Multivariable Generalized Estimating Equation Model Research Articles

Mortality risk factors for crush syndrome after an earthquake in Türkiye: Do systemic inflammatory parameters play any role?

The aim of our study is to assess the prognostic impact of the neutrophil-to-lymphocyte ratio (NLR), lymphocyte-to-monocyte ratio (LMR), and platelet-to-lymphocyte ratio (PLR) on 28-day mortality in patients admitted to the intensive care unit with crush syndrome following the Kahramanmaraş earthquake in Türkiye. A total of 63 adult patients with crush syndrome admitted to the intensive care unit after the earthquake were enrolled in this study. The medical records of the patients were examined using follow-up forms and the hospital data system. The mean age of the patients was 38.9±17.3 years, and the median time under debris was 31.5 hours. The 28-day mortality rate was 27%. In univariate generalized estimating equations (GEE) and other analyses, variables that are significant (or candidate variables) between 28-day mortality groups included age as a biological factor. These variables were included in the multivariate GEE model. The effects of continuous renal replacement therapy (CRRT), serum sodium concentration, Sequential Organ Failure Assessment (SOFA) score, and PLR on mortality were statistically significant. Elevated SOFA scores, the necessity for CRRT, increased serum sodium levels, and decreased PLR values are associated with increased 28-day mortality in patients with crush syndrome after an earthquake.

Higher CD4+ T Cell Counts after CAR-T in Large B-Cell Lymphomas Are Significantly Associated with Better Overall Survival

Introduction Chimeric antigen receptor T-cell (CAR-T) therapy has revolutionized the treatment of large B-cell lymphomas (LBCL), demonstrating remarkable efficacy. However, CAR-T therapy is also associated with significant toxicities, some arising from relative immunologic incompetence after treatment. A key knowledge gap persists regarding how the immune landscape changes post-CAR-T, defined as immune reconstitution (IR), and the influence of IR on patient outcomes. In this study, we set out to describe patterns and determinants of IR following CAR-T as well as impacts of IR on outcomes. Methods Adult patients with primary or transformed LBCL treated with CD19-CAR-T cells, namely axicabtagene ciloleucel (axi-cel), tisagenlecleucel (tisa-cel), and lisocabtagene maraleucel (liso-cel), between 2017 and 2022 were included in this single center retrospective study. Immune cell levels were measured at various timepoints after CAR-T. Median cell counts and interquartile range (IQR) at pre-specified time windows (using the first available observation) were used to describe temporal IR patterns. Multivariable generalized estimating equation (GEE) models with cubic time variables were used to investigate the association between cell count trajectory and baseline characteristics. A multivariable Cox model with longitudinal cell count as a time-varying covariate was utilized to assess the association between IR and overall survival (OS), adjusted for confounding factors. Kaplan-Meier (KM) curves were generated for visualization. Results A total of 210 patients were included, of whom 153 had at least one CD4+ T cell measurement from time of CAR-T. The median age was 65 (range 20-86) and most were male (65.7%) and diagnosed with diffuse large B-cell lymphoma (79.5%). Axi-cel (49.5%) was the most common product, followed by tisa-cel (28.6%) and liso-cel (21.9%). Lymphodepletion was mostly cyclophosphamide and fludarabine (82.4%). Of the 210 patients, 58.1% achieved a complete response (CR) and 17.1% a partial response by 100 days after CAR-T. Following CAR-T, CD3+ T cells were the most abundant circulating lymphocyte. By 30 days (±15 days) after CAR-T, median (IQR) values of CD3+ T cells, B cells, and NK cells were 445 (162-776), 0 (0-0), and 90 cells/ul (57-146), respectively; by 60 days (±15 days), median (IQR) values were 498 (220-706), 0 (0-0), and 104 cells/ul (69-169), respectively. Proportions of CD4+ and CD8+ T cells remained similar through one year after CAR-T, with CCR7-45RA+ TEMRA and CCR7-45RA- effector memory cells dominating among CD8's, and CCR7-45RA- effector memory cells being the most abundant among CD4's. Regulatory T cells (CD25+CD127-) comprised about 10% of the CD4+ T cell lymphocyte population, with a median (IQR) of 9 cells/ul (4.5-15.5) by day 30 (±15 days) after CAR-T. Finally, receipt of tisa-cel was associated with greater levels of CD4+ EM T cells compared with axi-cel and liso-cel (median [IQR]: 253 [158-406] vs 81 [51-138] and 113 cells/ul [88-154], respectively), by 30 days (±15 days) after CAR-T. According to a multivariable GEE model, patients receiving tisa-cel had significantly higher CD4+ count trajectories compared to those receiving axi-cel or liso-cel ( p<0.001 for both) ( Figure A). In addition, patients with more extensive pre-treatment (i.e., 6+ prior lines) had significantly lower CD4+ count trajectories compared to those who received 2-3 or 4-5 prior lines ( p=0.015 and p=0.004, respectively). A multivariable Cox model, moreover, indicated that a higher CD4+ count (in units of 50 cells/ul) was significantly associated with better OS (adjusted HR: 0.845; 95% CI, 0.745-0.960; p=0.010), adjusted for product and prior lines of therapy. KM curves by patients above or below the median of the last observed CD4+ count further demonstrated this significant association between longitudinal IR and OS ( Figure B). Conclusion In this single-center retrospective study, we extensively characterized the immune environment at time of and following CAR-T. To the best of our knowledge, this is the largest study of its kind. Additionally, we identified several factors that impact IR, which in turn shapes OS. In doing so, we show that CD4+ T lymphocytes are key players in disease response post-CAR-T and suggest that tracking CD4+ recovery following CAR-T may provide both diagnostic and therapeutic benefit.

