Work Experience

  • Senior Statistical Scientist, University of California San Francisco (April 1, 2018)


    San Francisco

  • Senior Statistician, University of California Davis, Center for Healthcare Policy and Research (June 1, 2017)



All Publications

  • Estimating Vitamin K Antagonist Anticoagulation Benefit in People With Atrial Fibrillation Accounting for Competing Risks: Evidence From 12 Randomized Trials. Circulation. Cardiovascular quality and outcomes Shah, S. J., van Walraven, C., Jeon, S. Y., Boscardin, J., Hobbs, F. D., Connolly, S. J., Ezekowitz, M. D., Covinsky, K. E., Fang, M. C., Singer, D. E. 2024: e010269


    Patients with atrial fibrillation have a high mortality rate that is only partially attributable to vascular outcomes. The competing risk of death may affect the expected anticoagulant benefit. We determined if competing risks materially affect the guideline-endorsed estimate of anticoagulant benefit.We conducted a secondary analysis of 12 randomized controlled trials that randomized patients with atrial fibrillation to vitamin K antagonists (VKAs) or either placebo or antiplatelets. For each participant, we estimated the absolute risk reduction (ARR) of VKAs to prevent stroke or systemic embolism using 2 methods-first using a guideline-endorsed model (CHA2DS2-VASc) and then again using a competing risk model that uses the same inputs as CHA2DS2-VASc but accounts for the competing risk of death and allows for nonlinear growth in benefit. We compared the absolute and relative differences in estimated benefit and whether the differences varied by life expectancy.A total of 7933 participants (median age, 73 years, 36% women) had a median life expectancy of 8 years (interquartile range, 6-12), determined by comorbidity-adjusted life tables and 43% were randomized to VKAs. The CHA2DS2-VASc model estimated a larger ARR than the competing risk model (median ARR at 3 years, 6.9% [interquartile range, 4.7%-10.0%] versus 5.2% [interquartile range, 3.5%-7.4%]; P<0.001). ARR differences varied by life expectancies: for those with life expectancies in the highest decile, 3-year ARR difference (CHA2DS2-VASc model - competing risk model 3-year risk) was -1.3% (95% CI, -1.3% to -1.2%); for those with life expectancies in the lowest decile, 3-year ARR difference was 4.7% (95% CI, 4.5%-5.0%).VKA anticoagulants were exceptionally effective at reducing stroke risk. However, VKA benefits were misestimated with CHA2DS2-VASc, which does not account for the competing risk of death nor decelerating treatment benefit over time. Overestimation was most pronounced when life expectancy was low and when the benefit was estimated over a multiyear horizon.

    View details for DOI 10.1161/CIRCOUTCOMES.123.010269

    View details for PubMedID 38525596

  • Development and External Validation of Models to Predict Need for Nursing Home Level of Care in Community-Dwelling Older Adults With Dementia. JAMA internal medicine Deardorff, W. J., Jeon, S. Y., Barnes, D. E., Boscardin, W. J., Langa, K. M., Covinsky, K. E., Mitchell, S. L., Lee, S. J., Smith, A. K. 2023


