Professional Education


  • Doctor of Philosophy, Stanford University, EPIDCR-PHD (2026)
  • Master of Science, Stanford University, STATS-MS (2026)
  • Master of Health Science, Johns Hopkins University, Epidemiology (2021)
  • Bachelor of Science, Peking University (2018)

Stanford Advisors


All Publications


  • Timing of mental disorders in relation to cancer-specific mortality in cancer patients. Journal of the National Cancer Institute Li, M., Deng, Z., Jung, M., L'Espérance, K., Baiocchi, M., Spiegel, D., Chung, B. I., Langston, M. E., Bondy, M. L. 2026

    Abstract

    To evaluate whether mental disorders identified before cancer diagnosis, within one year, or 2-5 years after cancer diagnosis are associated with cancer-specific mortality, and whether having both pre- and post-diagnosis disorders displays the highest mortality.We conducted a cohort study of 22,281 UK Biobank participants linked to cancer and death registries. Mental disorders were ascertained from inpatient and outpatient diagnoses and psychiatric prescriptions, then categorized by period. We applied period-specific longitudinal inverse probability weighting to preserve early-period information and estimate time-specific effects, with treatment and death censoring weights to reduce confounding and survival bias. Weighted Cox models estimated hazard ratios (HRs) and 95% confidence intervals (CIs) overall and by site (prostate, breast, colorectal, lung, and lymphoma). Additional models evaluated pre-only, post-only, and both pre- and post-diagnosis disorders versus none.Over median 5.3 year follow-up, 6,134 cancer-specific deaths occurred. Mental disorders were associated with higher overall cancer mortality: pre-diagnosis (HR [95% CI]: 1.19 [1.11,1.27]), first year post-diagnosis (1.38 [1.26,1.51]), and years 2-5 (2.19 [1.97,2.43]). Associations were strongest and most consistent during years 2-5 across sites (HRs, 1.58-2.27). Site-specific analyses additionally showed elevated hazards for pre-diagnosis and first-year disorders in lymphoma and first-year disorders in colorectal cancer. Patients with post-diagnosis mental disorders, regardless of pre-diagnosis status, had higher cancer mortality than those with pre-diagnosis disorders only or none.Mental disorders were associated with cancer-specific mortality throughout the continuum, with the strongest associations in years 2-5 across multiple sites. Findings support routine mental-health screening and intervention in survivorship care.

    View details for DOI 10.1093/jnci/djag289

    View details for PubMedID 42658001

  • Factors associated with mood and anxiety disorders in cancer survivors: A systematic review and meta-analysis. Li, M., Zhou, X., Deng, Z., Jung, M., Lu, J., Spiegel, D., Langston, M., Chung, B. I., Bondy, M. LIPPINCOTT WILLIAMS & WILKINS. 2026: e24106
  • Metabolic syndrome and risk of kidney cancer in the United States. Cancer Jung, M., Wang, X., Jeon, S. Y., Graff, R. E., Deng, Z., L'Espérance, K., Li, M., Langston, M. E., Chung, B. I. 2026; 132 (9): e70435

    Abstract

    Metabolic dysregulation may contribute to kidney cancer development through shared biological mechanisms and/or as an independent risk factor. However, evidence on this association in the US population remains limited.This study identifies associations between metabolic syndrome (MetS) and kidney cancer risk in the United States.This case-control study using administrative claims data from the MarketScan database (2007-2022) identified adult kidney cancer cases and 1:10 frequency-matched controls by age, index year, sex, insurance duration, and region. Metabolic exposures (obesity, hypertension, diabetes, and dyslipidemia) were assessed using diagnoses and prescriptions, with binary classification (MetS: ≥3 conditions; non-MetS: <3). Secondary analyses examined pre-MetS (1-2 conditions), metabolically healthy, dose-response relationships, combinations of conditions, and interaction between obesity and MetS. Multivariable logistic regression models were used to estimate odds ratios (ORs) and 95% confidence intervals (CIs).A total of 48,587 kidney cancer cases and 480,714 controls were included. MetS was associated with an increased risk of kidney cancer compared with non-MetS (OR, 1.43; 95% CI, 1.39-1.47). MetS (OR, 1.83; 95% CI, 1.77-1.88) and pre-MetS (OR, 1.47; 95% CI, 1.44-1.50) were associated with a higher risk compared with metabolically healthy. Each additional condition was associated with a 22% increase in the odds (95% CI, 1.21-1.23). Multiple conditions had higher odds than a single condition. Metabolically healthy obesity (OR, 1.48) and metabolically unhealthy obesity (OR, 1.77) showed stronger associations than metabolically healthy nonobesity.MetS was associated with an increased risk of kidney cancer, highlighting the importance of overall metabolic health in its prevention.

