Non-missing Data Research Articles

Risk Factors for Natural Hearing Evolution in Newborns With Congenital Cytomegalovirus Infection

Congenital cytomegalovirus (cCMV) is the major cause of congenital nonhereditary sensorineural hearing loss in children. Currently, criteria to identify infants at increased risk for unfavorable hearing outcome are lacking. To identify risk factors associated with cCMV-related hearing improvement, hearing deterioration, and late-onset hearing loss. This multicenter cohort study included patients from 6 secondary and tertiary hospitals enrolled in the Flemish CMV registry (Belgium). Newborns with untreated cCMV infection with at least 4-year audiological follow-up were included. Patients who presented with other possible causes of sensorineural hearing loss were excluded. Data were collected for 15 years (January 1, 2007, to February 7, 2022) and analyzed from September 26, 2022, to January 16, 2023. Primary outcome was hearing evolution (per-ear analysis; described as stable hearing, improvement, or deterioration). The association of gestational characteristics, clinical findings, timing of seroconversion, viral load, and hearing status at birth with hearing evolution was investigated using effect sizes (Cramer V, odds ratio [OR], or Hedges g). Of the 387 children, 205 of 385 with nonmissing data were male (53.2%), 113 (29.2%) had a symptomatic infection, and 274 (70.8%) had an asymptomatic infection. Every child was 4 years or older at final hearing evaluation. A total of 701 of 774 ears (90%) showed stable hearing (normal hearing or stable hearing loss since birth) over time. Late-onset hearing loss (normal hearing at birth followed by hearing loss) was present in 43 of 683 ears (6.3%). Among children with hearing loss present at birth, 24 of 34 ears (70.6%) had hearing deterioration, and 6 of 91 ears (6.6%) had hearing improvement. Prematurity was associated with a higher chance of hearing improvement (OR, 12.80; 95% CI, 2.03-80.68). Late-onset hearing loss was more prevalent in a first trimester infection (OR, 10.10; 95% CI, 2.90-34.48). None of the 104 ears of children with a third trimester seroconversion developed late-onset hearing loss. Findings of this cohort study support that ongoing audiological follow-up for untreated children with congenital hearing loss is important, as the majority of patients had hearing deterioration. The timing of seroconversion was associated with the risk of developing late-onset hearing loss. These insights can aid in parental counseling, patient stratification, and follow-up. Future research should focus on the effect of treatment, the influence of determined risk factors, and the study of eventual new risk factors in patients at high risk to develop hearing loss.

Rural-urban differences in social and emotional protective factors and their association with child health and flourishing.

Children in rural communities experience higher mortality rates and less access to health care services than those in urban communities. Protective factors like social support also vary by geography, but their contribution to differences in child health is understudied. Understanding geographic variation in protective health factors could provide insight into their impact on health and guide future intervention strategies. To examine protective factors' association with child flourishing and child health status in rural and urban communities. Publicly available data from the National Survey of Children's Health, 2018-2021, with nonmissing geographic data (N = 150,493) were used to assess the relationship between protective factors and child flourishing and health by rurality. Multivariate survey-weighted probit models examined these relationships, adjusting for child and caregiver characteristics. More than a third of children were not flourishing, according to parental report. An estimated 62% of rural children were flourishing compared to 66% of urban children (P<0.001). Urban caregivers also were more likely to report better adult mental and physical health status. Nevertheless, rural children were reported to receive more social support than urban children, while their caregivers reported more emotional support and living in supportive and safe neighborhoods (P<0.001). Rural caregivers reported more support from places of worship and less from counselors/other mental health care providers than urban caregivers. Despite higher reported caregiver emotional support and child social support, fewer rural children are flourishing. Health systems and community organizations able to leverage these existing social and emotional protective factors in rural communities could help close this gap.

Evaluation of Imputation Techniques for Genotypic Data of Soybean Crop under Missing Completely at Random Mechanism

Background: The issue of missing data is prevalent in all type of research work, which can diminish statistical power and lead to inaccurate results if not managed correctly. Missing data cannot be ignored because every piece of data, no matter how small, affects the outcome significantly. Imputation is a key component in dealing with missing data; however, the best way to impute missing values has not yet been identified. Methods: Our goal of this paper is to compare four more recently developed imputation techniques - MICE, MI, missForest and Amelia. In order to examine the performance of various imputation techniques, non-missing data were deleted from genotypic data of soybean crop with varied frequency of missingness by missing completely at random mechanism. The study compared different imputation techniques for solving missing values using the root mean square error and mean absolute error. Result: To fill in the dataset’s missing values, the imputation technique producing the lowest value of the RMSE and MAE will be taken into consideration. Finally it is observed that missForest technique performs best on the genotypic data of soybean at different proportion of missingness.

Assessment of racial and ethnic inequities in copay card utilization and enrollment in copay adjustment programs.

