Increased Mortality with Gallstone Disease: Results of a 20-Year Population-Based Survey in Pima Indians

Methods

Between 1966 and 1969 a cross-sectional study conducted in the Gila River Indian Community determined the prevalence of gallstone disease in Pima Indians [18]. An age- and sex-stratified random population sample of the census population enumerated in 1966, consisting of 50 men and 50 women in each of the six decades between 15 and 74 years of age, was examined. Medical records were reviewed, and oral cholecystograms were done on those without previously documented gallstone disease. Forty persons whose gallbladder status was unknown died before evaluation. The remaining 560 persons were classified into five groups: 1) radiographically normal gallbladder: n = 161; 2) gallstone disease discovered at the time of the oral cholecystogram but unknown before: n = 116; 3) gallstone disease already diagnosed before the study but no cholecystectomy by the time of the study: n = 29; 4) cholecystectomy before the survey: n = 77; and 5) unknown gallbladder status (cholecystography was contraindicated in 10, refused by 137, and inconclusive in 30): n = 177. Of the 116 persons with gallstone disease discovered by oral cholecystogram, 73 (63%) had nonvisualizing gallbladders. These persons were considered to have gallstone disease because previous studies in Pima Indians and in other groups have found that one nonvisualization after iopanoic acid is associated with the presence of cholelithiasis in at least 87% of the cases [18, 20, 21].

In our study, the 383 persons whose gallbladder status was ascertained during the initial survey were followed, and the death rates and underlying causes of death in this cohort were investigated. The date of entry into the study was either the date of the oral cholecystogram, or 1 January 1967 if a cholecystography was not done because of a previous diagnosis. The end point was 31 December 1986 or the date of death for persons who died before this date. The vital status of each person, with one exception, was verified through 31 December 1986. This person, last seen on 31 October 1986, was not known to have died and, for analytic purposes, was presumed to be alive during the last 2 months of the study.

Underlying causes of death were determined from death certificates, coded according to the International Classification of Diseases (ICD), 8th revision [22] through 1978 and 9th revision [23] from 1979 to the end of the study. Causes of death were classified into eight categories: 1) complications of gallstone disease, excluding gallbladder cancer [ICD codes 574.0 to 576.9]; 2) malignancy [ICD codes 140.0 to 239.9]; 3) trauma [ICD codes E800.0 to E999.9]; 4) cardiovascular disease [ICD codes 390 to 458.9 in the 8th revision and ICD codes 390.0-459.9 in the 9th]; 5) diabetes mellitus [ICD codes 250.0 to 250.9]; 6) infection [ICD codes 000.0 to 136; 320.0 to 324; 460 to 486; 540.0 to 543; 572; 590.0 to 590.9; 680.0 to 686.9 in the 8th revision and ICD codes 001.0 to 139.8; 320.0 to 326; 460.0 to 466.1; 480.0 to 487.8; 540.0 to 543.9; 572.0; 590.0 to 590.9; 599.0; 680.0 to 686.9 in the 9th]; 7) digestive disease other than malignant and infectious [ICD codes 520.0 to 537.9; 550.0 to 571.9; 573.0-577.9 in the 8th revision and ICD codes 520.0 to 537.9; 550.0 to 571.9; 572.1 to 579.9 in the 9th]; and 8) other (all ICD codes not listed above). Diabetes and trauma were included because of their high incidence in the population. The medical records of all persons whose deaths were attributed to diseases of the hepatobiliary system were reviewed, and the cause of death, based on this review of all available clinical information, was classified according to ICD conventions.

Age, sex, diabetes, serum cholesterol concentration, and obesity (estimated by the body mass index in kg/m²), were considered in addition to gallbladder disease as possibly confounding risk factors for death. These data were obtained at research examinations conducted at about 2-year intervals on each member of the Gila River Indian Community who was 5 years of age or older [24]. The data collected at the examination closest to the date of entry into the present study were used if that examination occurred within 3 years before or after study entry. Diabetes was diagnosed according to World Health Organization criteria [25] as described previously [19].

Statistical Analysis

Age-sex specific death rates were computed by dividing the number of deaths occurring in each age group by the person-years of follow-up in that age group. Persons were stratified by the presence of gallstone disease at baseline, and person-years of follow-up for each person were apportioned to the different age categories that he or she passed through. The number of deaths that occurred in each age category was then divided by the person-years of follow-up in that category to determine the death rate.

Death rate ratios according to gallbladder disease, death rate ratios for specific causes, and survival probability were computed using the proportional hazards model [26]. The rates and rate ratios were adjusted for sex and for age at entry by including age, sex, and an age-sex interaction term in the model. In addition, adjustment was made for the presence of diabetes, body mass index (including both linear and quadratic terms), and serum cholesterol concentration; interaction terms between these variables were also included when needed.

