk than those manufactured later.
There was a strong gradient in risk with body surface area (BSA); valves implanted in larger people (BSA ≥2.0 m2) were at 16 times greater risk than those implanted in smaller people (BSA <1.5 m2). Unfortunately, only a minority of the clinical records presented information on both height and weight, so that the apparent effect of BSA on risk, although strong, has considerable imprecision. Persons with unrecorded height and weight had a relative risk estimate of 0.27, intermediate between the lowest- and highest-risk groups. To test whether the effect of BSA was dependent on implant position, we explored interaction terms between BSA and position in the regression model. There was no interaction: the risk gradient for BSA was essentially the same for aortic and mitral valves.
Two patient characteristics had marginal associations with strut fracture (Table 1⇑). “Ischemia” was used to denote subjects with a medical record notation of ischemic disease, angina, papillary muscle dysfunction, or coronary artery disease. Patients with angina alone in the presence of aortic stenosis were not considered to have ischemia. Only 9% of the fractures occurred in patients with ischemia. The estimated fracture incidence ratio of 0.49 for ischemia was associated with a range of diminished risks (95% CI, 0.20 to 1.2). An indication of history of endocarditis marked valves that carried a 2.3-fold elevation in fracture risk. The effect estimates were similar for the subgroup of patients with active endocarditis defined as endocarditis present at or within 3 months of implant. Although the association between these factors and strut fracture was weaker than that between BSA and strut fracture, these terms were retained in the model to control for their possible confounding effects.
There were other patient characteristics for which it was possible to reliably abstract data from the medical record and which corresponded to plausible mechanisms that might have resulted in greater or lesser risk of valve failure. These included indications for valve implant (grouped as prior prosthetic failure, rheumatic heart disease, and congenital heart disease), left ventricular dysfunction, history of pulmonary hypertension or arrhythmia, race, and sex. Table 2⇓ contains summary information for these factors; none were associated with fracture risk. Case and control valves were matched according to surgeon but may have been implanted over a considerable time period, reflecting different levels of surgeon experience in placing CC60° valves. When valves were ranked according to chronological implant sequence within surgeon, there was no relation to fracture risk.
Following Brookmeyer (unpublished data, 1993), we compared predictors in the first 4 years after implant with those of subsequent years. The results were similar in the two time periods. The data are too sparse to support separate, time-specific investigations of all the risk factors identified in the main analysis.
On the basis of this analysis of 96 cases and 634 controls, we determined that young patients and large patients were at higher risk for strut fracture than were older and smaller patients. Fracture risk also varied dramatically with valve size and the date of manufacture.
Previous reports on the correlates of strut fracture in CC60° valves include a cohort analysis of outcomes in 2303 CC70° and CC60° valve recipients in the Netherlands and two analyses that combined fracture data reported to the manufacturer with data on the types and numbers of valves manufactured or implanted. Van der Graaf et al7 collated data from the medical records of Dutch valve recipients with follow-up information from a variety of sources to identify predictors of fracture for 42 valves. Grunkemeier et al6 incorporated terms for age, sex, valve size and position, and a surrogate variable representing weld date into detailed survival estimates from nearly 8000 patients in three clinics. They then estimated the survival of patients for whom the manufacturer’s data (the implant cards) contained all these elements. From the reported fractures with complete data in this group, they calculated the effects of age, sex, and valve characteristics on the risk of reported fracture. There were 46 failures in the group of valves with complete risk factor information, making the Grunkemeier study roughly comparable to the Dutch cohort in its statistical power and smaller than the present study.
Brookmeyer applied published survival figures for mitral and aortic valve recipients to the cohort to derive expected numbers of person-years at risk, which then served as the setting for describing the patterns of failure in 328 fractures of CC60° valves reported to the manufacturer (R. Brookmeyer, 1992 and 1993, and R. Brookmeyer, N. Weiss, Y. Yasui, E. Blackstone, unpublished data, 1993).
Large valves (≥29 mm) have consistently been found to be at increased risk, and when numbers have permitted analysis within this subgroup, the estimated risk has increased as valve size increased from 29 to 33 mm (R. Brookmeyer, N. Weiss, Y. Yasui, E. Blackstone, unpublished data, 1993).6 7 This is of particular interest because the designation of 29, 31, and 33 mm denotes the size of the sewing ring. The sizes of the disk and flange are the same for these three valve sizes. Implant position has also been identified as a predictor with mitral valves at greater risk (R. Brookmeyer, N. Weiss, Y. Yasui, E. Blackstone, unpublished data, 1993).7 Even though there were too few valves in some categories to allow for separate risk estimates for valve size and implant position, the result using a combination of these variables is compatible with information from other studies. The higher-risk manufacturing periods identified here also include those identified in the Brookmeyer et al study (R. Brookmeyer, N. Weiss, Y. Yasui, E. Blackstone, unpublished data, 1993), as reported by Schöndube et al,2 although our risk period was much broader and suggested a smaller differential in risk between the lower- and higher-risk periods. The results of our analyses of other manufacturing data are discussed in a separate article.