BACKGROUND: Increases in consumer cost sharing lead to decreases in the use of both high- and low-value care. Copay assistance was designed to reduce out-of-pocket (OOP) cost burden. Commercial insurers have recently instituted copay adjustment programs (CAPs), which exclude copay assistance from deductibles and OOP cost maximums, thereby effectively increasing the financial burden on patients. The utilization of these programs by specific demographic populations is unknown. OBJECTIVE: To assess utilization of copay assistance and CAP exposure in a commercially insured patient population and examine potential differences in the use of each of these programs by non-White and by White patients. METHODS: A retrospective, cross-sectional study using IQVIA Longitudinal Access and Adjudication Data, linked to Experian Marketing Solutions, LLC consumer data, identified unique patients who were younger than 65 years, covered by commercial insurance, had at least 1 pharmacy claim for treatment within prespecified therapeutic areas, and had full financial data visibility on paid claims (ie, nonmissing data on costs associated with the pharmacy claim and the secondary payer) between January 1, 2019, and September 30, 2021. Analyses of copay card use or CAP exposure (defined as the likelihood to be included in the accumulator or maximizer program) between non-White and White patient populations were adjusted for age, gender, household income, patient state of residence, pharmacy benefit manager, state-level CAP policy, and overall drug cost. RESULTS: In total, 4,073,599 unique patients (5.6% of the total database population) were included in the copay card analysis. In adjusted analyses, there were no significant differences in copay card utilization between non-White patients and White patients (odds ratio [OR] = 0.995, 95% CI = 0.99-1.00; P = 0.0964). However, among copay card users, non-White patients were significantly more likely to be exposed to CAPs, as either maximizers (OR = 1.27, 95% CI = 1.22-1.33; P < 0.0001) or accumulators (OR = 1.31, 95% CI = 1.26-1.36; P < 0.0001), compared with White patients. CONCLUSIONS: In an adjusted analysis of this selected sample of a commercially insured population, there was no difference in the use of copay cards between non-White and White patients. CAP exposure, however, was significantly higher among non-White patients. This increased exposure suggests a disproportionate effect due to this reduction in copay assistance benefits, which has the potential to exacerbate racial and ethnic disparities in access to medications. DISCLOSURES: This study was sponsored by Janssen Scientific Affairs, LLC. Mr Ingham, Dr Sadik, and Dr Song are employees of Janssen Scientific Affairs, LLC. Dr Zhao is an employee of IQVIA. Dr Fendrick is a consultant for AbbVie, Amgen, Bayer, CareFirst BlueCross BlueShield, Centivo, Community Oncology Association, Covered California, EmblemHealth, Exact Sciences, Freedman Health, GRAIL, Harvard University, Health & Wellness Innovations, Health at Scale Technologies, HealthCorum, Hygeia, MedsIncontext, MedZed, Merck, Mercer, Montana Health Cooperative, Pair Team, Penguin Pay, Phathom Pharmaceuticals, Proton Intelligence, Risalto Health, Risk International, Sempre Health, Silver Fern Health, State of Minnesota, Teladoc Health, US Department of Defense, Virginia Center for Health Innovation, Wellth, Wildflower Health, Yale New Haven Health System, and Zansors; received research funds from Agency for Healthcare Research and Quality (AHRQ), Boehringer-Ingelheim, Gary and Mary West Health Policy Center, Arnold Ventures, National Pharmaceutical Council, Patient-Centered Outcomes Research Institute (PCORI), Pharmaceutical Research and Manufacturers of America (PhRMA), Robert Wood Johnson (RWJ) Foundation, State of Michigan/The Centers for Medicare & Medicaid Services (CMS); and has an outside position at the American Journal of Managed Care (AJMC; co-editor), Medicare Evidence Development & Coverage Advisory Committee (MEDCAC) member, VBID Health (partner).

Exclusion of Families Who Speak Languages Other than English from Federally Funded Pediatric Trials

To determine whether individuals in the United States who speak languages other than English (LOE) are excluded from federally funded pediatric clinical trials and whether such trials adhere to the National Institutes of Health policy regarding inclusion of members of minority groups. Using ClinicalTrials.gov, we identified all completed, federally funded, US-based trials inclusive of children ≤17 and focused on one of 4 common chronic childhood conditions (asthma, mental health, obesity, and dental caries) as of June 18, 2019. We reviewed ClinicalTrials.gov online content, as well as published manuscripts linked to ClinicalTrials.gov entries, to abstract information about language-related exclusion criteria. Trials were deemed to exclude LOE participants/caregivers if explicit statements regarding exclusion were identified in the study protocol or published manuscript. Of total, 189 trials met inclusion criteria. Two-thirds (67%) did not address multilingual enrollment. Of the 62 trials that did, 82% excluded LOE individuals. No trials addressed the enrollment of non-English, non-Spanish-speaking individuals. In 93 trials with nonmissing data on ethnicity, Latino individuals comprised 31% of participants in trials that included LOE individuals and 14% of participants in trials that excluded LOE individuals. Federally funded pediatric trials in the United States do not adequately address multilingual enrollment, a seeming violation of federal and contractual requirements for accommodation of language barriers by entities receiving federal funding.

P74 AMGEVITA® persistence in patients with plaque psoriasis: a retrospective database analysis from the British Association of Dermatologists Biologics and Immunomodulators Register

