Clinical Manifestations of Gallstone Disease: Evidence From the Multicenter Italian Study on Cholelithiasis (MICOL) DAVIDE FESTI,1 SANDRA SOTTILI,2 ANTONIO COLECCHIA,2 ADOLFO ATTILI,3 GIUSEPPE MAZZELLA,2 ENRICO RODA,2 FERDINANDO ROMANO,4 AND THE MICOL RESEARCH GROUP* Despite the many efforts to delineate the clinical manifestations of gallbladder disease, the precise symptom complex associated with gallstones is still a matter of debate, and even the existence of gallstone-specific symptoms has been questioned. We carried out a large population-based cross-sectional study (MICOL) to identify symptoms significantly related to gallstones. Fourteen centers throughout Italy enrolled 29,504 subjects aged 30 to 69 years. All subjects were administered an ultrasonographic examination of the upper abdomen and a precoded questionnaire. All subjects were divided into 4 groups: 25,374 (86.0%) gallstone-free subjects (GF), 1,832 (6.2%) patients with gallstones not previously diagnosed (GNPD), 638 (2.2%) patients with gallstones previously diagnosed (GPD), 1,660 (5.6%) patients with a history of cholecystectomy for gallstones (CC). In logistic regression analysis, pain at epigastrium and, even more, pain at right hypocondrium were significantly associated with gallstones. For pain at right hypocondrium, this association progressively increased from GNPD (OR 5 1.60, 95% CI 5 0.97-2.65) to GPD (OR 5 8.77, 95% CI 5 5.27-14.61) to CC (OR 5 59.40, 95% CI 5 43.87-80.42). Absence of heartburn combined with right hypocondrium or epigastrium pain and intolerance to fried or fatty food were also significantly related to gallstones. We also found some pain characteristics significantly associated with gallstones, i.e., pain radiated to the right shoulder, forcing the patient to rest, occurring soon after meals or unrelated to meals, not relieved by bowel movements, and frequently accompanied by gallstonerelated morbidities. We developed a probability tree reporting the cumulative probability of having gallstones for each Abbreviations: US, ultrasonography; OR, odds ratio; CI, confidence intervals; BMI, body mass index; GF, gallstone-free subjects; GNPD, patients with gallstone not previously diagnosed; GPD, patients with gallstones previously diagnosed; CC, patients with a history of cholecystectomy for gallstones. From the 1Department of Medicine and Ageing, University of Chieti ‘‘G. d’Annunzio’’; 2Department of Internal Medicine and Gastroenterology, University of Bologna; 3Department of Clinical Medicine, University of Rome ‘‘La Sapienza’’; and 4Department of Biomedical Sciences, University of Chieti ‘‘G. d’Annunzio’’, Italy. Received January 5, 1999; accepted July 13, 1999. Supported in part by the Italian National Research Council (Targeted Projects ‘‘Preventive and Rehabilitative Medicine’’ 1982-1987 and ‘‘Prevention and Control of Risk Factors’’ 1990-1995). *For the composition of the MICOL Research Group see Attili.11 Address reprint requests to: Davide Festi, M.D., Cattedra di Gastroenterologia, Dipartimento di Medicina e Scienze dell’Invecchiamento, Palazzina SEBI, Università degli Studi di Chieti ‘‘G. d’Annunzio,’’ Via dei Vestini, 66013 Chieti, Italy. E-mail: dfesti@unich.it; fax:39-871-355-6706. Copyright r 1999 by the American Association for the Study of Liver Diseases. 0270-9139/99/3004-0003$3.00/0 combination of those symptoms and characteristics of pain significantly associated with gallstones. In conclusion, we have identified symptoms and signs significantly associated with gallstones. We have shown that there is an increase in frequency and severity of these symptoms and signs across the different stages of gallstone disease. We have proposed a complex of symptoms and signs significantly associated with gallstones that might help physicians in clinical decision making. (HEPATOLOGY 1999;30:839-846.) Despite the many efforts to delineate the clinical manifestations of gallbladder disease, the precise symptom complex associated with gallstones is still a matter of debate, and even the existence of gallstone-specific symptoms has been questioned.1-3 Most of the attention has been focused on pain in gallstone patients, but there is no general agreement on its clinical meaning3 and it is questioned whether pain occurs in gallstone disease with specific characteristics.1,2 The identification of symptoms associated with this disease might help physicians in clinical decision making.4-7 Moreover, the possibility of identifying patients who are more likely to have gallstones through a clinical assessment would limit the excessive and inappropriate use of diagnostic procedures for detection of gallstones and would lead to more specific and accurate therapeutic actions.5-7 This study is part of the MICOL project,8 and it is aimed at assessing symptoms associated with