|
|
| |||||||||||||||||||||||||||||||||||||||
Background Since 1989, laparoscopic cholecystectomy has been widely adopted as a treatment for gallstone disease. We analyzed the association between the introduction of this procedure and three variables: the rate at which cholecystectomy was performed in Maryland, the characteristics of patients undergoing cholecystectomy in routine clinical practice, and operative mortality.
Methods and Results We used 1985-1992 hospital-discharge data from all 54 acute care hospitals in Maryland to identify open and laparoscopic cholecystectomies, characteristics of patients undergoing these procedures, and deaths occurring during hospitalizations in which these procedures were performed. The annual rate of cholecystectomy, adjusted for age, rose from 1.69 per 1000 state residents in 1987-1989 to 2.17 per 1000 residents in 1992, an increase of 28 percent (P<0.001). As compared with patients undergoing open cholecystectomy, patients undergoing laparoscopic cholecystectomy tended to be younger, less likely to have acute cholecystitis or a common-duct stone, and more likely to be white and have private health insurance or belong to a health maintenance organization (P<0.001). Although the operative mortality associated with laparoscopic cholecystectomy was less than that with open cholecystectomy (adjusted odds ratio, 0.22; 95 percent confidence interval, 0.13 to 0.37) and the overall mortality rate for all cholecystectomies declined from 0.84 percent in 1989 to 0.56 percent in 1992, there was no significant change in the total number of cholecystectomy-related operative deaths because of the increase in the cholecystectomy rate.
Conclusions In Maryland, although the adoption of laparoscopic cholecystectomy has been accompanied by a 33 percent decrease in overall operative mortality per procedure, the total number of cholecystectomy-related deaths has not fallen because of a 28 percent increase in the total rate of cholecystectomy.
The rapid and widespread adoption of laparoscopic cholecystectomy has aroused concern about the safety of the new procedure. In addition, one wonders whether the availability of a less invasive approach to cholecystectomy has led to a change in the spectrum of patients undergoing the procedure and in the threshold for performing it3,4.
To address these issues, we analyzed hospital-discharge data from the state of Maryland for the period from 1985 through 1992. Our goals were to determine whether expanded use of laparoscopic cholecystectomy in routine clinical practice has been associated with changes in the rate at which cholecystectomy is performed and in the related operative mortality, and whether there are differences in the characteristics of patients undergoing laparoscopic cholecystectomy as compared with those undergoing open cholecystectomy.
Methods
Source of Data
Data were acquired on all hospital discharges in Maryland from January 1985 through December 1992 from the Maryland Health Services Cost Review Commission, which maintains a data base on all hospitalizations at the 54 nonfederal, short-stay hospitals in the state. For each hospitalization, the data base includes information on the patient's age, sex, race, marital status, and county of residence; the type of health insurance; the type of admission; the length of the hospital stay; and the disposition at discharge, as well as codes from the International Classification of Diseases, 9th Revision, Clinical Modification (ICD-9-CM) for up to five diagnoses and three procedures.
Criteria for Inclusion as a Case
To identify hospitalizations during which either open or laparoscopic cholecystectomy was performed, we used an algorithm based on a combination of the discharge diagnosis and the procedure codes. A case was classified as involving open cholecystectomy if it had a primary or secondary diagnosis code for gallbladder disease (ICD-9-CM codes 574.00 to 576.99) and a primary or secondary procedure code for total cholecystectomy (ICD-9-CM code 51.22), without a procedure code for either laparoscopy (code 54.21) or laparoscopic cholecystectomy (code 51.23).
Before 1991, there was no ICD-9-CM code for laparoscopic cholecystectomy. Hence, laparoscopic cholecystectomies performed before 1991 were coded with either ICD-9-CM codes 51.22 and 54.21 combined (total cholecystectomy and laparoscopy) or code 51.22 alone (total cholecystectomy). We classified a case as involving laparoscopic cholecystectomy if it had a primary or secondary diagnosis code for gallbladder disease (codes 574.00 through 576.99) and either a primary or secondary procedure code for laparoscopic cholecystectomy (code 51.23) or primary or secondary procedure codes for both total cholecystectomy (code 51.22) and laparoscopy (code 54.21), without a diagnosis code for inguinal hernia (codes 550.00 through 550.99), appendicitis (codes 540.00 through 543.99), or a gynecologic disorder (codes 614.00 through 639.99). The diagnosis codes for inguinal hernia, appendicitis, and gynecologic disorders were used to exclude cases in which the procedure code for laparoscopy may have referred to a procedure other than cholecystectomy.