    Most older adults living with dementia ultimately need nursing home level of care (NHLOC).To develop models to predict need for NHLOC among older adults with probable dementia using self-report and proxy reports to aid patients and family with planning and care management.This prognostic study included data from 1998 to 2016 from the Health and Retirement Study (development cohort) and from 2011 to 2019 from the National Health and Aging Trends Study (validation cohort). Participants were community-dwelling adults 65 years and older with probable dementia. Data analysis was conducted between January 2022 and October 2023.Candidate predictors included demographics, behavioral/health factors, functional measures, and chronic conditions.The primary outcome was need for NHLOC defined as (1) 3 or more activities of daily living (ADL) dependencies, (2) 2 or more ADL dependencies and presence of wandering/need for supervision, or (3) needing help with eating. A Weibull survival model incorporating interval censoring and competing risk of death was used. Imputation-stable variable selection was used to develop 2 models: one using proxy responses and another using self-responses. Model performance was assessed by discrimination (integrated area under the receiver operating characteristic curve [iAUC]) and calibration (calibration plots).Of 3327 participants with probable dementia in the Health and Retirement Study, the mean (SD) age was 82.4 (7.4) years and 2301 (survey-weighted 70%) were female. At the end of follow-up, 2107 participants (63.3%) were classified as needing NHLOC. Predictors for both final models included age, baseline ADL and instrumental ADL dependencies, and driving status. The proxy model added body mass index and falls history. The self-respondent model added female sex, incontinence, and date recall. Optimism-corrected iAUC after bootstrap internal validation was 0.72 (95% CI, 0.70-0.75) in the proxy model and 0.64 (95% CI, 0.62-0.66) in the self-respondent model. On external validation in the National Health and Aging Trends Study (n = 1712), iAUC in the proxy and self-respondent models was 0.66 (95% CI, 0.61-0.70) and 0.64 (95% CI, 0.62-0.67), respectively. There was excellent calibration across the range of predicted risk.This prognostic study showed that relatively simple models using self-report or proxy responses can predict need for NHLOC in community-dwelling older adults with probable dementia with moderate discrimination and excellent calibration. These estimates may help guide discussions with patients and families in future care planning.

    View details for DOI 10.1001/jamainternmed.2023.6548

    View details for PubMedID 38048097

    View details for PubMedCentralID PMC10696518

  • Frequency of Screening for Colorectal Cancer by Predicted Life Expectancy Among Adults 76-85 Years. JAMA Deardorff, W. J., Lu, K., Jing, B., Jeon, S. Y., Boscardin, W. J., Fung, K. Z., Lee, S. J. 2023; 330 (13): 1280-1282

    View details for DOI 10.1001/jama.2023.15820

    View details for PubMedID 37676665

    View details for PubMedCentralID PMC10485741

  • Social Frailty Index: Development and validation of an index of social attributes predictive of mortality in older adults. Proceedings of the National Academy of Sciences of the United States of America Shah, S. J., Oreper, S., Jeon, S. Y., Boscardin, W. J., Fang, M. C., Covinsky, K. E. 2023; 120 (7): e2209414120


    While social characteristics are well-known predictors of mortality, prediction models rely almost exclusively on demographics, medical comorbidities, and function. Lacking an efficient way to summarize the prognostic impact of social factor, many studies exclude social factors altogether. Our objective was to develop and validate a summary measure of social risk and determine its ability to risk-stratify beyond traditional risk models. We examined participants in the Health and Retirement Study, a longitudinal, survey of US older adults. We developed the model from a comprehensive inventory of 183 social characteristics using least absolute shrinkage and selection operator, a penalized regression approach. Then, we assessed the predictive capacity of the model and its ability to improve on traditional prediction models. We studied 8,250 adults aged ≥65 y. Within 4 y of the baseline interview, 22% had died. Drawn from 183 possible predictors, the Social Frailty Index included age, gender, and eight social predictors: neighborhood cleanliness, perceived control over financial situation, meeting with children less than yearly, not working for pay, active with children, volunteering, feeling isolated, and being treated with less courtesy or respect. In the validation cohort, predicted and observed mortality were strongly correlated. Additionally, the Social Frailty Index meaningfully risk-stratified participants beyond the Charlson score (medical comorbidity index) and the Lee Index (comorbidity and function model). The Social Frailty Index includes age, gender, and eight social characteristics and accurately risk-stratifies older adults. The model improves upon commonly used risk prediction tools and has application in clinical, population health, and research settings.

