Predicting failure to follow-up screened high blood pressure in Japan: a cohort study

Journal of Public Health Vol. 37, No. 3, pp. 498 505 doi:10.1093/pubmed/fdu056 Advance Access Publication August 7, 2014 Predicting failure to follow-up screened high blood pressure in Japan: a cohort study Akira Kuriyama 1, Yoshimitsu Takahashi 1, Yuka Tsujimura 1, Kikuko Miyazaki 1, Toshihiko Satoh 2, Shunya Ikeda 3, Takeo Nakayama 1 1 Department of Health Informatics, Kyoto University School of Public Health, Kyoto 606-8501, Japan 2 School of Social Informatics, Aoyama Gakuin University, Sagamihara 252-5258, Japan 3 Department of Pharmaceutical Sciences, School of Pharmacy, International University and Health Welfare, Ohtawara 324-8501, Japan Address correspondence to Akira Kuriyama, E-mail: nrk40448@nifty.com ABSTRACT Background This study aimed to determine the prevalence and predictors of working-age individuals who did not follow-up for possible hypertension that was detected in the population-based screening. Methods We conducted a retrospective cohort study, using the database of health insurance claims and health checkups from several health insurance societies for employees in Japan. Screened participants aged 20 years, with possible hypertension (systolic blood pressure 140 mmhg and/or diastolic blood pressure 90 mmhg) and without known antihypertensive treatment, were included. The outcome was lack of clinical follow-up for possible hypertension within 6 months of the latest screening. Multivariate logistic regression analysis was performed to identify predictors. Results Among 17 173 participants (15 793 males and 1380 females) who were identified as possible hypertensives, 89.7 and 82.3% of them, respectively, did not consult physicians for screened possible hypertension. Predictors of no clinical follow-up for males included younger age, lower body mass index (BMI), lower hemoglobin A1c and milder hypertension. Predictors for females included younger age, lower BMI and being insured. Conclusions Approximately 80% of participants failed to consult physicians even with positive screening results. Younger individuals with lower BMI are at high risk of no clinical follow-up. Keywords circulatory disease, health services, secondary and tertiary services Introduction Hypertension is one of the most common chronic diseases in the world, particularly in developed countries. In 2000, its prevalence was estimated to be at 26% for adults. 1 The United States Preventive Services Task Force recommended screening for high blood pressure (BP) in 1996, 2 but the Centers for Disease Control and Prevention recently reported that 55% of adults with uncontrolled hypertension were either unaware or aware but did not receive pharmacological treatment. 3 Forty million individuals in Japan, approximately more than one-third of the population, are estimated to have hypertension. 4 The disease and related complications account for Japanese 5.4 trillion yen (US 7.1 billion dollars) in total national medical costs. 5 Thus, secondary prevention of hypertension might have substantial merits from the clinical and societal perspectives, especially in working-age populations in whom organ damages due to hypertension can be still reversible. 6 Since the Health and Medical Service Act for the Aged Akira Kuriyama, Researcher Yoshimitsu Takahashi, Lecturer Yuka Tsujimura, Researcher Kikuko Miyazaki, Assistant Professor Toshihiko Satoh, Professor Shunya Ikeda, Professor Takeo Nakayama, Professor 498 # The Author 2014. Published by Oxford University Press on behalf of Faculty of Public Health. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com.