Validation
To assess the accuracy of our algorithm in classifying a case as involving open rather than laparoscopic cholecystectomy, we randomly selected 10 hospitals from the data base, contacted their surgical or operating-room departments, and requested counts of the open and laparoscopic cholecystectomies performed in 1990 and 1991. For these 10 hospitals there was a high correlation between the number of procedures we estimated from the data base and the number reported to us independently from the hospital departments, with regard to both laparoscopic (Spearman's r = 0.87) and open (Spearman's r = 0.89) cholecystectomy. The estimate of laparoscopic cholecystectomies derived from the data base with our algorithm was only slightly lower than the number obtained directly from the 10 hospitals (1026 vs. 1118), and the estimate of open cholecystectomies was slightly higher than the number obtained directly (1212 vs. 1178).
Identification of Coexisting Conditions
The patients' coexisting clinical conditions were considered to be the primary and secondary diagnoses listed in the data base that were not related to gallstone disease. A patient was considered to have had a bile-duct stone if any ICD-9-CM code from 574.30 through 574.59 was listed.
Operative Mortality
Operative mortality was defined as the percentage of patients who died during a hospitalization in which a cholecystectomy was performed. To exclude cases in which death may have been due to another major operation, such as a Whipple's operation, only cases with a primary-procedure code for open or laparoscopic cholecystectomy were included in our primary analysis of operative mortality.
Statistical Analysis
In our primary analysis of rates of performance of laparoscopic and open cholecystectomy, we included all cases involving either procedure that met the previously described criteria, whether or not cholecystectomy was listed as a primary or secondary procedure. In a secondary analysis, we excluded cases in which no primary procedure code was assigned for either open or laparoscopic cholecystectomy. In both analyses we determined crude and age-adjusted rates for each procedure for each year from 1985 through 1992. The denominators for the procedure rates were derived from the Maryland census figures for each year. Age-adjusted procedure rates were calculated by direct standardization, with figures from the 1990 state census used to represent the standard population5.
Differences in the characteristics of patients according to whether they underwent open or laparoscopic cholecystectomy and according to whether they died in the hospital or were discharged alive were analyzed with a chi-square test for categorical variables (i.e., sex, race, marital status, type of health insurance, type of admission, type of cholecystitis, presence of a bile-duct stone, and vital status at discharge) and Student's t-test for continuous variables (age and length of hospital stay). Multiple logistic-regression analyses were performed to identify characteristics of patients that were independently associated with either an increased likelihood of undergoing laparoscopic rather than open cholecystectomy or an increased likelihood of dying during the hospitalization.
Statistical calculations were performed with the Statistical Applications Software of the SAS Institute6.
Results
Numbers and Rates of Open and Laparoscopic Cholecystectomies
Table 1 shows the crude numbers and age-adjusted rates of open and laparoscopic cholecystectomies we identified for each year from 1985 through 1992. From 1985 through 1987, before laparoscopic cholecystectomy was introduced in the United States, our algorithm identified only two cases in which the technique was used, suggesting that the algorithm was generally successful in not counting other types of laparoscopic surgery as laparoscopic cholecystectomies. As Table 1 shows, before there was substantial use of laparoscopic cholecystectomy (i.e., from 1985 through 1989), the age-adjusted rate of cholecystectomy in Maryland was essentially constant, at 1.65 to 1.69 procedures per 1000 residents per year. From 1989 through 1992, in contrast, the overall age-adjusted rate of cholecystectomy in Maryland increased 28 percent, driven by a marked increase in laparoscopic cholecystectomies that more than offset a 70 percent decline in the rate of open cholecystectomies. By 1992, 76 percent of the procedures performed in Maryland were done laparoscopically. It is noteworthy that the overall rate of cholecystectomy plateaued during 1991 and 1992.
|
Patients undergoing laparoscopic cholecystectomy were more likely to be admitted electively, much less likely to have a diagnosis of acute cholecystitis or to have a common-bile-duct stone, and less likely to have one of several coexisting clinical conditions than patients undergoing open cholecystectomy (Table 2). Patients undergoing laparoscopic cholecystectomy were also more likely to be female and white and were more likely to have private health insurance or to be members of a health maintenance organization (HMO). Although the mean age of the patients undergoing laparoscopic cholecystectomy was lower than that of the patients undergoing open cholecystectomy and the percentage who were 65 years old or older was smaller, age was only slightly associated with the performance of laparoscopic rather than open cholecystectomy, after adjustment for other differences in the characteristics of the patients. During the period 1990-1992, the average hospital stay of patients undergoing laparoscopic procedures (2.7 days) was 65 percent shorter than that of patients undergoing open procedures (7.7 days). In this period, the proportion of patients with a diagnosis of acute cholecystitis increased for both procedures, but the proportion with the coexisting conditions shown in Table 2 increased only slightly.
