Association between dipeptidyl peptidase-4 inhibitors and urinary tract infection in elderly patients: A retrospective cohort study.

PURPOSE: Dipeptidyl peptidase-4 (DPP-4) inhibitors are a new class of antidiabetic drugs. Although they have been reported to increase the risk of infection, the findings are controversial. Given that urinary tract infections (UTIs) are common in the elderly, we conducted a retrospective cohort study by using health care insurance claims data, to elucidate the association between the DPP-4 inhibitors and the incidence of UTI in latter-stage elderly patients. METHODS: We analyzed 25,111 Japanese patients aged 75 years and older between the fiscal years 2011 and 2016. Patients using DPP-4 inhibitors and sulfonylureas (SUs) were matched at a 1:1 ratio using propensity scoring. The Incidence rate ratio (IRR) of UTI was compared between users of SUs and users of DPP-4 inhibitors by Poisson regression. Moreover, subgroup analyses stratified by sex were conducted to evaluate whether the combination of prostatic hyperplasia and DPP-4 inhibitors is associated with the incidence of UTI in male patients. RESULTS: The use of DPP-4 inhibitors was associated with an increased risk of UTI (adjusted IRR 1.23, 95% CI [1.04-1.45]). After propensity score matching, the association remained significant (adjusted IRR 1.28, 95% CI [1.05-1.56]). Moreover, elderly male patients with prostatic hyperplasia who received DPP-4 inhibitors had a higher risk of UTI than SU users without prostatic hyperplasia (Matched: crude IRR 2.90, 95% CI [1.78-4.71]; adjusted IRR 2.32, 95% CI [1.40-3.84]). CONCLUSIONS: The long-term use of DPP-4 inhibitors by elderly patients, particularly male patients with prostatic hyperplasia, may increase the risk of UTI.

Association of leptin and adiponectin levels with three-year weight change among Japanese male workers.

The purpose of this study was to clarify, through a prospective study, the relationship between leptin and adiponectin levels, and subsequent weight change. The study subjects were 2,485 male office workers aged 35-64 employed by a company in Fukuoka Prefecture, Japan. Of these men, 1,936 (77.9%) received health check-ups both in 2005-2007 and 3 years later, in 2008-2010. Of the subjects who received both health check-ups, 352 were excluded duo to cancer, cardiac infarction, stroke or diabetes mellitus, leaving a total of 1,584. We divided them into tertiles according to baseline leptin and adiponectin levels, and compared the subsequent change in body mass index (BMI) over 3 years. The subjects with the lowest leptin levels showed a significantly greater increase in BMI (difference in change in BMI=0.178 kg/m2, 95% CI:0.025-0.331 kg/m2) over 3 years when those with the highest leptin levels were regarded as the reference even after adjusting for age, baseline BMI, smoking status, drinking status and exercise. The subjects with the highest adiponectin levels showed a greater increase in BMI (difference in change in BMI=0.099 kg/m2, 95% CI: -0.034-0.231 kg/m2) over 3 years when those with the lowest adiponectin levels were regarded as the reference, but this association was not statistically significant after adjusting for age, baseline BMI, smoking status, drinking status and exercise. Our findings suggest that higher leptin levels may suppress weight gain in Japanese male workers.

Measurement of high-molecular-weight adiponectin is not useful in assessing coronary stenosis.

BACKGROUND: In many studies, high-molecular-weight (HMW) adiponectin has been considered the active form of adiponectin. However, whether HMW adiponectin is a good surrogate marker for coronary artery disease still needs to be elucidated. METHODS: We conducted a hospital-based cross-sectional study to examine the relationship between total, HMW or non-HMW adiponectin concentrations and coronary stenosis in 83 male patients and 138 male controls. RESULTS: Patients with coronary stenosis had significantly lower total adiponectin concentrations compared with controls. Non-HMW adiponectin concentrations in cases were significantly lower than the controls. However, there were no significant differences between cases and controls in HMW adiponectin concentrations. From the receiver operating characteristic (ROC) analysis, the area under the curve (AUC) for total and non-HMW adiponectin was significantly larger than that for HMW adiponectin concentrations. Of the three models, that for non-HMW adiponectin showed the largest AUC (total adiponectin 0.74, HMW adiponectin 0.54, and non-HMW adiponectin 0.79). CONCLUSIONS: Despite associations between total adiponectin levels and coronary stenosis, our data go against any apparent association between HMW adiponectin concentrations and coronary stenosis.