Van der Graaf et al7 found, as we did, that younger patients (<50 years) were at increased risk; if our results are adjusted to the dichotomous age categorization used in her study, the relative risk estimates are similar. The two studies based on the manufacturer’s data also concluded that the risk of strut fracture is higher in younger recipients (R. Brookmeyer, N. Weiss, Y. Yasui, E. Blackstone, unpublished data, 1993).6 Unfortunately, only 14% of the implant cards indicate patient age, so the earlier analyses were necessarily limited to a small fraction of the valves. Brookmeyer (unpublished data, 1992) found individuals 20 to 40 years old to be at greatest risk. Grunkemeier et al6 found that the risk was highest for patients implanted in their 20s and fell progressively, and by a factor of 10, to patients whose valves were implanted after age 60.
For the present study, age was drawn from the medical records and was available for a larger proportion of fractured valves than in the previous analyses of the manufacturer’s data and for a larger group than was available to Van der Graaf et al.7 We located documents with vital status information for the majority of controls in these analyses, and our results are substantially the same as those of the earlier analyses. Part of the apparent effect of age on risk may be mediated by a tendency not to identify or report valve failure as a cause of sudden death in elderly recipients. Lindblom et al8 found that most Swedish valve fractures were discovered unexpectedly at autopsy, and the relation between underascertainment and old age might be expected to follow as a result of any age-dependence of autopsy. However, in the Netherlands, where the absolute risk of reported fracture was higher than it was in Sweden (and twice that in the United States and Canada), the gradient of risk with age is similar to that of the present study.
BSA has not been examined in previous work; however, both Brookmeyer (unpublished data, 1993) and Grunkemeier6 found men to be substantially more likely to experience a fracture, after other risk factors were accounted for. Since neither author was able to control for body size, it may be that their results reflect the different BSAs of men and women. After controlling for BSA, we did not find a predictive effect for sex. Van der Graaf et al7 reported no influence of sex on the risk of failure.
Ischemic heart disease appeared to protect against strut fracture, and a history of endocarditis before implant appeared to increase risk in this study, both at marginal levels of statistical significance. No information on ischemic heart disease was available in other studies. Whatever the mechanism for the reduced risk, it may be distinct from poor left ventricular function. Neither we nor Van der Graaf et al7 found any important relation between left ventricular function and strut fracture, although our inability to do so could be attributed to the absence of adequate data on left ventricular function in most patients. Endocarditis was also examined in the Dutch cohort.7 Although measured differently from our study, endocarditis was not a correlate of fracture in the Dutch study. It was associated with postimplant mortality.
Our study is limited in that it is based only on US and Canadian fractures reported to the manufacturer. Although our review of the FDA information suggests that the manufacturer was aware of almost every fracture in the United States that was reported to the FDA, there was no way of ascertaining and studying either unrecognized or unreported fractures. This study was also restricted to valves for which an implant card was returned to the manufacturer. The proportion of implanted valves in the United States and Canada for which a card was returned is ≈85%. Although surgeons who elected to return implant cards could differ from those who did not, it is unlikely that risk factors for fracture might operate differently among the patients of such surgeons, and the requirement that the implant card be returned before fracture makes a biased pattern of exclusion unlikely. Moreover, all statistical computations were matched according to surgeon, so intersurgeon differences would not bias the risk estimate.
The medical variables we considered were restricted to those available at the time of implant, and their completeness varied with the record-keeping practices of the individual surgeons. All data, however, were recorded before fracture, so that although some data may be missing from too many records to allow analysis, the presence or absence of data is not associated with fracture and should not introduce any spurious associations.
Although several new patient characteristics were identified as potential predictors, the risk estimates associated with these are weak or imprecise or both. Taken with age, however, they do suggest an underlying theme: conditions that imply higher cardiac output increase the likelihood of strut fracture. Younger people tend to be more active. Younger hearts are capable of higher maximum heart rates, greater flow rates, and more rapid pressure changes (dP/dt) than those of older persons. Increasing BSA is directly proportional to increased cardiac output in healthy persons and engenders increasing values of flow and rate of pressure changes. Individuals with ischemic heart disease have reduced ability to generate high pressures and flow rates. Studies should be conducted to determine whether drugs that affect cardiac output might not modify the risk of strut fracture as well.
This study was supported by a research contract from Shiley Heart Valve Research Center Inc. We would like to express our gratitude to Drs Eugene Blackstone, Ronald Brookmeyer, Sándor J. Kovács, and Noel Weiss, who provided both scientific insight and critical review to this phase of our research; to Dale Fay, Patricia Hamilton, Dan Johnston, Mavis Miller, and the team of data abstractors from the Shiley Heart Valve Research Center; and to Dr Anne Prener for her work in the early stages of this project.
Reprint requests to Alexander M. Walker, MD, Epidemiology Resources Inc, One Newton Executive Park, Newton Lower Falls, MA 02162-1450.
- Received April 3, 1995.
- Revision received July 17, 1995.
- Accepted July 19, 1995.
- Copyright © 1995 by American Heart Association