gallstones. The MICOL project started in 1985 and was designed to investigate gallstone disease in terms of prevalence, incidence, natural history, risk factors, and disease markers. Data on prevalence and partial information on factors associated with the disease have already been published.9,10 The incidence study is in progress. PATIENTS AND METHODS Study Design. The MICOL project is a population-based crosssectional study, carried out in 8 Italian regions by 14 operative units. The project plan includes 2 cross-sectional surveys. The first survey started in 1985 and was completed in 1987. The second survey, planned to be carried out on the same subjects after 10 years to estimate the incidence of gallstone disease as well as its natural history, is still in progress. Complete details on the study protocol have been published elsewhere.9,10 Subjects. Forty-six thousand one hundred thirty-nine (46,139) subjects (25,443 men and 20,696 women aged 30-69 years) were eligible for the study. Subjects were recruited through the electoral rolls for the larger cities and through municipal census in the smaller towns. Twenty-nine thousand five hundred four (29,504) subjects agreed to participate (15,866 men and 13,638 women). The overall response rate was 63.9%. A slightly higher participation 839 840 FESTI ET AL. rate was recorded for females (65.9%), whereas no differences were observed between age groups. Screening Protocol. The screening protocol included: upper abdomen ultrasonography (US), physical examination, fasting blood specimen collection, administration of a precoded questionnaire regarding family and personal history, dietary habits, past and current medical history, use of medications, detailed and documented information on a previous diagnosis of gallstones. A preliminary reliability study was performed for US examination11 and questionnaire administration.9 Evaluation of Abdominal Symptoms. The questionnaire was administered by trained interviewers selected within the medical staff of each Operative Unit. Upper abdomen US was performed after the questionnaire by physicians different from the interviewers. The questionnaire included a specific section targeted to assess in the 5 years before the interview the occurrence of the following symptoms: belching, heartburn, nausea, vomiting, bloated feeling soon after meals, intolerance to fatty or fried foods, bitter taste in the morning, uncomfortable feeling at the right hypocondrium, epigastric discomfort, and abdominal pain. Patients reporting abdominal pain were asked to answer the following questions on the characteristics of the pain: localization (on a diagram of the abdomen); radiation (up, down, right, left, back, right shoulder, left shoulder); duration and tolerability (whether forcing to rest or not); disappearance with bowel movements; use of analgesics to relieve the pain; frequency in the last year; temporal relationship with meals (before or during, soon after, unrelated to); and concomitant occurrence of jaundice, dark urine, fever, as suggestive of gallstone disease complications. For patients with a history of cholecystectomy, this information had to be referred to the 5 years before the surgery. Statistical Analysis. Patients were classified into 4 groups according to their disease status: gallstone-free subjects, patients with gallstones not previously diagnosed, patients with previously diagnosed gallstones, and patients with a history of cholecystectomy for gallstones. For the last category, the questionnaire sought information concerning abdominal symptoms in the 5 years before cholecystectomy. This information could have been affected by a recall bias depending on the time elapsed from the surgery to the interview, leading to a nondifferential misclassification. An evaluation of the degree of this potential misclassification was preliminarily performed to assess the suitability of the cholecystectomized group for the analysis. This group was stratified according to the number of years elapsed between the surgery and the interview. The information reported by the patients was consistent across the strata, suggesting that the accuracy of the information was not substantially affected by the year when the cholecystectomy was performed. This finding did not rule out the possibility of a recall bias, but it suggested that a recall bias, if present, did not substantially affect our data. Therefore, the cholecystectomized group was included in the analysis. Because of the awareness of their disease status, patients with a previous documented diagnosis of gallstones might have produced biased information concerning the occurrence of pain and its characteristics, leading to an overestimation of the association between these symptoms and gallstones. The degree of this potential bias