    View details for DOI 10.1002/cncr.70435

    View details for PubMedID 42057471

  • Longitudinal Patterns of Cardiovascular Disease and Cancer Mortality in Renal Cell Carcinoma Patients by Tumor Stage JACC: CARDIOONCOLOGY Deng, Z., Jung, M., Li, M., Li, J., Langston, M. E., Chung, B. I. 2025; 7 (5): 627-640

    Abstract

    Cardiovascular disease (CVD) is the leading cause of non-cancer deaths among patients with renal cell carcinoma (RCC).The authors sought to investigate how CVD and RCC death rates change post-RCC diagnosis and time-dependent risk factors for death, across different tumor stages.Adults diagnosed with a first RCC from the Surveillance, Epidemiology, and End Results database (2004-2020) were included in this study. Poisson regression was used to model RCC and CVD death rates over 16-year follow-up, identifying a crossover time when the curves intersected, overall and by tumor stage. Fine-Gray competing risk models were implemented to evaluate the time-varying associations of risk factors with CVD and RCC deaths.Among overall 116,836 RCC patients, the CVD death rate overtook RCC death rate at 13.7 (95% CI: 12.0-15.4) years during follow-up. For stage I RCC patients, the crossover time was sooner at 2.8 (95% CI: 2.0-3.7) years, with all subgroups either experiencing a crossover or having a consistently higher CVD death rate than RCC death rate. No crossover time was identified for stage II-IV overall. Patients aged ≥75 years and non-Hispanic Black patients experienced a crossover time across all tumor stages. Several sociodemographic and clinical characteristics showed consistent associations with short- and long-term mortality from RCC and CVD.In RCC patients, stage I individuals showed a pronounced trend where CVD death became dominant over RCC death during survivorship, with consistent patterns by different sociodemographic and clinical characteristics. Management of both cancer and CVD during critical periods is essential for improving survival outcomes in RCC patients, with strategies tailored to tumor stage.

    View details for DOI 10.1016/j.jaccao.2025.06.004

    View details for Web of Science ID 001561955800015

    View details for PubMedID 40841112

  • Association between antihypertensive medication use and kidney cancer risk: a meta-analysis accounting for hypertension. BMC cancer Jung, M., Li, M., Shin, J., Chung, B. I., Langston, M. E. 2025; 25 (1): 1013

    Abstract

    Evidence that antihypertensive medication (AHTN) use is associated with an increased risk of kidney cancer (KC) is emerging. However, limited evidence is available on disentangling the effects of AHTN use on KC from hypertension, which is a risk factor for KC. We aimed to identify pooled estimates for the associations between AHTN use and KC risk, independent of hypertension.We searched for observational studies that investigated the associations between AHTN use and KC through January 2025. To identify the independent effects of AHTN from hypertension, we conducted stratified analyses with and without accounting for hypertension: any methods (matching, adjustment, or stratification/restriction) versus none. We conducted random-effects meta-analyses with robust variance estimation to calculate pooled relative risk (RR).In this meta-analysis consisting of 39 eligible studies, AHTN use was associated with an increased risk of KC based on estimates that accounted for hypertension (RR 1.19, 95% confidence interval (CI) 0.93-1.52 for angiotensin-converting enzyme inhibitor; RR 1.15, 95% CI 1.00-1.31 for angiotensin receptor blocker; RR 1.09, 95% CI 1.03-1.16 for beta-blocker, RR 1.40, 95% CI 1.12-1.75 for calcium channel blocker (CCB); RR 1.36, 95% CI 1.20-1.55 for diuretic; and RR 1.40, 95% CI 1.13-1.75 for non-classified AHTN). Findings from duration‒response relationships supported the main findings.AHTN use was associated with an increased risk of KC compared to no use, even after accounting for hypertension, with the highest risk observed for CCB. Our findings highlight the potential KC risks associated with different AHTN classes, with optimal cardiovascular care remaining an important consideration.