|
Table 3 shows the operative-mortality rate associated with cholecystectomies performed from 1985 through 1992, both overall and according to type of procedure, based on cases with a primary procedure code for cholecystectomy. From 1990 through 1992, the three years in which substantial numbers of laparoscopic cholecystectomies were performed in Maryland, operative mortality was significantly lower (P<0.05) for laparoscopic cholecystectomy than for open cholecystectomy. During this period, there was a significant increase (P<0.01) in operative mortality associated with open cholecystectomy. Despite the statistically significant decline in operative mortality associated with all cholecystectomies from 1989 through 1992, the number of patients who died in 1989 during a hospitalization in which a cholecystectomy was performed did not differ significantly from the number who died in 1992 (62 vs. 56 patients) because of the 28 percent increase in the rate of cholecystectomy during this period. The same findings obtained when we analyzed cholecystectomies listed as either primary or secondary procedures.
|
|
For decades, experts have debated the extent to which the introduction of new techniques has been responsible for the relentless increases in health care costs in the United States7,8. Recently published evidence suggests that despite the cost-saving potential of laparoscopic cholecystectomy,9 the introduction of this procedure in the United States has been associated with an increase in the overall rate of cholecystectomy and an increase in the cost of treating patients with gallstone disease3,4. We have used a statewide data base of hospital-discharge abstracts to evaluate the introduction of laparoscopic cholecystectomy in Maryland. Our analysis includes data through 1992 and allows a comparison of the effect of adopting laparoscopic cholecystectomy throughout an entire state with that of using it in a single large HMO3.
Our analysis yields several noteworthy findings. First, before laparoscopic cholecystectomy was introduced, annual cholecystectomy rates in Maryland were stable. There was an increase in the use of the laparoscopic technique in the state from 1990 through 1992 that exceeded the concurrent decrease in the use of open cholecystectomy and resulted in a 28 percent increase in the overall rate of cholecystectomy from 1989 through 1992. This increase in the cholecystectomy rate in Maryland was similar to the 29 percent increase observed in Connecticut from 1989 through 19914. The fact that the rate reached a plateau in Maryland in 1992, instead of declining, suggests that the increase is more likely to reflect a sustained lowering in the threshold of patients, physicians, or both for surgical intervention than to reflect the exhaustion of a one-time pool of patients who were postponing surgery because of their surgical risk or fears.
The difference between our experience in Maryland and that reported for a large HMO3 is worthy of comment. Although there was a comparable decline in the rate of open cholecystectomy from 1988 through 1992 in both Maryland and the U.S. Health Care HMO (reductions of 70 and 74 percent, respectively), the increase in the use of laparoscopic cholecystectomy has been greater in the HMO than in Maryland as a whole (57 percent vs. 29 percent). As a result, the overall rate of cholecystectomy among enrollees of U.S. Health Care, which was lower than the overall rate in Maryland in 1988 (1.37 per 1000 enrollees vs. 1.68 per 1000 state residents), was almost the same as the Maryland rate in 1992 (2.15 per 1000 enrollees vs. 2.17 per 1000 residents). The sharper increase in the use of laparoscopic cholecystectomy in the HMO than in the state population could partly be due to differences in demographic and clinical characteristics, such as a healthier or a younger population of HMO enrollees as compared with Maryland residents (mean age of enrollees in 1992, 29.5 years3; median age of state residents in 1991, 33 years). Multivariate analysis of patients undergoing cholecystectomy, however, showed that HMO enrollees in Maryland were more than 40 percent more likely to undergo laparoscopic rather than open cholecystectomy than were those with other types of insurance, even after adjustment for differences in age and other characteristics. This difference may have arisen either because HMO management advocated laparoscopic over open cholecystectomy in the belief that it was less expensive or because surgeons working in HMOs were more inclined to learn and perform laparoscopic cholecystectomy than surgeons providing care on a fee-for-service basis.
Our analysis also provides insight into the safety of laparoscopic cholecystectomy when performed widely by large numbers of surgeons in routine practice. Even after we attempted to control for differences in the clinical characteristics of patients undergoing open as compared with laparoscopic cholecystectomy, such as the greater likelihood that patients undergoing open cholecystectomy would have acute cholecystitis or a common-bile-duct stone, operative mortality was 80 percent lower for laparoscopic cholecystectomy. Our inability to adjust fully for differences in patients' clinical characteristics may have led to overestimation of the reduction in the risk of death attributable to laparoscopic cholecystectomy. Our estimate of the operative mortality of laparoscopic cholecystectomy, however, reflects routine clinical practice in Maryland and is very similar to rates reported in several other large series2,4,10,11,12,13,14. The fact that the operative mortality of patients undergoing open cholecystectomy continued to increase in 1992 suggests that for patients at lower risk, the technique chosen was still being shifted from open to laparoscopic cholecystectomy in 1992.
