Prognosis of gallstones with mild or no symptoms: 25 years of follow-up in a health maintenance organization

Prognosis of gallstones with mild or no symptoms: 25 years of follow-up in a health maintenance organization

08954356:89 J Clin Epidemiol Vol. 42, No. 2, pp. 127 -136, 1989 Printed in Great Britatn. All rights reserved Copyrtght 53.00 + 0.00 0 1989 Perga...

1021KB Sizes 0 Downloads 5 Views

08954356:89

J

Clin Epidemiol Vol. 42, No. 2, pp. 127 -136, 1989 Printed in Great Britatn. All rights reserved

Copyrtght

53.00 + 0.00

0 1989 Pergamon Press plc

PROGNOSIS OF GALLSTONES WITH MILD OR NO SYMPTOMS: 25 YEARS OF FOLLOW-UP IN A HEALTH MAINTENANCE ORGANIZATION GARY D. FRIEDMAN,’* CAROL A. RAVIOLA’ and BRUCE FIREMAN’ ‘Divtsion of Research, Kaiser Permanente Medical Care Program. Northern Department of Surgery. Kaiser Permanente Medical Cenier. Oakland.

California Region and California, U.S.A.

Abstract-Two-hundred ninety-eight patients wrth gallstones accompanied by mild or nonspecific symptoms, 123 with asymptomatic gallstones, and 46 with nonfunction on cholecystogram and mild or nonspecific symptoms were followed in the setting of a health maintenance organization for up to 25 years after diagnosis. In a life-table analysis, complications (severe events) developed in about 1% per year of patients with visualized gallstones and in about 2% per year of patients with nonfunction. During each of the first 5 years after diagnosis. all events. both severe and non-severe (including surgery for continuing mild symptoms) occurred in about 6% of the patients with mild symptoms accompanying either gallstones or nonfunction. and in about 4% of patients with asymptomatic gallstones. The annual probabilities for all events tended to decrease as length of follow-up increased. Among patients with stones and mild symptoms, women were more apt to develop any event than men (p = 0.02) and more obese patients were more likely to develop severe events than those who were thinner (p = 0.0s). The patients in this study are probably more representative of outpatients with gallstones seen by most primary care physicians than are most groups investigated previously for prognosis. Their event rates are sufficiently similar to those used in published decision analyses. that they should not alter previous conclusions that early elective cholecystectomy will have little positive or negative effect on average life expectancy of patients with gallstones. Gallbladder

diseases

Cholelithiasis

Follow-up

INTRODUCTION Decisions as to whether patients with gallstones should receive early elective cholecystectomy are based on weighing the risks and costs of the elective operation against those of no or delayed surgery [l-3]. An important factor in determining the risk of waiting is the rate at which complications or symptoms leading to cholecystectomy occur in patients with untreated gallstones. The few published studies of complication rates have focused on (1) patients with asymptomatic gallstones detected at surgery for other conditions [4-61, (2) an unclear mix of symptomatic and asymptomatic patients with ‘All correspondence should be addressed to: Dr Friedman. 3451 Piedmont Avenue, Oakland, CA 9461 I, U.S.A. Supported by the Community Services Program of Kaiser Foundation Hospitals.

studies

Prognosis

stones detected on cholecystogram [7,8], (3) university faculty members with gallstones discovered on screening oral cholecystogram, with either no or nonspecific symptoms [9], (4) placebo-treated subjects of a multicenter clinical trial of dissolution therapy, with radiolucent gallstones, both asymptomatic and symptomatic [lo], and (5) asymptomatic and symptomatic patients from a health maintenance organization, with little information about the temporal pattern of complications [I I]. The present study is concerned with the prognosis of gallstones accompanied by mild or nonspecific symptoms and detected largely in outpatients who received care in a health maintenance organization. These relatively heterogeneous and unselected outpatients are probably more representative than subjects of most previous studies, of patients considered by 127

128

GARY D. FRIEDMAN et al.

primary care physicians for referral for elective cholecystectomy. Follow-up data are also presented concerning patients with asymptomatic stones and symptomatic patients whose cholecystograms show nonfunction but no stones.

