Iron deficiency anaemia (IDA) is associated with adverse pregnancy outcomes globally. Women with inherited bleeding disorders are at increased risk, with scarce data on rates of IDA screening and correction during pregnancy. The impact of correction on outcomes is unclear. To assess the frequency of IDA screening, the frequency of IDA in pregnancies with and without bleeding disorders and the impact of correction on outcomes. This retrospective population-based cohort study includes all hospitalized pregnancies using the Alberta Pregnancy Birth Cohort. IDA was defined as haemoglobin <110 g/L in first/third trimester, <105 g/L in second trimester and ferritin <30 μg/L. Logistic regression was performed to assess the association between IDA, corrected IDA and pregnancy outcomes. 36 500/207 355 (18%) pregnancies and 14/58 (24%) pregnancies in the bleeding disorder subgroup had anaemia. Only one of three pregnancies with anaemia had a concurrent ferritin test, and over 80% demonstrated IDA. Ferritin screening was performed in 60% of the overall cohort and 79% in bleeding disorders. Young maternal age, multiparity, lower socioeconomic status and mothers from African and Asian countries significantly predict IDA. Despite the increased risk for adverse outcomes, only 43 (8%) first-trimester and 96 (9%) third-trimester IDA were corrected. IDA screening and correction remained suboptimal among pregnant women. Updated screening guidelines may promote better identification and outcomes.

Aims/hypothesisLarge-scale data on age- and sex-specific dementia mortality trends among people with diabetes remain limited, as most previous studies have been restricted to single countries or have not distinguished mortality by diabetes status. We estimated age- and sex-specific time trends in dementia mortality among individuals with and without diabetes from high-income jurisdictions.MethodsWe analysed aggregated mortality and demographic data using registries and administrative sources in Australia, Canada (Alberta and Ontario), France, Denmark, Finland and Scotland from 2000 to 2023. Poisson regression was used to estimate mortality rates for dementia as the underlying cause of death in people with and without diabetes at 60, 70, 80 and 90 years of age.ResultsA total of 114,559 and 589,706 dementia deaths were identified in over 42 and 244 million person-years of follow-up for individuals with and without diagnosed diabetes, respectively. Dementia mortality trends varied by age and jurisdiction but were generally consistent for both sexes. At younger ages (e.g. 60 and 70 years), the dementia mortality trends did not suggest any meaningful increases or decreases, except for in Scotland, which reported increasing dementia mortality over time only for those with diabetes. At older ages (e.g. 80 and 90 years), however, increases in dementia mortality were observed in most jurisdictions, ranging from 7.6% to 42.4% per 5 years. The magnitude of the increases was generally greater for those with diabetes. Mortality from dementia subtypes (e.g. Alzheimer's disease and vascular dementia) also increased over time in individuals aged 40-89 years, with greater increases in mortality rates for individuals with diabetes, specifically in Australia and Scotland.Conclusions/interpretationIncreases in dementia mortality were observed for those aged 80 years and above and were most marked for people with diabetes. These findings highlight the growing burden of dementia for health systems.

BACKGROUND:The significance of magnesium as a treatment or prognostic factor in heart failure (HF) remains uncertain despite frequent use. We evaluated the frequency and outcomes of magnesium testing, hypomagnesemia, and intravenous (IV) replacement in a large population-based cohort. METHODS:This retrospective cohort study used linked administrative data (April 2012 to March 2020). Patients with a primary diagnosis of HF in the emergency department (ED) or hospital were included. Outcomes included all-cause and cause-specific death and hospitalisation. Secondary outcomes included ED visits and physicians claims. We also examined rates of serum magnesium testing and hypomagnesemia. RESULTS:Among 78,957 acute HF episodes (in 42,763 patients), 58.7% included serum magnesium testing. Of those tested, serum magnesium levels were < 0.75 mmol/L in 31.7%, 0.75-0.95 mmol/L in 56.8% and > 0.95 mmol/L in 11.5%. Magnesium levels (per 0.02 mmol/L increase) were independently associated with mortality when < 0.70 mmol/L (hazard ratio [HR] 0.99, 95% confidence interval [CI] 0.98-0.99; P < 0.001) or > 0.86 mmol/L (HR 1.04, 95% CI 1.03-1.04; P < 0.001). IV magnesium was administered to 13.7% (n = 6333) of tested patients, including 29.7% without hypomagnesemia. After multivariable adjustment, IV magnesium was associated with a higher short-term mortality (HR 1.66, 95% CI 1.4-1.96; P < 0.0001) and hospitalisation risk (HR 1.36, 95% CI 1.13-1.63; P < 0.001). CONCLUSIONS:Serum magnesium testing is common in patients presenting to the ED or hospital with HF, and low or high magnesium is associated with worse outcomes. Replacement with IV magnesium was associated with worse outcomes even after adjustment, warranting further study.