    View details for DOI 10.1186/s12885-025-14406-3

    View details for PubMedID 40481406

    View details for PubMedCentralID PMC12143101

  • Incidence of Depression and Anxiety in Kidney Cancer Survivors: A Nationwide Population-Based Cohort Study. Cancer epidemiology, biomarkers & prevention : a publication of the American Association for Cancer Research, cosponsored by the American Society of Preventive Oncology Jung, M., Li, M., Choo, E., Lee, S., Spiegel, D., Baiocchi, M., Deng, Z., Li, J., Langston, M. E., Bondy, M. L., Chung, B. I. 2025: OF1-OF9

    Abstract

    Depression and anxiety have a high prevalence among kidney cancer survivors. We aimed to evaluate their cumulative incidence trajectories and associations between kidney cancer diagnosis and incidence of these conditions.This population-based cohort study used the Korean Nationwide Health Insurance and Medical Checkup Linkage Database. We included adults (≥20 years) diagnosed with kidney cancer (2010-2020; i.e., cancer survivors), along with age- and sex-matched noncancer comparators. We quantified the empirical risk trajectory of depression and anxiety for up to 5 years following diagnosis and conducted weighted Cox regressions to estimate time-dependent HRs with 95% confidence intervals (CI) within three time intervals: 0 to 1, 1 to 3, and 3 to 5 years.For the empirical risks, survivors (n = 24,310) had higher risks of depression (2.8% vs. 2.2%) and anxiety (3.3% vs. 2.6%) compared with comparators (n = 173,471). For the associations, survivors (n = 16,049) had an increased hazard of depression (HR = 1.92; 95% CI, 1.52-2.42) and anxiety (HR = 1.63; 95% CI, 1.31-2.02) compared with comparators (n = 100,782) in the first year. During the subsequent 1 to 3 years, survivors experienced an increased hazard of anxiety (HR = 1.32; 95% CI, 1.07-1.62). Trends of decreasing HRs for both disorders were observed across successive time intervals.Kidney cancer survivors had a higher rate of depression and anxiety, especially during the early phase following diagnosis, compared with the noncancer population.Our findings emphasize the need for early identification and treatment of psychiatric disorders, highlighting the integration of mental health care into oncology settings. They also inform future research on prevention and treatment strategies, focusing on timing and high-risk groups.

    View details for DOI 10.1158/1055-9965.EPI-24-1879

    View details for PubMedID 40287950

  • Disparities in Cancer and Cardiovascular Disease Mortality Among Asian Americans Diagnosed with Urologic Cancer JOURNAL OF RACIAL AND ETHNIC HEALTH DISPARITIES Deng, Z., Li, M., Li, J., Jung, M., Batai, K., Moore, J. X., Langston, M. E., Chung, B. I. 2025

    Abstract

    Asian Americans (AA) in the United States represent a heterogenous population from various ethnic backgrounds. We compared cancer and cardiovascular disease (CVD) mortality between various AA groups and Non-Hispanic White (NHW) patients diagnosed with urologic cancer.We assembled a population-based cohort that included 389,114 prostate cancer, 98,721 renal cell cancer, and 126,485 bladder cancer patients. Cumulative cancer and CVD mortality were compared between AA and NHW groups, accounting for competing risk of death. Multivariable Cox models were used to quantify the cause-specific hazard ratio (HR) with a 95% confidence interval (CI), comparing AA subgroups (Chinese, Japanese, Korean, Filipino, Vietnamese, Other Southeast Asian, and Indian/Pakistani) to NHW patients.AA ethnic subgroups had a lower or comparable mortality from prostate cancer compared with NHW patients (HR ranged 0.51-1.03). No overall difference was observed for renal cell cancer death, but an increased mortality was observed for Filipino (HR = 1.10; 95% CI, 1.00-1.22) and Other Southeast Asian (HR = 1.50; 95% CI, 1.06-2.12) patients that included Laotian, Hmong, Kampuchean, and Thai ethnicity. Although reduced mortality from bladder cancer (HR = 0.88; 95% CI, 0.83-0.93) was observed compared to NHW patients, an increased mortality was seen among Other Southeast Asians (HR = 1.63; 95% CI, 1.15-2.30). CVD mortality varied across AA ethnicities, with higher mortality observed in Filipino and Other Southeast Asian (HR ranged 1.23-2.40) compared with Chinese patients.Large heterogeneity exists in mortality among AA patients diagnosed with urologic cancer, with higher mortality from cancer and CVD observed in Filipino and Other Southeast Asian patients.