PREDICTING FAILURE TO FOLLOW-UP SCREENED HIGH BLOOD PRESSURE IN JAPAN 499 was enacted in 1983, a population-based strategy to prevent and manage hypertension and to eventually prevent cardiovascular diseases, including community or workplace screening, has been implemented. 7 Once BP higher than a predetermined cutoff is considered to be possible hypertensive, identified individuals are recommended to seek medical advice and physicians give them advice or medical treatment based on further evaluations. Full transition of participants from screening to medical consultation is necessary to achieve effective results, but the prevalence of susceptible individuals who fail to follow-up is unknown. The purpose of this study was to determine the prevalence and predictors of Japanese working-age individuals screened as possible hypertensives who did not follow-up with physicians. We particularly investigated whether the underlying diabetes mellitus, dyslipidemia, abnormal checkup data or depression would predispose the individuals to the follow-up with physicians on detected high BP. Methods Study population We conducted a retrospective cohort study of individuals identified as possible hypertensives in at least one health checkup. Participants were selected from an administrative database of health insurance claims ( January 2005 through March 2011) and health checkups (March 2005 through March 2010) from Japan Medical Data Center Co., Ltd. (Tokyo, Japan). The database contained 530 000 cumulative insured individuals, mainly company employees and their family members, who were mostly from working-age populations. 8 Briefly, the database provides encrypted personal identifiers, individual demographics and insurance information, prescribed drugs in Anatomical Therapeutic Chemical (ATC) codes, 9 diagnoses coded by the International Classification of Diseases, 10th revision (ICD-10) 10 and data as of the last health checkup. This study was approved by the Ethics Committee at Kyoto University Graduate School of Medicine (E1017). Design overview The Japanese Society of Hypertension Guidelines for the Management of Hypertension 2009 ( JSH-2009) recommended the following method of BP measurement, which is generally used in health checkups: first, the person is placed in a quiet room to sit for several minutes; next, a BP cuff is placed at heart level to measure BP at least twice every 1 2 min; and finally, the stable values are averaged to define BP. 7 We adopted the cutoff value to screen possible hypertension as systolic blood pressure (SBP) 140 mmhg and/or diastolic blood pressure (DBP) 90 mmhg. 11,12 Participants fulfilling all the following criteria were included: (i) aged 20 years; (ii) possible hypertension was identified in at least one health checkup and (iii) had health insurance claim data before the checkup when possible hypertension was identified. Exclusion criteria were as follows: (i) ATC codes suggesting prescribed antihypertensive medications (C01, C03, C04, C07, C08, C09) within 3 months before high BP was identified or (ii) did not receive health checkups. Study outcomes We focused on the last checkup when possible hypertension was identified. The primary outcome was set as no physician follow-up for possible hypertension within 6 months of the checkup; ICD-10 codes suggesting hypertension (I10 I15) were not generated during this period. We selected this time point, because not a few participants might presumably have checkups twice a year. Explanatory variables The following data were extracted from the database as of the last relevant checkup: sex, date of birth, ATC codes corresponding to prescribed antihyperglycemic (A10), antilipidemic (C10) and antidepressive agents (N06A) and ICD-10 codes corresponding to hypertension or suspected hypertension, BP values, body mass index (BMI; kg/m 2 ), hemoglobin A1c (HbA1c; %), high-density lipoprotein (HDL; mg/ml), low-density lipoprotein (LDL; mg/ml), proteinuria ( positive or negative), smoking habits (current or not), drinking habits (daily, occasionally or rarely) and number of times high BP was identified in health checkups. Age was determined by subtracting the date of birth from the last relevant checkup. Incident case was defined when high BP was identified for the first time in a certain individual. Based on JSH-2009 guidelines, hypertension was categorized as follows: Class I, SBP 140 159 mmhg and/or DBP 90 99 mmhg; Class II, SBP 160 179 mmhg and/or DBP 100 109 mmhg and Class III, SBP 180 mmhg and/or DBP 110 mmhg. 7 Participants were identified as having underlying diabetes mellitus, dyslipidemia or depression when the corresponding ATC codes were generated within 3 months of the relevant checkup. HbA1c values, which were determined by the Japanese Diabetes Society method, were 0.4 lower than those measured by National Glycohemoglobin Standardization Program methods. 13 Diabetes was defined as HbA1c 6.1%. 14 Dyslipidemia was defined as HDL 40 mg/dl or