    View details for DOI 10.1073/pnas.2209414120

    View details for PubMedID 36749720

    View details for PubMedCentralID PMC9963593

  • Development and External Validation of a Mortality Prediction Model for Community-Dwelling Older Adults With Dementia. JAMA internal medicine Deardorff, W. J., Barnes, D. E., Jeon, S. Y., Boscardin, W. J., Langa, K. M., Covinsky, K. E., Mitchell, S. L., Whitlock, E. L., Smith, A. K., Lee, S. J. 2022; 182 (11): 1161-1170


    Estimating mortality risk in older adults with dementia is important for guiding decisions such as cancer screening, treatment of new and chronic medical conditions, and advance care planning.To develop and externally validate a mortality prediction model in community-dwelling older adults with dementia.This cohort study included community-dwelling participants (aged ≥65 years) in the Health and Retirement Study (HRS) from 1998 to 2016 (derivation cohort) and National Health and Aging Trends Study (NHATS) from 2011 to 2019 (validation cohort).Candidate predictors included demographics, behavioral/health factors, functional measures (eg, activities of daily living [ADL] and instrumental activities of daily living [IADL]), and chronic conditions.The primary outcome was time to all-cause death. We used Cox proportional hazards regression with backward selection and multiple imputation for model development. Model performance was assessed by discrimination (integrated area under the receiver operating characteristic curve [iAUC]) and calibration (plots of predicted and observed mortality).Of 4267 participants with probable dementia in HRS, the mean (SD) age was 82.2 (7.6) years, 2930 (survey-weighted 69.4%) were female, and 785 (survey-weighted 12.1%) identified as Black. Median (IQR) follow-up time was 3.9 (2.0-6.8) years, and 3466 (81.2%) participants died by end of follow-up. The final model included age, sex, body mass index, smoking status, ADL dependency count, IADL difficulty count, difficulty walking several blocks, participation in vigorous physical activity, and chronic conditions (cancer, heart disease, diabetes, lung disease). The optimism-corrected iAUC after bootstrap internal validation was 0.76 (95% CI, 0.75-0.76) with time-specific AUC of 0.73 (95% CI, 0.70-0.75) at 1 year, 0.75 (95% CI, 0.73-0.77) at 5 years, and 0.84 (95% CI, 0.82-0.85) at 10 years. On external validation in NHATS (n = 2404), AUC was 0.73 (95% CI, 0.70-0.76) at 1 year and 0.74 (95% CI, 0.71-0.76) at 5 years. Calibration plots suggested good calibration across the range of predicted risk from 1 to 10 years.We developed and externally validated a mortality prediction model in community-dwelling older adults with dementia that showed good discrimination and calibration. The mortality risk estimates may help guide discussions regarding treatment decisions and advance care planning.

    View details for DOI 10.1001/jamainternmed.2022.4326

    View details for PubMedID 36156062

    View details for PubMedCentralID PMC9513707

  • Comparing Machine Learning to Regression Methods for Mortality Prediction Using Veterans Affairs Electronic Health Record Clinical Data. Medical care Jing, B., Boscardin, W. J., Deardorff, W. J., Jeon, S. Y., Lee, A. K., Donovan, A. L., Lee, S. J. 2022; 60 (6): 470-479


    It is unclear whether machine learning methods yield more accurate electronic health record (EHR) prediction models compared with traditional regression methods.The objective of this study was to compare machine learning and traditional regression models for 10-year mortality prediction using EHR data.This was a cohort study.Veterans Affairs (VA) EHR data.Veterans age above 50 with a primary care visit in 2005, divided into separate training and testing cohorts (n= 124,360 each).The primary outcome was 10-year all-cause mortality. We considered 924 potential predictors across a wide range of EHR data elements including demographics (3), vital signs (9), medication classes (399), disease diagnoses (293), laboratory results (71), and health care utilization (149). We compared discrimination (c-statistics), calibration metrics, and diagnostic test characteristics (sensitivity, specificity, and positive and negative predictive values) of machine learning and regression models.Our cohort mean age (SD) was 68.2 (10.5), 93.9% were male; 39.4% died within 10 years. Models yielded testing cohort c-statistics between 0.827 and 0.837. Utilizing all 924 predictors, the Gradient Boosting model yielded the highest c-statistic [0.837, 95% confidence interval (CI): 0.835-0.839]. The full (unselected) logistic regression model had the highest c-statistic of regression models (0.833, 95% CI: 0.830-0.835) but showed evidence of overfitting. The discrimination of the stepwise selection logistic model (101 predictors) was similar (0.832, 95% CI: 0.830-0.834) with minimal overfitting. All models were well-calibrated and had similar diagnostic test characteristics.Our results should be confirmed in non-VA EHRs.The differences in c-statistic between the best machine learning model (924-predictor Gradient Boosting) and 101-predictor stepwise logistic models for 10-year mortality prediction were modest, suggesting stepwise regression methods continue to be a reasonable method for VA EHR mortality prediction model development.