A cross-sectional study for prevalence and risk factors of peri-implant marginal bone loss

Progressive peri-implant marginal bone loss and peri-implantitis have become a growing problem, but cross-sectional studies on their prevalence and risk factors are sparse. The purpose of this cross-sectional clinical study was to investigate the prevalence of peri-implant marginal bone loss (MBL) and to identify systemic and local risk factors. All adult patients who had received dental implants at the National Taiwan University Hospital (NTUH) during 2009 or 2010 were included. Their medical records were collected from the NTUH-integrative Medical Database. Consecutive follow-up radiographs were accessed for severity of MBL. The influence of each factor on MBL was estimated by using generalized estimating equations (GEEs). A total of 732 participants with 1873 implants were analyzed (mean follow-up: 5.30 years). The prevalence of MBL was 59.15% at the individual level and 49.55% at the implant level. The risk indicators identified for the presence of MBL were follow-up period of more than 2 years, diagnosis of diabetes within 12 months, radiation therapy (2 years after implant placement), implant location at maxillary canine (compared with mandibular molar), and implants from the Nobel Biocare brands (Brånemark System and NobelActive). A second multivariate GEE model confirmed the association of progressive MBL with implant location at the maxillary canine and mandibular incisor and implant brand or design. The identified risk indicators for MBL were longer follow-up period, diagnosis of diabetes, radiation therapy, implant location at maxillary canine, and implant brand or design.

Association of Medicaid expansion with 2-year survival and time to treatment initiation in gastrointestinal cancer patients: A National Cancer Database study.

We evaluated whether Medicaid expansion (ME) was associated with improved 2-year survival and time to treatment initiation (TTI) among patients with gastrointestinal (GI) cancer. GI cancer patients diagnosed 40-64 years were queried from the National Cancer Database. Those diagnosed from 2010 to 2012 were considered pre-expansion; those diagnosed from 2014 to 2016 were considered post-expansion. Cox models estimated hazard ratios and 95% confidence intervals (CIs) for 2-year overall survival. Generalized estimating equations (GEE) estimated odds ratios (OR) and 95% CI of TTI within 30- and 90 days. Multivariable Difference-in-Difference models were used to compare expansion/nonexpansion cohorts pre-/post-expansion, adjusting for patient, clinical, and hospital factors. 377,063 patients were included. No significant difference in 2-year survival was demonstrated across ME and non-ME states overall or in site-based subgroup analysis. In stage-based subgroup analysis, 2-year survival significantly improved among stage II cancer, with an 8% decreased hazard of death at 2 years (0.92; 0.87-0.97). Those with stage IV had a 4% increased hazard of death at 2 years (1.04; 1.01-1.07). Multivariable GEE models showed increased TTI within 30 days (1.12; 1.09-1.16) and 90 days (1.22; 1.17-1.27). Site-based subgroup analyses indicated increased likelihood of TTI within 30and 90 days among colon, liver, pancreas, rectum, and stomach cancers, by 30 days for small intestinal cancer, and by 90 days for esophageal cancer. In subgroup analyses, all stages experienced improved odds of TTI within 30 and 90 days. ME was not associated with significant improvement in 2-year survival for those with GI cancer. Although TTI increased after ME for both cohorts, the 30- and 90-day odds of TTI was higher for those from ME compared with non-ME states. Our findings add to growing evidence of associations with ME for those diagnosed with GI cancer.

Hurricanes Irma and Maria and Diabetes incidence in Puerto Rico

ObjectiveTo evaluate the impact of Hurricanes Irma/Maria on diabetes incidence in Puerto Rico. Mortality increased substantially after the hurricanes, but morbidity was not assessed.MethodsWe recruited 364 participants from the San Juan Overweight Adults Longitudinal Study (SOALS) aged 40–65 years who completed a three-year follow-up and were free of diabetes. We conducted additional questionnaires 1.7–2.5 years after hurricanes. Glycosylated hemoglobin (HbA1c), fasting glucose and insulin were assessed at all three visits. We compared diabetes incidence between pre-hurricane visits and between visits spanning the hurricanes using Generalized Estimating Equation (GEE) adjusting for within person repeated measures, age, and body mass index (BMI).ResultsDiabetes incidence was significantly higher spanning the hurricanes than pre-hurricane (multivariate GEE model: IRR = 2.1; 95% CI: 1.4–3.1). There was a significantly higher increase spanning the hurricanes compared to pre-hurricanes for Homeostatic Model Assessment of Insulin Resistance (HOMA-IR) (median: 0.3 uIU/mL vs. 0.2 uIU/mL). HbA1c levels increased by 0.4% spanning the hurricanes.ConclusionIncreases in diabetes incidence, HOMA-IR and HbA1c were higher spanning the hurricanes compared to the pre-hurricanes period. The increase in diabetes incidence remains significant after adjusting for age and BMI.