surgery or severe symptoms or manifestations (such as acute cholecystitis) before the index visit, and no subsequent gallbladder surgery that was connected with the index visit (e.g. index visit led to referral to surgeon who operated 3 months later). Altogether 499 persons met these criteria: 315 had gallstones with mild or nonspecific symptoms, 132 had asympMETHODS tomatic gallstones, 49 had nonfunction with IdentiJication and classiJication of subjects mild or nonspecific symptoms; three had asympIn the Kaiser Permanente Medical Care tomatic nonfunction and are not considered Program (KPMCP) of Northern California we further. (The major reasons for exclusion from the original group of 6160 with records reviewed sought persons who had documented gallstones that were not promptly removed and who could were: no objective evidence of gallstones in the and evidence of gallbladder be followed up after an identification date, record-59.2%, surgery or complications of gallstones before described below as the “index visit”. Subscribers to the KPMCP comprise approximately one the index visit-l 7.7%). We further excluded 17 fourth of the urban and suburban population in patients with mildly symptomatic stones, nine a broad area surrounding San Francisco Bay with asymptomatic stones and three with mildly and are quite diverse with respect to ethnic symptomatic nonfunction because their records background and socioeconomic status [ 121. indicated a history of earlier first objective confirmation, but the year in which this took After a search for data sets that contained substantial numbers of patients with gallstones place was quite uncertain. Repeated data analand after a pretest of selection criteria [13] we ysis, with these patients included, gave results used two computer-stored record sets for initial very similar to the findings presented here. The presence or absence of symptoms refers screening. The first contained diagnoses of gallbladder disease made in outpatient clinics at to the subject’s status at the time of the first the KPMCP’s San Francisco Medical Center objective confirmation. The nature and severity between November 1967 and August 1973 of symptoms at that time were often difficult to [14, 151. The second contained responses to determine from the clinic notes. They no doubt ranged from the mildest indigestion to biliary questions on self-administered questionnaires colic. With the added uncertainty as to what used from July 1964 to August 1973 in multiactually constitute specific gallstone symptoms, phasic health checkups given at both the San we did not attempt to divide the symptomatic Francisco and Oakland facilities [16]. From the group into those with specific and nonspecific latter we listed persons who indicated that they symptoms. We simply classified them according had been told by a doctor that they had gallbladder disease and did not indicate that they to whether right upper quadrant abdominal had had gallbladder surgery. pain or tenderness was mentioned, in the belief In this way, 6365 persons were identified as that many of those with such mentions and possibly having gallstones and the medical few of those without, had specific gallbladder records of 6160 (96.8%) were reviewed in the disease symptoms. time allotted. To be included in one of the follow-up study groups a patient had to meet Follow -up and analysis the following criteria: gallstones demonstrated Rates of occurrence of complications or by X-ray (cholecystogram or radio-opaque surgery and the cumulative probabilities of deshadows on other X-ray judged by the radioloveloping these events were determined by the gist to be gallstones) or by palpation at surgery, life-table method [17] based on information in or nonfunction on double-dose oral chole- the medical records. The date of first objective cystogram unexplained by extraneous factors confirmation, referred to hereafter as the date of such as diarrhea. This objective confirmation diagnosis, or just “diagnosis,” was considered must have occurred no later than 2 years after to be “time zero.” However, person-years of the “index visit,” which was the first computerobservation were not begun until the date of the recorded clinic visit or questionnaire response index visit, since subjects were first identified by that led to the initial listing of the patient. means of that computer-recorded visit. In order Further, there could have been no gallbladder to enter this study persons had to survive with

Prognosis

gallbladders intact and free of severe symptoms or complications until the index visit. The inclusion of their experience before that visit would bias the study toward finding a lower complication rate. Of course, inferences from these “left-truncated” data are based on the assumption that the experience of the patients who are observed during the early years of follow-up is representative of that of all patients of interest in the early years, just as it is assumed that the experience of patients remaining after some are lost to follow-up, or “right-censored,” is representative of that of all patients of interest in the later years. Follow-up ended when the subject experienced the event under consideration, died, had his or her last visit in our records, or when the gallbladder was surgically removed for any reason. whichever came first. The logrank test was used to compare probabilities of events in different subgroups of patients [18]. Definition