was assessed by comparing the frequency of the different pain characteristics between gallstone patients with and without a previous diagnosis, matched for disease severity and pain location. Because no differences were observed between these 2 groups, the information concerning both the occurrence of pain and its relevant characteristics was very unlikely to be biased by the patient’s awareness of having gallstones. After this preliminary validation of the information provided with the questionnaire, the frequency of each symptom was compared among the different groups in terms of unadjusted odds ratios (ORs) and 95% confidence interval (CI) using the gallstone-free subjects as the comparison group. Further analyses were performed using HEPATOLOGY October 1999 multivariate techniques to adjust for age, sex, body mass index (BMI), and concomitant diseases previously shown to be significantly related to gallstone disease (diabetes, cirrhosis, myocardial infarction, angina, peptic ulcer).7 Three logistic regression models were created, 1 for each of the gallstone groups, using the gallstonefree subjects as the comparison group. As a consequence of the results of the aforementioned analyses, patients referring pain located at the right hypocondrium or epigastrium were selected and further evaluated. The distribution of the different pain characteristics was compared among the 4 disease categories, and unadjusted ORs and 95% CI were computed. A multivariate analysis was subsequently performed using logistic regression, and for each gallstone group a model was created using the gallstone-free subjects as the comparison group. We carried out further analyses to evaluate whether those symptoms and characteristics of pain that were significantly associated with gallstone might contribute to the diagnostic process in clinical practice. Therefore, we identified from our study population a subset of subjects that might simulate the population of subjects who seek medical advice for pain at the right hypocondrium or at the epigastrium and who have not been yet diagnosed with gallstone disease, if any. Accordingly, we selected those subjects who reported seeking medical advice for pain at the right hypocondrium or at the epigastrium during the year preceding the interview, who had not had a diagnosis of gallstone before this visit, and who had or had not diagnosed with gallstones at the time of this visit. A logistic regression analysis was performed on this subset of subjects to identify those symptoms and characteristics of pain that might be significant predictors of gallstone in this simulation of a clinical setting. Using the ORs estimated from the logistic regression model we computed the probability of having gallstone for each combination of those symptoms and characteristics of pain significantly associated with gallstones.12 RESULTS Of the 29,504 subjects enrolled in the study, 25,374 (86.0%) were gallstone-free (GF), 1,832 (6.2%) were patients with gallstones not previously diagnosed (GNPD), 638 (2.2%) were patients with previously diagnosed gallstones (GPD), 1,660 (5.6%) were patients with a history of cholecystectomy for gallstones (CC). Table 1 shows the distribution of abdominal symptoms in the 4 different categories, as well as the crude estimates of the association (unadjusted ORs) between each symptom and gallstones assessed for each disease category. The frequency of each symptom increased from GF up to CC, with higher increases in nausea, vomiting, uncomfortable feeling at the right hypocondrium, pain at the epigastrium, and pain at the right hypocondrium. Notably, using the GF patients as the reference category, the OR for pain at the epigastrium was 1.42 (GNPD), 3.53 (GPD), and 19.99 (CC). A more striking increase was observed for pain at the right hypocondrium, moving from 1.76 (GNPD) to 8.53 (GPD) to 48.63 (CC). No differences were observed among the 4 groups for the frequencies of pain in sites other than epigastrium and right hypocondrium, and just a slight increase across the groups for the relevant ORs. Table 2 shows the results of the logistic regression analysis. The ORs were adjusted for age, sex, BMI, and concomitant diseases. After this adjustment, only a few symptoms were still significantly associated with gallstone disease. Comparing GNPD and GF, pain at the epigastrium (OR 5 1.64, 95% CI 5 1.12-2.38), pain at the right hypocondrium (OR 5 1.60, 95% CI 5 0.97-2.65), uncomfortable feeling at the right HEPATOLOGY Vol. 30, No. 4, 1999 FESTI ET AL. 841 TABLE 1. Distribution of Symptoms (and Unadjusted ORs) in Gallstone-Free Subjects, Patients With Gallstones not Previously Diagnosed, Patients With a Previous Diagnosis of Gallstones, and Patients With a History of Cholecystectomy for Gallstones Symptoms Belching (%) OR (95% CI) Heartburn (%) OR (95% CI) Nausea (%) OR (95% CI) Vomiting (%) OR (95% CI) Bloated feeling after meals (%) OR (95% CI) Intolerance to fatty or fried foods (%) OR (95% CI) Bitter taste in the morning (%) OR (95% CI) Uncomfortable feeling at the right hypocondrium (%) OR (95% CI) Epigastric discomfort (%) OR (95% CI) Pain at other sites (%)* OR (95% CI) Pain at the epigastrium (%)* OR (95% CI) Pain at the right hypocondrium (%)* OR (95% CI) Gallstone-Free Subjects (n 5 25,374) Gallstones not Previously Diagnosed (n 5 1,832) Gallstones Previously Diagnosed (n 5 638) Cholecystectomized (n 5 1,660) 19.6 20.8 1.08 (0.96-1.22) 29.5 0.94 (0.85-1.05) 14.0 1.05 (0.92-1.21) 7.6 1.10 (0.92-1.32) 38.0 1.11 (1.00-1.22) 36.8 1.14 (1.03-1.26) 36.6 1.05 (0.95-1.16) 24.1 1.29 (1.15-1.44) 22.8 1.00 (0.89-1.12) 4.8 0.91 (0.73-1.14) 6.2 1.42 (1.17-1.74) 6.2 1.76 (1.44-2.15) 28.9 1.67 (1.40-1.99) 31.8 1.05 (0.89-1.25) 22.7 1.90 (1.57-2.30) 11.4 1.73 (1.35-2.22) 51.4 1.91 (1.63-2.23) 57.1 2.61 (2.22-3.06) 51.9 1.96 (1.67-2.30) 47.9 3.72 (3.17-4.36) 31.0 1.51 (1.28-1.80) 5.5 1.43 (1.01-2.03) 11.4 3.53 (2.73-4.58) 22.4 8.53 (6.97-10.45) 38.3 2.55 (2.30-2.83) 44.9 1.84 (1.66-2.03) 41.7 4.63 (4.17-5.14) 36.2 7.62 (6.81-8.51) 59.2 2.62 (2.37-2.90) 65.6 3.74 (3.36-4.15) 64.4 3.28 (2.96-3.64) 58.1 5.60 (5.06-6.21) 45.8 2.84 (2.57-3.14) 4.0 2.75 (2.11-3.59) 24.7 19.99 (17.18-23.26) 49.1 48.63 (42.37-55.82) 30.7 13.4 6.9 35.6 33.8 35.5 19.8 22.9 5.4 4.5 3.7 NOTE. In computing the unadjusted odds ratios the gallstone-free subjects have been used as the comparison group. *The category ‘‘no pain’’ has been taken as baseline. hypocondrium (OR 5 1.19, 95% CI 5 1.04-1.36), and interactions between no heartburn and both pain at the epigastrium and pain at the right hypocondrium were significantly associated with gallstones. Comparing GPD and GF, the association with gallstones became significant for intolerance to fried and fatty food (OR 5 2.03, 95% CI 5 1.65-2.51), increased markedly for pain at the right hypocondrium (OR 5 8.77, 95% CI 5 5.27-14.61), slightly increased for pain at the epigastrium (OR 5 2.48, 95% CI 5 1.37-4.47), and for uncomfortable feeling at the right hypocondrium (OR 5 2.60, 95% CI 5 2.10-3.22). These associations were confirmed in CC, with a further increase of the ORs for pain at the right hypocondrium (OR 5 59.40, 95% CI 5 43.8780.42) and at the epigastrium (OR 5 17.14, 95% CI 5 12.5023.50). In all the models, previously reported pain located either at the epigastrium or at the right hypocondrium was consistently associated with gallstones, whereas pain at other sites was unrelated to the disease. Therefore, the 3,732 patients reporting pain at the epigastrium or at the right hypocondrium were selected and further analyzed to investigate if specific pain characteristics could be related to gallstones. Some characteristics were significantly associated with gallstones, showing a progressive increase in the ORs throughout the 3 disease categories (Table 3). The ORs for pain radiated to the right shoulder were 3.75 (95% CI 5 2.12-6.64) for GNPD, 5.31 for GPD (95% CI 5 3.14-9.00), and 6.13 (95% CI 5 4.34-8.66) for CC. The ORs for pain forcing to rest were 1.68 (95% CI 5 1.27-2.21) for GNPD, 3.35 for GPD (95% CI 5 2.50-4.46), and 7.38 (95% CI 5 6.26-8.60) for CC. An TABLE 2. Results of the Multivariate Analysis Performed to Identify the Abdominal Symptoms Associated With Gallstones Variable Localization of pain (no pain as baseline) (1) Epigastrium (2) Right hypocondrium Intolerance to fried or fatty food (no intolerance as baseline) Uncomfortable feeling at the right hypocondrium (no feeling as baseline) Interactions Pain at the epigastrium 3 no heartburn Pain at the right hypocondrium 3 no heartburn Gallstones not Previously Diagnosed OR (95% CI) Gallstones Previously Diagnosed OR (95% CI) Cholecystectomized OR (95% CI) 1.64 (1.12-2.38) 1.60 (0.97-2.65) 1.09 (0.97-1.22) 1.19 (1.04-1.36) 2.48 (1.37-4.47) 8.77 (5.27-14.61) 2.03 (1.65-2.51) 2.60 (2.10-3.22) 17.14 (12.50-23.50) 59.40 (43.87-80.42) 2.40 (1.95-2.96) 2.75 (2.23-3.38) 1.47 (0.98-2.21) 1.46 (0.96-2.21) 3.58 (2.14-5.99) 2.14 (1.44-3.18) 2.06 (1.59-2.67) 1.38 (1.12-1.70) NOTE. Three logistic regression models are reported, comparing patients with gallstones not previously diagnosed, patients with a previous diagnosis of gallstones, and patients with a history of cholecystectomy for gallstones with gallstone-free subjects, respectively. All the models have been adjusted for age, sex, BMI, and concomitant diseases. 