    View details for DOI 10.1007/s40615-025-02354-4

    View details for Web of Science ID 001441641000001

    View details for PubMedID 40067651

    View details for PubMedCentralID 6104025

  • Calcium Channel Blocker Versus Renin-Angiotensin System Inhibitor in Risk of Kidney Cancer Among Patients With Hypertension: A Propensity Score-Matched Cohort Study. Cancer medicine Jung, M., Li, S., Deng, Z., Li, J., Li, M., Basran, S., Langston, M. E., Chung, B. I. 2024; 13 (22): e70429

    Abstract

    Use of antihypertensive medications could be associated with an increased risk of kidney cancer. Despite their various mechanisms of action, whether this association differs between different classes of medications remains unclear.The objective of this study is to compare the risk of kidney cancer between first-line treatment options of antihypertensive medications in a hypertensive population.In this retrospective cohort study, we used the MarketScan Databases (2007-2021). We included individuals older than 30 years of age with a diagnosis of hypertension who received first-line medications for hypertension, which included three classes: angiotensin-converting enzyme inhibitors (ACEI), angiotensin receptor blockers (ARB), and dihydropyridine calcium channel blockers (dCCB). We applied a propensity score matching method and created three separate cohorts: (1) ARB versus ACEI, (2) dCCB versus ACEI, and (3) dCCB versus ACEI. For non-dCCB, we repeated the analyses. The primary outcome was kidney cancer incidence. To assess kidney cancer risk, we applied multivariable conditional Cox proportional hazards models.In the first cohort, ARB use was associated with an increased risk of kidney cancer compared to ACEI use (hazard ratio [HR] 1.10, 95% confidence interval [CI] 1.02-1.18). In the second cohort, dCCB use was associated with an increased risk of kidney cancer compared to ACEI use (HR 1.29, 95% CI 1.18-1.40). In the third cohort, dCCB use was associated with a higher risk of kidney cancer compared to ARB use (HR 1.17, 95% CI 1.08-1.28). Null association was shown when comparing non-dCCB with ACEI or ARB use.Use of dCCB showed a higher risk of kidney cancer compared to ACEI or ARB use in patients with hypertension.

    View details for DOI 10.1002/cam4.70429

    View details for PubMedID 39548764

    View details for PubMedCentralID PMC11568363

  • Increased risk of cardiovascular disease among kidney cancer survivors: a nationwide population-based cohort study. Frontiers in oncology Jung, M., Choo, E., Li, S., Deng, Z., Li, J., Li, M., Basran, S., Lee, S., Langston, M. E., Chung, B. I. 2024; 14: 1420333

    Abstract

    Cardiovascular disease (CVD) is a major concern of morbidity and mortality among cancer survivors. However, few evidence exists on the short- and long-term risk of CVD in kidney cancer (KCa) survivors.In this nationwide, large population-based retrospective cohort study, we used the Korean national health insurance and medical checkup survey linkage database (2007-2021), drawn from the entire Korean population. We included adults diagnosed with KCa as the first primary cancer and matched them to an individual without KCa at a 1:5 ratio. The primary outcome was CVD incidence, including myocardial infarction, stroke, atrial fibrillation, heart failure, peripheral arterial occlusion, and venous thromboembolism (VTE). We evaluated CVD risk at 6 months, 1 year, and 5 years following cancer diagnosis, using Fine-Gray competing risk models that accounted for death as a competing factor.A total of 149,232 participants were included (KCa survivors: N=20,093 and matched non-KCa individuals: N=129,139). After 6-month follow-up, KCa survivors showed an increased risk of CVD compared to the general population (subdistribution hazard ratio (HR) 2.70, 95% confidence interval (CI) 2.31-3.15). After 1 year, KCa survivors had a higher risk of CVD (HR=1.77, 95% CI: 1.56-2.00). After 5 years, this elevated CVD risk remained (HR=1.10, 95% CI: 1.03-1.18), with VTE identified as the primary contributing disease (HR=3.05, 95% CI:2.59-3.59).KCa survivors had an increased risk of CVD up to 5 years after cancer diagnosis compared to the general population. Our findings emphasize the importance of comprehensive healthcare management for both CVD and KCa throughout cancer survivorship.