Abstract While multiple studies have evaluated the persistence of biologic agents in patients with plaque psoriasis (Yiu ZZN, Becher G, Kirby B et al. Drug survival associated with effectiveness and safety of treatment with guselkumab, ixekizumab, secukinumab, ustekinumab and adalimumab in patients with psoriasis. JAMA Dermatol 2022; 158:1131–41), persistence data for adalimumab biosimilars are limited. From BADBIR, we assessed AMGEVITA persistence in UK patients aged ≥ 18 years with a diagnosis of chronic plaque psoriasis, who had received at least one dose of AMGEVITA between October 2018 and July 2021 with at least 6 months of follow-up. Study outcomes included AMGEVITA persistence at 6, 12, 18 and 24 months (primary); reasons for AMGEVITA discontinuation (secondary) and Psoriasis Area Severity Index (PASI) score from baseline to 6 months after AMGEVITA start date (exploratory). Outcomes were analysed overall and separately for biologic-experienced (transitioned to AMGEVITA from another biologic) vs. biologic-naïve (started AMGEVITA with no prior biologics) patients. Persistence was defined as the continuous use of AMGEVITA from first administration to discontinuation using a permissible gap of &amp;lt; 90 days assessed using Kaplan–Meier survival methodology. Of 1432 eligible patients, most (n = 1,183; 82.6%) were biologic experienced, 40.1% (n = 574) were female and 56% were aged 35–54 years. At AMGEVITA initiation, mean (SD) time since psoriasis diagnosis overall was 25.4 (12.9) years, and 20.6 (12.6) and 26.6 (12.7) years in biologic-naïve and biologic-experienced patients, respectively; mean (SD) PASI was 8.1 (8.7) in the overall population, and 14.8 (7.6) and 3.2 (5.6) in biologic-naïve and biologic-experienced patients, respectively. Overall, AMGEVITA persistence at 6, 12, 18 and 24 months was 0.94 [95% confidence interval (CI) 0.93–0.95], 0.87 (95% CI 0.86–0.89), 0.84 (95% CI 0.82–0.86) and 0.82 (95% CI 0.80–0.84), respectively; 17.7% (n = 253) of patients discontinued AMGEVITA and mean (SD) time to discontinuation was 9.4 (6.0) months. The most common reasons for AMGEVITA discontinuation were ineffectiveness (n = 102; 40.3%) and adverse events (n = 74; 29.2%). Persistence at 6, 12, 18 and 24 months was higher in biologic-experienced patients than in biologic-naïve patients: 0.94 (95% CI 0.92–0.95), 0.88 (95% CI 0.86–0.90), 0.85 (95% CI 0.83–0.87) and 0.84 (95% CI 0.81–0.86) vs. 0.94 (95% CI 0.91–0.97), 0.84 (95% CI 0.78–0.88), 0.77 (95% CI 0.71–0.82) and 0.74 (95% CI 0.68–0.80), respectively. Mean (SD) time to AMGEVITA discontinuation was similar in biologic-naïve and biologic-experienced patients: 9.0 (6.1) and 10.5 (5.6) months, respectively. Among patients with non-missing data (n = 94 in both groups), mean (SD) change in PASI score at 6 months was −11.3 (8.6) in biologic-naïve and −1.0 (4.4) in biologic-experienced patients. In conclusion, patients with chronic plaque psoriasis initiating AMGEVITA in UK clinical practice demonstrated high drug persistence over 2 years, irrespective of prior biologic use. Funding sources: this analysis was funded by Amgen.

Use of Intergovernmental Transfers-based Medicaid Supplemental Payments to Boost Nursing Home Finances: Evidence From Indiana Nursing Homes.

Indiana provides intergovernmental transfers-based supplemental payment to nursing homes owned/operated by the nonstate governmental organization (NSGO) but NSGOs may divert substantial supplemental payments away from participating nursing homes. The aim of this study was to estimate the effect of participation in the intergovernmental transfers-based Medicaid supplemental payment program on nursing home revenue and expenditures. Difference-in-differences regressions using Callaway and Sant'Anna method accounting for treatment effect heterogeneity across groups and over time. All 410 Medicare and Medicaid-certified nursing homes in Indiana from 2009 to 2017 with nonmissing data (N=3170). The key independent variable is a binary variable indicating NSGO ownership. Outcome variables include total revenue, total operating, clinical, hotel, and administrative expenses as well as profit margins extracted from Medicare Cost Report. Control variables include facility and resident characteristics from Nursing Home Compare and LTCfocus data. Supplemental payments increased nursing home revenues by about $0.58 million on average but payments were larger in later years. On a per-person per day basis, nursing home revenue increased by $21.9 dollars with an increase in administrative ($11.3), and hotel ($6.9) expenses but a decrease in clinical ($4.67) expenses. NSGO-owned/operated nursing homes received only a fraction of the total supplemental payments on average, but we observed increased payments to nursing homes in later years. Participating nursing homes did not increase clinical expenses. Our findings raise questions on the transparency of the financing arrangements between NSGOs and nursing homes and the need to link supplemental payments to clinical expenses.

Assessment of sociodemographic disparities in outcomes among pts with metastatic NSCLC (mNSCLC) receiving select immuno-oncology (IO) regimens: CORRELATE.

9055 Background: IO treatment (tx) has transformed the care of pts with mNSCLC, persistently demonstrating significant survival benefits over chemotherapy. Prior analyses have shown OS and PFS are shorter in the real-world (rw) setting compared with randomized clinical trials (RCTs). A contributing factor may be the more diverse sociodemographic characteristics of pts in the rw. Race and socio-economic status (SES), in particular, have been associated with disparities in cancer outcomes, which may be related to inequalities in healthcare access. One aim of CORRELATE was to assess whether outcomes differ based on sociodemographic characteristics in pts receiving IO in the rw setting. Methods: Eligible pts from the US Flatiron Enhanced Datamart database had mNSCLC (progressed or de novo), started 1L IO tx between Nov 1 2016 and May 31 2021, and met eligibility criteria based on 6 RCTs (KEYNOTE [KN]-024, KN-189, KN-407, IMpower150, CheckMate [CM] 9LA and CM 227) for IO regimens approved in the US for mNSCLC. Flatiron’s area-level measure of SES, available for pts treated at community oncology centers, was based on a validated composite index and has 5 levels (highest [5] to lowest [1]). Multivariate Cox regression models were used to determine the association of sociodemographic variables (age, sex, race, US region and SES) with OS and rwPFS separately; smoking history, ECOG performance status (PS) and PD-L1 expression were also included in the models, as were any pairwise interactions that were statistically significant ( P&lt; 0.05). Results: 2070 pts with non-missing data on the variables of interest were included in the dataset: 60% were aged ≥65 y, 86% white, 55% male and 53% were from the US Southern region. SES was relatively evenly distributed (SES 1–5: 18%, 22%, 21%, 24% and 15%, respectively); 41% of pts had PD-L1 ≥50% (17% unknown PD-L1 expression), 61% had ECOG PS 1 and 92% were former/current smokers. Median OS was 13.3 mos (95% CI 12.3–14.7); median rwPFS was 5.9 mos (5.6–6.2). In the multivariate analyses, race, region and SES did not have a statistically significant association with either OS or rwPFS. Older age (≥65 y; P= 0.0091) and male sex ( P&lt; 0.0001) were associated with a higher risk of death (HR [95% CI] 1.16 [1.04–1.29] and 1.28 [1.15–1.43], respectively). For rwPFS, the interaction between sex and smoking status was statistically significant ( P= 0.0312); specifically, female non-smokers had a higher risk of progression or death vs female smokers (HR 1.45 [95% CI 1.15–1.82]). Conclusions: Our results from a nationally representative dataset of pts with mNSCLC treated at non-academic facilities show that, in pts who had equivalent access to IO tx, the OS and rwPFS benefits derived from IO were similar, irrespective of race, region or SES. More efforts are required to identify and address barriers of access to care to ensure that all pts receive appropriate cancer tx.