A multi-subgroup predictive model based on clinical parameters and laboratory biomarkers to predict in-hospital outcomes of plasma exchange-centered artificial liver treatment in patients with hepatitis B virus-related acute-on-chronic liver failure.

Postoperative risk stratification is challenging in patients with hepatitis B virus-related acute-on-chronic liver failure (HBV-ACLF) who undergo artificial liver treatment. This study characterizes patients' clinical parameters and laboratory biomarkers with different in-hospital outcomes. The purpose was to establish a multi-subgroup combined predictive model and analyze its predictive capability. We enrolled HBV-ACLF patients who received plasma exchange (PE)-centered artificial liver support system (ALSS) therapy from May 6, 2017, to April 6, 2022. There were 110 patients who died (the death group) and 110 propensity score-matched patients who achieved satisfactory outcomes (the survivor group). We compared baseline, before ALSS, after ALSS, and change ratios of laboratory biomarkers. Outcome prediction models were established by generalized estimating equations (GEE). The discrimination was assessed using receiver operating characteristic analyses. Calibration plots compared the mean predicted probability and the mean observed outcome. We built a multi-subgroup predictive model (at admission; before ALSS; after ALSS; change ratio) to predict in-hospital outcomes of HBV-ACLF patients who received PE-centered ALSS. There were 110 patients with 363 ALSS sessions who survived and 110 who did not, and 363 ALSS sessions were analyzed. The univariate GEE models revealed that several parameters were independent risk factors. Clinical parameters and laboratory biomarkers were entered into the multivariate GEE model. The discriminative power of the multivariate GEE models was excellent, and calibration showed better agreement between the predicted and observed probabilities than the univariate models. The multi-subgroup combined predictive model generated accurate prognostic information for patients undergoing HBV-ACLF patients who received PE-centered ALSS.

Longitudinal impact of stressful life events on HIV-related risk and psychosocial problems among MSM in Chennai and Mumbai, India.

Studies show that stressful life events (SLE) (e.g., discrimination, financial problems) can lead to psychosocial problems and exacerbate condomless anal sex (CAS) without protection via pre-exposure prophylaxis (PrEP) among men who have sex with men. However, few studies have examined this relationship among men who have sex with men in India, and none have examined this longitudinally. As a part of an HIV-prevention intervention, 608 MSM from Chennai and Mumbai, India, completed behavioral surveys at baseline, 4, 8, and 12months. We used longitudinal generalized estimating equations (GEE) modeling to examine the relationship between SLE and its severity and subsequent psychosocial problems, CAS, and history of diagnosed sexually transmitted infection (STI). All models are adjusted for age, sexual identity, intervention arm, human immunodeficiency virus status, and recruitment city. The number of SLE and their corresponding perceived impact score remained consistent at each time point. In multivariable GEE models, the number of SLE was predictive of CAS, depression, and harmful drinking. Similarly, the ratio of the impact of SLE was predictive of CAS, depression, and diagnosed STI. However, harmful drinking was not predictive in this model. These findings provide evidence that can inform future interventions, which can be used to enhance self-acceptance, coping skills, and other forms of resiliency.

Abstract A109: Colorectal cancer screen-up-to-date: 10-year prevalence in a regional safety-net health system

Abstract Introduction: Colorectal cancer (CRC) screening is an effective secondary prevention method with an increased probability of diagnosing CRC at an earlier stage, and a consequent improvement in survival post-treatment. This is especially true for individuals who undergo guideline recommended screening at appropriate intervals. Studies have reported a consistent rise in long-term trends of guideline-adherent screen-up-to-date (SUTD) rates among predominantly White and insured individuals. Here we use longitudinal data from 2011-2020 and report 10-year prevalence and correlates of CRC SUTD among patients in a safety-net health system. Methods: All patients aged 50-74 years who had a primary care encounter in any of the 12 community clinics in a large county safety-net health system were included. An individual was considered to be SUTD if he/she had a stool test during the calendar year, flexible sigmoidoscopy in the past 5 years, or colonoscopy in the past 10 years. To obtain a population health estimate (and not only examine frequencies), we included in the denominator patients with a primary care visit in the past 3 years. Multivariable generalized estimating equations (GEE) model was used to examine the association of SUTD status with time-varying demographic and clinical characteristics over the 10-year period. Results: Our analytical cohort had 50,647 patients in 2011, of which 40.9% (20,708) patients were SUTD. Annual rates of SUTD were largely unchanged until 2019, when the prevalence increased to 46.8% after initiation of a population health outreach mailed FIT program. The SUTD rate fell to the baseline level of 40.8% in 2020 after the pandemic-induced suspension of the mailed FIT program. Multivariable GEE model demonstrated that older patients, females, and Hispanics had higher odds of being SUTD compared to younger patients, males, and non-Hispanics, respectively. Additionally, patients who had prior interaction with the healthcare system (had prior stool tests or prior primary care encounters) had higher odds of being SUTD than those with no prior experience with the healthcare system (no prior stool tests or no prior primary care encounters). Conclusion: This study establishes contemporary evidence about the 10-year prevalence and correlates of CRC SUTD status among patients in a safety-net health system. Prevalence remained constant for most of the decade, except in 2019, when a population-based mailed FIT outreach program was implemented to complement usual visit-based screening. Despite the disruptions caused by the COVID-19 pandemic, screening rates in 2020 did not drop below pre-2019 levels (~40%), though the prior increases due to the mailed FIT program were lost. We believe that effective implementation of broad population-based, screening outreach efforts are instrumental in improving and sustaining CRC SUTD rates in safety-net health systems, and can consequently help to decrease CRC incidence and related mortality. Citation Format: Rasmi G. Nair, Ellen Hu, Lei Wang, Eric Kim, Cynthia Ortiz, Jacquelyn Lykken, Celette S. Skinner, Ethan Halm. Colorectal cancer screen-up-to-date: 10-year prevalence in a regional safety-net health system [abstract]. In: Proceedings of the 15th AACR Conference on the Science of Cancer Health Disparities in Racial/Ethnic Minorities and the Medically Underserved; 2022 Sep 16-19; Philadelphia, PA. Philadelphia (PA): AACR; Cancer Epidemiol Biomarkers Prev 2022;31(1 Suppl):Abstract nr A109.

