Bidirectional association between gallstones and renal stones: Two longitudinal follow-up studies using a national sample cohort

The present study evaluated the associations between gallstones and renal stones using a national sample cohort of the Korean population. The Korean National Health Insurance Service-National Sample Cohort was collected from 2002 to 2013. We designed two different longitudinal follow-up studies. In study I, we extracted gallstone patients (n = 20,711) and 1:4-matched control I subjects (n = 82,844) and analyzed the occurrence of renal stones. In study II, we extracted renal stone patients (n = 23,615) and 1:4-matched control II subjects (n = 94,460) and analyzed the occurrence of gallstones. Matching was performed for age, sex, income, region of residence, and history of hypertension, diabetes mellitus, and dyslipidemia. Crude and adjusted hazard ratios (HRs) were calculated using a Cox proportional hazards model, and the 95% confidence intervals (CIs) were calculated. Subgroup analyses were performed according to age and sex. The adjusted HR of renal stones was 1.93 (95% CI = 1.75–2.14) in the gallstone group (P < 0.001). The adjusted HR of gallstones was 1.97 (95% CI = 1.81–2.15) in the renal stone group (P < 0.001). The results were consistent in all subgroup analyses. Gallstones increased the risk of renal stones, and renal stones increased the risk of gallstones.

one study, we extracted gallstone patients and 1:4-matched control subjects and analyzed the occurrence of renal stones. In the other study, we extracted renal stone patients and 1:4-matched control subjects and analyzed the occurrence of gallstones.
Results study I. The mean follow-up periods were 65.74 (standard deviation, SD = 41.60) months and 64.87 (SD = 41.80) months for the gallstone group and the control I group, respectively. The rate of renal stones was higher in the gallstone group (2.7% [563/20,711]) than in the control I group (1.4% [1,180/82,844], P < 0.001, Table 1). The general characteristics (age, sex, income, region of residence, and hypertension, diabetes, and dyslipidemia histories) of participants were exactly the same due to the matching (P = 1.000).
The crude and adjusted HRs of gallstones were 1.97 (95% CI = 1.81-2.14) and 1.97 (95% CI = 1.81-2.15) in the renal stone group, respectively (each P < 0.001, Table 4). In the subgroup analyses, all crude and adjusted HRs of gallstones were higher in the renal stone group (each P < 0.05, Table 5

Discussion
This study revealed a bidirectional association between gallstones and renal stones. Gallstones increased the risk of renal stones, and renal stones increased the risk of gallstones. In both studies, these relationships were consistent in all subgroups according to age and sex.
The association between gallstones and renal stones could be explained by common pathophysiology. First, obesity and metabolic syndrome are risk factors for both gallstones and renal stones 17,18 . Hyperinsulinemia may increase hepatic cholesterol secretion and cholesterol supersaturation by activating hydroxymethylglutaryl coenzyme A reductase or by upregulating hepatocyte LDL receptors 19,20 . Obesity and insulin resistance result in defective ammoniagenesis 21 , so diabetes can increase the risk of uric acid renal stones due to a low urinary pH 22 . Second, obesity is an independent risk factor for infection 23 . Infection of the gallbladder can result in liver cell damage and bile acid excretion 24 . Hemolysis and chronic bacterial infections can produce pigment stones in the gallbladder 25 . Obesity can increase the risk of urinary tract infections 26 . Third, intestinal malabsorption can result in decreased bile acid resorption and increased urinary oxalate 27 . Finally, water and ion channels in the gallbladder might affect urine composition 28 .
The advantages of this study are consistent with those of our previous studies using the national sample cohort [29][30][31] . We used a large, representative, nationwide population. Because National Health Insurance Service (NHIS) data include all citizens in the nation without exception, no participants were missing during the follow-up periods. The control groups were randomly selected by matching for age, sex, income, region of residence, and past medical histories to avoid confounding effects. Furthermore, an adjusted hazard model was used to minimize the confounders. We designed two different studies to analyze the direction of the effect.
This study has the several limitations. Despite the cohort study design, we could not exclude the effects of possible confounders that might affect both gallstones and renal stones. Because we did not have data for body mass index or smoking or alcohol history, we could not adjust for these factors. These lifestyle factors could influence or mediate the association between gallstones and renal stones. Patients who could not consult with a clinic might have been missed. The possibility of detection bias exists. Visits for one disease could increase the detection rates of the other disease. Therefore, we performed an additional analysis confined to >6 months after the detection of one disease (Supplementary Table S1). In study I, we analyzed the occurrence of renal stones >6 months after detection of a gallstone. The adjusted HR was 1.43 (95% CI = 1.27-1.61, P < 0.001). In study II, we analyzed the occurrence of gallstones >6 months after detection of a renal stone. The adjusted HR was 1.53 (95% www.nature.com/scientificreports www.nature.com/scientificreports/ CI = 1.39-1.69, P < 0.001). Therefore, the association between gallstones and renal stones was consistent, even considering the possibility of detection bias.
In conclusion, gallstones increased the risk of renal stones, and renal stones increased the risk of gallstones.