500 JOURNAL OF PUBLIC HEALTH LDL 140 mg/dl, 15 with triglycerides excluded because not all participants were presumably examined after fasting. Statistical analysis Characteristics and prevalence of primary outcomes were described in proportion to total participants, and data were examined within 3, 6 and 12 months of the last relevant checkup. For the characteristics of the baseline data, continuous variables were analyzed with the Wilcoxon-signed rank test, and categorical variables were analyzed with the x 2 test. Sex-specific, age-adjusted logistic regression analysis was performed to estimate odds ratios (ORs) and 95% confidence intervals (CIs) of explanatory variables for lack of follow-up. Sex-specific, logistic multivariate regression analysis, with variables with P, 0.20 in age-adjusted analysis, was also conducted to determine predictors for no clinical follow-up within 6 months of the last relevant checkup. The data regarding drinking habits were substantially missing, and were, therefore, excluded from multivariate analysis. Multiple imputation methods were used to handle missing variables. 16 Two sensitivity analyses were conducted. The first analysis assessed the frequency of no clinical follow-up, at different BP cutoffs of 150/90, 140/100 or 150/100 mmhg. It was on the assumption that health checkup facilities presumably might set hypertension cutoffs.140/90 mmhg as shown in the JHS 2009 guideline, and because clinical follow-up might have been hindered by the belief of the regression to the mean phenomenon. The second sensitivity analysis was performed, on the assumption that all of individuals, who had already been treated for diabetes mellitus or dyslipidemia and were screened to have possible hypertension, visited physicians for their potential hypertension. We assumed that individuals with these underlying diseases would visit the same physicians who also see the co-morbid diseases, but we suspected that ICD-10 codes may not be given unless a specific diagnosis or prescription for hypertension was made. Statistical significance tests were two-sided; P, 0.05 was considered significant. Analyses were performed with Stata, version 11.2 (Stata Corp., College Station, TX, USA). Results Participant characteristics Of the 22 369 individuals screened as possible hypertensives, 17 173 (15 793 males and 1380 females; age ranged 20 68 years old) were eligible for this study (Fig. 1). Males are younger than females. Females were more likely to have underlying lipid abnormality than males, while no differences were found in the frequency of underlying diabetes mellitus and depressive state. Males were more likely to drink alcohol and to have smoking habits. At the checkup, females have a higher SBP and DBP, respectively. Males had a higher BMI, whereas females had a higher HbA1c. Males were more likely to have proteinuria, and to have high BP pointed out for the first time (Table 1). Primary analysis Among participants, 14 161 males (89.7%) and 1135 females (82.3%) screened as hypertensives did not follow-up 6-month later (Table 2). The proportion remained high 12 months after checkup (86.5% of males and 78.0% of females). Age-adjusted logistic regression analyses of male participants revealed that lack of clinical follow-up was associated with smoking, underlying diabetes, dyslipidemia and depression, lower HbA1c, lower BMI, incident cases screened as possible hypertensives for the first time, negative proteinuria and lipid abnormality as of the last relevant checkup (Table 3). No clinical follow-up for possible hypertensives in females was associated with lower BMI, being insured, incident cases and lipid abnormality as of the last relevant checkup (Table 4). Multivariate regression analyses are also shown in Tables 3 and 4. Males with an increased risk for no clinical follow-up tended to be of younger age (OR: 0.96, 95% CI: 0.95 0.96), had lower BMI (OR: 0.98, 95% CI: 0.97 0.99), had lower HbA1c (OR: 0.92, 95% CI: 0.86 0.99), had underlying dyslipidemia (OR: 0.72, 95% CI: 0.56 0.92), had a smoking habit (OR: 1.34, 95% CI: 1.19 1.50) and had negative proteinuria (OR: 0.62, 95% CI: 0.49 0.79). Females with an increased risk for no clinical follow-up were younger (OR: 0.96, 95% CI: 0.95 0.98), had a lower BMI (OR: 0.96, 95% CI: 0.93 0.99), were insured (OR: 0.66, 95% CI: 0.48 0.92) and incident cases of screened as possible hypertension (OR: 1.47, 95% CI: 1.09 1.98). Sensitivity analysis The number of significant predictors decreased as BP cutoffs changed (the first sensitivity analysis). A younger age and lower hypertension category were significant in every analysis for males, while smokers and proteinuria were significant except in only one analysis (Supplementary data, Table S5). A younger age was significant in both methods for females, while incident cases and lower BMI was significant in some of the analyses (Supplementary data, Table S5). The second sensitivity analysis described the frequency of no clinical follow-up, on the assumption that all of individuals who had already been treated for diabetes mellitus or dyslipidemia visited physicians for their potential hypertension. Among participants, 13 885 males (87.9%) and 1118 females