Association Between Material Hardship in Families With Young Children and Federal Relief Program Participation by Race and Ethnicity and Maternal Nativity

Even brief periods of hardship during early childhood may have lifelong consequences. Prior cross-sectional research limited to respondents with English proficiency and internet access during the COVID-19 crisis documented families with young children that struggled to afford basic needs like food and rent. Few studies have examined experiences of families with young children by race and ethnicity and maternal nativity. To examine the association of COVID-19 relief programs with the mitigation of household food insecurity among families with young children, as well as being behind on rent and disparities in program receipt. This cohort study used data from the ongoing repeat cross-sectional Children's HealthWatch study comprising families surveyed at baseline (January 1, 2018, to March 20, 2020) and during the COVID-19 crisis (September 1, 2020, to June 30, 2021). Baseline questionnaires were administered face to face to caregivers of children aged 48 months or younger in English or Spanish in emergency departments or primary care clinics in 5 US cities. The follow-up questionnaires were administered via telephone. Supplemental Nutrition Assistance Program (SNAP) participation or receipt of at least 1 Economic Impact Payment (EIP; ie, stimulus check) during the COVID-19 crisis. The primary outcomes were household food insecurity (assessed via the US Household Food Security Survey Module: Six-Item Short Form) and being behind on rent. Logistic and binomial regression models were used to calculate adjusted odds ratios (aORs) and unadjusted and adjusted prevalence ratios (PRs). Of 1396 caregiver-child dyads (20.3% response rate), race and ethnicity data were available for 1357 caregivers: 514 (37.9%) were Black, non-Latino; 558 (41.1%) were Latino; 230 (16.9%) were White, non-Latino; and 55 (4.1%) were of other non-Latino race or ethnicity. Among 1390 responses with nonmissing data, 417 children (30.0%) had an immigrant mother, and among 1388 responses, 1238 (33.5%) were publicly insured. During the crisis, 467 of 1395 respondents (33.5%) reported household food insecurity, and 567 of 1391 respondents (40.8%) reported being behind on rent. Families with immigrant mothers had lower odds of EIP and SNAP participation than families with US-born mothers (eg, aOR, 0.07 [95% CI, 0.05-0.12] for both EIP and SNAP vs neither), despite being more likely to report household food insecurity (adjusted PR [aPR], 1.48 [95% CI, 1.28-1.71]) and being behind on rent (aPR, 1.14 [95% CI, 1.00-1.30]). Families with Black (unadjusted PR [uPR], 1.40 [95% CI, 1.08-1.82]) or Latino (uPR, 1.54 [95% CI, 1.19-1.98]) caregivers or caregivers of other race and ethnicity (uPR, 1.67 [1.12-2.49]) were also more likely than families with White, non-Latino caregivers to experience household food insecurity or being behind on rent (families with Black caregivers: uPR, 2.02 [95% CI, 1.58-2.58]; families with Latino caregivers: 1.68 [95% CI, 1.30-2.16]; families with caregivers of other race or ethnicity: uPR, 1.94 [95% CI, 1.34-2.80]). Adjustment for covariates and differential participation in relief programs did not entirely account for these disparities. The results of this cohort study suggest that the COVID-19 crisis exacerbated preexisting inequities in food insecurity and difficulty paying rent according to race and ethnicity and maternal nativity and that equity-focused policy changes are needed to ensure that all children and their families in the US can afford basic needs for optimal health.

A marginal modelling approach for predicting wildfire extremes across the contiguous United States

This paper details a methodology proposed for the EVA 2021 conference data challenge. The aim of this challenge was to predict the number and size of wildfires over the contiguous US between 1993 and 2015, with more importance placed on extreme events. In the data set provided, over 14% of both wildfire count and burnt area observations are missing; the objective of the data challenge was to estimate a range of marginal probabilities from the distribution functions of these missing observations. To enable this prediction, we make the assumption that the marginal distribution of a missing observation can be informed using non-missing data from neighbouring locations. In our method, we select spatial neighbourhoods for each missing observation and fit marginal models to non-missing observations in these regions. For the wildfire counts, we assume the compiled data sets follow a zero-inflated negative binomial distribution, while for burnt area values, we model the bulk and tail of each compiled data set using non-parametric and parametric techniques, respectively. Cross validation is used to select tuning parameters, and the resulting predictions are shown to significantly outperform the benchmark method proposed in the challenge outline. We conclude with a discussion of our modelling framework, and evaluate ways in which it could be extended.

