Reproductive Episodes and Asymptomatic Gallstones in Females : Findings from A Nested Case-Control Study in Hokkaido, Japan

To explore reproductive risk factors for asymptomatic gallstones in females, we conducted a nested case-control study. The study comprised 126 cases (women with asymptomatic gallstones) and 378 controls ( women free from gallstones and other diseases of the liver and biliary tract) among 3, 927 women aged 40 years or more in Y town, Hokkaido, Japan. Controls were randomly selected with an allocation ratio of 1 : 3 from 3, 599 cohort women, matching to cases for sex(female), age (2 years), residential area, and year of examination. Odds ratio analysis revealed the following major findings. (1) Women married for the first time at 22 years old or later are likely to be at greater risk of asymptomatic gallstones with odds ratios of 1.60 (95% confidence interval : 0.97-2.64) ; 1.68 (0.96-2.94) for postmenopausal women. (2) First full-term delivery at 23 years old or later tended to increase the risk by 1.63 (0.97-2.77) ; 1.63 (0.90-2.9) when postmenopausal. and (3) Pregnancy of 5 times or more also appeared to elevate the risk by 1.47 (0.97-2.22) ; 1.55 (0.94-2.57) when postmenopausal. Two odds ratios obtained by logistic regression analysis are significant: 1.67 (1.14-3.32) and 1.59 (1.01-2.52) for age at first marriage and total number of pregnancy, respectively. Finally, we emphasize that this is the first epidemiological investigation in Japan which explored an association between reproductive episodes and asymptomatic gallstone formation in women. J Epidemiol, 1993; 3 : 91-97.

To our knowledge, this is the first epidemiological study in Japan, in which reproductive factors were investigated in relation to asymptomatic gallstones in women.

MATERIALS AND METHODS
Population at risk, screening for gallstones, and collection of relevant data This study was conducted at Y town, which is inhabited by approximately 20,000 population and located in the southern part of Hokkaido Island, with major industrial makeup of agricultural and diary farming and fishery including scallop cultivation.
In the period from 1983 to 1989, 3,927 women aged 40 years or more were examined by ultrasonography using a Aloka SSD-256, realtime, linear-array scanner with a 3.5-MHz probe (Aloka Co., Ltd., Japan) under fasting condition.
Such information were simultaneously collected or measured as date of birth, residence, height, weight, skinfold, urinalysis, electrocardiogram, eye fundus photograph, and chest x-ray examination. Biochemical examination was conducted for blood collected by venopuncture under fasting condition. To obtain personal information on past history, lifestyle habits, dietary habits, and other epidemiological items, a self-administered questionnaire was used, which was distributed about one month before examination and reviewed at examination by five public health nurses who were specifically trained for this study. When answers were missing, vague or contradictory, collection or correction was made by their direct interview. The review of each questionnaire by nurses was performed irrelevantly of the findings obtained by abdominal scanning. These data collection and examination were annually conducted every summer since 1983.
Among various information thus collected, we can only use some of them in this particular study, because study design is a nested case-control study, in which information to be discussed should be restricted to those that existed sometime before the detection of asymptomatic gallstones. Reproductive episodes to be discussed in this paper are representatively such kind of information.
Obesity is regarded as one of relevant factors to symptomatic gallstones mostly in cross-sectional analysis1-4,7,8,13-15,18-24) Past episode of obesity was, however, not available in our data collection, and, therefore, it was not included as a variable in the analysis.

Case identification and eligibility
Among a total of 3,927 women examined, we identified 201 women with gallstones and/or past history of cholecystectomy. Among 201 women, those underwent cholecystectomy were 63 women, followed by 7 stone formers not cholecystectomized, but experienced three typical symptoms of biliary colic, jaundice and fever in the past years, and 5 stone formers who had no symptoms at detection, but later developed above three symptoms in at least two years after detection of gallstones.
Excluding these 75 women, as a consequence, we could finally identified 126 asymptomatic gallstone formers, who are defined as cases in this study.

Control selection
We derived a source of controls from 3,927 women by excluding 201 stone formers and 127 women who demonstrated abnormal findings in the liver and biliary tract at examination. From 3,599 women, we randomly selected three controls per case by a pair matching procedure.
Matching variables are sex (females), age (±2 years), residential area, and year of examination (all examinations were conducted in summer every year). Finally we could successfully selected 378 women, who could satisfy above inclusion and exclusion criteria and who were also proved to be free from gallstones either symptomatic or asymptomatic for at least two years after examination.

Data analysis
Data on 126 cases and 378 controls were analyzed using matched-set odds ratio with 95% confidence interval and Mantel-Haenszel chi-square statistics to assess the significance of differences using two-sided test of sign ificance26). For multivariate analysis, we used a conditional logistic regression mode 117).
To examine the relationship between the risk of asymptomatic gallstone and reproductive episodes by menopausal status, cases and controls were divided into women still menstruating (premenopausal) and women who had stopped menstruating naturally or artificially (postmenopausal).
In this analysis, we calculated age-adjusted odds ratio with 95% confidence interval by Mantel-Haenszel Extension Method 18) and used an unconditional logistic regression mode 121). Data analysis was all processed at Nagoya University Computation Center and Aichi Medical University Information Processing Center, using SAS and related programs.      In premenopausal women, pregnancy of 5 times or more alone demonstrated an odds ratio of a little less than two without any statistical significance. In postmenopausal women, age at first marriage of 22 years or later, age at first full-term delivery of 23 years or later, and pregnancy of 5 times or more appeared to increase the risk of asymptomatic gallstones, with age-adjusted odds ratio of 1.68 (0.96-2.94), 1.63 (0.90-2.95), and 1.55 (0.94-2.57), correspondingly.