of ezvnts

The outcome events that we looked for were the following: acute cholecystitis, i.e. a clinical diagnosis supported by severe right upper quadrant pain or tenderness with accompanying fever or leukocytosis; acute biliary colic (similar symptoms without fever or leukocytosis) that led to an immediate medical visit, usually to an emergency room; extrahepatic obstructive jaundice believed to be due to gallstones; gallstone ileus; or acute biliary pancreatitis. We also noted cholecystectomy for chronic or milder symptoms as an outcome event; this may have occurred without a worsening in the patient’s symptoms, but because the patient decided to have something definitive done about them. We labeled all the above events as “severe” except acute biliary colic and surgery for chronic symptoms. RevieMt procedure Two trained medical record analysts screened the medical records for eligibility to be included in the follow-up study. When patients were probably eligible. one of the analysts abstracted their charts and made tentative classifications. Then, one of us (G.D.F.) reviewed all of these cases with the abstracting analyst and made a final decision about inclusion and the occurrence of events. All records with ambiguous or complex clinical histories were reviewed by another one of us (C.A.R.), a surgeon with extensive experience in biliary tract disease, before final classification.

129

of Gallstones RESULTS

Characteristics

of the subjects

As shown in Table 1 the patients studied were predominantly white and female. The minimum, mean, and maximum ages at diagnosis were: for gallstones with mild symptoms-29, 55, and 84 years; for asymptomatic gallstones-17, 53, and 77 years; and for nonfunction with mild symptoms-20, 52, and 77 years, respectively. There were 1 l-15% diabetics and very few alcoholics. Most patients had multiple rather than single stones. Most asymptomatic stones were radiopaque, as expected, since most (103 or 83.7% of 123) were incidental findings on X-rays ordered for investigation of other organs. (Four or 3.3% were discovered on oral cholecystogram and 16 or 13.0% were palpated at laparotomy. of which 14 were hysterectomies.) Also, as expected, the patients tended to be overweight, the median body mass index (weight/height’) falling in the second and third highest decile for women and the fourth highest decile for men of the distribution of body mass indexes of a large group of Kaiser Permanente subscribers who took multiphasic checkups [19]. The records of most symptomatic subjects did not mention right upper quadrant pain or tenderness. Complications mild symptoms

in patients

with gallstones

and

During follow-up, which ranged up to 37 years after diagnosis of gallstones, 26 (8.7% of 298) patients with mild symptoms initially, developed severe complications including 17 with acute cholecystitis, seven with obstructive jaundice and two with biliary pancreatitis. Two of the acute cholecystitis patients also had obstructive jaundice, one had biliary pancreatitis, one had a gangrenous gallbladder, one had a duodenal fistula and one had a carcinoma of the gallbladder discovered incidentally at surgery. (Presumed metastases became clinically evident about 20 months later when the patient developed obstructive jaundice and liver scan showed irregular labeling.) All but four of these patients had surgical treatment. Non-severe events developed in another 73 (24.5%) patients including 11 with cholecystectomy following acute biliary colic, four with acute biliary colic and no surgery, and 58 with surgery for chronic symptoms. Figure 1 shows a life-table analysis of the cumulative probability of developing a severe

GARY D. FRIEDMAN er al.

130

Table 1. Characteristics of the subjects Nonfunction with mild symptoms

Gallstones with mild symptoms

Asymptomatic gallstones

298 75.2

123 71.5

46 73.9

Race White (%) Black (%) Asian (%) Hispanic (%) Other and unknown (%)

73.5 11.7 5.7 1.3 7.7

75.6 10.6 3.3 1.6 8.9

71.7 17.4 4.3 2.2 4.3

Age* c45 years (%) 45-54 (%) 5564 (%) 65-74 (%) 75+ (%)

21.1 23.8 32.9 17.8 4.4

23.6 31.7 28.5 13.8 2.4

23.9 34.8 28.3 8.7 4.3

Education High school or less (%) Beyond high school (%) Unknown (%)

44.3 35.6 20.1

47.2 40.7 12.2

50.0 30.4 19.6

Alcoholismt(%) Diabetes mellitusf(%)

4.7 15.4

0.0 11.4

4.3 15.2

Stone count Single (%) Multiple (%) Unknown (%)

33.6 62.7 3.7

26.8 65.9 7.3

Stone radiolucency Opaque (%) Lucent (%) Both (%) Unknown (%)

29.9 46.6 1.0 22.5

78.1 7.3 2.4 12.2

Right upper quadrant pain* (%)

29.5

0.0

32.6

26.9 26.8

25.7 26.6

26.6 26.5

Total number Female (%)

Body mass indexemedian

WI

[w$FJ[t Men

*At time of diagnosis. tAt or after time of diagnosis. IEver recorded in medical record. $Close.st to time of index visit. 100