Is there an optimal window of time for transferring single frozen-thawed euploid blastocysts? A cohort study of 1170 embryo transfers.

Is there an optimal window of time when the transfer of single frozen-thawed euploid blastocysts is associated with a maximal live birth rate (LBR)? Performing a single frozen-thawed euploid blastocyst transfer at 160 ± 4 h post-hCG trigger in modified-natural frozen-thawed embryo transfer (FET) cycles was independently associated with a higher LBR as compared to transfers outside this window; however, in natural FET cycles, LBRs were comparable across a wider range of time intervals. There is compelling evidence for maintaining embryo-endometrial synchrony to optimize clinical outcomes following FETs, which could potentially be achieved by matching the transfer time of an embryo post-ovulation to its developmental age post-oocyte retrieval. For modified-natural cycles, ovulation is widely accepted to occur ∼40 h following the hCG trigger, whilst ovulation following spontaneous LH surge onset is thought to vary from 24 to 56 h. This is a multicentered retrospective cohort study analyzing 1170 single frozen-thawed euploid blastocyst transfers following trophectoderm biopsy and preimplantation genetic testing (PGT) between May 2015 and February 2019. Limiting the analysis to single euploid embryo transfers allowed for a more accurate estimation of the endometrial synchrony factor by controlling for the developmental stage of the embryo (full blastocyst or more advanced) and its genetic composition. LBR per FET was the primary outcome measure. Patients underwent natural or gonadotrophin-induced preparation of the endometrium, with serial serum oestradiol, LH and progesterone measurements. Optimally timed transfers were predefined as those conducted 120 ± 4 h post-ovulation since biopsy and subsequent cryopreservation of full blastocysts which is usually performed at 116-124 h post-oocyte retrieval. This was considered the equivalent of 160 ± 4 h post-hCG trigger in modified-natural cycles (n = 253), as ovulation was assumed to occur ∼40 h after the hCG trigger. For natural cycles (n = 917), this was also considered to be, on average, 160 ± 4 h post the spontaneous LH surge. Thus, study groups were determined as those with optimal timing or not, and additional exploratory and subgroup analyses were performed, varying the time window in terms of onset and width, both overall and per endometrial preparation protocol. Statistical analysis was performed using the generalized estimating equations (GEE) framework to control for the clustered nature of the data while adjusting for potential confounders. Overall, LBRs were significantly higher when the transfer had been performed at 160 ± 4 h post-hCG trigger or LH surge onset compared to when it had been performed outside this window (44.7% vs 36.0%; P = 0.008). A multivariable regression GEE model including the cycle type (natural versus modified-natural), previtrification embryo quality (top versus good quality), embryo stage (fully hatched versus hatching or earlier blastocyst), vitrification day (D5 versus D6) and survival rate (>90% versus <90%) as covariates, confirmed that, overall, embryo transfers conducted 160 ± 4 h post-hCG trigger or LH surge onset (the assumed equivalent of 120 ± 4 h post-ovulation) were associated with a significantly higher LBR (relative risk (RR) 1.21, 95% CI 1.04-1.41). Subgroup exploratory analyses per endometrial preparation protocol demonstrated that these findings were primarily present in the modified-natural cycle group (RR 1.52, 95% CI 1.15-1.99), whilst the natural cycle group showed comparable LBRs across a wider range of time intervals. Moreover, the overall LBR for the natural group (36.8%; 95% CI 33.7-39.9%) was lower than that of the modified-natural group (41.3%; 95% CI 35.4-47.1%), suggesting that there likely remains a greater potential to further optimize the timing of natural cycle embryo transfers. As with all retrospective studies, the presence of residual unknown bias cannot be excluded. Additionally, patients included in this study were a selected group who underwent PGT for specific reasons and hence the results obtained might not be directly applicable to the general population or embryos that have not undergone embryo biopsy. Furthermore, the criteria utilized to interpret hormonal data from natural cycles were specifically adopted for the present study and need to be validated in further studies. The results of this study highlight the significance of embryo-endometrial synchrony for the optimization of frozen embryo transfer outcome. However, it also clearly supports that the implantation window is in most cases wide and the achievement of live birth is possible with relatively high success rates even outside the optimal window of 160 ± 4 h post-trigger for modified-natural cycles and across a range of time intervals for natural cycles. Additionally, this study suggests that implantation rates could be further optimized in natural cycles by improving methods of assessing embryo-endometrial synchrony. C. V. is supported by a National Health and Medical Research Council Early Career Fellowship (GNT1147154). No other funding was received for this study and there are no competing interests. N/A.