BACKGROUND:Cardiovascular disease has historically been the most common cause of death (COD) among people with and without diabetes. However, substantial progress has been made in the management of cardiovascular disease. We conducted a multinational analysis to establish whether this trend is still the case. METHODS:In this multinational, population-based study, we assembled aggregated annual mortality data collected during routine clinical care from nationally or regionally representative administrative datasets in high-income jurisdictions between 2000 and 2023. For inclusion, datasets must have ongoing enrolment of new patients with diabetes, cause-specific death counts in people with and without diabetes, and sex-specific and age-specific data. We collected population size, counts of prevalent diabetes (type 1 and type 2), death counts, and person-years of follow-up in people with and without diagnosed diabetes by sex and 10-year age group. We estimated cause-specific trends in mortality rates, proportional mortality, and mortality rate ratios (MRR) for people with versus those without diabetes (type 1 and type 2) using Poisson models standardised for age and sex. FINDINGS:Using data from 11 jurisdictions, we identified 2·7 million deaths in people with diabetes and 11·0 million deaths in people without diabetes during a total of 1·7 billion person-years of follow-up. Cardiovascular disease mortality decreased in all jurisdictions in populations with and without diabetes. Mean 5-year declines in cardiovascular disease mortality among people with diabetes ranged from 8·3% (95% CI 5·9 to 10·7) to 25·4% (22·8 to 28·0). Mortality due to diabetes declined in most jurisdictions. Dementia mortality increased in people with and without diabetes in six (86%) of seven jurisdictions. Cancer mortality declined in people with diabetes in three (33%) of nine jurisdictions and in people without diabetes in six (67%). At the end of the observation period, cancer was the leading COD in people with diabetes in four (36%) of 11 jurisdictions. MRRs were generally stable for all CODs. Exceptions include Lithuania, where the mean 5-year change in MRR for cardiovascular disease was -7·6% (-10·1 to -5·1), indicating a more rapid fall in cardiovascular disease mortality in people with diabetes than in people without. For dementia, the MRR increased in Denmark (5-year change 8·0% [5·0 to 11·1]) and Scotland (11·4% [8·5 to 14·3]). INTERPRETATION:Mortality from cardiovascular disease and diabetes has declined among people with diabetes in most jurisdictions, whereas mortality from dementia has increased markedly, independent of age. Cardiovascular disease is no longer universally the most common COD among people with diabetes in high-income countries. FUNDING:US Centers for Disease Control and Prevention, Diabetes Australia Research Program, and Victoria State Government Operational Infrastructure Support Program.

ABSTRACT Introduction Missing data frequently occurs in health databases and can bias analyses if not correctly dealt with. Objectives Using real-world data where missing values were present in as much as 30% of our sample, we compared complete-case and multiple-imputation methods for recovering true parameters of a multivariable logistic regression model for the association between maternal glucose levels during pregnancy and child excess weight at preschool age. Methods This study utilized a cohort of 130,424 children from two Canadian urban health zones with complete preschool-age body mass index (BMI) measurements linked to administrative health data. We introduced missingness through deletion following three distinct mechanisms: missing completely at random (MCAR), at random (MAR), and not at random (MNAR). We employed complete-case and multiple-imputation methods to handle the introduced missingness. The associations between five categories of maternal glucose levels during pregnancy with child excess weight at pre-school age were determined from logistic regression models using the full observed data (true values), observed data without deletions (complete-case estimates), and imputed data (multiple-imputation estimates). Accuracy of complete-case and multiple-imputation estimates were evaluated against true values. Finally, we conducted a sensitivity analysis for the MNAR mechanism using pattern-mixture models with an additive shift. Results Under MCAR and MAR, multiple-imputation generally yielded larger bias and relative bias, but smaller or similar mean square error and reached higher significance than complete-case analysis. Both methods showed high significance (≥ 0.96) for most effects and high coverage (≥ 0.99) consistently. Under MNAR, both complete-case and multiple-imputation showed poor performance regarding bias and statistical significance. Sensitivity analysis indicated performance varied by specific effect. Conclusions When faced with missing data, researchers should assess missingness mechanisms, report both complete-case and multiple-imputation estimates under MCAR/MAR while accounting for power-versus-bias tradeoffs, and employ pattern-mixture sensitivity analyses to test robustness when MNAR is plausible.