Retrospective longitudinal study on the relationship between 8-year weight change and current eating speed.

Most of the studies that have examined the relationship between the speed of eating and obesity have been cross-sectional. We investigated 529 male workers who received health check-ups provided by the employer in 2000 and 2008. We obtained information on the subjects' alcohol consumption, smoking status, self-reported speed of eating, and exercise in 2008, and height and weight in both 2000 and 2008. We compared weight change from 2000 to 2008 between 2 groups classified according to the speed of eating: a group of fast eaters, and a combined group of medium and slow eaters. The fast-eating group had a higher average weight gain (1.9 kg) than the medium and slow eating group (0.7 kg). Although statistically significant only for the 20-29-year age group, weight gain was greater in the fast-eating group for all age groups and was statistically significant when the age groups were combined. The relationship between eating fast and weight change was statistically significant even after adjusting for age and body mass index in 2000, drinking, smoking, and exercise. Our results suggested that the speed of eating is related to the rate of weight gain.

Urbanization and physician maldistribution: a longitudinal study in Japan.

BACKGROUND: The relative shortage of physicians in Japan's rural areas is an important issue in health policy. In the 1970s, the Japanese government began a policy to increase the number of medical students and to achieve a better distribution of physicians. Beginning in 1985, however, admissions to medical school were reduced to prevent a future oversupply of physicians. In 2007, medical school entrants equaled just 92% of their 1982 peers. The urban annual population growth rate is positive and the rural is negative, a trend that may affect denominator populations and physician distribution. METHODS: Our data cover six time points and span a decade: 1998, 2000, 2002, 2004, 2006, and 2008. The spatial units for analysis are the secondary tier of medical care (STM) as defined by the Medical Service Law and related legislation. We examined trends in the geographic disparities in population and physician distribution among 348 STMs in Japan. We compared populations and the number of physicians per 100,000 populations in each STM. To measure maldistribution quantitatively, we calculated Gini coefficients for physician distribution. RESULTS: Between 1998 and 2008, the total population and the number of practicing physicians for every 100,000 people increased by 0.95% and 13.6%, respectively. However, the inequality of physician distribution remained constant, although small and mostly rural areas experienced an increase in physician to population ratios. In contrast, as the maldistribution of population escalated during the same period, the Gini coefficient of population rose. Although the absolute number of practicing physicians in small STMs decreased, the fall in the denominator population of the STMs resulted in an increase in the number of practicing physicians per population in those located in rural areas. CONCLUSIONS: A policy that increased the number of physicians and the physician to population ratios between 1998 and 2008 in all geographic areas of Japan, irrespective of size, did not lead to a more equal geographical distribution of physicians. The ratios of physicians to population in small rural STMs increased because of concurrent trends in urbanization and not because of a rise in the number of practicing physicians.

The effect of the interaction between obesity and drinking on hyperuricemia in Japanese male office workers.

BACKGROUND: Obesity and drinking are acknowledged risk factors for hyperuricemia. However, the effect of the interaction between obesity and drinking on hyperuricemia is not well understood. METHODS: The cases comprised 3028 male office workers with hyperuricemia (serum uric acid level >or=7.0 mg/dL); the controls were 5348 men with a serum uric acid level less than 6.0 mg/dL. Logistic regression analysis was used to assess risk factors for hyperuricemia and the interaction between obesity and drinking in hyperuricemia. Participants were divided into 2 groups according to their BMI: individuals with a BMI of 25 or higher were classified as obese and those with a BMI less than 25 were classified as non-obese. In addition, participants were divided into 5 groups based upon their ethanol consumption. The 2 variables were then combined to create 10 groups. RESULTS: With non-obese non-drinkers as the reference category, the odds ratio for hyperuricemia was 1.80 for non-obese drinkers of less than 25 mL/day of ethanol, 2.15 for non-obese drinkers of 25-49 mL/day, 2.60 for non-obese drinkers of 50-74 mL/day, 2.56 for non-obese drinkers of 75+ mL/day, 4.40 for obese non-drinkers, 5.74 for obese drinkers of less than 25 mL/day, 6.57 for obese drinkers of 25-49 mL/day, 5.55 for obese drinkers of 50-74 mL/day, and 7.77 for obese drinkers of 75+ mL/day. The interaction between obesity and drinking in hyperuricemia was statistically significant. CONCLUSION: Our results suggest that although combining the effects of obesity and drinking did not result in a multiplicative increase in the risk for hyperuricemia, the combined risk was greater than the sum of the effects of obesity and drinking.