    View details for DOI 10.3389/fonc.2024.1420333

    View details for PubMedID 39070148

    View details for PubMedCentralID PMC11272517

  • Frequency and distribution of ophthalmic surgical procedures among patients with inherited retinal diseases. Ophthalmology. Retina Mees, L., Li, M., Antonio-Aguirre, B., Liu, T. Y., Wu, A., Kong, X., Singh, M. S. 2024

    Abstract

    OBJECTIVE OR PURPOSE: In this study, we aimed to characterize the frequency and distribution of ocular surgeries in patients with inherited retinal diseases (IRDs) and evaluate associated patient and disease factors.DESIGN: Retrospective cohort.PARTICIPANTS: Subjects 18 years and older who were followed at the Johns Hopkins Genetic Eye Disease (GEDi) Center.METHODS: We studied a retrospective cohort of patients with an IRD diagnosis to analyze the occurrence of laser and incisional surgeries. Subjects were categorized into two groups: central dysfunction (macular/cone/cone-rod dystrophy, "MCCRD group") and panretinal or peripheral dysfunction (retinitis pigmentosa-like, "RP group"). Genetic testing status was recorded. The association of patient and disease factors on the frequency, distribution, and timing of surgeries was analyzed.MAIN OUTCOME MEASURE: Prevalence, prevalence odds ratio (POR), hazard ratio (HR) of ophthalmic procedures by phenotype.RESULTS: A total of 1472 eyes of 736 subjects were evaluated. Among them, 31.3% (n = 230) had undergone ocular surgery, and 78.3% of those (n=180/230) had a history of more than one surgery. A total of 602 surgical procedures were analyzed. Cataract extraction with intraocular lens implantation (CEIOL) was the most common (51.2%), followed by YAG capsulotomy, refractive surgery, retinal surgery, and others. CEIOL occurred more frequently in RP than in MCCRD subjects (POR 2.59, p = 0.002). RP subjects underwent CEIOL at a younger age than MCCRD patients (HR = 2.11, p < 0.001).CONCLUSION: Approximately one-third of IRD patients had a history of laser or incisional surgery. CEIOL was the most common surgery; its frequency and timing may be associated with IRD phenotype. This data may inform the design of prospective research. Such efforts may illuminate routine clinical decision-making and contribute to surgical strategy development for cell and gene therapy delivery.

    View details for DOI 10.1016/j.oret.2024.03.005

    View details for PubMedID 38485090

  • Greater Gestational Vitamin D Status is Associated with Reduced Childhood Behavioral Problems in the Environmental Influences on Child Health Outcomes Program JOURNAL OF NUTRITION Melough, M. M., Li, M., Hamra, G., Palmore, M., Sauder, K. A., Dunlop, A. L., Lewinn, K. Z., Zhao, Q., Kelly, R. S., Switkowski, K. M., Hipwell, A. E., Korrick, S. A., Collett, B. R., Mackenzie, D., Nozadi, S. S., Kerver, J. M., Schmidt, R. J., Mcgrath, M., Sathyanarayana, S. 2023; 153 (5): 1502-1511