AIM:To estimate the prevalence of medication use in nonhospitalized pregnant women with COVID-19. METHODS:A prospective two-stage individual patient meta-analysis across 10 data sources in Europe and North America studied medication use among nonhospitalized pregnant women with COVID-19 between January 2020 and December 2022. Comparisons were made between medication use within 30 days pre- and post-COVID-19 diagnosis in this cohort and two comparator groups: pregnant women without COVID-19 and nonpregnant women with COVID-19. Prevalence estimates were pooled using a random-effects model stratified by trimester. RESULTS:50 335 nonhospitalized pregnant women with COVID-19 were identified. The pooled prevalence of antibacterial use in the third trimester was higher post-COVID-19 diagnosis (6.8%, 95% confidence interval [CI] = 5.5-8.4, I2 = 94%) compared with the same women pre-COVID-19 (3.9%, 95% CI = 3.1-4.9, I2 = 89%). Overall, pregnant women with COVID-19 had higher medication use compared to pregnant women without COVID-19, although the CIs of the prevalence overlapped. Post-COVID-19, antithrombotic prevalence was 4.5% (95% CI = 1.1-16.5, I2 = 100%) among pregnant women with COVID-19 in the third trimester, compared to 2.1% (95% CI = 1.2-3.6, I2 = 99%) among those without COVID-19 in the third trimester. Compared to nonpregnant women with COVID-19, pregnant women with COVID-19 were less likely to be prescribed analgesics, antiprotozoals, corticosteroids, psychoanaleptics and psycholeptics, and more likely to be prescribed antithrombotics, cough and cold and nasal preparations, and drugs used in diabetes across all trimesters. High heterogeneity existed in nearly all analyses. CONCLUSION:This international meta-analysis reveals low medication use and country-specific variations, enhancing insight into the management of COVID-19 in nonhospitalized pregnant women. Higher antithrombotic use post-COVID-19 suggests prophylactic treatment in this population, but variation between countries emphasizes the challenges of combining multinational data.

Background Heart failure (HF) management guidelines continue to evolve with emerging evidence from landmark trials defining new standards for guideline-directed medical therapy (GDMT). However, data on HF drug costs in Canada remain limited. This study examined long-term historical costs of drugs associated with HF management among elderly patients hospitalized with HF between 2013/14 and 2018/19. Methods Prescription claims from the National Prescription Drug Utilization Information System were linked for patients aged >65 years with a HF hospitalization between April 1, 2013 – March 31, 2019 in Canada, except Quebec, Nova Scotia, and territories. HF management drugs included GDMT and diuretics; costs were inflated to 2025 Canadian dollars. Annual total and per-patient drug costs were calculated. Results The total drug cost for managing HF in 163,973 elderly hospitalized patients over the five-year period was $252.2 million. While average annual per-patient costs remained relatively stable ($567 in 2013/14 to $593 in 2018/19), total annual costs rose from $31.5 million to $50.8 million, largely driven by an increase in the number of patients hospitalized with HF. GDMT accounted for the majority of drug costs, and increased from 72.2% to 76.0%. In 2018/19, costs were higher in males ($632) than females ($558), and highest among patients aged 65–74 years ($680) and lowest among those 90+ ($493) years. Conclusions The costs of drugs to manage HF is substantial and is increasing over time, primarily as a result of increasing number of patients hospitalized with HF. There were modest shifts toward greater use of GDMT, with variations across province, sex, and age groups.