Inverse correlation between adiponectin and the risk of metabolic syndrome in middle-aged Japanese male workers.

Despite a close association between adiponectin and both hypertension and type 2 diabetes, the relationship between adiponectin and metabolic syndrome has not yet been well-investigated. To examine and evaluate the association between serum adiponectin levels and metabolic syndrome based on Japanese diagnostic criteria, we analyzed adiponectin and anthropometric parameters in 869 male employees aged 40-59 who belonged to a health insurance society in Fukuoka Prefecture and who underwent annual health check-ups from August 2006 to July 2007. Two hundred and thirty-two of the 869 subjects (26.7%) were diagnosed with metabolic syndrome. The serum adiponectin levels were significantly higher in the non-metabolic syndrome group. In a multiple logistic regression analysis, the subjects in the top quartile of serum adiponectin (adjusted odds ratio:0.36;95% confidence interval:0.21-0.63) and the second (adjusted odds ratio:0.51;95% confidence interval:0.31-0.84) quartile had a significantly decreased risk for metabolic syndrome in comparison to the bottom quartile. The dose-response relationship between serum adiponectin levels and metabolic syndrome was significant (p for trend 0.0001) after adjusting for age, body mass index, smoking status, and drinking status. The current findings suggest that hypoadiponectinemia is inversely correlated with the risk of metabolic syndrome in middle-aged Japanese male workers.

Adiponectin levels associated with the development of hypertension: a prospective study.

Adiponectin is a recently discovered protein that seems to be exclusively secreted by adipocytes and is the most abundant adipose tissue-derived protein. While some recent studies have demonstrated an association between adiponectin levels and hypertension, these studies were cross-sectional in design, and the results have been inconsistent. Therefore we performed a prospective study to elucidate the role of adiponectin in the development of hypertension. The results of this study showed that serum adiponectin levels were significantly lower in hypertensive subjects than in normotensive subjects. Moreover, in logistic regression analysis, the subjects in the lowest quartile had a 3.72-fold higher risk than those in the highest quartile. Even after adjusting for potential confounding factors, this association was found to be significant. Low serum adiponectin levels were found to be independently associated with a higher risk for the development of hypertension. Our results therefore suggest that hypoadiponectinemia is a novel predictor of hypertension.

Hyperleptinemia is associated with hypertension in Japanese males.

Leptin is a hormone which is predominantly secreted by adipose tissue. Recent studies have shown that leptin increases arterial blood pressure. Although data from available animal studies clearly indicate an association between leptin and hypertension, results of human studies have been less definitive. We conducted a case-control study to examine the association between serum leptin levels and hypertension in 111 hypertensive subjects and 222 male controls, using conditional logistic regression analyses. Mean serum leptin levels were found to be marginally higher in the case subjects than in the control subjects (3.3 ng/ml versus 3.0 ng/ml), however, conditional logistic regression analysis revealed that subjects in the highest quartile had a significantly increased risk of hypertension compared with those in the lowest quartile, even after adjusting for drinking status and diabetes mellitus (adjusted OR, 2.11;95% CI, 1.01-4.39). Our findings suggest that leptin plays an important role in the development of hypertension.

Risk factors for nosocomial infection in the neonatal intensive care unit by the Japanese Nosocomial Infection Surveillance (JANIS).

We evaluated the infection risks in the neonatal intensive care unit (NICU) using data of NICU infection surveillance data. The subjects were 871 NICU babies, consisting of 465 boys and 406 girls, who were cared for between June 2002 and January 2003 in 7 medical institutions that employed NICU infection surveillance. Infections were defined according to the National Nosocomial Infection Surveillance (NNIS) System. Of the 58 babies with nosocomial infections, 15 had methicillin-resistant Staphylococcus aureus (MRSA) infection. Multiple logistic regression analysis demonstrated that the odds ratio for nosocomial infections was significantly related to gender, birth weight and the insertion of a central venous catheter (CVC). When the birth weight group of more than 1, 500 g was regarded as the reference, the odds ratio was 2.35 in the birth weight group of 1,000-1,499 g and 8.82 in the birth weight group of less than 1,000g. The odds ratio of the CVC (+) for nosocomial infection was 2.27. However, other devices including artificial ventilation, umbilical artery catheter, umbilical venous catheter, and urinary catheter were not significant risk factors. The incidence of MRSA infection rapidly increased from 0.3% in the birth weight group of more than 1,500 g to 2.1% in the birth weight group of 1,000-1,499 g, and to 11.1% in the birth weight group of less than 1,000g. When the birth weight group of more than 1,500 g was regarded as the reference, multiple logistic regression analysis demonstrated that the odds ratio was 7.25 in the birth weight group of 1,000-1,499 g and 42.88 in the birth weight group of less than 1,000g. These odds ratios were significantly higher than that in the reference group. However, the application of devices did not cause any significant differences in the odds ratio for MRSA infection.