Economic-related inequalities in hepatitis B virus infection among 115.8 million pregnant women in China from 2013 to 2020.

SummaryBackgroundHepatitis B virus (HBV) infection is a serious global health problem and China has the largest disease burden. Literatures focusing on economic-related inequalities in HBV infection among pregnant women are scarce. We aimed to quantify the economic-related inequalities and the change over time in HBV infection among pregnant women in mainland China from 2013 to 2020 to inform strategies considering economic-related inequalities.MethodsWe used national cross-sectional secondary data of pregnant women in 30 provinces from the National Integrated Prevention of Mother-to-Child Transmission of HIV, Syphilis and Hepatitis B Programme (iPMTCT Programme) from 2013 to 2020. We calculated concentration index and adjusted difference between the rich and the poor in the multivariable generalized estimating equation (GEE) model to measure economic-related inequality, after adjusted other risk factors.FindingsIn this study, a total of 115,789,148 pregnant women of mainland China from 2013 to 2020 were included, the overall hepatitis B surface antigen (HBsAg) prevalence was 6.27% (95%CI: 6.26%-6.28%). The curve lay above the equality line, with the negative value of the concentration index of -0.027, which indicated that economic-related health disparities exist in the distribution of HBV infection and the inequality disadvantageous to the poor (pro-poor). The concentration index showed a trend of fluctuating decline, indicating that economic-related inequalities in HBsAg prevalence were narrowing. The adjust difference between counties with lowest economic level and counties with highest economic level (reference group) were 46.3% in HBsAg prevalence (all p<0.05) in the multivariable GEE model, after controlling other confounders. A significant dose-response relationship was observed between low economic level and high HBsAg prevalence that the adjust difference increased from 15.6% (aOR=1.156, 95% CI: 1.064–1.257) in the high-economic group to 46.3% (aOR=1.463, 95% CI: 1.294–1.824) in the lowest-economic group, compared with the highest-economic group. The association between low economic level and high HBsAg prevalence was stable in the sensitivity analysis.InterpretationHBV infection was more concentrated among population with lower economic status. Economic-related inequalities in HBV infection decreased in the past decade. Our findings highlight the importance of developing equity-oriented policies and targeted interventions to reduce HBV infection among the poor and hard-to-reach populations to achieve the 2030 HBV elimination goals on time.Funding10.13039/501100001809National Natural Science Foundation of China.

Network impact score is an independent predictor of post-stroke cognitive impairment: A multicenter cohort study in 2341 patients with acute ischemic stroke

BackgroundPost-stroke cognitive impairment (PSCI) is a common consequence of stroke. Accurate prediction of PSCI risk is challenging. The recently developed network impact score, which integrates information on infarct location and size with brain network topology, may improve PSCI risk prediction. AimsTo determine if the network impact score is an independent predictor of PSCI, and of cognitive recovery or decline. MethodsWe pooled data from patients with acute ischemic stroke from 12 cohorts through the Meta VCI Map consortium. PSCI was defined as impairment in ≥ 1 cognitive domain on neuropsychological examination, or abnormal Montreal Cognitive Assessment. Cognitive recovery was defined as conversion from PSCI < 3 months post-stroke to no PSCI at follow-up, and cognitive decline as conversion from no PSCI to PSCI. The network impact score was related to serial measures of PSCI using Generalized Estimating Equations (GEE) models, and to PSCI stratified according to post-stroke interval (<3, 3–12, 12–24, >24 months) and cognitive recovery or decline using logistic regression. Models were adjusted for age, sex, education, prior stroke, infarct volume, and study site. ResultsWe included 2341 patients with 4657 cognitive assessments. PSCI was present in 398/844 patients (47%) <3 months, 709/1640 (43%) at 3–12 months, 243/853 (28%) at 12–24 months, and 208/522 (40%) >24 months. Cognitive recovery occurred in 64/181 (35%) patients and cognitive decline in 26/287 (9%). The network impact score predicted PSCI in the univariable (OR 1.50, 95%CI 1.34–1.68) and multivariable (OR 1.27, 95%CI 1.10–1.46) GEE model, with similar ORs in the logistic regression models for specified post-stroke intervals. The network impact score was not associated with cognitive recovery or decline. ConclusionsThe network impact score is an independent predictor of PSCI. As such, the network impact score may contribute to a more precise and individualized cognitive prognostication in patients with ischemic stroke. Future studies should address if multimodal prediction models, combining the network impact score with demographics, clinical characteristics and other advanced brain imaging biomarkers, will provide accurate individualized prediction of PSCI. A tool for calculating the network impact score is freely available at https://metavcimap.org/features/software-tools/lsm-viewer/.