Despite the decrease in overall operative mortality from cholecystectomy that has resulted from the adoption of the laparoscopic technique, the total number of deaths during hospitalizations in Maryland in which cholecystectomy was performed has not changed. This is true because of the increase in the overall rate of cholecystectomy. Even with no significant reduction in the number of operative deaths related to cholecystectomy, there is still substantial benefit to patients from the use of laparoscopic rather than open cholecystectomy, since those who undergo laparoscopic cholecystectomy can resume full duties, including employment outside the home, much sooner than those who undergo open cholecystectomy.
Our study has several limitations. First, the accuracy of the coding of diagnoses and procedures in the data base requires mention. Many Maryland hospitals perform internal audits to ensure the accuracy of data they submit to the Health Services Cost Review Commission. A quarterly audit of discharge records by the Maryland Peer Review Organization, involving the same data that are submitted to Medicare, found that the overall rate of coding errors was less than 9 percent (Allen P: personal communication). Other studies have shown that many such errors involve improper sequencing of codes or failure to code complications,15 errors that would not have affected our results significantly.
Second, a unique ICD-9-CM code was not assigned to laparoscopic cholecystectomy until 1991. We therefore used an algorithm that combined the procedure codes for laparoscopy and total cholecystectomy in order to identify cases of laparoscopic cholecystectomy in the years before 1991. To reduce our false positive rate in the detection of cases of laparoscopic cholecystectomy, we required the cases selected to have a gallbladder-related diagnosis and excluded cases in which laparoscopy may have been performed for a procedure other than gallbladder removal. The fact that our algorithm identified few or no cases of laparoscopic cholecystectomy from 1985 through 1989 suggests that we are unlikely to have substantially overestimated the number of laparoscopic cholecystectomies. Our small validation study supports this conclusion.
Third, it is possible that in the case of some patients cholecystectomy was begun laparoscopically, converted to an open procedure, and coded as such. This limitation would not have influenced our estimate of the total number of cholecystectomies, but it might influence our findings with regard to the characteristics of patients undergoing laparoscopic as compared with open cholecystectomy and our analysis of procedure-specific operative mortality.
A fourth area of concern relates to differences between cases in which the cholecystectomy was coded as a primary procedure and those in which it was coded as a secondary procedure. To address this concern, we performed a secondary analysis of trends in surgical rates that was limited to cases in which cholecystectomy was listed as the primary procedure. This sensitivity analysis did not alter our conclusions about the characteristics of patients who had laparoscopic as compared with open cholecystectomy.
A further limitation relates to the definition of operative mortality. It is possible that not all the deaths that occurred during hospitalizations in which a cholecystectomy was performed were related to that procedure. Also, our estimates of operative mortality do not include cholecystectomy-related deaths occurring after discharge. Nevertheless, the operative mortality estimated under our definition was very similar to that in other published reports2,4,9,10.
Finally, our analysis did not address nonfatal complications of laparoscopic cholecystectomy, such as bile-duct injury and retained stones2,10,12,16,17,18,19. These are also pertinent to conclusions about the overall safety and cost of this procedure.
We are indebted to Michael Klag, M.D., M.P.H., for statistical advice; to Joel Gibson for assistance with programming; to Paul Allen, Patrick Kegley, and Cindy Kegley for assistance in obtaining and using the data base and interpreting discharge-coding practices; to Sean Tunis, M.D., M.Sc., for assistance in developing the coding algorithm; and to Lee Bruchey and Donna Lea for assistance in the preparation of the manuscript.
Source Information
From the Departments of Medicine (C.A.S., E.B.B., E.P.S.) and Surgery (M.A.T., H.A.P.), School of Medicine, and the Department of Health Policy and Management (E.B.B., E.P.S.), School of Hygiene and Public Health, Johns Hopkins University, Baltimore.
Address reprint requests to Dr. Steinberg at Johns Hopkins University, Program for Medical Technology and Practice Assessment, 1830 E. Monument St., Rm. 8068, Baltimore, MD 21205.
References
| |||||||||||||||||||||||||||||||||||||||
Related Letters:
Laparoscopic Cholecystectomy
Zahedi T., Steiner C. A., Bass E. B., Steinberg E. P.
Extract |
Full Text
N Engl J Med 1994;
331:54-55, Jul 7, 1994.
Correspondence
This article has been cited by other articles:
HOME | SUBSCRIBE | SEARCH | CURRENT ISSUE | PAST ISSUES | COLLECTIONS | PRIVACY | HELP | beta.nejm.org Comments and questions? Please contact us. The New England Journal of Medicine is owned, published, and copyrighted © 2008 Massachusetts Medical Society. All rights reserved. |
Previous