    Abstract

    Vitamin D deficiency is common in pregnancy. Vitamin D plays an important role in the developing brain, and deficiency may impair childhood behavioral development.This study examined the relationship between gestational 25(OH)D concentrations and childhood behavior in the Environmental influences on Child Health Outcomes (ECHO) Program.Mother-child dyads from ECHO cohorts with data available on prenatal (first trimester through delivery) or cord blood 25(OH)D and childhood behavioral outcomes were included. Behavior was assessed using the Strengths and Difficulties Questionnaire or the Child Behavior Checklist, and data were harmonized using a crosswalk conversion. Linear mixed-effects models examined associations of 25(OH)D with total, internalizing, and externalizing problem scores while adjusting for important confounders, including age, sex, and socioeconomic and lifestyle factors. The effect modification by maternal race was also assessed.Early (1.5-5 y) and middle childhood (6-13 y) outcomes were examined in 1688 and 1480 dyads, respectively. Approximately 45% were vitamin D deficient [25(OH)D < 20 ng/mL], with Black women overrepresented in this group. In fully adjusted models, 25(OH)D concentrations in prenatal or cord blood were negatively associated with externalizing behavior T-scores in middle childhood [-0.73 (95% CI: -1.36, -0.10) per 10 ng/mL increase in gestational 25(OH)D]. We found no evidence of effect modification by race. In a sensitivity analysis restricted to those with 25(OH)D assessed in prenatal maternal samples, 25(OH)D was negatively associated with externalizing and total behavioral problems in early childhood.This study confirmed a high prevalence of vitamin D deficiency in pregnancy, particularly among Black women, and revealed evidence of an association between lower gestational 25(OH)D and childhood behavioral problems. Associations were more apparent in analyses restricted to prenatal rather than cord blood samples. Interventions to correct vitamin D deficiency during pregnancy should be explored as a strategy to improve childhood behavioral outcomes.

    View details for DOI 10.1016/j.tjnut.2023.03.005

    View details for Web of Science ID 001054463100001

    View details for PubMedID 37147034

    View details for PubMedCentralID PMC10367223

  • Trends in Screen Time Use Among Children During the COVID-19 Pandemic, July 2019 Through August 2021. JAMA network open Hedderson, M. M., Bekelman, T. A., Li, M., Knapp, E. A., Palmore, M., Dong, Y., Elliott, A. J., Friedman, C., Galarce, M., Gilbert-Diamond, D., Glueck, D., Hockett, C. W., Lucchini, M., McDonald, J., Sauder, K., Zhu, Y., Karagas, M. R., Dabelea, D., Ferrara, A., Environmental Influences on Child Health Outcomes Program, Smith, B. P., Newby, K. L., Jacobson, L. P., Parker, C. B., Gershon, R. C., Cella, D., Alshawabkeh, A. N., Aschner, J. L., Merhar, S. L., Pryhuber, G. S., Moore, P. E., Lampland, A. L., Wadhawan, R., Wagner, C. L., Hudak, M. L., Mayock, D. E., Washburn, L. K., Croen, L. A., Karr, C. J., Mason, A., Lester, B. M., Carter, B. S., Marsit, C. J., Pastyrnak, S. L., Neal, C., Smith, L. M., Helderman, J. B., McEvoy, C., Tepper, R. S., Kerver, J., Barone, C., McKane, P., Paneth, N., Elliott, M. R., Trasande, L., Sathyanarayana, S., Bush, N., Nguyen, R. H., Barrett, E. S. 2023; 6 (2): e2256157

    Abstract

    Importance: The COVID-19 pandemic led to widespread lockdowns and school closures that may have affected screen time among children. Although restrictions were strongest early in the pandemic, it is unclear how screen time changed as the pandemic progressed.Objective: To evaluate change in children's screen time from before the pandemic to during the pandemic, from July 2019 through August 2021.Design, Setting, and Participants: This is a longitudinal cohort study with repeated measures of screen time collected before the pandemic and during 2 pandemic periods. Children aged 4 to 12 years and their parent were enrolled in 3 pediatric cohorts across 3 states in the US participating in the Environmental Influences of Child Health Outcomes (ECHO) Program. Data analysis was performed from November 2021 to July 2022.Exposures: COVID-19 pandemic period: prepandemic (July 2019 to March 2020), pandemic period 1 (December 2020 to April 2021), and pandemic period 2 (May 2021 to August 2021).Main Outcomes and Measures: The primary outcomes were total, educational (not including remote school), and recreational screen time assessed via the ECHO Child Media Use questionnaire. Linear mixed-effects models were used for screen time adjusted for child's age, number of siblings, sex, race, ethnicity, and maternal education.Results: The cohort included 228 children (prepandemic mean [SD] age, 7.0 [2.7] years; 100 female [43.9%]) with screen time measured during the prepandemic period and at least once during the pandemic period. Prepandemic mean (SD) total screen time was 4.4 (3.9) hours per day and increased 1.75 hours per day (95% CI, 1.18-2.31 hours per day) in the first pandemic period and 1.11 hours per day (95% CI, 0.49-1.72 hours per day) in the second pandemic period, in adjusted models. Prepandemic mean (SD) recreational screen time was 4.0 (3.5) hours per day and increased 0.89 hours per day (95% CI, 0.39-1.39 hours per day) in the first pandemic period and 0.70 hours per day (95% CI, 0.16-1.25 hours per day) in the second pandemic period. Prepandemic mean (SD) educational screen time was 0.5 (1.2) hours per day (median [IQR], 0.0 [0.0-0.4] hours per day) and increased 0.93 hours per day (95% CI, 0.67-1.19 hours per day) in the first pandemic period and 0.46 hours per day (95% CI, 0.18-0.74 hours per day) in the second pandemic period.Conclusions and Relevance: These findings suggest that screen time among children increased during the COVID-19 pandemic and remained elevated even after many public health precautions were lifted. The long-term association of increased screen time during the COVID-19 pandemic with children's health needs to be determined.