842 FESTI ET AL. HEPATOLOGY October 1999 TABLE 3. Distribution of Pain Characteristics (and Unadjusted ORs) in Subjects Reporting Pain at the Right Hypocondrium or at the Epigastrium Characteristics of Pain Tolerability Not forcing to rest % Forcing to rest % OR (95% CI) Radiation to the right shoulder Not radiated to the right shoulder % Radiated to the right shoulder % OR (95% CI) Relation with meals Before or during meals %* Unrelated to meals % OR (95% CI) Soon after meals % OR (95% CI) Duration Less than 1⁄2 h % More than 1⁄2 h % OR (95% CI) Frequency in the last year More than 1/month % Less than 1/month % OR (95% CI) Analgesic consumption No % Yes % OR (95% CI) Relieved by bowel movements Yes % No % OR (95% CI) Clinical signs of gallstone complications No % Yes % OR (95% CI) GallstoneFree Subjects (n 5 2,084) Gallstones not Previously Diagnosed (n 5 227) Gallstones Previously Diagnosed (n 5 214) Cholecystectomized (n 5 1,207) 68.90 31.10 56.90 43.10 1.68 (1.27-2.21) 39.80 60.20 3.35 (2.50-4.46) 23.10 76.90 7.38 (6.26-8.60) 97.9 2.1 92.5 7.5 3.75 (2.12-6.64) 89.7 10.3 5.31 (3.14-9.00) 88.3 11.7 6.13 (4.34-8.66) 21.50 37.30 11.80 35.70 1.74 (1.11-2.79) 52.50 2.33 (1.50-3.65) 7.10 39.00 3.17 (1.80-5.56) 53.80 3.96 (2.27-6.87) 11.40 35.20 1.78 (1.42-2.27) 53.50 2.45 (1.98-3.11) 25.30 74.70 20.90 79.10 1.28 (0.92-1.81) 19.20 80.80 1.43 (1.00-2.05) 11.70 88.30 2.56 (2.10-3.18) 44.50 55.50 31.00 69.00 1.78 (1.33-2.41) 22.50 77.50 2.76 (1.98-3.85) 58.20 41.80 0.58 (0.50-0.67) 57.30 42.70 53.30 46.70 1.18 (0.89-1.54) 40.40 59.60 1.98 (1.49-2.64) 30.20 69.80 3.10 (2.67-3.60) 27.10 72.90 23.10 76.90 1.24 (0.89-1.73) 19.40 80.60 1.54 (1.08-2.22) 14.70 85.30 2.16 (1.80-2.65) 71.10 28.90 58.10 41.90 1.77 (1.34-2.36) 46.30 53.70 2.85 (2.12-3.81) 34.20 65.70 4.73 (4.03-5.43) 41.10 NOTE. In computing the unadjusted odds ratios the gallstone-free subjects have been used as the comparison group. *The category ‘‘before or during meals’’ has been taken as baseline. analogous trend was observed for clinical signs of gallstone complications (jaundice, fever, dark urine), a duration of pain greater than 30 minutes, analgesic consumption, and pain not relieved by bowel movements. A weaker association, but still statistically significant, was observed for pain unrelated to meals or occurring soon after meals. A low frequency of pain episodes in the last year (,1/month) was positively associated with gallstone disease, both in GPD (OR 5 1.78, 95% CI 5 1.33-2.41) and GNPD (OR 5 2.76, 95% CI 5 1.983.85). On the contrary, in the CC group, a low frequency (,1/month) of pain episodes in the last year (intended as the year before cholecystectomy) was negatively associated with gallstone disease (OR 5 0.58, 95% CI 5 0.50-0.67). Table 4 shows the results of the logistic regression analysis. The ORs were adjusted for age, sex, BMI, and concomitant diseases (diabetes, cirrhosis, myocardial infarction, angina, peptic ulcer). Comparing GF with GNPD, the variables significantly associated with gallstones were pain radiated to the right shoulder (OR 5 3.55, 95% CI 5 1.84-6.82), pain onset unrelated to meals (OR 5 1.61, 95% CI 5 1.00-2.61) or soon after meals (OR 5 2.22, 95% CI 5 1.40-3.53), pain not relieved by bowel movements (OR 5 1.46, 95% CI 5 1.022.08), and clinical signs of gallstone complications (OR 5 1.56, 95% CI 5 1.14-2.14). The model comparing GPD with GF confirmed these associations, with a slight increase in the relevant ORs, and evidenced a further significant association for pain forcing patients to rest (OR 5 2.25, 95% CI 5 1.59-3.17). The last model, comparing CC with GF, was in line with the earlier findings, although a stronger association of gallstones with pain forcing to rest was observed (OR 5 4.06, 95% CI 5 3.31-4.97), and consumption of analgesics was significantly related to the disease (OR 5 1.68, 95% CI 5 1.38-2.06). Table 5 shows the results of the logistic regression analysis performed to identify symptoms and characteristics of pain predictive of gallstones in a subset of 1,887 subjects that emulates a real clinical setting, where subjects with pain at the right hypocondrium or at the epigastrium who were not previously diagnosed with gallstones seek medical advice. After adjusting for sex, age, BMI, and concomitant diseases, we found that absence of heartburn, pain radiated to the right shoulder, pain forcing the patients to rest, pain occurring soon after meals or unrelated to meals, pain unrelieved by bowel movements, and clinical signs of gallstone complications were significantly associated with gallstones. Using the HEPATOLOGY Vol. 30, No. 4, 1999 FESTI ET AL. ORs estimated from the logistic regression model, we created a probability tree that reports the cumulative probability of having gallstone for each combination of those symptoms and characteristics of pain significantly associated with gallstones (Fig. 1). According to the probability tree, the concomitant presence of all symptoms and signs associated with gallstones raises the probability of detecting gallstones up to 95.9%. DISCUSSION Over the past decades many investigators have sought to