Materials and Methods study population and Data Collection. This study was approved by the ethics committee of Hallym
University (2014-I148). The ethics committee of Hallym University waived the written informed consent from the study participants. All analyses adhered to the guidelines and regulations of the ethics committee of Hallym University. This national cohort study used data of the Korean Health Insurance Review and Assessment Service-National Sample Cohort (HIRA-NSC). The sample cohort was directly extracted from the mother population from the Korean NHIS to minimize non-sampling errors. The sample cohort consisted of approximately 2% of the entire Korean population (50 million). The sampling was performed based on the 1,476 levels (age [18 categories], sex [2 categories], and income level [41 categories]) using randomized stratified systematic sampling methods via proportional allocation to represent the entire population. After data selection, a statistician verified the appropriateness of the sample by comparing the data from the entire Korean population to the sample data. The National Health Insurance Sharing Service provided the details of the sampling procedures on their website 32     www.nature.com/scientificreports www.nature.com/scientificreports/ database is composed of (i) personal information, (ii) health insurance claim codes (procedures and prescriptions), (iii) diagnostic codes using the International Classification of Disease-10 (ICD-10), (iv) death records from the Korean National Statistical Office (using the Korean Standard Classification of disease), (v) socio-economic data (residence and income), and (vi) medical examination data for each participant over 2002 to 2013.
The exact population statistics were available using the NHIS database because all Korean citizens are identified by a 13-digit resident registration number from birth to death. All Koreans have to enroll in the NHIS. All the medical records of all Korean hospitals and clinics are registered using the 13-digit resident registration number to register individual patients in the medical insurance system. Thus, the risk of duplicated medical records is minimal, even if a patient visits different hospitals or clinics. In addition, all medical treatments in Korea can be traced without exclusion using the HIRA system. In Korea, reporting notice of death to an administrative entity is legally obligatory before a funeral can be held, and the reason of death and date are documented by medical doctors on a death certificate. participant selection. Of 1,125,691 patients with 114,369,638 medical claim codes, the participants who were diagnosed with gallstones (ICD-10: K80; Cholelithiasis) were included. Among them, the participants who visited hospitals or clinics ≥2 times for gallstones (n = 21,501) were selected. Histories of renal stones were classified using ICD-10 codes (N20; Calculus of kidney and ureter). We selected participants who visited hospitals or clinics ≥2 times for renal stones (n = 24,123).
We designed study I and II. These studies are independent from each other. For study I, the patients with gallstone were followed up for the presence of renal stone (Fig. 1a). Thus, gallstone was the first stone event, followed by renal stone. In contrast, study II investigated the subsequent occurrence of gallstone after renal stone (Fig. 1b). In these cases, the renal stone was the first stone event, followed by gallstone.
Study I. The 1:4 matching was performed between the gallstone patients and the control I group who were never diagnosed with gallstones from 2002 through 2013. The control group was selected from the total population (n = 1,104,190). The age, group, sex, income group, region of residence, and past medical histories (hypertension, diabetes, and dyslipidemia) were matched between the gallstone and control groups. The selection bias was minimized by sorting the control I participants using a random number order and then selecting them from top to bottom. It was presumed that the matched control I participants were involved at the same time as each matched gallstone participant (index date). Thus, participants in the control group who died before the index date were excluded. Participants who had a previous history of renal stones were excluded from both the gallstone and control groups. A total of 703 participants were excluded in the gallstone group. A total of 87 gallstone patients were additionally excluded due to the insufficient matching of participants. Finally, 20,711 gallstone patients and 82,844 control I participants were included in this study (Fig. 1a).
Study II. Renal stone patients were matched 1:4 with the control II participants who were not diagnosed with renal stones from 2002 through 2013. The control group was extracted from the total population (n = 1,101,568). The matching factors were identical to those of study I (age, group, sex, income group, region of residence, and  www.nature.com/scientificreports www.nature.com/scientificreports/ past medical histories [hypertension, diabetes, and dyslipidemia]). The matching procedures and exclusion criteria were also identical to those of study I. A total of 473 participants were excluded in the renal stone group. An additional 35 renal stone patients were excluded due to insufficient matching of participants. Eventually, = 23,615 renal stone patients and 94,460 control II participants were analyzed in study II (Fig. 1b).
The past medical histories were investigated based on ICD-10 codes. For the strict disease criteria, hypertension (I10 and I15), diabetes (E10-E14), and dyslipidemia (E78) were included if the participants had visited a hospital or clinical with that diagnosis ≥2 times.
statistical Analyses. The rates of general characteristics were compared between the gallstone and control groups (study I) and between the renal stone and control groups (study II) using the chi-square test.
In study I, the hazard ratio (HR) of gallstones (independent variable) for renal stones (dependent variable) was analyzed using a Cox proportional hazards model. In study II, the HR of renal stones (independent variable) for gallstones (dependent variable) was analyzed using another Cox proportional hazards model. For each analysis, crude (simple) and adjusted (age, sex, income, region of residence, hypertension, diabetes, and dyslipidemia) models were applied, and the 95% confidence intervals (CIs) were calculated.
For the subgroup analysis, the participants were divided according to age and sex (0-29 years old, 30-59 years old, 60+ years old; men, and women).
Two-tailed analyses were performed, and P values less than 0.05 were considered to designate significance. SPSS v. 21.0 (IBM, Armonk, NY, USA) was used for statistical analyses.