The BSCM score: a guideline for surgical decision-making for brainstem cavernous malformations

Microsurgical resection of brainstem cavernous malformations (BSCMs) can be performed today with acceptable morbidity and mortality. However, in this highly eloquent location, the indication for surgery remains challenging. We aimed to elaborate a score system that may help clinicians with their choice of treatment in patients with BSCMs in this study. A single-center series of 88 consecutive BSCMs patients with 272 follow-up visits were included in this study. Univariable and multivariable generalized estimating equations (GEE) were constructed to identify the association of variables with treatment decisions. A score scale assigned points for variables that significantly contributed to surgical decision-making. Surgical treatment was recommended in 37 instances, while conservative treatment was proposed in 235 instances. The mean follow-up duration was 50.4 months, and the mean age at decision-making was 45.9 years. The mean BSCMs size was 14.3 ml. In the multivariable GEE model, patient age, lesion size, hemorrhagic event(s), mRS, and axial location were identified as significant factors for determining treatment options. With this proposed score scale (grades 0–XII), non-surgery was the first option at grades 0–III. The crossover point between surgery and non-surgery recommendations lay between grades V and VI while surgical treatment was found in favor at grades VII–X. In conclusion, the proposed BSCM operating score is a clinician-friendly tool, which may help neurosurgeons decide on the treatment for patients with BSCMs.

Sociodemographic and clinical factors associated with transdermal alcohol concentration from the SCRAM biosensor among persons living with and without HIV.

Transdermal alcohol biosensors can objectively monitor alcohol use by measuring transdermal alcohol concentration (TAC). However, it is unclear how sociodemographic and clinical factors that influence alcohol metabolism are associated with TAC. The main aim of this study was to examine how sociodemographic factors (sex, age, race/ethnicity) and clinical factors (body mass index, liver enzymes: alanine aminotransferase [ALT] and aspartate transaminase [AST]), alcohol use disorder, and HIV status were associated with TAC while controlling for level of alcohol use. We analyzed data from a prospective study involving contingency management for alcohol cessation among persons living with and without human immunodeficiency virus (HIV) that used the Secure Continuous Remote Alcohol Monitoring (SCRAM) biosensor. Forty-three participants (Mage =56.6years; 63% male; 58% people living with HIV) yielded 183 SCRAM-detected drinking days. Two indices derived from SCRAM: peak TAC (reflecting level of intoxication) and TAC area under the curve (TAC-AUC; reflecting alcohol volume)-were the main outcomes. Self-reported alcohol use (drinks/drinking day) measured by Timeline Followback was the main predictor. To examine whether factors of interest were associated with TAC, we used individual generalized estimating equations (GEE), followed by a multivariate GEE model to include all significant predictors to examine their associations with TAC beyond the effect of self-reported alcohol use. Number of drinks per drinking day (B=0.29, p<0.01) and elevated AST (B=0.50, p=0.01) were significant predictors of peak TAC. Positive HIV status, female sex, elevated AST, and number of drinks per drinking day were positively associated with TAC-AUC at the bivariate level, whereas only self-reported alcohol use (B=0.85, p<0.0001) and female sex (B=0.67, p<0.05) were significant predictors of TAC-AUC at the multivariate level. HIV status was not independently associated with TAC. Future studies should consider the sex and liver function of the participant when using alcohol biosensors to measure alcohol use.

The More the Merrier: Integrating Multiple Models of Skeletal Maturity Improves the Accuracy of Growth Prediction.

Multiple systems using radiographic skeletal markers to measure development have been described, including the Greulich and Pyle Atlas (GP), the Fels Method (Fels), and the Sanders Hand Classification (Sanders). The purpose of this study was to quantitatively assess whether the integration of skeletal maturity assessment methods and demographic variables improves the accuracy of pediatric growth predictions over the use of skeletal markers or chronologic age alone. The Brush Inquiry contains prospectively collected longitudinal data on children who lived in Cleveland, Ohio between 1926 and 1942. A total of 16 boys and 29 girls were selected for study. All had age, height, and an anteroposterior radiograph of the hand at each of 3 visits. Those visits occurred at 85%, 90%, and 95% of final height. We determined the growth completed at each visit by dividing the height observed by the final height at skeletal maturity. Boys and girls were analyzed separately using chronologic age, height, GP, Fels, and Sanders. The residual difference between the height predicted and actual height, as well as the SD of the prediction error of the cohort at each time point was calculated. To account for multiple visits from each subject, all linear models were produced using the generalized estimating equations (GEEs) procedure. For boys, age, GP, and Fels performed similarly in predicting growth remaining at all 3 time points. For girls, age, GP, and Fels performed similarly in predicting growth remaining at the 85% and 95% time points; however, the Fels Method demonstrated improved performance at the 90% time point compared with chronologic age (P = 0.0076) and GP alone (P = 0.0155). For both boys and girls, the most accurate multivariate GEE model with the lowest SD of prediction error integrated Fels, age, GP, Sanders, and height. The most accurate multivariate GEE model of growth prediction for both boys and girls integrated Fels, age, GP, Sanders, and height. When calculating the amount of growth remaining, it is prudent to integrate multiple systems for greater predictive accuracy. Level III.