RESULTS
No material effects were, however, noted by age at menarche and total number of live birth in pre-and post-menopausal women, and by ages at first marriage and at first full-term delivery in premenopausal women.
We further analyzed a whole data set by a conditional logistic regression model and two menopausal data sets by an unconditional logistic regression model, including all reproductive episodes : ages at menarche, menopause, first marriage and first full-term delivery, and total numbers of pregnancy and live birth. Three logistic regression analyses yielded only one significant result, as summarized in Table 5. Separate multivariate analysis by menopausal status did not detect any significant reproductive episodes which either increase or decrease the risk of asymptomatic gallstone formation.
Significant odds ratios, detected by multivariate analysis on a whole data set, were 1.67 (95% CI : 1.14-3.32) for age at first marriage of 22 years or later and 1.59 (95% Cl : 1.01-2.52) for pregnancy of 5 times or more.

DISCUSSION
Gallstones have usually been investigated among symptomatic patients or those already experienced cholecystectomy, but rarely on those with asymptomatic stones. In Japan, in particular, few studies have been undertaken by contrasting those with and those without asymptomatic gallstones, which was, therefore, a main design of the present study.
We could clarify two major reproductive risk factors in our study, which significantly increase the probability of gallstone formation : first marriage at 22 years old or later and pregnancy of 5 times or more.
To our knowledge, age at first marriage has hardly ever been assessed in relating with either symptomatic or asymptomatic gallstones.
Two studies, however, examined age at first birth, which may be in generally parallel with age at first marriage, in association with symptomatic gallstones. One study) reported that the risk of symptomatic gallstones was about two-fold when age at first birth was 31 years or more ; suggesting an association between older age at first birth and symptomatic gallstones.
Another study, a case-control study), obtained an opposite finding that average age at first birth was significantly lower in patients with symptomatic gallstones (21.8 years old) than in controls (23.1 years old) among women aged less than 50 years,but average age at first birth was not significantly different at all between cases and controls when they aged 50 years or older.
An association of parity or frequency of pregnancy has been relatively often examined with symptomatic gallstones, but rarely with asymptomatic gallstones. According to an investigation 14), which examined 612 middle-aged women with symptomatic gallstones, no significant risk differential was detected between nulliparous and parous wemen. And also no elevation of risk for symptomatic gallstones was found by parity30) among 632 female symptomatic patients aged 40-69 years. In contrast to these investigators, frequent pregnancy or parity was reported as a significant Reproductive Episodes and Asymptomatic Gallstones 95 a risk factor incriminated for symptomatic gallstones"" or cholecystectomy18), though not for asymptomatic gallstones.
Effects of pregnancy and menstrual cycle are reported in relation to gallstone formation as follows32): (1) both fasting and residual volumes of the gallbladder after meal are increased in pregnancy33), (2) during the luteal phase of the menstrual cycle, gallbladder emptying is impaired 34), (3) bile acid metabolism is altered by pregnancy35), (4) cholelithiasis is experimentally demonstrated by prolonged progesterone treatment in rabbits36), and (5) contractile properties of the gallbladder are impaired in guinea pig in vitro in the presence of progesterone37).
Transient obesity at pregnancy may also be related to gall-stone formation.
Becoming obese gradually with advancing months of pregnancy might lead to poor contracting of the gallbladder due to poor elevation of the diaphragm, which finally results in development of biliary stasis3,18). Biliary stasis is considered as a biochemical basis of favouring cholesterolgallstone formation by secreting cholesterol-saturated bile, nucleation of cholesterol crystals and growth of stones in the gallbladder 38).
We conducted this study at Y town, inhabitants of which have never been screened for gallstones by ultrasonography.
It means that women found to have asymptomatic gallstones are all detected for the first time and those who have not sought for any medical attention or treatment.
Referent group was selected from 3,599 women without any abnormalities in the liver and biliary tract and gallstones proven by ultrasonography, with a ratio of 1 : 3 by individually matching to cases for sex, age, residence and year of examination.
Besides this matching procedure, we traced the study subjects for at least two years, and confirmed that all cases remained as asymptomatic and controls as non-stone formers. This implies further that the two groups are reasonably comparable. Selection bias of the study subjects is also not applicable, since they are not subjectively selected by any means, and all eligible cases are included, and all controls are randomly selected.
Information on reproductive episodes collected by a self-administered questionnaire are reviewed and corrected by direct interview by five public health nurses. They are also inhabitants in Y town and have close contacts with examinees. This indicates that examinees are not likely to answer incorrectly to private questions such as reproductive episodes. Besides this unlikeli-ness, all information are collected without knowing whether an interviewee will be included as a study subject or not, since the present study is later constructed as a nested case-control study. Despite these reliabilities, possibility of misclassification bias particularly by menopausal status can not be excluded, because whether postmenopausal or not is solely dependent upon the answer of each woman. Inter-and intra-interviewer variations, however, are believed to be minimal, since all interviewers were trained beforehand by standardizing whole interview procedures.
Finally, we could identify the two major reproductive risk factors for asymptomatic gallstones in females, though their medical implications in gallstone formation are not yet fully understood.
It should, however, be particularly emphasized that the present findings are obtained for asymptomatic gallstones, but not for symptomatic gallstones.
Epidemiologic risk factors for gallstone formation are believed to be not only these reproductive factors but also various life-style factors not readily assessed in the present nested casecontrol study. And therefore other types of epidemiological investigations should be undertaken to evaluate an association of diets, body weight changes, or serum levels of total cholesterol and triglyceride.