Prevalence and predictors of outcomes among ESRD patients with COVID-19

Background End-stage renal disease patients on hemodialysis (ESRD) patients are at high risk for contracting COVID-19. In this propensity matched cohort study, we examined the prevalence of COVID-19 in emergency room (ER) patients and examined whether clinical outcomes varied by ESRD status.MethodsPatients who visited George Washington University Hospital ER from April 2020 to April 2021 were reviewed for COVID-19 and ESRD status. Among COVID-positive ER patients, the propensity for ESRD was calculated using a logistic regression model to create a propensity-matched sample of ESRD vs non-ESRD COVID-19 patients. A multivariable model examined whether ESRD was an independent predictor of death and other outcomes in COVID-19 patients.ResultsAmong the 27,106 ER patients, 2689 of whom were COVID-positive (9.9%). The odds of testing positive for COVID-19 were 0.97 ([95% CI: 0.78–1.20], p = 0.76) in ESRD vs non-ESRD patients after adjusting for age, sex, and race. There were 2414 COVID-positive individuals with non-missing data, of which 98 were ESRD patients. In this COVID-positive sample, ESRD patients experienced a higher incidence of stroke, sepsis, and pneumonia than non-ESRD individuals. Significant independent predictors of death included age, race, sex, insurance status, and diabetes mellitus. Those with no insurance had odds of death that was 212% higher than those with private insurance (3.124 [1.695–5.759], p < 0.001). ESRD status was not an independent predictor of death (1.215 [0.623–2.370], p = 0.57). After propensity-matching in the COVID-positive patients, there were 95 ESRD patients matched with 283 non-ESRD individuals. In this sample, insurance status continued to be an independent predictor of mortality, while ESRD status was not. ESRD patients were more likely to have lactic acidosis (36% vs 15%) and length of hospital stay ≥ 7 days (48% vs 31%), but no increase in odds for any studied adverse outcomes.ConclusionsIn ER patients, ESRD status was not associated with higher odds for testing positive for COVID-19. Among ER patients who were COVID positive, ESRD was not associated with mortality. However, insurance status had a strong and independent association with death among ER patients with COVID-19.

Predicted Health Literacy Disparities Between Immigrant and US-Born Racial/Ethnic Minorities: a Nationwide Study.

Racial/ethnic minorities in the USA exhibit reduced health literacy (HL) proficiency, leading to increased health disparities. It is unclear how the effect of birth status (immigrant/US-born) affects HL proficiency among racial/ethnic minorities. To identify the direct, indirect, and total effects of birth status on HL proficiency among a nationally representative population of racial/ethnic minority adults in the USA. A cross-sectional study of 2019 data from the Medial Expenditure Panel Survey. Participants aged 18 or older reporting as racial/ethnic minorities (Black, Asian, or Hispanic) with non-missing data. We predicted HL proficiency for each participant using a previously published model. Path analysis was used to estimate the direct, indirect, and total effects of birth status on HL proficiency, accounting for several other covariates. Prevalence ratios were estimated using adjusted Poisson regression to evaluate differences in the "Below Basic" HL category. An estimated weighted 81,092,505 participants were included (57.5% US-born, 42.5% immigrant). More racial/ethnic minority immigrant participants fell into the lowest category of HL proficiency, "Below Basic" (14.3% vs 5.5%, p < 0.05). Results of the path analysis indicated a significant, negative direct effect of birth status on HL proficiency (standardized coefficient = - 0.24, SE = 0.01, 95%CI: - 0.26, - 0.23) in addition to an indirect effect mediated through insurance status, health-system resource use, and English proficiency. The total effect of birth status on HL proficiency was found to be - 0.29. The immigrant participant group had 81% higher prevalence of falling into the "Below Basic" HL category compared to US-born participants (prevalence ratio = 1.81, 95%CI: 1.52, 2.16). Immigrant status has a strong, negative, direct effect on HL proficiency among racial/ethnic minorities in the USA. This may be a result of barriers that prevent equitable access to resources that improveproper HL proficiency. US policymakers may consider several methods to reduce this disparity at the health-system-, provider-, and patient-levels.

Abstract WP35: Neighborhood Effects On Knowledge Of Stroke Risk Factors And Stroke Warning Signs

Background: Low socio-economic status has been associated with worse stroke outcomes and longer arrival times to the ED. Objective: To determine whether stroke knowledge is different by neighborhood and if neighborhood socio-economic status (NSES) predicts knowledge of stroke risk factors (RF) or stroke warning signs (WS). Methods: A survey was done in the Greater Cincinnati Northern Kentucky Region in 2016 where participants were targeted to match those of ischemic stroke patients within the same population. The participants were asked open-ended questions regarding stroke WS and RFs. Participants also provided their zip codes. NSES was determined based on zip code and census data. NSES was determined using a previously validated index and divided into quartiles. We used regression to determine the association between NSES and only being able to name 0-1 RF or WS. Results: There were 1902 participants in the surveys with non-missing geographic data who lived in 96 different zip codes. Participants were 66.8±16.4 years old, 56.2% were female, 24.4% were black and 36.0% had less than a HS education. About 20% of participants were unable to name one stroke risk factor or one warning sign. There was substantial heterogeneity of stroke knowledge by zip code (Figure 1). There was no significant association between NSES and knowledge of stroke WS. After adjustment for age, sex, and race, participants who lived in neighborhoods in the lowest quartile of NSES were more likely to be only able to name 0-1 RFs when compared to all other neighborhoods (Lowest Quartile of NSES OR 1.45 (95% CI 1.06-1.98)). Conclusion: We identified geographic disparities of stroke knowledge in a well-defined population, emphasizing the need for targeting public health interventions towards areas most in need. NSES accounted for some of the disparities in stroke RF knowledge, but not stroke WS knowledge. 1

Evaluating Linear Ion Trap for MS3-Based Multiplexed Single-Cell Proteomics.

There is a growing demand to develop high-throughput and high-sensitivity mass spectrometry methods for single-cell proteomics. The commonly used isobaric labeling-based multiplexed single-cell proteomics approach suffers from distorted protein quantification due to co-isolated interfering ions during MS/MS fragmentation, also known as ratio compression. We reasoned that the use of MS3-based quantification could mitigate ratio compression and provide better quantification. However, previous studies indicated reduced proteome coverages in the MS3 method, likely due to long duty cycle time and ion losses during multilevel ion selection and fragmentation. Herein, we described an improved MS acquisition method for MS3-based single-cell proteomics by employing a linear ion trap to measure reporter ions. We demonstrated that linear ion trap can increase the proteome coverages for single-cell-level peptides with even higher gain obtained via the MS3 method. The optimized real-time search MS3 method was further applied to study the immune activation of single macrophages. Among a total of 126 single cells studied, over 1200 and 1000 proteins were quantifiable when at least 50 and 75% nonmissing data were required, respectively. Our evaluation also revealed several limitations of the low-resolution ion trap detector for multiplexed single-cell proteomics and suggested experimental solutions to minimize their impacts on single-cell analysis.