PREDICTING FAILURE TO FOLLOW-UP SCREENED HIGH BLOOD PRESSURE IN JAPAN 501 Insured individuals and family members (n = 230 604) Individuals who did not visit medical checkups (n = 125 616) Received medical checkups (n = 104 988) Individuals with SBP<140 mmhg or DBP <90 mmhg (n = 82 619) Individuals with SBP 140 mmhg and/or DBP 90 mmhg (n = 22 369) Individuals without antihypertensive treatment (n = 19 076) Individuals with data available from April 2005 to March 2010 (n = 17 448) Individuals under antihypertensive treatment (n = 3293) Individuals whose insurance claims before March 2005 or after April 2010 were missing (n = 1628) Individuals aged 19 or 69 (n = 275) Individuals, aged 20 to 68 (n = 17 173) Fig. 1 Study flow diagram. (81.0%) screened as hypertensives did not follow-up with physicians (Supplementary data, Table S6). The predictors of no clinical follow-up for males included a younger age, a lower BMI, a lower HbA1c and positive smoking habits and negative proteinuria. The predictors of no clinical follow-up for females were a younger age, a lower BMI, had a lower BMI, being insured and incident cases of screened as possible hypertension. These results were consistent with the primary analysis. The prevalence of no clinical follow-up remained high for each sex but tended to decrease as the BP cutoff increased, as described in the primary analysis. Lower BMI and lower HbA1c for males and incident cases in females proved to be independently significant, in addition to the significant predictors shown in primary analysis. Underlying depression independently and significantly predicted no clinical follow-up when the SBP cutoff was 140 mmhg. Discussion Main finding of this study This study is the first one to trace working-age adults, identified as possible hypertension during population-based screening. Approximately 80% of study participants between 20 and 68 years of age did not follow-up with physicians within 6 months of a screening where possible hypertension was identified, using 140/90 mmhg as the cutoff. The proportion of no clinical follow-up remained high (.50%) throughout sensitivity analyses where high BP cutoffs, definition of follow-up and follow-up period were assessed in different combinations. Males with milder hypertension, younger age, proteinuria and smoking habits did not follow-up with physicians; younger females were also less likely to follow-up. Lower BMI and lower HbA1c for males and incident cases for females were found to be additional significant predictors.

502 JOURNAL OF PUBLIC HEALTH Table 1 Baseline characteristics of study participants Variables Male Female P-value Participants 15 793 1380 Median age (IQR) (years) 40 (8.5) 48 (8.5),0.01 Age category, n (%) 20 29 years old 2765 (17.5) 112 (8.1),0.01 30 39 years old 4773 (30.2) 267 (9.4) 40 49 years old 4353 (27.6) 379 (27.5) 50 59 years old 2917 (18.5) 467 (33.8) 60 68 years old 985 (6.1) 155 (11.2) Underlying disorders, n (%) Diabetes mellitus 297 (1.9) 22 (1.6) 0.45 Dyslipidemia 538 (3.4) 73 (5.3),0.01 Depression 329 (2.1) 30 (2.2) 0.82 Drinking habits, n (%) Daily 2223 (20.1) 76 (7.5),0.01 Occasionally 4927 (44.6) 219 (2.2) Rarely 3897 (35.3) 718 (70.9) Positive smoking habits, n (%) 6202 (48.3) 116 (10.7),0.01 BP Median SBP (IQR) 144 (140, 149) 146 (141, 153),0.01 Median DBP (IQR) 87 (80, 92) 89 (83, 94),0.01 Hypertension category a, n (%) Class I 14 109 (89.3) 1149 (83.3),0.01 Class II 1401 (8.9) 194 (14.0) Class III 283 (1.8) 37 (2.7) Mean BMI (IQR), kg/m 2 24.3 (2.4) 23.3 (2.8),0.01 Median hemoglobin A1c (IQR), % 5.1 (4.9, 5.3) 5.2 (4.9, 5.4),0.01 Dyslipidemia, n (%) 6244 (39.5) 544 (39.4) 0.93 Proteinuria, n (%) 2023 (12.8) 119 (8.6),0.01 Incident of identification as possible hypertension, n (%) 7535 (47.7) 939 (68.0),0.01 IQR, inter-quartile range; BP, blood pressure; SBP, systolic BP; DBP, diastolic BP. a Hypertension categories: Class I, SBP 140 mmhg and/or DBP 90 mmhg; Class II, SBP 160 mmhg and/or DBP 100 mmhg; Class III, SBP 180 mmhg and/or DBP 110 mmhg. All proportions denote number of relevant. Table 2 Frequency and proportion of no follow-up for possible hypertension BP cutoff Males Females Sample size 3 months 6 months 12 months Sample size 3 months 6 months 12 months 140/90 mmhg 15 793 14 484 (91.7) 14 161 (89.7) 13 666 (86.5) 1380 1184 (85.8) 1135 (82.3) 1077 (78.0) 150/90 mmhg 7726 6714 (86.9) 6466 (83.7) 6076 (78.6) 852 691 (81.1) 655 (76.8) 610 (71.6) 140/100 mmhg 13 897 12 677 (91.6) 12 380 (89.4) 11 956 (86.3) 1225 1046 (85.4) 1003 (81.8) 951 (77.6) 150/100 mmhg 3908 3200 (81.9) 3037 (77.8) 2803 (71.7) 508 399 (78.5) 376 (74.0) 346 (68.1) BP, blood pressure. Figures in parentheses represent the proportion of no follow-ups as a percentage.