clearly delineate the clinical manifestations of gallstone disease.1 Nevertheless, the epidemiological studies addressing this issue have produced controversial results.2 In part, these contradictory findings are related to differences in research design. Some studies are uncontrolled,13-15 other reports present findings from patients seeking medical care,16-18 from case-control studies,15,19-21 or from community-based studies on selected22-27 or unselected28-31 populations. A recent meta-analysis, which takes into account all the studies issued on this topic from 1983 to 1992, concludes that only upper abdominal pain (with no preference for the right side) TABLE 4. Results of the Multivariate Analysis Performed to Identify Pain Characteristics Associated With Gallstones in Subjects Reporting Pain at the Right Hypocondrium or at the Epigastrium Variable Radiated to the right shoulder (not radiated to the right shoulder as baseline) Duration (,1⁄2 h as baseline) Tolerability (not forcing to rest as baseline) Relation with meals (before or during meals as baseline) Unrelated to meals Soon after meals Unrelieved by bowel movements (relieved as baseline) Clinical signs of gallstone complications Analgesic consumption Gallstones not Previously Diagnosed OR (95% CI) Gallstones Previously Diagnosed OR (95% CI) Cholecystectomized OR (95% CI) 3.55 (1.84-6.82) 4.20 (2.21-7.95) 4.11 (2.60-6.48) 1.17 (0.80-1.71) 0.87 (0.58-1.30) 1.29 (0.99-1.66) 1.36 (0.98-1.87) 2.25 (1.59-3.17) 4.06 (3.31-4.97) 1.61 (1.00-2.61) 2.63 (1.44-4.81) 2.22 (1.40-3.53) 3.68 (2.04-6.64) 1.38 (1.04-1.85) 2.12 (1.60-2.81) 1.46 (1.02-2.08) 1.69 (1.14-2.51) 2.31 (1.81-2.94) 1.56 (1.14-2.14) 1.99 (1.44-2.76) 2.78 (2.29-3.37) 1.06 (0.78-1.45) 1.25 (0.89-1.76) 1.68 (1.38-2.06) NOTE. Three logistic regression models are reported comparing patients with gallstones not previously diagnosed, patients with a previous diagnosis of gallstones, and patients with a history of cholecystectomy for gallstones with gallstone-free subjects, respectively. All the models have been adjusted for age, sex, BMI, and concomitant diseases. 843 TABLE 5. Results of the Multivariate Analysis Performed to Identify Signs and Symptoms Associated With Gallstones in a Subset of 1,887 Subjects That Emulates a Real Clinical Setting, Where Subjects With Pain at the Right Hypocondrium or at the Epigastrium and Without a Previous Diagnosis of Gallstones Seek Medical Advice Variable OR (95% CI) No heartburn (heartburn as baseline) Radiated to the right shoulder (not radiated to the right shoulder as baseline) Tolerability (not forcing to rest as baseline) Relation with meals (before or during meals as baseline) Unrelated to meals Soon after meals Unrelieved by bowel movements (relieved as baseline) Clinical signs of gallstone complications 1.64 (1.22-2.19) 5.09 (2.68-9.65) 1.97 (1.44-2.69) 2.16 (1.33-3.50) 2.53 (1.57-4.06) 1.84 (1.26-2.70) 1.83 (1.35-2.47) NOTE. Only those subjects who reported seeking medical advice for pain at the right hypocondrium or at the epigastrium during the year preceding the interview, who had not had a diagnosis of gallstone before this visit, and who had or had not diagnosed with gallstone at the time of this visit have been included in the model. The model has been adjusted for age, sex, BMI, and concomitant diseases. and no other abdominal symptom shows a consistent association with gallstones.2 We have studied a large general population with the objective of identifying symptoms significantly associated with gallstones. As a general consideration, in the assessment of symptoms or signs associated with a given disease, the choice of the groups to compare is of crucial importance. Because almost all diseases have a subclinical stage, the comparison of all diseased patients taken as a single category with undiseased subjects leads to an underestimation of symptoms and/or signs associated with the disease. The higher the prevalence of the subclinical stage, as in the case of gallstone disease, the higher the dilution of the association, if any. Based on these considerations, we have split the subjects enrolled in the study into 4 categories reproducing the different stages of gallstone disease: absence of disease, silent disease, overt disease, severe disease. This approach let us not only to identify symptoms and signs associated with gallstones but also let us assess whether frequency and severity of these symptoms and signs would increase throughout the 3 disease categories. We found that pain at the right hypocondrium, epigastric pain, and, to a lesser extent, intolerance to fried or fatty food were associated with gallstones. In patients reporting pain, whether at the epigastrium or at the right hypocondrium, the absence of heartburn increased the likelyhood of having gallstones. Our finding that