Importance Clinical risk factors, lifestyle factors, and social determinants of health have established sex-specific associations with new-onset cardiovascular disease (CVD). However, little is known about the role of patient-reported health measures in individuals without CVD. Objective To determine whether self-rated health is independently associated with the development of CVD and whether this association differs by sex. Design, Setting, and Participants This retrospective cohort study included community-dwelling adults enrolled in the Ontario Health Study from March 1, 2009, to December 31, 2017, with no prior CVD or active cancer. Outcome ascertainment was via linkage to administrative databases for follow-up outcomes to March 31, 2024, and data were analyzed from January 5, 2025, to January 16, 2026. Exposure Excellent, very good to good, and fair to poor self-rated health. Main Outcomes and Measures The primary outcome was CVD events (hospitalization for myocardial infarction, stroke, heart failure, and cardiovascular death). Cause-specific hazard models with an interaction between self-rated health and sex were adjusted for age, traditional risk factors, lifestyle factors, social determinants of health, and family history of CVD. Results The cohort consisted of 170 197 participants (104 789 [61.6%] women; median age, 48 [IQR, 36-58] years) followed up for a median of 12.1 (IQR, 12.0-12.3) years. Fair to poor health was reported in 11 661 women (11.1%) and 6381 (9.8%) men; very good to good health, in 75 819 (72.4%) women and 47 865 (73.2%) men; and excellent health, in 17 309 (16.5%) women and 11 162 (17.1%) men. After adjustment, poorer self-rated health was associated with a higher rate of CVD in both sexes. Compared with excellent self-rated health, the fully adjusted hazard ratios (HRs) for fair to poor self-rated health were 2.08 (95% CI, 1.80-2.40) in women and 1.45 (95% CI, 1.29-1.64) in men; the HRs for very good to good self-rated health compared with excellent health were 1.26 (95% CI, 1.11-1.43) in women and 1.04 (95% CI, 0.94-1.14) in men (P < .001 for interaction). Conclusions and Relevance In this cohort study of individuals without CVD, as many as 1 in 10 rated their health as fair to poor. Fair to poor self-rated health was an independent risk factor associated with new CVD, and its relative hazard was greater in women than in men. These findings support the use of a simple self-assessment of health to aid in risk stratification in the primary prevention of CVD.

We sought to determine the frequency of new psychoactive medication prescriptions after hospital discharge in COVID-19 intensive care unit (ICU) survivors and describe associated 1-year clinical outcomes. In a multicentre population-based cohort study using linked provincial health data sets from Alberta, Canada to identify comorbidities, ICU interventions, laboratory values, and medications, we included all adults admitted to an ICU with COVID-19 who survived to hospital discharge between January 2021 and July 2022. We identified new psychoactive medication recipients at discharge and persistent recipients after 1-year. An inverse probability of treatment weighting model quantified the association with 1-year all-cause mortality, hospital readmission, and emergency department (ED) visits. We found that 1,486 psychoactive-naïve adults (mean, 56 yr; 67.5

BACKGROUND:Preterm birth, defined as delivery before 37 weeks of gestation, is a major cause of neonatal morbidity and mortality and places a substantial burden on healthcare systems. Many existing prediction tools depend on clinical measurements, imaging, or laboratory results that may be unavailable or inconsistently recorded early in pregnancy. Routinely collected administrative health records may offer a scalable alternative for early population-level risk prediction. OBJECTIVE:This study aimed to develop and evaluate machine learning models for predicting preterm birth at 26 weeks of gestation using routinely collected administrative health records. STUDY DESIGN:We conducted a retrospective population-based cohort study of 328,834 singleton live-birth pregnancies in Alberta, Canada, from 2009 through 2018. Maternal inpatient and outpatient records, physician claims, and prescription dispensations were linked to construct pregnancy-level features from 1 year before conception through 26 weeks of gestation. We developed and compared logistic regression, random forest, tabular neural network, gradient-boosted tree, and transformer-based models. Data were split chronologically into a training set from 2009 through 2016 (261,648 pregnancies) and an independent holdout set from 2017 through 2018 (67,186 pregnancies). Model performance was assessed on the holdout set using the area under the receiver operating characteristic curve, area under the precision-recall curve, calibration, Brier score, and threshold-based classification measures. RESULTS:The gradient-boosted tree model achieved the best overall performance, with an area under the receiver operating characteristic curve of 0.7704 (95% confidence interval, 0.7625-0.7778) and an area under the precision-recall curve of 0.4403, outperforming the other models. Transformer-based models also showed strong discrimination and performed better than several traditional approaches. Predicted risks showed good alignment with observed outcomes, and risk stratification identified clinically meaningful groups with progressively higher observed preterm birth rates across increasing predicted risk categories. CONCLUSION:Machine learning models trained on large-scale administrative health records can predict preterm birth with good discrimination by 26 weeks of gestation. These models support early, scalable, population-level risk stratification using routinely collected health system data and may help identify pregnancies that could benefit from closer follow-up and preventive planning. Such models are intended to complement, not replace, clinical assessment.