-J

60 g

50

1 0

5

10

15

20

25

Years After First Documentation

Fig. 1. Life table cumulative probabilities of developing a severe event or any event, according to time since diagnosis (first documentation) of gallstones (time zero) in patients with mild symptoms. Each subject contributed person-years of observation starting with his or her domputer-stored index visit, which was usually later than the date of diagnosis.

event and any event. The analysis is carried through 25 years of follow-up after diagnosis of the stones at which time only 22 patients were still under observation. Table 2 provides the probabilities of having an event for each 5 year interval plus the probability of having an event _ . during each year assuming equal probabilities for each year in a particular 5 year interval. Severe events developed in about 1% of patients at risk each year with no consistent increase or decrease in risk apparent over time. In contrast, all events occurred in about 6% of patients at risk each year in the first 5 years (the highest actual observed rate was 8.4% in the second year) and the rate declined to about l&1.7% after 15 years. Factors aficting prognosis. Of the possible prognostic factors studied, only sex and body

Prognosis Table

of Gallstones

131

2. Cumulative and annual probabilities of any and severe events for each five years follow-up after diagnosis. Stones with mild symptoms; 298 patients

Interval from diagnosis (yr)

Number of patients under observation at end of interval*

Any event

Severe event

Any event

Severe event

5 IO IS 20 25

150 135 82 45 22

0.28 0.41 0.47 0.52 0.56

0.05 0.12 0.15 0.18 0.22

0.063 0.038 0.024 0.016 0.017

0.010 0.015 0.007 0.007 0.012

Annual probability of event during last 5 years in interval+

Cumulative probability of an event at end of interval

*Later subjects are not necessarily subsets of earlier ones. smce additional observation as follow-up time Increases. *Assuming equal probability for each year of a 5-year interval.

mass index showed statistically significant (p d 0.05) associations with the probability of developing events (Table 3). Women tended to have more events than men, but the difference was due to non-severe rather than severe events. Patients at or above their sex-specific median (for the entire study group) body mass index had a higher risk for developing severe events than thinner patients but not for events as a whole. Age, educational level. presence or absence of right upper quadrant pain at time zero and single vs multiple stones were not related (p > 0.05) to the subsequent occurrence of either severe, or any, events. Size of stones was not recorded in the records of enough patients for meaningful analysis. Complications

in pcrtknts

ri,ith usj~mptomatic

stones

Seven (5.7% of 123) patients with asymptomatic stones developed severe events. Of the five who were treated surgically, two had ascending cholangitis (one fatal and the other with acute cholecystitis). one had obstructive jaundice, one had acute cholecystitis and one had empyema of the gallbladder. The two nonsurgical severe events were one acute choleTable 3. Relation of cumulative the two characteristics studied

probability

come under

cystitis and one obstructive jaundice. Of the 16 (13.0%) patients with non-severe events, 13 had cholecystectomy for chronic symptoms, and three had acute biliary colic of whom two were treated surgically. The cumulative probability of occurrence of events tended to be lower among persons with asymptomatic (Table 4) than among those with symptomatic stones (Table 2) after 5, 10, 15 and 20 years (p = 0.03 by logrank test comparing the entire distributions). Annual probabilities showed less consistent differences, probably because of random variability due to small numbers of subjects. Like symptomatic stones, there was a tendency for a decrease over time of the probability of any event, but not for the probability of a severe event. As with stones with mild symptoms women were more apt to experience all events than men (cumulative probability 0.36 vs 0.11 at 10 years), but not severe events (both 0.05 at 10 years). Compared to thinner patients, those at or above the median for body mass index had a lower IO-year probability of any event: 0.21 vs 0.36, and a higher IO-year probability of a severe event: 0.07 vs 0.04. The relationship to prognosis of neither sex nor obesity was statistically

of event Probability

Severe event

Any event

Women Men BM I > median BMI < median

subjects

life-table probability of events to sex and body mass index. that had a statistically significant (p GO.05) relation to prognosis

Cumulative

of

5 yr

IO yr

15 yr

5 yr

IO yr

IS yr

0.31 0.19 0.27 0.28

0.45 0.27 0.39 0.42

0.51 0.36 0.47 0.48

0.05 0.05 0.07 0.03

0.13 0.10 0.15 0.09

0.14 0.18 0.18 0.12

*From logrank test comparing the entire survival-till-event which extended up to 37 years.

Any event

(p)’

Serious

event

0.02

0.95

0.64

0.05

curves

of the two subgroups

GARY D. FRIEDMAN er al.