    View details for DOI 10.1097/MLR.0000000000001720

    View details for PubMedID 35352701

    View details for PubMedCentralID PMC9106858

  • Predicting Life Expectancy to Target Cancer Screening Using Electronic Health Record Clinical Data. Journal of general internal medicine Lee, A. K., Jing, B., Jeon, S. Y., Boscardin, W. J., Lee, S. J. 2022; 37 (3): 499-506


    Guidelines recommend breast and colorectal cancer screening for older adults with a life expectancy >10 years. Most mortality indexes require clinician data entry, presenting a barrier for routine use in care. Electronic health records (EHR) are a rich clinical data source that could be used to create individualized life expectancy predictions to identify patients for cancer screening without data entry.To develop and internally validate a life expectancy calculator from structured EHR data.Retrospective cohort study using national Veteran's Affairs (VA) EHR databases.Veterans aged 50+ with a primary care visit during 2005.We assessed demographics, diseases, medications, laboratory results, healthcare utilization, and vital signs 1 year prior to the index visit. Mortality follow-up was complete through 2017. Using the development cohort (80% sample), we used LASSO Cox regression to select ~100 predictors from 913 EHR data elements. In the validation cohort (remaining 20% sample), we calculated the integrated area under the curve (iAUC) and evaluated calibration.In 3,705,122 patients, the mean age was 68 years and the majority were male (97%) and white (85%); nearly half (49%) died. The life expectancy calculator included 93 predictors; age and gender most strongly contributed to discrimination; diseases also contributed significantly while vital signs were negligible. The iAUC was 0.816 (95% confidence interval, 0.815, 0.817) with good calibration.We developed a life expectancy calculator using VA EHR data with excellent discrimination and calibration. Automated life expectancy prediction using EHR data may improve guideline-concordant breast and colorectal cancer screening by identifying patients with a life expectancy >10 years.

    View details for DOI 10.1007/s11606-021-07018-7

    View details for PubMedID 34327653

    View details for PubMedCentralID PMC8858374

  • Long-term individual and population functional outcomes in older adults with atrial fibrillation. Journal of the American Geriatrics Society Parks, A. L., Jeon, S. Y., Boscardin, W. J., Steinman, M. A., Smith, A. K., Fang, M. C., Shah, S. J. 2021; 69 (6): 1570-1578


    Older adults with atrial fibrillation (AF) have multiple risk factors for disablement. Long-term function and the contribution of strokes to disability have not been previously characterized. Our objective was to determine long-term function among older adults with AF and the relative contribution of stroke.We used data from the nationally representative Health and Retirement Study (1992-2014) with participants ≥65 years with incident AF. We examined the association of incident stroke with three outcomes: independence with activities of daily living (ADL), instrumental activities of daily living (IADL), and residence outside a nursing home (community-dwelling). We fit logistic regression models with repeated measures adjusting for comorbidities and demographics to estimate the effect of stroke on function. We estimated the contribution of strokes to the overall population burden of disability using the method of recycled predictions.Among 3530 participants (median age 79 years, 53% women), 262 had a stroke over 17,396 person-years. Independent of stroke and accounting for comorbidities, annually, ADL independence decreased by 4.4%, IADL independence decreased by 3.9%, and community dwelling decreased by 1.2% (p < 0.05 for all). Accounting for comorbidities, of those who experienced a stroke, 31.9% developed new ADL dependence, 26.5% developed new IADL dependence, and 8.6% newly moved to a nursing home (p < 0.05 for all). Considering all causes of function loss, 1.7% of ADL disability-years, 1.2% of IADL disability-years, and 7.3% of nursing home years could be attributed to stroke over 7.4 years.Older adults lose substantial function over time following AF diagnosis, independent of stroke. Stroke was associated with a significant functional decline and increase in the likelihood of nursing home move, but stroke did not accelerate subsequent disability accrual. Because of the high background rate of disability, stroke was not the dominant determinant of population-level disability in older adults with AF.