ADJUSTING FOR COUNTRY-LEVEL VARIATION IN DEMENTIA PREVALENCE WITH CLASSIFICATION ALGORITHMS IN SHARE

Abstract Background. Population-level dementia prevalence depends on societal factors and individual-level risk and protective factors. To improve our understanding of how these factors interact, we can use cross-national surveys such as the Survey of Health, Ageing and Retirement in Europe (SHARE). However, in absence of validated cognitive assessments, adjusting for underdiagnosis of dementia is needed. The present study sought to explore the usefulness of the Langa-Weir and alternative algorithms to detect probable dementia while accounting for country-level variation in estimated prevalence and underdiagnosis of dementia.Method. Data from 57,880 respondents aged 60 years and older to wave 7 of SHARE (2017) with non-missing data on variables related to sociodemographics and cognition were used. Adaptations of the Langa-Weir classification algorithm were compared to a weighted logistic regression model and an XGBoost classifier applying the synthetic minority oversampling technique. Different specifications of algorithms were tested globally and for individual countries and compared with the World Alzheimer’s Report (2015)’s country-level projections of dementia prevalence for 2018. Results. All algorithms accurately classified self-reported diagnosis of dementia (accuracy = 0.90-0.96), with the Langa-Weir classification based on recall and a cutoff reflecting country-specific prevalence outperforming other algorithms regarding compensation of underdiagnosis. Algorithmically detected probable dementia is associated with newly self-reported dementia diagnosis, drop-out and death two years later. Discussion. Identifying probable dementia through classification algorithms can increase statistical power and improve validity in cross-national investigations. Further research is needed to replicate the findings in validated cognitive assessments and to identify causes of cross-national variation in dementia underdiagnosis.

441. Association between Incident HIV-Associated Wasting/Low Weight and All-Cause Mortality in the OPERA® Cohort

Abstract Background HIV-associated wasting (i.e., progressive, involuntary weight loss with both fat and lean tissue loss; HIVAW) is an under-appreciated AIDS-defining illness; the 2012-2018 period prevalence was reported as 18% in a recent claims study in the United States. We aimed to assess the association between incident HIVAW/low weight and all-cause mortality in the era of modern combination antiretroviral therapy (ART). Methods In the Observational Pharmaco-Epidemiology Research &amp; Analysis (OPERA®) cohort, PWH without (a) any prior HIVAW/low weight, (b) malignancy within 3 years, and (c) opportunistic infection within 1 year who were active in care between 2016 and 2020 were followed through death, loss to follow-up, or study end (31OCT2021). HIVAW/low weight included a wasting or low BMI/underweight diagnosis (ICD codes, title search) or BMI &amp;lt; 20 kg/m2. Hazard ratios (HR) and 95% confidence intervals (CI) for the association between time-dependent incident HIVAW/low weight (exposure) and all-cause mortality (outcome) were estimated with extended Cox regression models. The adjusted model included age at baseline, race, ethnicity, and time-dependent covariates (log10 viral load, Veterans Aging Cohort Study [VACS] Mortality Index score). Viral load and VACS score were included as surrogate markers for ART use and comorbidities, respectively. Linear and quadratic terms of continuous variables were included. Results Of 67,119 PWH without prior HIVAW/low weight in OPERA®, 62,314 (93%) PWH had non-missing covariate data and were included in the models; baseline characteristics did not differ between the full and model study populations (Table 1). Over a median follow-up of 45 months (interquartile range: 27, 65), there were 4,755 (8%) cases of incident HIVAW/low weight and 1,354 (2%) deaths. In the adjusted model, PWH who experienced incident HIVAW/low weight had a significantly increased risk of death over follow-up than those who did not experience HIVAW/low weight (HR: 1.96; 95% CI: 1.68, 2.27) (Table 2). Conclusion In this analysis of 62,314 PWH in care, incident HIVAW/low weight was associated with twice the risk for all-cause mortality in the modern ART era. Particular attention needs to be paid to HIVAW/low weight among PWH to restore health and potentially reduce the risk of death. Disclosures Rachel P. Weber, PhD, AIDS Healthcare Foundation: Client of my employer|EMD Serono: Client of my employer|Gilead Sciences: Client of my employer|Janssen: Client of my employer|Merck &amp; Co.: Client of my employer|TheraTechnologies: Client of my employer|ViiV Healthcare: Client of my employer Laurence Brunet, PhD, AIDS Healthcare Foundation: Client of my employer|EMD Serono: Client of my employer|Gilead Sciences: Client of my employer|Janssen: Client of my employer|Merck &amp; Co: Client of my employer|TheraTechnologies: Client of my employer|ViiV Healthcare: Client of my employer Javeed Siddiqui, MD, MPH, Abbvie: Advisor/Consultant|Abbvie: Honoraria|Cumberland pharmaceuticals: Advisor/Consultant|Cumberland pharmaceuticals: Honoraria|EMD serono: Advisor/Consultant|EMD serono: Honoraria|Merck: Advisor/Consultant|Merck: Honoraria Javeed Siddiqui, MD, MPH, Abbvie: Advisor/Consultant|Abbvie: Honoraria|Cumberland pharmaceuticals: Advisor/Consultant|Cumberland pharmaceuticals: Honoraria|EMD serono: Advisor/Consultant|EMD serono: Honoraria|Merck: Advisor/Consultant|Merck: Honoraria Michael Harbour, MD, MPH, FACP, EMD Serono: Employee Amy L. Phillips, PharmD, EMD Serono, Inc.: Employment Brooke Hayward, SM, MBA, EMD Serono, Inc.: Employee Jennifer S. Fusco, BS, AIDS Healthcare Foundation: Client of my employer|EMD Serono: Client of my employer|Gilead Sciences: Client of my employer|Janssen: Client of my employer|Merck &amp; Co.: Client of my employer|TheraTechnologies: Client of my employer|ViiV Healthcare: Client of my employer Ricky K. Hsu, MD, Gilead: Honoraria|Merck: Honoraria|ViiV: Advisor/Consultant|ViiV: Grant/Research Support|ViiV: Honoraria Gregory P. Fusco, MD, MPH, AIDS Healthcare Foundation: Client of employer|EMD: Grant/Research Support|Gilead Sciences: Client of employer|Janssen: Client of employer|Merck &amp; Co.: Client of employer|Theratechnologies: Client of employer|ViiV Healthcare: Client of employer.