Results

Death Rates

During the 20-year follow-up, 186 of the 383 (49%) persons with known gallbladder status died. Of the 222 persons with either gallstones or previous cholecystectomy, deaths occurred in 18 of 29 (62%) persons with gallstones diagnosed before the survey, 70 of 116 (60%) persons with gallstones discovered at the time of the survey, and 45 of 77 (58%) persons with cholecystectomy before the survey. Overall, 133 of 222 (60%) persons with a positive gallstone history died compared to 53 of 161 (33%) without gallstone disease.

Death rates were similar between persons with previously diagnosed gallstones and those whose gallstones were discovered at the time of the survey (age-sex-adjusted death rate ratio, 0.9; 95% CI, 0.5 to 1.5). Further, the death rate in this combined group was the same as that in the 77 persons with previous cholecystectomy (age-sex-adjusted death rate ratio, 1.0; CI, 0.7 to 1.5) but was higher than that of the group with normal gallbladders (age-sex-adjusted death rate ratio, 1.9; CI, 1.2 to 3.2). Therefore, persons with gallstones diagnosed at or before the survey and those with previous cholecystectomy were considered together as a combined group of persons with gallstone disease. These persons were compared with the persons with normal gallbladders.

Table 1 presents death rates by age, sex, and presence of gallstone disease. Persons with gallstone disease had a higher death rate than those with normal gallbladders; the age-sex-adjusted death rate ratio was 1.9 (CI, 1.3 to 2.7). When adjusted for age, sex, body mass index, diabetes, and serum cholesterol concentration, the ratio was 1.8 (CI, 1.3 to 2.7). Figure 1 shows the probability of survival as a function of gallstone disease, age, and sex. Although death rates were higher in the men, the association of gallbladder disease with mortality rates did not differ statistically by sex.

View larger version:

• In this window
• In a new window

Figure 1. Survival probability was computed from a proportional hazards model of survival time as a function of gallbladder disease, age, sex, and age-sex interaction. To construct the figure, the model was evaluated for each sex at an entry age of 35 years (top panel) and 55 years (bottom panel). Survival according to presence or absence of gallstone disease at entry.

The death rate ratio for persons with gallstone disease, diabetes, or both diseases relative to persons with neither condition at entry into the study is shown in Figure 2. In both men and women, those with gallstone disease had almost twice the mortality of those without gallstone disease, and among women, those with diabetes had twice the mortality of those without diabetes. The presence of both gallstone disease and diabetes greatly increased mortality in women.

View larger version:

• In this window
• In a new window

Figure 2. Death rates for persons with these diseases were examined relative to persons with neither condition at entry into the study (for whom a ratio of 1.0 was assigned). The ratios were derived from a proportional hazards model of survival time as a function of gallstone disease, diabetes, age, and sex and evaluated for each sex at the mean age (48 years) of all persons. D = diabetes; GS = gallstone disease. Effect on death rate ratios of gallstone disease and diabetes.

Causes of Death

Two deaths were directly attributable to gallstone disease or its treatment; one person died from postoperative aspiration pneumonia following cholecystectomy, and one died from cholangitis. Two persons without gallstone disease at baseline died of acute pancreatitis; one of them had a solitary stone in the gallbladder during the episode of pancreatitis, but no dilatation of intrahepatic or cystic ducts was detected by ultrasound examination.

Table 2 shows death rate ratios by cause of death. The rate ratios were adjusted for age and sex in all 383 persons, and for age, sex, body mass index, serum cholesterol, and diabetes in the 360 persons for whom these data were available. When smoking was added to the model, only 334 persons were included because the smoking questionnaire was introduced after some persons had been enrolled. None of the adjusted rate ratios or their confidence intervals was substantially changed (data not shown). A higher death rate due to malignancies was found in persons with gallstone disease than in those without (age-sex-adjusted death rate ratio, 6.6; CI, 1.3 to 33.1). Malignancies, however, did not account for all the excess mortality among those with gallbladder disease because their rate of death due to other causes was also statistically higher (age-sex-adjusted death rate ratio, 1.7; CI, 1.1 to 2.4), but this excess was distributed among each of the other categories of causes.

Table 3 lists the 20 fatal malignancies in patients with gallstone disease and the two in those without. Among persons with gallstone disease, 11 were attributed to digestive tract malignancies, of which 6 involved the gallbladder or bile ducts. Only two deaths were attributed to malignancy (primary liver cancer and ovarian cancer) in the other persons.

Discussion

Total Mortality

Persons with gallstone disease, defined as gallstones or previous cholecystectomy, had statistically higher death rates than those without gallstone disease at the baseline examination. Little difference in mortality could be discerned among the subgroups of gallstone disease. Only two deaths were directly related to acute complications of gallstones or cholecystectomy. Persons with gallstone disease had a substantially higher death rate due to malignancies; six deaths were attributed to malignancies involving the gallbladder or bile ducts. The excess mortality was not affected by adjustment for potential confounding factors such as age, sex, obesity, serum cholesterol concentration, and diabetes, a frequent condition in this population [19, 24].

Other studies of mortality in patients with gallstone disease have found, like ours, that few patients die directly from gallstone disease [1, 2, 27-32]. None compared the experience of persons with and without gallstones within a specific population.