    View details for DOI 10.1001/jamanetworkopen.2022.56157

    View details for PubMedID 36790805

  • Prenatal Antidepressant Exposures and Autism Spectrum Disorder or Traits: A Retrospective, Multi-Cohort Study RESEARCH ON CHILD AND ADOLESCENT PSYCHOPATHOLOGY Brennan, P. A., Dunlop, A. L., Croen, L. A., Avalos, L. A., Salisbury, A. L., Hipwell, A. E., Nozadi, S. S., Sathyanarayana, S., Crum, R. M., Musci, R., Li, M., Li, X., Mansolf, M., O'Connor, T. G., Elliott, A. J., Ghildayal, N., Lin, P. D., Sprowles, J. L. N., Stanford, J. B., Bendixsen, C., Ozonoff, S., Lester, B. M., Shuster, C. L., Huddleston, K. C., Posner, J., Paneth, N. 2022

    Abstract

    Prenatal antidepressant exposure has been associated with increased risk for neurodevelopmental disorders in childhood, including autism spectrum disorder (ASD). The current study utilized multi-cohort data from the Environmental influences on Child Health Outcomes (ECHO) program (N = 3129) to test for this association, and determine whether the association remained after adjusting for maternal prenatal depression and other potential confounders. Antidepressants and a subset of selective serotonin reuptake inhibitors (SSRIs) were examined in relation to binary (e.g., diagnostic) and continuous measures of ASD and ASD related traits (e.g., social difficulties, behavior problems) in children 1.5 to 12 years of age. Child sex was tested as an effect modifier. While prenatal antidepressant exposure was associated with ASD related traits in univariate analyses, these associations were statistically non-significant in models that adjusted for prenatal maternal depression and other maternal and child characteristics. Sex assigned at birth was not an effect modifier for the prenatal antidepressant and child ASD relationship. Overall, we found no association between prenatal antidepressant exposures and ASD diagnoses or traits. Discontinuation of antidepressants in pregnancy does not appear to be warranted on the basis of increased risk for offspring ASD.

    View details for DOI 10.1007/s10802-022-01000-5

    View details for Web of Science ID 000886839800002

    View details for PubMedID 36417100

    View details for PubMedCentralID 8504533

  • Impact of the COVID-19 pandemic on children's sleep habits: an ECHO study. Pediatric research Lucchini, M., Bekelman, T. A., Li, M., Knapp, E. A., Dong, Y., Ballard, S., Deoni, S., Dunlop, A. L., Elliott, A. J., Ferrara, A., Friedman, C., Galarce, M., Gilbert-Diamond, D., Glueck, D., Hedderson, M., Hockett, C. W., Karagas, M. R., LeBourgeois, M. K., Margolis, A., McDonald, J., Ngai, P., Pellerite, M., Sauder, K., Ma, T., Dabelea, D., Environmental influences on Child Health Outcomes, Smith, P. B., Newby, K. L., Jacobson, L. P., Catellier, D. J., Gershon, R., Cella, D., Alshawabkeh, A. N., Aschner, J., Merhar, S., Lampland, A., Reynolds, A., Hudak, M., Pryhuber, G., Moore, P., Washburn, L., Croen, L., Karr, C., Mason, A., Lester, B., Carter, B., Marsit, C., Pastyrnak, S., Neal, C., Smith, L., Helderman, J., McEvoy, C., Tepper, R., Kerver, J., Barone, C., McKane, P., Paneth, N., Elliott, M., Trasande, L., Sathyanarayana, S., Bush, N., Nguyen, R., Barrett, E. 2022