Identification of Social and Racial Disparities in Risk of HIV Infection in Florida using Causal AI Methods.

Florida -the 3rd most populous state in the USA-has the highest rates of Human Immunodeficiency Virus (HIV) infections and of unfavorable HIV outcomes, with marked social and racial disparities. In this work, we leveraged large-scale, real-world data, i.e., statewide surveillance records and publicly available data resources encoding social determinants of health (SDoH), to identify social and racial disparities contributing to individuals' risk of HIV infection. We used the Florida Department of Health's Syndromic Tracking and Reporting System (STARS) database (including 100,000+ individuals screened for HIV infection and their partners), and a novel algorithmic fairness assessment method -the Fairness-Aware Causal paThs decompoSition (FACTS)- merging causal inference and artificial intelligence. FACTS deconstructs disparities based on SDoH and individuals' characteristics, and can discover novel mechanisms of inequity, quantifying to what extent they could be reduced by interventions. We paired the deidentified demographic information (age, gender, drug use) of 44,350 individuals in STARS -with non-missing data on interview year, county of residence, and infection status- to eight SDoH, including access to healthcare facilities, % uninsured, median household income, and violent crime rate. Using an expert-reviewed causal graph, we found that the risk of HIV infection for African Americans was higher than for non- African Americans (both in terms of direct and total effect), although a null effect could not be ruled out. FACTS identified several paths leading to racial disparity in HIV risk, including multiple SDoH: education, income, violent crime, drinking, smoking, and rurality.

HIV Diagnoses Through Partner Services in the United States in 2019 and Opportunities for Improvement.

HIV partner services in 2019 resulted in 1214 diagnoses (3.5% of diagnoses in the United States) with 22.7% new positivity among partners without prior diagnosis. Areas for improvement are identified. Background HIV partner services (HIV PS) is an effective strategy for diagnosing HIV infection. Sex/needle-sharing partners of individuals diagnosed with HIV are notified about potential exposure and offered HIV testing and other services. We assessed the HIV PS contribution to HIV diagnoses in the United States and assessed priority areas for improvements. Methods National HIV Monitoring and Evaluation Partner Services and case surveillance data reported to the Centers for Disease Control and Prevention for 2019 were used for this analysis. The percentage of all new diagnoses that HIV PS programs reported is described nationally and by state. Linkage to HIV medical care among newly diagnosed partners is described. Potential increases in diagnosing HIV infection are assessed by HIV PS step to identify priority areas for improvement. Results HIV PS contributed 1214 of 35,164 (3.5%) of all diagnoses nationally in 2019, and contributions ranged from 0% to 31.8% by state. Of partners tested with nonmissing data, 22.7% were newly diagnosed. An estimated 1692 new partner diagnoses were lost during HIV PS steps. Steps resulting in the highest losses included index patients not being interviewed, partners not being tested for HIV, and index patients not being located. Seventy-two percent of partners newly diagnosed with HIV were linked to HIV medical care. Conclusions HIV PS is an effective strategy for diagnosing HIV, and a high percent of sex/needle-sharing partners was newly diagnosed with HIV. Expanded HIV PS in some states and targeted improvements in HIV PS steps can enhance the contribution of HIV PS toward achieving national goals.

Sensitivity Analysis on the Primary Efficacy Results of SB12 (Eculizumab Biosimilar) Phase III Study in Paroxysmal Nocturnal Hemoglobinuria Patients