an uncomfortable feeling at the right hypocondrium is significantly related to gallstones might reflect a true association or it could represent a misclassification of pain from the interviewed patients. We also found that pain occurring in gallstone patients, whether at the right hypocondrium or at the epigastrium, is preferentially radiated to the right shoulder, forcing patients to rest, occurring soon after meals or unrelated to meals, not relieved by bowel movements, and frequently accompanied by gallstone-related morbidities. These symptoms and signs increased in frequency and severity across the 3 disease categories. In fact, when GNPD patients were compared with GF subjects, only a weak association was found between gallstones and pain located at the right hypocondrium, pain at the epigastrium, and the absence of heartburn in presence of pain. In the comparison 844 FESTI ET AL. HEPATOLOGY October 1999 HEPATOLOGY Vol. 30, No. 4, 1999 of GPD with GF, these associations became much stronger, and a relation between intolerance to fried or fatty food and gallstones was also evidenced. The last category (CC), representative of the most severe stage of gallstone disease, showed the highest ORs for pain at the right hypocondrium and pain at the epigastrium. We considered the possibility that the information bias for GPD, partly discussed in the Patients and Methods section, might have lead to a spurious association of pain at the epigastrium or pain at the right hypocondrium with gallstones. Nonetheless, if this bias had been present, it would have had the same effect on both the aforementioned associations, with a similar increase in the relevant ORs as compared with the ORs observed for GNPD. The OR for pain at the right hypocondrium was much higher than that for pain at the epigastrium, suggesting that this bias, if any, had only minimally affected our findings. On the contrary, it is likely that an underestimation of these associations has occurred because of the presumable inclusion in the GPD category of patients with a silent disease accidentally diagnosed. As far as the characteristics of pain are concerned, we have found that for those characteristics that are not an expression of the degree of disease severity (radiation to the right shoulder, occurrence soon after meals or occurrence unrelated to meals, and unrelief with bowel movements), the strength of the association with gallstones remained stable across the 3 disease categories. On the contrary, for those characteristics that are an expression of the degree of disease severity (pain forcing patients to rest, drug consumption for pain relief, and clinical signs of gallstone complications) the association with gallstones strengthened progressively from GNPD to GPD, and to CC. These findings support the hypothesis that the natural history of gallstone disease moves from a silent to a clinically evident stage. Finally, we have made an attempt to provide physicians with a positive clinical format for diagnosing gallstones. We have proposed a probability tree developed on the basis of data provided from a subset of patients that closely approaches the context of a real clinical setting. Each branch of the probability tree includes a unique combination of symptoms with an associated probability of having gallstones given that combination of symptoms. All possible combinations of symptoms are reported in the probability tree. The probability of having gallstones ranges from 14.2%, when only pain at the epigastrium or at the right hypocondrium is reported, to 95.9%, when all symptoms found to be related to gallstones are reported. In conclusion, we have identified symptoms and signs significantly associated with gallstones. We have shown that there is an increase in frequency and severity of these symptoms and signs across the different stages of gallstone disease. We have proposed a complex of symptoms and signs significantly associated with gallstones that might help physicians in clinical decision making. Acknowledgment: The authors thank Dr. Franca Daniele for her technical assistance in the preparation of the manuscript. REFERENCES 1. Diehl AK. Symptoms of gallstone disease. Bailliere’s Clin Gastroenterol 1992;6:635-657. FESTI ET AL. 845 2. Kraag N, Thijs C, Knipschild P. Dyspepsia—How noisy are gallstones? A meta-analysis of epidemiologic studies of biliary pain, dyspeptic symptoms, and food intolerance. Scand J Gastroenterol 1995;30:411-421. 3. Jorgensen T. Abdominal symptoms and gallstone disease: an epidemiological investigation. HEPATOLOGY 1989;9:856-890. 4. Hussaini SH. Clinical economics review: the management of gallstone disease. Aliment Pharmacol Ther 1996;10:699-705. 5. Strasberg SM, Clavien PA. Cholecystolithiasis: lithotherapy for the 1990s. HEPATOLOGY 1992;16:820-839. 