    Abstract

    BACKGROUND: Sleep in childhood is affected by behavioral, environmental, and parental factors. We propose that these factors were altered during the COVID-19 pandemic. This study investigates sleep habit changes during the pandemic in 528 children 4-12 years old in the US, leveraging data from the Environmental Influences on Child Health Outcomes (ECHO) Program.METHODS: Data collection occurred in July 2019-March 2020 (pre-pandemic) and two pandemic periods: December 2020-April 2021 and May-August 2021. Qualitative interviews were performed in 38 participants.RESULTS: We found no changes in sleep duration, but a shift to later sleep midpoint during the pandemic periods. There was an increase in latency at the first pandemic collection period but no increase in the frequency of bedtime resistance, and a reduced frequency of naps during the pandemic. Qualitative interviews revealed that parents prioritized routines to maintain sleep duration but were more flexible regarding timing. Children from racial/ethnic minoritized communities slept less at night, had later sleep midpoint, and napped more frequently across all collection periods, warranting in-depth investigation to examine and address root causes.CONCLUSIONS: The COVID-19 pandemic significantly impacted children sleep, but parental knowledge of the importance of sleep might have played a significant protective role.IMPACT: During the COVID-19 pandemic, US children changed their sleep habits, going to bed and waking up later, but their sleep duration did not change. Sleep latency was longer. Parental knowledge of sleep importance might have played a protective role. Regardless of data collection periods, children from racial/ethnic minoritized communities slept less and went to bed later. This is one of the first study on this topic in the US, including prospective pre-pandemic qualitative and quantitative data on sleep habits. Our findings highlight the pandemic long-term impact on childhood sleep. Results warrants further investigations on implications for overall childhood health.

    View details for DOI 10.1038/s41390-022-02309-z

    View details for PubMedID 36195633

  • Comparing Generalized Estimating Equation and Linear Mixed Effects Model for Estimating Marginal Association with Bivariate Continuous Outcomes OPHTHALMIC EPIDEMIOLOGY Li, M., Kong, X. 2023; 30 (3): 307-316

    Abstract

    Both linear regression with generalized estimating equations (GEE) and linear mixed-effects models (LMEM) can be used to estimate the marginal association of an exposure with clustered continuous outcomes. This study compares their performance for bivariate continuous outcomes which are common in eye studies.Parametric and non-parametric simulations were used to compare the GEE models including independent, exchangeable, and unstructured working correlation structures and LMEM including random intercept only and random intercept and slope models in R and SAS. Data generation referenced the data distributions from a real-world study for estimating ocular structure-visual function relationships in patients with retinitis pigmentosa.From both parametric and non-parametric simulations, comparing the random intercept LMEM and GEE exchangeable model, bias was similar; coverage probability of the 95% confidence interval (CI) from the random intercept LMEM was often closer to 95%, especially when the sample size was small; the power for testing the association of the exposure was higher from the GEE exchangeable model, but its type-I error rate might be inflated especially when the sample size was small. The type-I error rate from the random intercept LMEM was closer to 0.05, but it might be under 0.05 and coverage probability might be over 95%. The GEE independent model performed worst and the LMEM with both random intercept and slope might not converge.To estimate marginal exposure-outcome association with bivariate continuous outcomes, the random intercept LMEM may be preferred. It has the best coverage probability of 95% CI and is the only model with correct type-I error rates in this study. However, it may have low power and overly wide CI in studies with small sample size or low inter-eye correlation.

    View details for DOI 10.1080/09286586.2022.2098984

    View details for Web of Science ID 000826036400001

    View details for PubMedID 35838059

    View details for PubMedCentralID PMC9840717