Background SB12 is a humanized monoclonal antibody that has developed as a proposed biosimilar to reference eculizumab (ECU) and its pharmacokinetics (PK) similarity compared with ECU in healthy subjects has demonstrated. Also, SB12 Phase III study has demonstrated equivalent clinical efficacy by evaluating lactate dehydrogenase (LDH) and comparable safety, PK, pharmacodynamics (PD), and immunogenicity between SB12 and ECU in paroxysmal nocturnal hemoglobinuria (PNH) patients (Jun Ho Jang et al., EHA 2022). Objective To explore the robustness of the primary efficacy results of SB12 Phase III clinical study (NCT04058158) by conducting sensitivity analysis. Methods This was a Phase III randomized, double-blind, multi-national, cross-over study in patients with PNH. A total of 50 patients who aged ≥ 18 years with a confirmed diagnosis of PNH and ≥ 1.5 upper limit of normal range (ULN) of LDH without previous exposure to a complement inhibitor were included. Patients were randomized in 1:1 ratio to either treatment sequences and received either SB12 or ECU by IV infusion every 7 days for the first 4 weeks (initial phase) and 900 mg for the fifth week, followed by 900 mg every 14 days (maintenance phase). The treatment was switched to ECU or SB12 at Week 26, and switched treatment was provided until Week 50. The primary endpoints of this study were LDH level at Week 26 (equivalence declared if the two-sided 95% confidence interval (CI) of the least squares means (LSMeans) difference in LDH level at Week 26 between SB12 and ECU lied within the pre-defined equivalence margin of [−1.2 × ULN, 1.2 × ULN] = [−337.2, 337.2], where ULN = 281 U/L) and time-adjusted area of under the effect curve (AUEC) of LDH from Week 14 to Week 26 and Week 40 to Week 52 (equivalence declared if the two-sided 90% CI of the ratio of geometric LSMeans in time-adjusted AUEC of LDH between SB12 and ECU lied within the pre-defined equivalence margin of [0.77, 1.29]). Sensitivity analysis was performed using complete case analysis or multiple imputation method in modified full analysis set (M-FAS). Complete case analysis is based on complete LDH data, without imputation on missing LDH data. With regard to multiple imputation that assumes missing at random (MAR), missing LDH data was imputed for patients who dropped out of the study prior to the primary analysis timepoint. Missing LDH data was imputed by fully conditional specification (FCS), with the non-missing LDH data at previous visits, treatment sequence, and gender as covariates. In case of sensitivity analysis of LDH level at Week 26, results of linear model based on loge-transformed LDH values were combined first using Rubin's rule and then delta method was used later to report mean difference and 95% CI in original scale. Results In the sensitivity analysis of LDH level at Week 26, for the complete case analysis in M-FAS, estimated LSMeans difference (SB12 − ECU) of LDH level at Week 26 between SB12 and ECU was 34.48 (95% CI [−47.66, 116.62]). In the multiple imputation analysis, estimated LSMeans difference (SB12 − ECU) of LDH level at Week 26 between SB12 and ECU was 26.91 (95% CI [−56.24, 110.05]). Sensitivity analysis results of LDH level at Week 26 provided consistent result with primary analysis (SB12 − ECU: estimated LSMeans difference 34.48, 95% CI [−47.66, 116.62]). Meanwhile, sensitivity analysis of time-adjusted AUEC of LDH from Week 14 to Week 26 and from Week 40 to Week 52, estimated ratio of geometric LSMeans in time-adjusted AUEC of LDH (SB12/ECU) between SB12 and ECU was 1.10 (90% CI [0.99, 1.21]) for the complete case analysis in M-FAS. In the multiple imputation analysis that assumed missing data were MAR, estimated ratio of geometric LSMeans in time-adjusted AUEC of LDH (SB12/ECU) between SB12 and ECU was 1.07 (90% CI [0.93, 1.22]). Sensitivity analysis results of time-adjusted AUEC of LDH from Week 14 to Week 26 and from Week 40 to Week 52 showed consistent result with primary analysis (SB12/ECU: estimated ratio of geometric LSMeans 1.08, 90% CI [0.95, 1.23]). Conclusion This sensitivity analysis results showed consistent results and validated robustness of the primary analysis and it supports the therapeutic equivalence between SB12 and ECU. Figure 1View largeDownload PPTFigure 1View largeDownload PPT Close modal

LBMON178 2nd Year Outcomes Of The Open-label Extension Of Chiasma Optimal, A Phase 3 Study Of Oral Octreotide Capsules In Acromegaly

Abstract Background Oral octreotide capsules (OOC) are an approved treatment option in the United States for patients with acromegaly who have previously responded to injectable somatostatin receptor ligands (iSRLs). Safety and efficacy of OOC was demonstrated in the CHIASMA OPTIMAL trial (NCT03252353), showing maintenance of response and a safety profile consistent with iSRLs. Results from the first 48 weeks of the open-label extension (OLE) of this trial suggested that this effect was durable for the long-term. Objective Report efficacy and safety through the second year of the OLE. Methods Eligible patients had the option to enroll in the OLE of CHIASMA OPTIMAL following the double-blind placebo-controlled (DPC) period. Endpoints in the second year of the OLE were exploratory, including: the proportion of patients completing week 96 of the OLE, the proportion of responders entering year 2 who remained responders of those with evaluable data (response defined as average Insulin-like growth factor I [IGF-I] ≤1. 0 × upper limit of normal [ULN] at last 2 visits between weeks 84-96), and changes in IGF-I and growth hormone (GH) from OLE baseline to OLE week 96 in those completing the DPC period on study drug. Results Thirty-two patients from the core period (DPC groups: n=14, placebo; n=18, OOC) entered year 2 of the OLE and 31 (97%) completed. OOC dose at the start of year 2 was 40 mg/d, n=3; 60 mg/d, n=8; and 80 mg/d, n=21. For patients who had non-missing data response rate at 96 weeks was 100% (17/17) for patients who entered year 2 as responders, and 93% (27/29) for patients overall. Mean IGF-I for patients from the OOC group who completed the DPC period on study drug (n=14) was 0.81 × ULN at baseline of OLE and 0.78 × ULN at week 96, (change, −0. 03 × ULN). Mean GH (n=12) was 0.43 ng/mL and 0.44 ng/mL at baseline and week 96, respectively (change, 0. 01 ng/mL). For patients who completed the DPC period on placebo (n=8), mean IGF-I was 1. 05 and 0.77 × ULN at OLE baseline and week 96, respectively (change, −0.28); mean GH (n=5) was 1. 06 and 0.34 ng/mL at OLE baseline and week 96, respectively (change, −0.72 ng/mL). Median exposure to OOC was 2.1 years, with exposure &amp;gt;3 years for 5 patients. Throughout the OLE period, 50% of patients experienced adverse events (AEs), with no serious AEs. One patient discontinued treatment due to an AE (headache). The safety profile was consistent with that reported for OOC; no new patterns were observed with increased exposure. Conclusion Maintenance of biochemical response with OOC was durable up to 96 weeks. The safety profile of OOC was consistent with that of iSRLs. No new safety signals were observed with increased exposure duration. Presentation: Monday, June 13, 2022 12:30 p.m. - 2:30 p.m.