BACKGROUND:Understanding the varied impact of COVID-19 severity on pregnancy outcomes is crucial for informed clinical management and targeted interventions. OBJECTIVE:To evaluate the impact of COVID-19 on pregnancy outcomes, distinguishing between pregnant women managed in primary care and those requiring hospitalization. SEARCH STRATEGY:Regulatory authorities actively promoted global cooperation on COVID-19's impact during pregnancy. Data were obtained through these regulatory bodies and direct researcher communication rather than through systematic searches. SELECTION CRITERIA:Data sources required secondary population-based data to identify pregnancies with COVID-19, along with hospitalization, diagnostic and medication codes. Eligibility for the meta-analysis was determined through protocol evaluation and researcher consultations. DATA COLLECTION AND ANALYSIS:PRISMA-IPD and Cochrane guidelines for prospective meta-analysis were followed. Protocols and definitions were standardized across sources, and a common R script was developed. Initially, crude and adjusted relative risks (aRR) with 95% confidence intervals (CI) were calculated to assess adverse outcomes in pregnant women with and without COVID-19 in each data source. Estimates were stratified by trimester at infection and hospitalization status. Subsequently, data were pooled using a random-effects meta-analysis. MAIN RESULTS:Data from 10 sources across seven countries contributed to the meta-analysis, including 86 210 pregnant women diagnosed with COVID-19, of whom 4.4% were hospitalized. Non-hospitalized pregnant women with COVID-19 had no increased risks of adverse outcomes compared to pregnant women without COVID-19. However, hospitalized women with COVID-19 in each trimester had higher risks of cesarean section, preterm birth, and LBW compared to pregnant women without COVID-19. Hospitalization due to COVID-19 in the third trimester was associated with increased risk of stillbirth (aRR 5.90, 95% CI: 2.22-15.71, I2 = 0%). First-trimester hospitalizations due to COVID-19 did not show heightened risks of GDM (aRR 2.08, 95% CI: 0.93-4.64, I2 = 65%), pre-eclampsia (aRR 1.79, 95% CI: 0.48-6.66, I2 = 71%), or major congenital anomalies (aRR 1.30, 95% CI: 0.55-3.06, I2 = 0%). CONCLUSIONS AND RELEVANCE:COVID-19 requiring hospitalization is associated with adverse pregnancy outcomes, emphasizing the need to prevent severe illness during pregnancy. This study also highlights the importance of international collaboration for gathering pregnancy data and shows that building global research networks is essential for responding to future health crises.

Background: Better data are necessary to determine whether baseline level of kidney function affects the rate of progressive kidney disease following pregnancy. Objective: The objective was to determine whether the baseline (pre-pregnancy) estimated glomerular filtration rate (eGFR) modifies the association between becoming pregnant and the subsequent rate of progressive kidney disease. Design: Population-based cohort study using provincial administrative health care databases in Ontario and Alberta, Canada. Setting: The sample will be accrued from April 1, 2007, to March 31, 2023, in Ontario and from April 1, 2012, to March 31, 2023, in Alberta. Follow-up for study outcomes will occur until March 31, 2024. Participants: The pregnant group will include adult female residents of Ontario or Alberta with a record of a pregnancy of 20 to 46 weeks’ gestation during the accrual period, and the non-pregnant group will include adult female residents with no prior record of pregnancy. The cohort entry dates in those in the pregnant group will be the estimated date of conception; the entry dates for those in the non-pregnant group will be randomly assigned following the distribution of dates in the pregnant group. To be eligible, individuals must be between 18 and 45 years old at cohort entry. They require at least 1 serum creatinine measurement within 2 years before entry and should not have received maintenance dialysis or a prior kidney transplant. Both groups will be categorized into one of 3 levels of baseline eGFR (≥60, 45–59, and <45 mL/min per 1.73 m 2 ). Inverse probability of treatment weighting on a propensity score will be used to balance the pregnant and non-pregnant groups on baseline characteristics (including age, proteinuria, hypertension, and diabetes) within the 3 categories of baseline eGFR. Measurements: The primary outcome, progressive kidney disease, will be defined as a composite of a persistent ≥40% drop in eGFR from the baseline value, a new persistent eGFR <15 mL/min per 1.73 m 2 , receipt of maintenance dialysis, or receipt of a kidney transplant. The secondary outcomes will be the components of the primary composite outcome examined separately and the annualized change in eGFR in mL/min per 1.73 m 2 from baseline. Methods: We will test for statistical interaction to determine whether the baseline category of eGFR modifies the rate of long-term progressive kidney disease after pregnancy. We hypothesize that a statistical interaction will be present. We will present weighted cause-specific hazard ratios (HRs) and cumulative incidence function (CIF) curves for up to 10 years of follow-up for the pregnant and non-pregnant groups stratified by each eGFR category. We will perform additional pre-specified analyses to confirm whether the findings are robust and examine associations that account for baseline proteinuria. Results: Based on a feasibility analysis using ICES data in Ontario, we expect the cohort to include over 400 000 pregnant females and 1.2 million non-pregnant females. This includes at least 395 000 pregnant females with baseline eGFR ≥60 mL/min/1.73 m 2 , 300 with eGFR 45 to 59 mL/min/1.73 m 2 , and 110 with eGFR <45 mL/min/1.73 m 2 . The median follow-up is anticipated to be 5 years (range = 1-17 years) with minimal loss to follow-up. Limitations: Measures of kidney function will be obtained as part of routine care (not according to a research schedule). Measures of baseline proteinuria are frequently missing from routine care data, even in up to 15% of those with an eGFR <45 mL/min per 1.73 m 2 . Conclusion: This study will investigate whether the level of baseline eGFR modifies the rate of progressive kidney disease after pregnancy and will estimate the cumulative incidence of progressive kidney disease in pregnant and non-pregnant females across 3 categories of baseline eGFR.