132

Table 4. Cumulative and annual probabilities of any and severe events for each 5 years of follow-up after diagnosis. Asymptomatic stones; 123 patients

Interval from diagnosis (yr)

Number of patients under observation at end of interval*

Any event

Severe event

Any event

Severe event

5 10 15 20

51 59 38 21

0.18 0.30 0.34 0.41

0.04 0.05 0.10 0.16

0.039 0.030 0.014 0.020

0.008 0.003 0.009 0.014

Cumulative probability of an event at end of interval

Annual probability of event during last 5 years in interval?

*Later subjects are not necessarily subsets of earlier ones as additional subjects come under observation as follow-up time increases. tAssuming equal probability for each year of a 5-year interval.

significant. Nor were the findings for age, educational level, right upper quadrant pain or multiplicity of stones.

unreliable. Because of small numbers, an analysis of prognostic factors would not yield meaningful results.

Complications in patients with gallbladder nonfunction and mild symptoms

DISCUSSION

The six (13.0%) severe events in the 46 patients with nonfunction on cholecystogram were acute cholecystitis in three patients (one gangrenous), one ascending cholangitis, one obstructive jaundice and one gastric outlet obstruction due to carcinoma of the gallbladder which was fatal. All were treated surgically. The ten (21.7%) nonsevere events were seven cholecystectomies for chronic symptoms, and three cases with severe biliary colic, one treated surgically. At exploratory laparotomy, an additional patient proved to have agenesis of the gallbladder which was not counted as an event. Probabilities of occurrence of all events during the first 15 years of follow-up of patients with mild symptoms and a cholecystogram showing non-function (Table 5) were similar to those observed in patients with gallstones and mild symptoms (Table 2). Severe events appeared to be somewhat more frequent in the non-function group, but numbers were small and probability estimates were therefore

In our primary study group, patients with gallstones and mild or nonspecific symptoms, the annual probability of developing a severe event or complication averaged about 1% per year with little change over time. For any event, including surgery for continuing symptoms, the annual probability averaged about 6% during the first 5 years after diagnosis and decreased to about l-1/2O/, per year after 15 years. Because of differing or unclear definitions of the symp tomatic status of patients studied or of the complications counted, it is difficult to compare previous studies with ours. In a study of 112 patients at the Mayo Clinic with asymptomatic stones discovered at laporatomy for other conditions, Comfort et al. [6] found that 19% developed colic in a 10-20 year follow-up giving roughly a 1.4% average annual rate of symptom development. Lund [7] reported on a 5-20 year follow-up of about 95 patients who were asymptomatic or had slight symptoms. About 30% of the men and about 52% of the women later

Table 5. Cumulative and annual probabilities of any and severe events for each 5 years of follow-up after diagnosis. Nonfunction with mild symptoms; 46 patients

Interval from diannosis (vrl

Number of patients under observation at end of interval*

Anv event

Severe event

Anv event

Severe event

5 10 15

26 24 17

0.27 0.37 0.49

0.09 0.15 0.24

0.060 0.030 0.041

0.018 0.015 0.021

Cumulative probability of an event at end of interval

Annual probability of event during last 5 years in interval?

*Later subjects are not necessarily subsets of earlier ones as additional subjects come under observation as follow-up time increases. tAssuming equal probability for each year of a 5-year interval.

Prognosis

of Gallstones

developed severe or frequent attacks of pain or complications. If one assumes that follow-up averaged 12 years, the annual rate of developing these endpoints was about 3% in men and about 6% in women. In Wenckert and Robertson’s 1 l-year follow-up of 781 (apparently mixed symptomatic and asymptomatic) patients with abnormal cholecystograms [8], we estimated average annual rates of developing complication or severe symptoms of about 4% or about 6% based on differing statements in the paper. Gracie and Ransohoff [9] found an average annual rate of development of biliary pain of about 1.3% per year in 123 university faculty members, almost all men, who had asymptomatic gallstones detected on screening cholecystogram. Like our findings for all events combined, the average annual probability declined over time. in their group from 2.1% in the first 5 years to I .l% in the second 5 years to 0.7% in the third 5 years. Thistle et al. followed up 305 placebo-treated patients with radiolucent gallstones for 2 years as part of the National Collaborative Gallstone Study [lo]. Although a much higher percentage developed biliary colic or prolonged biliary pain, 12, or 4%, required cholecystectomy giving an annual probability of about 2%. Among the prognostic factors also studied by us, in Thistle’s study a history of biliary pain in the previous 12 months and age less than 55 years were predictive of the development of biliary colic, prior biliary pain and higher body weight were predictive of prolonged biliary pain and three or more stones was predictive of clinically indicated cholecystectomy. The recent data from McSherry rt al. [l l] suggest, approximately, that cholecystectomy was required in about 1.6% of patients with asymptomatic gallstones per year and in 7.9% of patients with symptomatic gallstones per year. Two other follow-up studies do not provide sufficient information to permit estimation of annual complication rates [4, 51. We believe that, compared to the subjects of most previous studies, our main study group is more representative of patients with gallstones usually encountered by the practicing physician. Our subjects were largely outpatients who came in complaining of pain or other gastrointestinal symptoms for whom a cholecystogram was ordered. While there is considerable doubt about whether nonspecific gastrointestinal symptoms are referable to the gallbladder [9], most recognized gallstone patients are not totally asymptomatic with stones detected incidentally on