PREDICTING FAILURE TO FOLLOW-UP SCREENED HIGH BLOOD PRESSURE IN JAPAN 503 Table 3 Logistic regression analyses of no follow-up predictors in males Age-adjusted Multivariate OR (95% CI) P-value OR (95% CI) P-value Age 0.96 (0.95 0.96),0.01 BMI (kg/m 2 ) 0.95 (0.94 0.96),0.01 0.98 (0.97 0.99) 0.02 Hemoglobin A1c 0.85 (0.80 0.90),0.01 0.92 (0.86 0.99) 0.03 SBP 0.95 (0.94 0.95),0.01 DBP 0.93 (0.92 0.93),0.01 Hypertension category a Class I 7.96 (6.21 10.21),0.01 7.20 (5.59 9.28),0.01 Class II 1.66 (1.27 2.17),0.01 1.57 (1.20 2.05),0.01 Class III 1 1 Lipid abnormality as of the last checkup 0.87 (0.78 0.97) 0.01 0.99 (0.88 1.11) 0.84 Incident case 1.28 (1.16 1.43),0.01 1.04 (0.93 1.17) 0.45 Drinking habits Daily 1 Occasionally 0.97 (0.83 1.14) 0.73 Rarely 1.34 (1.16 1.54),0.01 Positive smoking habits 1.40 (1.25 1.56),0.01 1.34 (1.19 1.50),0.01 Underlying disorders Diabetes mellitus 0.71 (0.53 0.96) 0.03 0.97 (0.67 1.39) 0.85 Dyslipidemia 0.71 (0.56 0.89),0.01 0.72 (0.56 0.92) 0.01 Depression 0.75 (0.54 1.05) 0.09 0.83 (0.56 1.16) 0.28 Proteinuria 1.37 (1.15 1.63),0.01 0.62 (0.49 0.79),0.01 OR, odds ratio; CI, confidence interval; SBP, systolic blood pressure; DBP, diastolic blood pressure. a Hypertension categories: Class I, SBP 140 mmhg and/or DBP 90 mmhg; Class II, SBP 160 mmhg and/or DBP 100 mmhg; Class III, SBP 180 mmhg and/or DBP 110 mmhg Among 22 369 individuals with possible hypertension, 3293 individuals who had been under antihypertensive treatment were excluded. This excluded population might have received medical checkups and initiated antihypertensive treatment before the study period. Therefore, the present examined population (n ¼ 19 076) might be less-conscious for clinical follow-up than the excluded ones. However, population at risk of our cohort study and research purpose needs to comprise individuals screened as possible hypertension but not concurrently treated. What is already known on this topic The National Health and Nutrition Survey conducted by the Japanese Ministry of Health, Labour and Welfare recently reported that 69.2% of males and 78.1% of females aged 30 years and screened as hypertensives did not receive antihypertensive treatment. 17 The survey was a self-administered questionnaire, which can be subject to information bias 18 and may have overestimated the prevalence of follow-up. What this study adds We had several a priori hypotheses for this study. First, patients with underlying diabetes mellitus or dyslipidemia would tend to visit the same physician who treated these diseases to seek medical advice for possible hypertension. The proportion of non-consultations with physicians remained high in both primary and sensitivity analyses, and thus the underlying medical disorders did not predict the follow-up in this cohort. Second, abnormal checkup data such as proteinuria, high HbA1c and dyslipidemia could motivate participants to seek medical advice, a suggestion not described in previous studies. We found proteinuria, high HbA1c and higher BMI for males and high BMI for females as predictors for clinical follow-up, and thus this hypothesis was confirmed in part. Third, underlying depression might deter participants from visiting physicians because a depressive state predisposes individuals to avoid taking new actions. A cross-sectional survey in Germany suggested that mental disorders could be associated with acknowledged but untreated hypertension. 19 In the present study, underlying depression