Prevalence and correlates of intentional substance use to reduce illicit opioid use in a Canadian setting

ABSTRACT Background: While preliminary evidence has begun to document intentional use of one substance to reduce the use of another, the phenomenon of drug substitution among people who use illicit opioids remains understudied. Therefore, we sought to estimate the prevalence and correlates of intentional substance use to reduce illicit opioid use among persons who use drugs (PWUD). Methods: We analyzed data from three prospective cohorts of PWUD in Vancouver, Canada, using multivariable generalized estimating equations (GEE). Results: Between June 2012 and June 2016, 1527 participants were recruited and contributed 4991 interviews. Of those, 336 (22%) illicit opioid-using participants self-reported substitution to reduce illicit opioid use at least once during study period contributing 467 (9.4%) interviews. Among those interviews, substances substituted for opioids were alcohol (15 participants, 3.2%), stimulants (235, 50.3%), cannabis (129, 27.6%), benzodiazepines (21, 4.5%), and others (20, 4.3%). In multivariable GEE model adjusted for socio-demographic factors, reporting substitution to reduce illicit opioid use was positively associated with greater likelihood of daily cannabis use (Adjusted Odds Ratio = 1.56, 95% Confidence Interval: 1.24–1.96]. Conclusions: While daily cannabis use was associated with reporting opioid substitution attempts, additional study is needed to examine potential of cannabis/cannabinoids to reduce illicit opioid use.

DOP90 Physician global assessment of IBD Activity is associated with COVID-19 severity in patients aged less than 50 years

BackgroundSECURE-IBD is an international database that monitors COVID-19 outcomes in IBD patients. We examined the influence of IBD activity (reflected by Physician Global Assessment (PGA)) on COVID-19 severity, while exploring the modifying effect of age.MethodsWe analyzed 3,028 cases reported to SECURE-IBD from March 13-November 17, 2020. PGA was categorized as remission/mild (reference) vs moderate vs severe. Outcomes used as surrogates of severe SARS-CoV2 infection were hospitalization and a composite of ICU/ventilation/death. Using generalized estimating equations (GEE) to account for clustering by country, we determined the unadjusted association between disease activity and COVID-19 outcomes, overall and by age decade. We included the following covariates in multivariable GEE models a priori: time of COVID-19 diagnosis, age, sex, comorbidities (0, 1, ≥2), IBD type, systemic corticosteroids (CS), anti-TNF monotherapy, anti-TNF plus immune modulator, immune modulator monotherapy, and mesalamine. We included additional covariates (country GDP, race, BMI, anti-integrins and ustekinumab) if they changed the estimate by ≥10%. We also built models stratified by age; the age cut-off for stratification was determined using Joinpoint regression (selecting the age at which outcome rates significantly changed).ResultsCOVID-19 was more severe in patients with more active IBD: hospitalization rates 19% (remission/mild), 26% (moderate) and 45% (severe); ICU/ventilation/death rates 5% (remission/mild), 6% (moderate) and 12% (severe) (p<0.05, Figure 1). In unadjusted analyses, higher PGA was associated with an increased risk of hospitalization (moderate: OR 1.53, 95% CI 1.24–1.88; severe: OR 3.30, 95% CI 2.44–4.46) and ICU/ventilation/death (moderate: OR 1.44, 95% CI 1.10–1.87; severe: OR 2.58, 95% CI 1.71–3.90). As per Figure 2, the magnitude of association between severe PGA (vs remission/mild) and adverse outcomes was greatest in patients ≤50 years. Joinpoint regression also supported a significant change in the rate of ICU/ventilation/death at 50 years. In multivariable analyses (Figure 3), PGA remained significantly associated with hospitalization, but not ICU/ventilation/death. However, in multivariable analyses stratified by age, PGA was consistently associated with both hospitalization and ICU/ventilation/death in patients ≤50, but not >50 years. ConclusionOur data suggest an association between IBD activity, as assessed by PGA, and severe COVID-19 outcome with an increased risk in younger patients. Adequate IBD treatment despite the pandemic is needed to avert poor outcome to symptomatic SARS-CoV2 infections.

Socioeconomic and behavioral determinants of cardiovascular diseases among older adults in Belgium and France: A longitudinal analysis from the SHARE study.

Despite advances in the healthcare system, cardiovascular diseases (CVDs) are still an important public health problem with disparities in the burden within and between countries. Studies among the adult population documented that socioeconomic and environmental factors play a role in the incidence and progression of CVDs. However, evidence is scarce on the socioeconomic determinants and the interplay with behavioral risks among older adults. Therefore, we identified socioeconomic and behavioral determinants of CVDs among older adults. Our sample consisted of 14,322 people aged 50 years and above from Belgium and France who responded to the waves 4, 5, 6 and/or 7 of the Survey of Health Ageing and Retirement in Europe. The effect of determinants on the occurrence of CVD was examined using a Generalized Estimating Equation (GEE) approach for binary longitudinal data. The overall rate of heart attack was 8.3%, which is 7.6% in Belgium and 9.1% in France. Whereas, 2.6% and 2.3% in Belgium and France, respectively, had experienced stroke. In the multivariable GEE model, older age [AOR: 1.057, 95%CI: 1.055-1.060], living in large cities [AOR: 1.14, 95%CI: 1.07-1.18], and retirement [AOR: 1.21, 95%CI: 1.16-1.31] were associated with higher risk of CVD. Furthermore, higher level of education [AOR: 0.82, 95%CI: 0.79-0.90], upper wealth quantile [AOR: 0.82, 95%CI: 0.76-0.86] and having social support [AOR: 0.81, 95%CI: 0.77-0.84] significantly lowers the odds of having CVD. A higher hand grip strength was also significantly associated with lower risk of CVD [AOR: 0.987, 95%CI: 0.984-0.990]. This study demonstrated that older adults who do not have social support, live in big cities, belong to the lowest wealth quantile, and have a low level of education have a higher likelihood of CVD. Therefore, community-based interventions aimed at reducing cardiovascular risks need to give more emphasis to high-risk retired older adults with lower education, no social support and those who live in large cities.