[Reliability of health insurance claim statistical data based on the principal diagnosis method].

OBJECTIVE: The estimation of disease-specific medical expenses in Japan is based on the assumption that all medical care expenses in a given claim are spent on the principal diagnosis even though health insurance claims actually include multiple diagnoses. The purpose of this study was therefore to evaluate the validity of this methodology. METHODS: Medical expenses and the diagnosis based on the International Classification of Diseases, which are presently used for the Japanese National Health Insurance program, were obtained from 8471 outpatient medical expense claims from the National Health Insurance for Medical services for the Aged in Okaya city, Nagano Prefecture in May 2004. We calculated the frequency of each specific disease and estimated the disease-specific medical expenses using only the principal diagnosis in a claim and 10 diagnoses (principal diagnosis and nine additional diagnoses). Disease-specific medical expenses using 10 diagnoses was estimated according to the Proportional Disease Magnitude method. RESULTS: The proportion that the principal diagnosis method reflected the frequency based on the method using ten diagnoses differed depending on the diagnosis. The proportion for hypertension was 59.7%, whereas it was 33.3% for diabetes mellitus. Hypertension and diabetes mellitus were estimated to be 18.32% and 5.38%, respectively, of all medical care expenses, as determined using the principal diagnosis method. However, when using ten diagnoses, hypertension and diabetes mellitus were estimated to be 8.50% and 5.16%, respectively, of all medical care expenses. CONCLUSIONS: The above described principal diagnosis method is therefore considered to overestimate the medical care expenses of diseases that are often selected as the principal diagnosis.

The treatment of infectious waste arising from home health and medical care services: present situation in Japan.

The management of waste materials arising from home health and medical care services (HHMC wastes) in Japan is now receiving greater attention from governmental workers dealing with general household waste materials. In general, HHMC waste materials are collected in a mixed form, transported and disposed of along with municipal solid wastes. As a result, municipal workers are suffering needle stick accidents so that infections associated with HHMC waste materials may occur. The collection and transportation by patients and their families of HHMC waste materials with sharp-edges, such as injection needles, to medical-related facilities can prevent municipal workers from experiencing needle-prick accidents. One of the most important strategies for medical-related facilities is hence the education of patients and their families. Improved rules for handling HHMC waste materials are essential for the safe and effective management.

Interaction of low serum adiponectin levels and smoking on coronary stenosis in Japanese men.

BACKGROUND: The relationship between serum adiponectin levels and coronary stenosis and the interaction of low serum adiponectin levels and smoking on coronary stenosis have not been clarified. METHODS: We conducted a hospital-based case-control study to examine serum adiponectin levels and smoking in 86 male patients with coronary stenosis and 145 male controls, using multiple logistic regression analyses. Serum adiponectin levels were grouped into quartiles for the analysis of the relationship between serum adiponectin levels and coronary stenosis, and into dichotomy for the analysis of the interaction of low serum adiponectin levels and smoking status. RESULTS: Decreasing serum adiponectin levels were significantly associated with increased risk of coronary stenosis in a dose-dependent manner. As for the interaction of low serum adiponectin levels and smoking on coronary stenosis, adjusted odds ratios were 14.00 (95% Confidence Interval 5.45-37.48) among ever-smokers with low serum adiponectin levels, 3.84 (95% CI: 1.44-10.21) among ever-smokers with high serum adiponectin levels and 3.79 (95% CI: 1.38-10.45) among never-smokers with low serum adiponectin levels, compared with the reference group of never-smokers with high serum adiponectin levels. CONCLUSIONS: The interaction of low serum adiponectin levels and smoking on coronary stenosis was likely to be a synergic effect. This study suggests that the combination of low serum adiponectin levels and smoking is the most useful predictor for coronary stenosis among Japanese men.

Effects of Eating Fast and Eating Before Bedtime on the Development of Nonalcoholic Fatty Liver Disease.