Ii3

routine screening examinations or on laparotomy for other conditions. Nor are they necessarily similar to persons who agree to participate in a randomized controlled trial of stone dissolution [lo]. Nevertheless, our rate of development of complications is not strikingly different from previous reports. Our smaller group with unequivocally asymptomatic stones, usually discovered incidentally in X-rays ordered for the investigation of other organs (e.g. back, kidneys), had a lower complication rate more consistent with that of the asymptomatic subjects of Comfort et al. [6] and of Gracie and Ransohoff [9]. Recently, ultrasonography has become the most common method for diagnosing gallstones. A group of patients with ultrasonographically-detected stones would contain some persons with functioning and some with nonfunctioning gallbladders if oral cholecystograms were to be given. Thus, such a group would be expected to experience complication rates intermediate between those found in our study groups with stones and those found in our study group with nonfunction. It will take several years before long-term follow-up data will be available on large groups of patients with gallstones detected by ultrasound. The setting for our study was a health maintenance organization, which may perform less elective surgery than the fee-for-service sector [20]. In this setting there was certainly no withholding of surgery as needed for complications and, in reviewing the clinical charts, we noted that surgery was often recommended when mild symptoms were present or to prevent later complications when gallstones were asymptomatic. Both internists and surgeons have varying degrees of enthusiasm for cholecystectomy when asymptomatic or mildly symptomatic stones are present and the entire spectrum of opinion was apparent in our medical records. As expected in a retrospective review of clinical records that were not prepared primarily for research purposes, it was often difficult to be sure of the symptom status of the patient, either when stones were discovered or later when symptoms or complications developed. Not only do physicians vary in the degree to which they record details but, from the records, we suspect that patients vary in the way they report symptoms depending on whether they would like to have, or to avoid surgery. For example, we sometimes found only what appeared to be very mild or nonspecific symptoms recorded for

134

GARY D. FRIEDMAN et al.

a patient in medical clinic notes, but more classical biliary pain symptoms during the same time period described upon admission to the hospital for cholecystectomy. This could have been due to poor recollection by, or to bias of, the patient or to bias on the part of the physician. A more complete and objective ascertainment of symptoms could be accomplished by prospective data collection in which all patients are asked a standardized set of questions about gastrointestinal symptoms at each visit. However, this would probably not solve the problem entirely, especially if the patient believes that his or her answers would influence the decision about cholecystectomy. Other important limitations of our records were the frequent absence of a precise count of the stones present on X-ray or a clear description of their size or appearance. Although our subjects with mild symptoms showed higher probabilities of complications than the asymptomatic group, we could not demonstrate that those whose records mentioned right upper quadrant pain or tenderness had a higher complication rate than those with more nonspecific symptoms. In fact, there was a slight though not statistically significant trend in the opposite direction (e.g. 36 vs 43%, respectively, with events by the end of 10 years of follow-up). Given the limitations of routine medical records already discussed, we did not believe that we could go further than this simple dichotomy in trying to differentiate specific from nonspecific symptoms. We believe that most physicians would agree that the events that we included as “severe” were indeed serious and the kinds of complications one would wish to prevent by prophylactic cholecystectomy. Most of the other events that we included under “any event” were cholecystectomy performed for chronic symptoms. Here, the limitations of medical records were again noted because it was not always possible to determine whether the patient’s symptoms truly worsened before surgery. In some cases they may not have, but the patient chose to undergo cholecystectomy either because he or she became fed up with recurring symptoms or because the physician finally persuaded the patient that surgery was indicated. Whether or not all physicians would agree to regard these operations as complications of gallstones, they are certainly important events from the viewpoint of the patient and the health-care system.