    View details for DOI 10.1111/jgs.17087

    View details for PubMedID 33675093

    View details for PubMedCentralID PMC8442883

  • Geriatric Syndromes and Atrial Fibrillation: Prevalence and Association with Anticoagulant Use in a National Cohort of Older Americans. Journal of the American Geriatrics Society Shah, S. J., Fang, M. C., Jeon, S. Y., Gregorich, S. E., Covinsky, K. E. 2021; 69 (2): 349-356


    Although guidelines recommend focusing primarily on stroke risk to recommend anticoagulants in atrial fibrillation (AF), physicians report that geriatric syndromes (e.g., falls and disability) are important when considering anticoagulants. Little is known about the prevalence of geriatric syndromes in older adults with AF or the association with anticoagulant use.We performed a cross-sectional analysis of the 2014 Health and Retirement Study, a nationally representative study of older Americans. Participants were asked questions to assess domains of aging, including function, cognition, and medical conditions. We included participants 65 years and older with 2 years of continuous Medicare enrollment who met AF diagnosis criteria by claims codes. We examined five geriatric syndromes: one or more falls within the last 2 years, receiving help with activities of daily living (ADLs) or instrumental ADLs (IADL), experienced incontinence, and cognitive impairment. We determined the prevalence of geriatric syndromes and their association with anticoagulant use, adjusting for ischemic stroke risk (i.e., CHA2 DS2 -VASc score [congestive heart failure, hypertension, age, diabetes mellitus, stroke, vascular disease, and sex]).In this study of 779 participants with AF (median age = 80 years; median CHA2 DS2 -VASc score = 4), 82% had one or more geriatric syndromes. Geriatric syndromes were common: 49% reported falls, 38% had ADL impairments, 42% had IADL impairments, 37% had cognitive impairments, and 43% reported incontinence. Overall, 65% reported anticoagulant use; guidelines recommend anticoagulant use for 97% of participants. Anticoagulant use rate decreased for each additional geriatric syndrome (average marginal effect = -3.7%; 95% confidence interval = -1.4% to -5.9%). Lower rates of anticoagulant use were reported in participants with ADL dependency, IADL dependency, and dementia.Most older adults with AF had at least one geriatric syndrome, and geriatric syndromes were associated with reduced anticoagulant use. The high prevalence of geriatric syndromes may explain the lower than expected anticoagulant use in older adults.

    View details for DOI 10.1111/jgs.16822

    View details for PubMedID 32989731

    View details for PubMedCentralID PMC8174581

  • Fingerstick Glucose Monitoring in Veterans Affairs Nursing Home Residents with Diabetes Mellitus. Journal of the American Geriatrics Society Jeon, S. Y., Shi, Y., Lee, A. K., Hunt, L., Lipska, K., Boscardin, J., Lee, S. 2021; 69 (2): 424-431