Few studies have evaluated the effects of lifestyle habits, such as eating behaviors, on the development of nonalcoholic fatty liver disease (NAFLD). It is known that NAFLD increases the risk of type 2 diabetes, prediabetes, cardiovascular disease, and chronic kidney disease. Therefore, a retrospective cohort study was conducted to evaluate the effect of eating behaviors and interactions between these behaviors on the development of NAFLD among health insurance beneficiaries without NAFLD. Study subjects were 2254 male and female insurance beneficiaries without NAFLD who had attended specific health checkups during fiscal years 2009 and 2012 among health insurance societies located in Fukuoka and Shizuoka Prefectures (Japan). The incidence of NAFLD was defined as Fatty Liver Index scores ≥60 or visiting medical organizations for fatty liver disease treatment according to claims data. Eating behaviors, including eating speed and eating before bedtime, were evaluated by a self-administered questionnaire. During the study period, 52 (2.3%) subjects progressed to NAFLD. Subjects who ate before bedtime but did not eat fast had a higher risk of NAFLD (adjusted odds ratio [AOR] = 2.15; 95% confidence interval [CI]: 1.03-4.46). Those with both negative eating habits had a significantly higher risk of NAFLD (AOR = 2.48; 95% CI: 1.09-5.63). Subjects who habitually ate before bedtime, and those who ate fast and before bedtime, tended to have an increased risk of NAFLD. Earlier intervention to modify these poor eating behaviors could be useful to prevent NAFLD. (Population Health Management 2016;19:279-283).

[Risk factors for requiring long-term care among middle-aged and elderly people].

OBJECTIVE: The purpose of this study is to clarify risk factors for requiring long-term care and all-cause mortality among middle-aged and elderly people. METHODS: The subjects were 2292 residents (759 males and 1,533 females) living in rural areas and attending health checkups during 1989-1993. Using the data obtained, we examined risk factors for long-term care needs and all-cause mortality. The observation period was from health checkups to March 2002 for all-cause mortality and to September 2002 for long-term care. The Cox' proportional hazards model was used to assess for both outcomes. RESULTS: Although all-cause mortality was two and a half times as high among males as among females, there were no differences between the sexes in the rate for requiring long-term care. In thirty five percent of cases needing long-term care, this was attributable to cerebrovascular diseases, in 24% to dementia, and in 9% to fracture. Risk factors significantly associated with higher all-cause mortality were age, low BMI, low total cholesterol, liver dysfunction, and smoking among males and females, as well as urine sugar among males and anemia among females. Risk factors significantly associated with requiring long-term care were age, hypertension and urine sugar among males, and age, anemia and urine sugar among females. CONCLUSIONS: This study shows that control of hypertension and diabetes mellitus is important for avoiding necessity for long-term care.

[An epidemiological study of the relationship between the number of remaining teeth and the disabled elderly receiving long-term care].

This study attempts to clarify the relationship between the long-term care of the elderly and the number of remaining teeth by comparing the remaining teeth of the healthy with that of the disabled elderly. A survey was conducted in a town, Fukuoka Prefecture. The disabled group consisted of 62 persons with level of disability of 4-5 according to the Long-term Care Insurance categorization. Controls were selected randomly from healthy elderly residents, one for each in the disabled group, matching gender and age (less than +/- 1 year old). Public health nurses conducted interviews using a questionnaire to determine the cause of disease, medical history, lifestyle and period of long-term care, and counted the remaining teeth. The mean number of remaining teeth presented a significant difference with 3.7 for the disabled group and 9.1 for the control group (p < 0.01). A logistic regression analysis revealed that the odds ratio for long-term care to that of the reference with 20 or more teeth was 7.03 (95% CI: 1.15-42.85) for 10 to 19 teeth, 15.61 (2.89-84.26) for 1 to 9 teeth, and 15.11 (2.84-80.48) for no teeth. The age group of 65-69 years and 70-79 years in the disabled group had significantly smaller mean numbers of remaining teeth than their control group counterparts by approximately 14 and 12 respectively, but in the age group of 80 years or more, no significant difference was observed between the two groups. In conclusion, our results suggested that elderly people with fewer remaining teeth have a high risk for long-term care. Especially, the risk was thought to be strongly associated with the loss of some teeth in middle age.

Evaluation of the fatty liver index as a predictor for the development of diabetes among insurance beneficiaries with prediabetes.