Gallbladder cancer has long been a feared complication of gallstones, based on the frequent finding of gallstones associated with this malignancy, but apparently its incidence is low among patients with gallstones [21]. We observed two cases. One occurred in the 2495 person-years of follow-up of stones with mild symptoms, and the other in the 445 personyears of follow-up of the patients with gallbladder nonfunction on cholecystogram. Apparently both were fatal. With so few cases in our study it is difficult to estimate accurately the rate of development of this complication. The risk appears to be small but not inconsequential. Our study provided little data to help identify persons at higher risk of complications. Women were more apt than men to experience nonsevere events, which consisted largely of surgery for continuing chronic symptoms. This may simply represent a greater willingness on the part of women to have surgery or to do something about the symptoms. Fatter patients were at greater risk of serious events. Unfortunately, however, they are also more apt to experience complications of elective cholecystectomy [22]. Decision analysis has been employed to determine whether prophylactic cholecystectomy is preferable to watchful waiting after gallstones are discovered. One of the key elements in this decision analysis is the rate at which complications develop in persons with gallstones. Other important considerations are mortality rates associated with elective cholecystectomy and with the development of complications, treated surgically or otherwise. Ransohoff et al. [l] based their primary analysis on the low rates of biliary pain and complications observed in their study of University of Michigan faculty men with asymptomatic stones and found that, on average, at age 50, prophylactic cholecystectomy would lose 18 days of life for a man and 12 days of life for a woman. In a sensitivity analysis applied to 30-year-olds they increased their yearly rate of development of biliary pain and complications to 5.4 and 0.6%, respectively, and this led to a slight advantage in average survival for early prophylactic cholecystectomy. (No results were given for 50-year-olds which would be more representative of our study group.) Our observed rate of complications (i.e. severe events), which have more effect on mortality than does surgery for uncomplicated pain, was somewhat higher (about 1%) than their assumed 0.6%, so it is possible that the relative survival advantage of early elective

Prognosis

cholecystectomy would be somewhat greater in our group, but probably still relatively small. Fitzpatrick er al. [2] performed a different decision analysis with fairly similar results, i.e. the patients who were good surgical risks gained, on average, two weeks of life by early elective cholecystectomy whereas those who were poor surgical risks lost a month of life. They used the 6.3% annual complication rate from Wenckert and Robertson [8] which is larger than ours. In sensitivity analysis, 49-year-old good-risk females would gain some days of life from the wait-and-see policy only if the rate of complications were 0.8% or less. somewhat lower than what we observed. Kottke er ul. [3] compared the estimated life expectancy in persons with asymptomatic gallstones who had elective cholecystectomy vs a “wait and see” approach. The difference varied from a few days to about one-half year depending on age and the annual complication rate, assumed to be I, 4, or 5%. The authors appeared not to believe that even the one-half year average gain in a 25year-old to be worth the small risk of immediate death with elective cholecystectomy. Thus, when our event-rate data are used in these decision analyses there is little reason either to accept or reject early elective cholecystectomy for asymptomatic or mildly symptomatic gallstones purely on the basis of average survival. A related concern, that cholecystectomy might increase the risk of developing colorectal cancer, now seems unlikely [21, 231 and need not be a factor in the decision. Other considerations then become important. Both Ransohoff et al. [l] and Fitzpatrick et al. [2] pointed out the high costs of treating gallstones routinely with early elective cholecystectomy. The economic interests of society as a whole would not favor this policy. The other major consideration is the preference of the patient. Some patients may greatly value the peace of mind provided by removal of their gallstones or the relief of symptoms that respond to cholecystectomy. and would be willing to undergo the slight mortality risk of elective surgery, or even to pay for it if they must. Others would prefer to live with their stones, even if they produce symptoms. rather than undergo the anxiety, transient pain. and small risk of early mortality associated with elective cholecystectomy.

of Gallstones

I

2

3

4

5.

6

I.

8.

9.

IO.

II.

12.

13.

14.

15.

16. 17.

IX.

19. Ackno~~,led~emenrs-We are grateful to May Joan Thomas for medical record abstraction Jaffe, Ph.D., for computer programming.

Kuwatani and and to Jay

20.

Ransohoff

135

REFERENCES DF. Gracle WA.