    Guidelines recommend less intensive glycemic treatment and less frequent glucose monitoring for nursing home (NH) residents. However, little is known about the frequency of fingerstick (FS) glucose monitoring in this population. Our objective was to examine the frequency of FS glucose monitoring in Veterans Affairs (VA) NH residents with diabetes mellitus, type II (T2DM).National retrospective cohort study in 140 VA NHs.NH residents with T2DM and older than 65 years admitted to VA NHs between 2013 and 2015 following discharge from a VA hospital.NH residents were classified into five groups based on their highest hypoglycemia risk glucose-lowering medication (GLM) each day: no GLMs; metformin only; sulfonylureas; long-acting insulin; and any short-acting insulin. Our outcome was a daily count of FS measurements.Among 17,474 VA NH residents, mean age was 76 (standard deviation (SD) = 8) years and mean hemoglobin A1c was 7.6% (SD = 1.5%). On day 1 after NH admission, 49% of NH residents were on short-acting insulin, decreasing slightly to 43% at day 90. Overall, NH residents had an average of 1.9 (95% confidence interval (CI) = 1.8-1.9) FS measurements on NH day 1, decreasing to 1.4 (95% CI = 1.3-1.4) by day 90. NH residents on short-acting insulin had the most frequent FS measurements, with 3.0 measurements (95% CI = 2.9-3.0) on day 1, decreasing to 2.6 measurements (95% CI = 2.5-2.7) by day 90. Less frequent FS measurements were seen for NH residents receiving long-acting insulin (2.1 (95% CI = 2.0-2.2) on day 1) and sulfonylureas (1.7 (95% CI = 1.5-1.8) on day 1). Even NH residents on metformin monotherapy had 1.1 (95% CI = 1.1-1.2) measurements on day 1, decreasing to 0.5 (95% CI = 0.4-0.6) measurements on day 90.Although guidelines recommend less frequent glucose monitoring for NH residents, we found that many VA NH residents receive frequent FS monitoring. Given the uncertain benefits and potential for substantial patient burdens and harms, our results suggest decreasing FS monitoring may be warranted for many low hypoglycemia risk NH residents.

    View details for DOI 10.1111/jgs.16880

    View details for PubMedID 33064879

    View details for PubMedCentralID PMC8139138

  • Patterns and Trends in Advance Care Planning Among Older Adults Who Received Intensive Care at the End of Life. JAMA internal medicine Block, B. L., Jeon, S. Y., Sudore, R. L., Matthay, M. A., Boscardin, W. J., Smith, A. K. 2020; 180 (5): 786-789


    This study uses Medicare claims data to identify and assess disparities in the use of advance care planning among adults 65 years or older who died between 2000 and 2015 and received intensive care during the last 30 days of life.

    View details for DOI 10.1001/jamainternmed.2019.7535

    View details for PubMedID 32119031

    View details for PubMedCentralID PMC7052782

  • Clinical Outcomes and Costs Following Unplanned Excisions of Soft Tissue Sarcomas in the Elderly. The Journal of surgical research Bateni, S. B., Gingrich, A. A., Jeon, S. Y., Hoch, J. S., Thorpe, S. W., Kirane, A. R., Bold, R. J., Canter, R. J. 2019; 239: 125-135


    Surgical guidelines for soft tissue sarcoma (STS) emphasize pretreatment evaluation and reports of the perils of unplanned excision exist. Given the paucity of population-based data on this topic, our objective was to analyze clinical outcomes and costs of planned versus unplanned STS excisions in the Medicare population.We analyzed 3913 surgical patients with STS ≥66 y old from 1992 to 2011 using the Surveillance, Epidemiology, and End Results-Medicare datafiles. Planned excisions were classified based on preoperative MRI and/or biopsy, whereas unplanned excisions were classified by excision as the first procedure. Inverse probability of treatment weighting with propensity scores was used to adjust for clinicopathologic differences. Re-excisions, complications, and Medicare payments were compared with multivariate models. Overall survival and disease-specific survival were analyzed using Cox proportional hazards and competing risk models.Before the first excision, 24.3% had an MRI and biopsy, 27.3% had an MRI, 11.4% had a biopsy, and 36.9% were unplanned. Re-excision rates were highest for unplanned excisions: 46.3% compared to 18.1%, 36.4%, and 29.7% for other groups (P < 0.0001). There was no difference in disease-specific survival or overall survival between groups (P > 0.05). Planned excisions were associated with increased Medicare costs (P < 0.05), with the first resection contributing to the majority of costs. Subgroup analyses by histologic grade and tumor size revealed similar results.Survival was comparable with greater health care costs in elderly patients undergoing planned STS excision. Although unplanned excisions remain a quality of care issue with high re-excision rates, these data have important implications for the surgical management of STS in the elderly.