504 JOURNAL OF PUBLIC HEALTH Table 4 Logistic regression analyses of no follow-up predictors in females Age-adjusted Multivariate OR (95% CI) P-value OR (95% CI) P-value Age 0.96 (0.95 0.98),0.01 BMI (kg/m 2 ) 0.96 (0.93 0.99) 0.01 0.96 (0.93 0.99) 0.02 Hemoglobin A1c 0.96 (0.81 1.15) 0.66 SBP 0.97 (0.96 0.98),0.01 DBP 0.95 (0.93 0.96),0.01 Hypertension category a Class I 3.79 (1.90 7.56),0.01 Class II 1.32 (0.64 2.76),0.01 Class III 1 Lipid abnormality as of the last checkup 0.80 (0.60 1.07) 0.13 0.88 (0.65 1.19) 0.40 Incident case 1.37 (1.02 1.83) 0.04 1.47 (1.09 1.98) 0.01 Drinking habits Daily 1 Occasionally 1.82 (0.88 3.75) 0.10 Rarely 1.21 (0.81 1.81) 0.35 Positive smoking habits 1.38 (0.77 2.44) 0.28 Underlying disorders Diabetes mellitus 1.29 (0.43 3.88) 0.65 Dyslipidemia 0.75 (0.44 1.30) 0.31 Depression 1.10 (0.41 2.92) 0.86 Proteinuria 0.90 (0.52 1.54) 0.69 0.64 (0.35 1.16) 0.14 Being insured 0.67 (0.51 0.96) 0.03 0.66 (0.48 0.92) 0.01 OR, odds ratio; CI, confidence interval; SBP, systolic blood pressure; DBP, diastolic blood pressure. a Hypertension categories: Class I, SBP 140 mmhg and/or DBP 90 mmhg; Class II, SBP 160 mmhg and/or DBP 100 mmhg; Class III, SBP 180 mmhg and/or DBP 110 mmhg. proved to be a significant predictor of non-consultations only in the sensitivity analysis, which was inconsistent with the primary analysis. Our findings suggest that underlying depression might predict no clinical follow-up consultations in limited situations. According to the analytic framework of the U.S. Preventive Services Task Force, How can we most effectively identify patients with hypertension for whom treatment will be beneficial? is the most relevant question for strategies that bridge screening and treatment of hypertension. 20 To our knowledge, the present study is the first to focus on this issue and clarify predictors of no clinical follow-ups for possible hypertension. Unlike the USA, however, physicians who screen individuals and physicians who treat patients are not always the same in Japan; Japanese individuals screened as possible hypertensives thus need to overcome an additional obstacle to receive treatment. Because screening systems differ across countries, key questions should be established considering their own systems and problems unique to them. Limitation of this study This study has some limitations. First, the database lacked information on individual socioeconomic status, such as academic and occupational background. This made it impossible to assess factors that might have potentially affected their health behaviors. Second, the original database included a moderate amount of missing data. Drinking habits, for example, constitute an important facet of lifestyle-related diseases; however, over 20% of such data were missing and thus excluded from multivariate analysis. Other analyses yielded similar results for other variables before and after application of multiple imputation methods (data not shown), implying the robustness of our findings. Third, our cohort excluded individuals over 68 years of age. Further research in elderly population is warranted to assess whether they follow-up with physicians on potential hypertension. Despite these limitations, our study is the first to trace individuals identified as possible hypertension during populationbased screening in Japan, a unique secondary prevention

PREDICTING FAILURE TO FOLLOW-UP SCREENED HIGH BLOOD PRESSURE IN JAPAN 505 system against cardiovascular diseases. Continuous emphasis on individuals screened as possible hypertensives, considering the predictors identified by our study, will facilitate follow-up with physicians, improve management of hypertension among the target population, prevent cardiovascular events and eventually reduce related mortality. Conclusions In conclusion, approximately 80% of study participants did not follow-up with physicians after identification of possible hypertension in the population-based screening, and the proportion remained consistently high even with higher BP cutoffs. Predictors of no clinical follow-ups included young age, proteinuria, smoking habits and mild hypertension for males and young age for females. To optimize the outcome of a population-based screening