The impact of high-dose opioid prescription on mortality rates among people living with HIV: A retrospective cohort study

ObjectivesHigh-dose opioid use is associated with increased morbidity, mortality, and healthcare utilization. People living with HIV (PLHIV) are frequently prescribed these medications to manage their pain. However, little is known about the relationship between being prescribed high doses of opioids (> 90 MME/d) and mortality risk among this population. The objective of this study was to examine the trends in mortality and the relationship between high-dose opioid analgesic prescribing and mortality among PLHIV. MethodsUtilizing the STOP HIV/AIDS cohort––a population-level linked database of treatment of PLHIV in British Columbia––we conducted bivariable and multivariable generalized estimating equation (GEE) models with a Poisson distribution to examine the relationship between high-dose opioid prescription and all-cause mortality rates in the study sample. ResultsBetween 1996 and 2015, 9272 PLHIV were included in the study. Age- and sex-adjusted mortality rate (using the 2011 Canadian population as the reference) was 30.99 per 1000 person-years (95% confidence interval [CI]: 28.11–33.88). In a multivariable GEE model with adjustment for various demographic and clinical confounders, there was a positive and independent association between being prescribed high-dose opioids and all-cause mortality rates (adjusted rate ratio [ARR] = 3.01; 95%CI: 2.47–3.66). DiscussionWe found that mortality rates were significantly higher among PLHIV who were prescribed high-dose opioids compared to those who were prescribed lower doses. Our results highlight the risk associated with the prescribing of high-dose opioids to manage HIV-related pain and emphasize the need to explore non-opioid approaches to pain management.

Association of Serum Tocilizumab Trough Concentrations with Clinical Disease Activity Index Scores in Adult Patients with Rheumatoid Arthritis.

To determine whether serum trough concentrations of tocilizumab (TCZ) administered as a fixed-dose subcutaneous (SC) injection for the treatment of rheumatoid arthritis (RA) are associated with disease activity responses. We analyzed datasets from the Israeli branch of the multinational TOZURA study, which evaluated a weekly subcutaneous TCZ treatment regimen in a real-life clinical setting. Generalized estimating equations (GEE) were used to evaluate associations between the TCZ levels and the study outcomes. Linear models and GEE were used to evaluate associations between patient characteristics and TCZ levels. A significant association between the TCZ concentrations and the change in the Clinical Disease Activity Index (CDAI) score was observed. In a multivariate binary GEE model, every increase of 10 µg/ml in the concentration of TCZ was associated with being in a state of CDAI remission or low disease activity (OR 1.41) versus a moderate/high disease activity state. An OR of 1.52 was associated with being in a state of Health Assessment Questionnaire-Disability Index remission. In univariate linear models, there was an inverse association between body mass index (BMI) and improvement in the CDAI score, and the BMI score was associated with lower TCZ concentrations. Patients who weighed > 100 kg had lower TCZ concentrations. In the first 24 weeks of treatment with SC TCZ injections, TCZ concentrations were associated with clinical improvement, while body weight and BMI were inversely associated with TCZ concentrations. Personalizing the dose of SC TCZ to body weight may improve outcomes of clinical disease activity in patients with RA.

Association of Baseline Peptidylarginine Deiminase 4 Autoantibodies With Favorable Response to Treatment Escalation in Rheumatoid Arthritis.

To determine if the baseline presence of autoantibodies to peptidylarginine deiminase 4 (PAD4) predicts therapeutic response to biologic and conventional disease-modifying antirheumatic drugs (DMARDs) in patients with rheumatoid arthritis (RA) in whom methotrexate (MTX) monotherapy was unsuccessful. Baseline serum from 282 RA patients in whom MTX monotherapy was unsuccessful was screened for the presence of anti-PAD4 antibodies by immunoprecipitation. Clinical response to either triple DMARD (MTX, sulfasalazine, and hydroxychloroquine) or MTX/etanercept combination therapy was determined at 24 and 48 weeks post-treatment initiation. Disease activity was measured using the Disease Activity Score 28-joint assessment (DAS28), and erosive disease was quantified using the Sharp/van der Heijde scoring method. Generalized estimating equations (GEEs) were used to model the clinical responses to treatment in patients with and those without baseline anti-PAD4 antibodies. Anti-PAD4 antibody positivity was associated with male sex, a history of never smoking, and anti-citrullinated protein antibodies. At baseline, patients with anti-PAD4 antibodies had longer disease duration and significantly more radiographic joint damage than anti-PAD4-negative patients, but did not differ in disease activity according to the DAS28. In unadjusted analyses and multivariable GEE models, patients with anti-PAD4 antibodies exhibited greater improvements in DAS28 (adjusted P = 0.02 and P = 0.008, respectively) and less radiographic progression (adjusted P = 0.01 and P = 0.002, respectively) compared to anti-PAD antibody-negative patients, independent of treatment received. Although anti-PAD4 antibodies were associated with worse baseline radiographic joint damage, suggesting a history of active or undiagnosed disease, treatment escalation therapy was more effective in reducing disease activity and slowing the progression of joint damage in this patient subset.