    View details for DOI 10.1016/j.jss.2019.01.055

    View details for PubMedID 30825757

    View details for PubMedCentralID PMC6488355

  • Nonstandard Employment and Health in South Korea: The Role of Gender and Family Status SOCIOLOGICAL PERSPECTIVES Lim, S., Jeon, S., Kim, J., Woo, H. 2018; 61 (6): 973-999
  • Playing With the Rules and Making Misleading Statements: A Response to Luo, Hodges, Winship, and Powers AMERICAN JOURNAL OF SOCIOLOGY Land, K. C., Fu, Q., Guo, X., Jeon, S. Y., Reither, E. N., Zang, E. 2016; 122 (3): 962-973

    View details for DOI 10.1086/689853

    View details for Web of Science ID 000390414800008

  • A population-based analysis of increasing rates of suicide mortality in Japan and South Korea, 1985-2010. BMC public health Jeon, S. Y., Reither, E. N., Masters, R. K. 2016; 16: 356


    In the past two decades, rates of suicide mortality have declined among most OECD member states. Two notable exceptions are Japan and South Korea, where suicide mortality has increased by 20 % and 280 %, respectively.Population and suicide mortality data were collected through national statistics organizations in Japan and South Korea for the period 1985 to 2010. Age, period of observation, and birth cohort membership were divided into five-year increments. We fitted a series of intrinsic estimator age-period-cohort models to estimate the effects of age-related processes, secular changes, and birth cohort dynamics on the rising rates of suicide mortality in the two neighboring countries.In Japan, elevated suicide rates are primarily driven by period effects, initiated during the Asian financial crisis of the late 1990s. In South Korea, multiple factors appear to be responsible for the stark increase in suicide mortality, including recent secular changes, elevated suicide risks at older ages in the context of an aging society, and strong cohort effects for those born between the Great Depression and the aftermath of the Korean War.In spite of cultural, demographic and geographic similarities in Japan and South Korea, the underlying causes of increased suicide mortality differ across these societies-suggesting that public health responses should be tailored to fit each country's unique situation.

    View details for DOI 10.1186/s12889-016-3020-2

    View details for PubMedID 27107481

    View details for PubMedCentralID PMC4841973

  • Clarifying hierarchical age-period-cohort models: A rejoinder to Bell and Jones. Social science & medicine (1982) Reither, E. N., Land, K. C., Jeon, S. Y., Powers, D. A., Masters, R. K., Zheng, H., Hardy, M. A., Keyes, K. M., Fu, Q., Hanson, H. A., Smith, K. R., Utz, R. L., Yang, Y. C. 2015; 145: 125-8


    Previously, Reither et al. (2015) demonstrated that hierarchical age-period-cohort (HAPC) models perform well when basic assumptions are satisfied. To contest this finding, Bell and Jones (2015) invent a data generating process (DGP) that borrows age, period and cohort effects from different equations in Reither et al. (2015). When HAPC models applied to data simulated from this DGP fail to recover the patterning of APC effects, B&J reiterate their view that these models provide "misleading evidence dressed up as science." Despite such strong words, B&J show no curiosity about their own simulated data--and therefore once again misapply HAPC models to data that violate important assumptions. In this response, we illustrate how a careful analyst could have used simple descriptive plots and model selection statistics to verify that (a) period effects are not present in these data, and (b) age and cohort effects are conflated. By accounting for the characteristics of B&J's artificial data structure, we successfully recover the "true" DGP through an appropriately specified model. We conclude that B&Js main contribution to science is to remind analysts that APC models will fail in the presence of exact algebraic effects (i.e., effects with no random/stochastic components), and when collinear temporal dimensions are included without taking special care in the modeling process. The expanded list of coauthors on this commentary represents an emerging consensus among APC scholars that B&J's essential strategy--testing HAPC models with data simulated from contrived DGPs that violate important assumptions--is not a productive way to advance the discussion about innovative APC methods in epidemiology and the social sciences.

    View details for DOI 10.1016/j.socscimed.2015.07.013

    View details for PubMedID 26277370

    View details for PubMedCentralID PMC4673395