Background:Studies have suggested that the COVID-19 pandemic negatively impacted patient adherence with chronic medications. We explored whether adherence patterns changed in patients chronically treated with cardiovascular drugs after onset of the COVID-19 pandemic. Methods:In this retrospective cohort study we examined drug dispensation data for all adult Albertans who were chronic users of at least 1 cardiovascular drug class between 2017 and 2023. We calculated each patient's proportion of days covered (PDC) for each drug class in the prepandemic phase (March 15, 2018 to March 14, 2020) and the pandemic phase (March 15, 2020 to March 14, 2022), and used generalized estimating equation logistic regression to estimate the effect of time period on achievement of good adherence (PDC >0.8) after adjusting for age, sex, socioeconomic status, and comorbidities. Results:Of 548,601 chronic users of at least 1 cardiovascular drug class between March 15, 2018 and March 14, 2022, 47.2% were women, the mean age was 62.3 years, and 55.4% had Charlson Comorbidity Index (CCI) scores of 0. The most frequently dispensed cardiovascular drugs were angiotensin-converting enzyme inhibitors or angiotensin II receptor blockers (67.6%) and statins (53.8%); the most frequent diagnoses were hypertension (77.2%), diabetes mellitus (30.6%), and ischemic heart disease (19.6%). Chronic users of cardiovascular drugs were more likely to have PDC >0.8 during the pandemic than in the prepandemic period: 75.4% vs 72.8%, with adjusted odds ratios ranging from 1.05 (95% confidence interval 1.00-1.11) for mineralocorticoid receptor antagonists to 1.16 (95% confidence interval 1.15-1.17) for statins. Conclusions:Chronic users of cardiovascular drugs exhibited better adherence during the COVID-19 pandemic than before the pandemic.

Over 100 million pregnant people worldwide remain at risk of COVID-19. We compared the prevalence of severe COVID-19 in pregnancy and in people of reproductive age, and the risk of adverse pregnancy/neonatal outcomes in those with/without COVID-19 during gestation. In the Canadian Mother–Child Cohort, two sub-cohorts were identified using medical services, prescription medication fillings, hospitalizations, and COVID-19 surveillance testing programs data (28 February 2020–2021). The first included all pregnant people with at least one completed trimester of pregnancy during the study period, stratified on COVID-19 status. The second included all non-pregnant people (aged 15–45) with a positive COVID-19 test during the same period. COVID-19 severity was categorized based on hospital admissions before the end of pregnancy. Associations between COVID-19 during pregnancy and adverse perinatal outcomes were quantified using log-binomial regressions. A total of 150,345 pregnant people (3464 (2.3%) had COVID-19), and 112,073 non-pregnant people with COVID-19 were included. Maternal age at the time of COVID-19 diagnosis/positive test was statistically significantly lower among pregnant individuals compared to those who were not pregnant (96% had less than 40 years vs. 80%, p < 0.001). In pregnancy, COVID-19 was associated with the risk of spontaneous abortions (adjRR 1.76, 95%CI 1.37, 2.25), gestational diabetes (adjRR 1.52, 95%CI 1.18, 1.97), prematurity (adjRR 1.30, 95%CI 1.01, 1.67), and NICU (adjRR 1.32, 95%CI 1.10, 1.59); COVID-19 treatment with medications reduced risks. Severe COVID-19 was more prevalent in pregnancy and was associated with higher risks of adverse maternal/neonatal outcomes. As some countries are pulling back preventive strategies for COVID-19, this study highlights the importance of continued surveillance during pregnancy to prevent adverse pregnancy outcomes.