6. Paumgartner G, Carr-Locke D, Dubois F, Roda E, Thistle JL. Strategies in the treatment of gallstone disease. Gastroenterol Int 1994;5:177-188. 7. Ransohoff DF, Gracie WA. Treatment of gallstones. Ann Intern Med 1993;119:606-619. 8. The MICOL GROUP. Prevalence of gallstone disease in 18 italian population samples: first results from the MICOL study. In: Capocaccia L, Ricci G, Angelico F, Angelico M, Attili AF, Lalloni L, eds. Recent Advances in the Epidemiology and Prevention of Gallstone Disease. Dordrecht: Kluwer Academic Publishers, 1991;37-44. 9. Attili AF, Carulli N, Roda E, Barbara L, Capocaccia L, Menotti A, Okolicsanyi L, et al. Epidemiology of gallstone disease in Italy: prevalence data of the Multicenter Italian Study on Cholelithiasis (M.I.COL.). Am J Epidemiol 1995;141:158-165. 10. Attili AF, Capocaccia R, Carulli N, Festi D, Roda E, Barbara L, Capocaccia L, et al. Factors associated with gallstone disease in the MICOL experience. HEPATOLOGY 1997;26:809-818. 11. Festi D, Lalloni L, Taroni F, Barbara L, Menotti A, Ricci G. Inter and intra-observer variation in ultrasonographic detection of gallstones: the Multicenter Italian Study on Epidemiology of Cholelithiasis (M.I.COL.). Eur J Epidemiol 1989;5:51-57. 12. Sackett DL, Haynes RB, Tugwell P. Clinical Epidemiology. A Basic Science for Clinical Medicine. Boston/Toronto: Little, Brown and Company, 1995:108-119. 13. French EB, Robb WAT. Biliary and renal colic. Br Med J 1963;2:135-138. 14. Gunn A, Keddie N. Some clinical observations on patients with gallstones. Lancet 1972;2:239-241. 15. Rigas B, Torosis J, McDougall CJ, Vener KJ, Spiro HM. The circadian rhythm of biliary colic. J Clin Gastroenterol 1990;12:409-414. 16. Hinkel CL, Moller GA. Correlation of symptoms, age, sex, and habitus with cholecystographic findings in 1000 consecutive examinations. Gastroenterology 1957;32:807-815. 17. Persson G, Sloth M, Skold S, Thulin A. Evaluation of anamnestic data in patients referred for oral cholecystography. Scand J Gastroenterol 1989;24:550-556. 18. Diehl AK, Sugarek NJ, Todd KH. Clinical evaluation for gallstone disease: usefulness of symptom and signs in diagnosis. Am J Med 1990;89:29-33. 19. Talley NJ, Piper DW. Comparison of the clinical features and illness behaviour of patients presenting with dyspepsia of unknown cause (essential dyspepsia) and organic disease. Austral New Zealand J Med 1986;16:352-359. 20. Talley NJ, McNeil D, Piper DW. Discriminant value of dyspeptic symptoms: a study of the clinical presentation of 221 patients with dyspepsia of unknown cause, peptic ulceration, and cholelithiasis. Gut 1987;28:40-46. 21. Bates E, Ebbs SR, Harrison M, A’Hern RP. Influence of cholecystectomy on symptoms. Br J Surg 1991;78:964-967. 22. Wilbur RS, Bolt RJ. Incidence of gallbladder disease in ‘‘normal’’ men. Gastroenterology 1959;36:251-255. 23. Price WH. Gall-bladder dyspepsia. Br Med J 1963;2:138-141. 24. Bainton D, Davies GT, Evans KT, Gravelle IH. Gallbladder disease. Prevalence in a South Wales industrial town. N Engl J Med 1976;294:1147-1149. 25. Rome Group for the Epidemiology and Prevention of Cholelithiasis (GREPCO): Prevalence of gallstone disease in an Italian adult female population. Am J Epidemiol 1984;119:796-805. 26. Janzon L, Aspelin P, Eriksson S, Hildell J, Trell E, Ostberg H. Ultrasonographic screening for gallstone disease in middle-aged women. Detection rate, symptoms, and biochemical features. Scand J Gastroenterol 1985;20:706-710. 27. Rome Group for the Epidemiology and Prevention of Cholelithiasis (GREPCO). The epidemiology of gallstone disease in Rome, Italy. I. Prevalence data in men. HEPATOLOGY 1988;8:904-906. ; FIG. 1. Probability tree reporting the probability of having gallstones associated to each of the combinations of symptoms reported in Table 5. Clinical sign of gallstone complications are reported as complications. (A) The starting node represents patients with pain at the right hypocondrium or at the epigastrium and heartburn. (B) The starting node represents patients with pain at the right hypocondrium or at the epigastrium and without heartburn. 846 FESTI ET AL. 28. Barbara L, Sama C, Labate AMMM, Taroni F, Rusticali G, Festi D, Sapio C, et al. A population study on the prevalence of gallstone disease: the Sirmione study. HEPATOLOGY 1987;7:913-917. 29. Berndt H, Nurnberg D, Pannwitz H. Pravalenz der Cholelithiasis. Ergebnisse einer epidemiologischen Studie mittels Sonographic in der DDR. Z Gastroenterol 1989;27:662-666. HEPATOLOGY October 1999 30. Glambek I, Arnesjo B, Soreide O. Correlation between gallstones and abdominal symptoms in a random population. Results from a screening study. Scand J Gastroenterol 1989;24:277-281. 31. Heaton KW, Braddon FEM, Mountford RA, Hughes AO, Emmett PM. Symptomatic and silent gallstones in the community. Gut 1991;32:316320.