system, our findings should be highlighted to trace, enlighten and motivate individuals with possible hypertension. Supplementary data Supplementary data are available at the Journal of Public Health online. Acknowledgements The authors would like to sincerely thank Japan Medical Data Center Co., Ltd, Tokyo, Japan for providing the data. Funding This study was partly supported by a grant-in-aid from the Ministry of Education, Culture, Sports, Science, and Technology. References 1 Kearney PM, Whelton M, Reynolds K et al. Global burden of hypertension: analysis of worldwide data. Lancet 2005;365:217 23. 2 United States Preventive Services Task Force. Guide to Clinical Preventive Services, 2nd ed. Alexandria, VA: International Medical Publishing, Inc, 1996. 3 Centers for Disease Control and Prevention. Vital signs: awareness and treatment of uncontrolled hypertension among adults - United States, 2003 2010. MMWR 2012;61:703 9. 4 Statistics and Information Department Minister s Secretariat Ministry of Health Labour and Welfare. The Fifth National Survey of Cardiovascular Diseases. Tokyo: Health and Welfare Statistics Association, 2002. 5 Statistics and Information Department Minister s Secretariat Ministry of Health Labour and Welfare. Estimates of National Medical Care Expenditure 2011. Tokyo: Health and Welfare Statistics Association, 2011. 6 Zanchetti A. Evidence-based medicine in hypertension: what type of evidence? J Hypertens 2005;23:1113 20. 7 Ogihara T, Kikuchi K, Matsuoka H et al. The Japanese Society of Hypertension Guidelines for the Management of Hypertension ( JSH 2009). Hypertens Res 2009;32:3 107. 8 Kimura S, Sato T, Ikeda S et al. Development of a database of health insurance claims: standardization of disease classifications and anonymous record linkage. J Epidemiol 2010;20:413 9. 9 WHO Collaborating Centre for Drug Statistics Methodology, ATC/ DDD index 2013. http://www.whocc.no/atc_ddd_index/ (29 June 2014, date last accessed). 10 World Heath Organization. International Classifications of Diseases (ICD). http://www.who.int/classifications/icd/en/ (29 October 2012, date last accessed). 11 Chobanian AV, Bakris GL, Black HR et al. Seventh report of the Joint National Committee on Prevention, Detection, Evaluation, and Treatment of High Blood Pressure. Hypertension 2003;42:1206 52. 12 World Health Organization-International Society of Hypertension Guidelines for the Management of Hypertension. Guidelines Subcommittee. J Hypertens 1999;17:151 83. 13 The Committee of Japan Diabetes Society on the diagnostic criteria of diabetes mellitus. Report of the Committee on the classification and diagnostic criteria of diabetes mellitus. J Japan Diab Soc 2010;53:450 67. 14 The Commitee of Japan Diabetes Society on the Diagnostic Criteria of Diabetes Mellitus. Report of the Commitee on the classification and diagnostic criteria of diabetes mellitus. Diabetol Int 2010;1:2 20. 15 Japan Atherosclerosis S. Japan Atherosclerosis Society ( JAS) guidelines for prevention of atherosclerotic cardiovascular diseases. J Atheroscler Thromb 2007:5 57. 16 Horton NJ, Lipsitz SR. Multiple imputation in practice: comparison of software packages for regression models with missing variables. Am Stat 2001;55:244 54. 17 Ministry of Health Labour and Welfare. National Health and Nutrition Survey. http://www.mhlw.go.jp/stf/houdou/2r98520000020qbb.html (29 June 2014, date last accessed). 18 Delgado-Rodriguez M, Llorca J. Bias. J Epidemiol Community Health 2004;58:635 41. 19 Schmitz N, Thefeld W, Kruse J. Mental disorders and hypertension: factors associated with awareness and treatment of hypertension in the general population of Germany. Psychosom Med 2006;68:246 52. 20 Sheridan S, Pignone M, Donahue K. Screening for high blood pressure: a review of the evidence for the U.S. Preventive Services Task Force. Am J Prev Med 2003;25:151 8.