Wolfenson LB. Neuhauser D. Prophylactic cholecystectomy or expectant management for silent gallstones: a decision analysis to assess survival. Ann Intern Med 1983: 99: 199-204. Fitzpatrick G. Neutra R, Gilbert JP. Costeffectiveness of cholecystectomy for silent gallstones. In: Bunker JP, Barnes BA. Mosteller F. eds. Costs, Risks and Benefits of Surgery. New York: Oxford University Press; 1977: 24&26l. Kottke TE. Feldman RD. Albert DA. The risk ratio is Insufficient for climcal decisions: the case of Med Decis Making prophylactic cholecystectomy. 1984: 4: ill- 194. Peterson R. Gall-stones during the course of 1066 abdominal sectlons for pelvic disease. Surg Gynecol Obstet I91 5; 70: 784 291. Truesdell ED. The frequency and future of gallstones believed lo be quiescent or symptomless. Ann Surg 1944; 119: 232 245. Comfort MW. Gray HK. Wilson JM. The silent gallstone: a ten to twent) year follow-up study of II2 cases. Ann Surg 1948: 128: 931 937. Lund J. Surgical indicatlona in cholelithiasis: prophylactlc cholecystectomy elucidated on the basis of long-t%1 follow-up on 526 nonoperated cases. Ann Surg 1960: 151: 153 162. Wenckert A, Robertson B. The natural course of gallstone disease: eleven-year review of 781 nonoperated cases. Gastroenterology 1966; 50: 376.38 I. Gracie WA. Ransohoff DF. The natural history of silent gallstones: the innocent gallstone is not a myth. N Engl J Med 1982; 307: 798-800. Thistle JL. Clear) PA. Lachin JM. CI al. The natural history of cholelithiasis: the National Cooperative Gallstone Study. Ann Intern Med 1984; 101: I71 175. McShcrry CK, Ferstenberg H, Calhoun WF, Lahman E. Vlrshup M. The natural history of diagnosed gallstclne disease in symptomatic and asymptomatic patients. Ann Surg 1987: 202: 59-63. Hiatt RA, Friedman CD. The frequency of kidney and urinary tract diseases m a defined population. Kidney Int 1982: 22: 63-68. Frledman CD. Problems In determlnmg the prognosis of asymptomatic gallstones. In: Capocaccla L. Ricci G. Angelic0 F. Angelic0 M. Attili AF. eds. Epidemiology and Prevention of Gallstone Disease. Boston: MTP Press Ltd; 1984: 158-167. Friedman GD. Collen MF, Harris LE, Van Brunt EE, Davis LS. Experience In monitoring drug reactions m outpatients. The Kaiser-Permanente Drug Monitoring System. JAMA 1971; 217: 567-572. Van Brunt EE, Davis LS. Terdiman JF. Collen MI-. The Kaiser-Permanente Pilot Medical Data System. In: Collen MF, ed. Proc Int Conf on Health Technology Systems. Potomac. Mayland: Operations Research Society of America. Health Applications Section; 1974: 261~285. Collen MF. Davis LF. The multitest laboratory m health care. J Occup Med 1969; II: 355-360. Lee ET. Statistical Methods for Survival Data Analysis. Belmont, Cahfornia: Lifetlme Learning Publications: 1980: 8% 95. Peto R. Pike MC. Armitage P. cf uf. Design alld analysis of randomized clinical trials requiring prolonged observation of each patient. II. Analysis and examples. Br J Cancer 1977; 35: I-39. Sidney S. Friedman CD. Siegelaub AB. Thinness and mortality. Am J Public Health 1987; 77: 317-322. LoGerfo JP. Organizational and financial influences on patterns of surgical care. Surg Clin North Am 1981; 62: 677-684.

136 21.

GARY D. FRIEDMAN et al.

Marenghini A, Moreau JA, Melton LJ, Hench VS, Zinsmeister AR, DiMagno EP. Gallstones, gallbladder cancer, and other gastrointestinal malignancies. Ann Intern Med 1987; 107: 30-35. 22. Schwartz SI. Complications. In: Schwartz SI, Shires

GT, Spencer FC, Storer EH, eds. RIndpIes of Surgery, 4th edn. New York: McGraw-Hill; 1984: 455484. 23. Friedman GD, Goldhaber MK, Quesenberry CP Jr. Cholecystectomy and large bowel cancer. Lancet 1987; 1: 906-908.