OBJECTIVE:Our aim in this work was to examine postpartum diabetes screening and rates in women diagnosed with gestational diabetes mellitus (GDM) using standard vs modified criteria during the COVID-19 pandemic. METHODS:Women with GDM pregnancies between January 1, 2020, and December 31, 2021, in Alberta, Canada, were stratified by the GDM diagnosis criteria and followed for 18 months postpartum diabetes screening. Proportions of prediabetes and diabetes were compared between the standard vs modified GDM criteria groups at 6 and 18 months. Multivariable logistic regression analysis was used to examine differences in prediabetes and diabetes rates between the 2 GDM criteria groups after adjusting for baseline differences. RESULTS:Among 10,238 individuals with GDM, 780 were diagnosed using the modified criteria and 9,458 were diagnosed using the standard criteria. There was no difference in the proportion of individuals who underwent postpartum screening by 6 months (27.1% vs 28.9%, p=0.29) or by 18 months (43.1% vs 45.3%, p=0.24) among the modified and standard groups, respectively. Diabetes proportions were higher in women diagnosed with GDM using the modified criteria compared with those diagnosed using the standard criteria (27.0% vs 4.2%, p<0.0001; adjusted odds ratio 8.18, 95% confidence interval 5.76 to 11.6). Proportions of prediabetes and diabetes at 18 months were 15.2% and 20.8% (p=0.014) and 29.8% and 6.1% (p<0.0001) for modified and standard groups, respectively. CONCLUSIONS:Regardless of the GDM diagnostic method, postpartum diabetes screening among women with GDM was suboptimal during the COVID-19 pandemic. The modified criteria for GDM identified a group of women who were at higher risk for conversion to diabetes.

Background: Certain patients with diabetes and COVID-19 are at high risk of severe outcomes. Identification of risk factors among this group is required to risk-stratify those who may benefit from further surveillance. We aimed to develop machine learning (ML) models predicting severe outcomes among individuals with diabetes and COVID-19 in Alberta, Canada. Methods: Patients with diabetes and COVID-19 determined by PCR test administered in community and/or emergency department (ED) settings (March 2020-March 2021) were included. Outcomes were ED visit, hospitalization or death for those tested in the community (“Community cohort”) and hospitalization or death for those tested in ED (“ED cohort”), and in the combined cohorts (“Community+ED cohort”). Outcomes and features (sociodemographics, drug/healthcare utilization, health history) were identified using healthcare administrative data (2008-2021). Calibration plots, areas under the receiver operating curve, precision-recall curves (AUC, AUPRC), and threshold analyses were used to assess the models. Results: The Community cohort included 11,247 individuals (1,665 ED visits; 756 hospitalizations; 421 deaths). AUCs for models predicting ED/hospitalization/death were 0.65/0.70/0.93. The AUCs for predicting death in ED (1,495 individuals; 169 deaths) and Community+ED (12,410 individuals; 582 deaths) cohorts were 0.82 and 0.93. Models predicting hospitalization in these cohorts performed poorly and are not reported. Of all models, that predicting death from the Community performed best (sensitivity 0.77, specificity 0.91, positive predictive value 0.26, negative predictive value 0.99), and improved the prediction of death at a 10% risk threshold (compared to the pre-test probability, positive likelihood ratio 9.06 and negative likelihood ratio 0.25). Conclusion: Identifying diabetes patients at the highest risk of the worst outcomes would assist in triaging patients to ensure appropriate resource use in times of high demand. Overall, the model predicting death among patients with diabetes and COVID-19 in the community could be useful in identifying who requires additional care.