AIMS/INTRODUCTION: Non-alcoholic fatty liver disease (NAFLD) is the most common liver disease in developed countries, and it was required to monitor patients with prediabetes. However, there have been few reports establishing the risk for diabetes mellitus (DM) among patients with prediabetes. The purpose of the present study was to evaluate the effect of NAFLD on the progression of DM among insurance beneficiaries with prediabetes, using data from specific health check-ups and the fatty liver index (FLI). MATERIALS AND METHODS: We used a retrospective cohort study that enrolled 967 insurance beneficiaries with prediabetes who had rarely drunk or could not drink alcohol, or whose alcohol consumption was <19 g/day from two health insurance societies. We divided insurance beneficiaries into FLI <30, intermediates FLIs and FLI ≥60, and compared the incidence rate of DM among the groups after 3 years' follow up, using multiple logistic regression models. RESULTS: During 3 years' follow up, progression of diabetes was seen in 65 men (11.5%) and 24 women (6.0%). Logistic regression analyses showed that those with NAFLD had significantly higher risks of developing DM; this was the case in both men (odds ratio 2.68, 95% confidential interval 1.29-5.56) and women (odds ratio 10.35, 95% confidential interval 3.22-33.31). CONCLUSIONS: Among insurance beneficiaries with prediabetes, those with NAFLD had a significantly higher risk of DM than those without NAFLD. The FLI might be useful for detecting individuals who have an especially higher risk for DM, and developing more effective guidance for delivering healthcare services in Japan.

Does antihypertensive treatment with renin-angiotensin system inhibitors prevent the development of diabetic kidney disease?

BACKGROUND: Diabetic kidney disease (DKD) is the leading cause of end-stage renal disease worldwide. Renin-angiotensin system (RAS) inhibitors are the first-line treatment for diabetic patients with hypertension. However, whether RAS inhibitors prevent the development of DKD remains controversial. We conducted a retrospective cohort study quantifying the preventive effect of antihypertensive treatment with RAS inhibitors on DKD, using data from specific health check-ups and health insurance claims. METHODS: The study subjects were 418 patients with diabetes and hypertension, drawn from health insurance societies located in Fukuoka and Shizuoka prefectures in Japan. The subjects were divided into three groups, according to the type of antihypertensive treatment they received. They were then compared in terms of the development of DKD, using the diagnostic codes from ICD-10. RESULTS: Thirty subjects (6.2%) developed DKD during the study period between April 2011 and September 2013. RAS inhibitor treated group showed a significantly lower risk of DKD [adjusted odds ratio (AOR) = 0.35; 95% confidential interval (CI): 0.16-0.76] compared with the no treatment group. CONCLUSION: We conclude that antihypertensive treatment with RAS inhibitors is potentially useful for preventing the development of DKD.

Does seropositivity for Helicobacter pylori antibodies increase outpatient costs for gastric and duodenal ulcer or inflammation?

BACKGROUND: Helicobacter pylori is regarded as an important cause of both peptic ulcer and chronic gastritis. In particular, seropositivity is highest in patients with duodenal ulcer. No studies have determined whether there are differences in the direct medical costs associated with gastric/duodenal ulcer or inflammation, between seropositive and seronegative patients. OBJECTIVE: To examine the relationship between seropositivity for H. pylori and outpatient visits and direct medical costs for gastric/duodenal ulcer or inflammation in Japan from the perspective of the payor and patients. METHODS: Participants were males (n = 653) who worked for an agricultural co-operative in Fukuoka Prefecture, attended an annual health examination (including a written lifestyle and medical survey), belonged to the same health insurance society consistently for 4 years from April 1996 to March 2000, and provided a blood sample. The survey asked about lifestyle, including smoking and drinking, and past medical history. We retrospectively analysed the annual number of outpatient visits per person and outpatient medical cost (Yen, 2000 values) per person for visits relating to gastric or duodenal ulcer or inflammation using International Classification of Diseases (9th edition) -- Clinical Modification codes. We assessed for potential confounding factors using analysis of covariance and the chi-square test. RESULTS: The annual outpatient incidence of disease, the number of visits to physicians, and the medical costs for gastric or duodenal ulcer or inflammation were about 2-fold greater in individuals with antibodies to H. pylori compared with those without antibodies. CONCLUSION: Population-based studies and/or randomised controlled clinical trials that target high-risk groups and account for the unique way in which data are collected in Japan are needed to determine whether medical costs for gastric and duodenal ulcer might be reduced by treating asymptomatic patients who have antibodies to H. pylori.