The persons without gallstones at entry into this study were not systematically re-examined to determine if they subsequently developed gallstones. Thus, persons who may have developed gallstones after the baseline examination were, for the purpose of this analysis, classified in the group without gallstone disease. Because any potential misclassification would reduce the difference between the persons with gallbladder disease and those with normal gallbladders, our results probably underestimate the importance of gallstone disease as a risk factor for death. These factors also reduced the power of the study to detect possible differences between groups for deaths due to specific causes. The effect of subsequent treatment, for example, cholecystectomy, on mortality was not assessed.

Cause-Specific Mortality

The underlying causes of death were determined from death certificates. Although in a previous study of Pima Indians, 22% of 677 death certificates were reclassified after review of available medical records, most reclassifications occurred for heart disease, diabetes, nephritis, nephrotic syndrome and nephrosis, and chronic liver disease [33], and only 1 of 54 deaths attributed to malignant neoplasms was reclassified.

We found the death rate from malignancy to be statistically higher in those with gallstone disease. Digestive tract malignancies accounted for 11 of the 20 cancer deaths; six were cancers of the gallbladder or biliary ducts. Our findings contrast with those reported from a longitudinal survey in Rochester, Minnesota [34], where no association between gallbladder disease and pancreatic, colonic, or gastric cancer was found. However, the diagnosis of gallstone disease was made clinically in that study rather than by population screening.

In a 15-country mortality study [5], asymptomatic gallstones at necropsy were associated with cancers of the uterus, large bowel, and stomach. In a case–control study from Sweden [6], the prevalence of gallstone disease at necropsy in patients who died from cancer was higher than the prevalence in those who died from other causes, but only among women under 50 years of age.

The relationship of gallstone disease with other malignancies of the digestive tract was evaluated in a case–control study by Hyvarinen and Partanen [7], who found a high rate of previous cholecystectomies among patients with esophageal or gastric cancers. By contrast, in an autopsy survey, Kalima and colleagues [35] found a lower prevalence of gallstones among patients with gastric cancer. In a longitudinal study, Linos and colleagues [8] found a statistical association between asymptomatic gallstones and colon cancer but concluded that it could have resulted from selection bias. Because patients with colon cancer are more likely to have their gallbladders studied than are healthy persons, asymptomatic cholelithiasis may be discovered more frequently, and an artifactual association may be found. The contention that cholecystectomy may lead to the development of colorectal cancer [9-13] has been disputed [36, 37].

An association between gallstones and gallbladder cancer was found in the previously cited Rochester study [34], but only among men. However, six persons whose adenocarcinoma of the gallbladder was diagnosed within 15 days of the initial diagnosis of gallstones were excluded, possibly affecting the conclusions [38]. Kimura and colleagues [39] found statistically higher rates of carcinoma of the gallbladder or extrahepatic bile ducts among either sex with gallstones in an autopsy survey of 4482 persons in Tokyo. Further, in a U.S. multicenter case–control study, a higher rate of gallstones was found among 131 patients with gallbladder cancer than in the control group [40], with an odds ratio of 4.4 among the non-Indian patients and 20.9 among the American Indian patients.

Persons who develop gallstone disease, especially at an early age, may experience higher mortality and may be at greater risk of developing malignancies at sites in addition to the biliary tract. Gallstone disease in Pima Indians is characterized by an early age of onset [18], which may provide sufficient time for deleterious effects of the disease, especially the carcinogenic ones, to manifest. For example, bacterial transformation of bile components in persons with gallstone disease, even after cholecystectomy, may induce cancer in the biliary tract or the gut [41-47]. Malignancies elsewhere in the body may also be explained if some of the active carcinogens contained in the bile undergo intestinal absorption and escape hepatic detoxification.

It is important to stress that malignancy did not account for all the excess mortality associated with gallstone disease. The age-sex-adjusted death rate for causes other than malignancy was statistically higher in persons with gallstone disease than in those without. The reasons are unknown, but the higher rate of traumatic death in persons with gallstone disease may be attributable to alcohol consumption. Alcoholism and cirrhosis have been associated with pigment gallstone formation [48]. Other studies, however, have reported a negative association between alcohol consumption and gallstone formation [49-51]. In addition, previous studies have found an association between gallstone disease and cardiovascular diseases [15, 16]. The lack of an association in our study may be due to the low rate of coronary artery disease in this population [52-54].

Although plausible explanations exist for the increased cancer mortality in Pima Indians with gallstone disease, the increased death rates due to other causes are unexplained. Whether primary prevention of gallstone disease or cholecystectomy alters risk for either is unknown and deserves further investigation. The immediate prospects for prevention of gallstone disease, however, are uncertain. On the other hand, surgery for the disease is readily available in developed countries but is not generally recommended in persons with silent disease because of its expense and morbidity and because few persons with gallstones develop symptoms [1, 55], but our findings suggest the need for continuing examination of the management of silent gallstone disease.