Assessing the research landscape and clinical utility of large language models: a scoping review.

IMPORTANCE: Large language models (LLMs) like OpenAI's ChatGPT are powerful generative systems that rapidly synthesize natural language responses. Research on LLMs has revealed their potential and pitfalls, especially in clinical settings. However, the evolving landscape of LLM research in medicine has left several gaps regarding their evaluation, application, and evidence base. OBJECTIVE: This scoping review aims to (1) summarize current research evidence on the accuracy and efficacy of LLMs in medical applications, (2) discuss the ethical, legal, logistical, and socioeconomic implications of LLM use in clinical settings, (3) explore barriers and facilitators to LLM implementation in healthcare, (4) propose a standardized evaluation framework for assessing LLMs' clinical utility, and (5) identify evidence gaps and propose future research directions for LLMs in clinical applications. EVIDENCE REVIEW: We screened 4,036 records from MEDLINE, EMBASE, CINAHL, medRxiv, bioRxiv, and arXiv from January 2023 (inception of the search) to June 26, 2023 for English-language papers and analyzed findings from 55 worldwide studies. Quality of evidence was reported based on the Oxford Centre for Evidence-based Medicine recommendations. FINDINGS: Our results demonstrate that LLMs show promise in compiling patient notes, assisting patients in navigating the healthcare system, and to some extent, supporting clinical decision-making when combined with human oversight. However, their utilization is limited by biases in training data that may harm patients, the generation of inaccurate but convincing information, and ethical, legal, socioeconomic, and privacy concerns. We also identified a lack of standardized methods for evaluating LLMs' effectiveness and feasibility. CONCLUSIONS AND RELEVANCE: This review thus highlights potential future directions and questions to address these limitations and to further explore LLMs' potential in enhancing healthcare delivery.

Evaluating the coding accuracy of type 2 diabetes mellitus among patients with non-alcoholic fatty liver disease.

BACKGROUND: Non-alcoholic fatty liver disease (NAFLD) describes a spectrum of chronic fattening of liver that can lead to fibrosis and cirrhosis. Diabetes has been identified as a major comorbidity that contributes to NAFLD progression. Health systems around the world make use of administrative data to conduct population-based prevalence studies. To that end, we sought to assess the accuracy of diabetes International Classification of Diseases (ICD) coding in administrative databases among a cohort of confirmed NAFLD patients in Calgary, Alberta, Canada. METHODS: The Calgary NAFLD Pathway Database was linked to the following databases: Physician Claims, Discharge Abstract Database, National Ambulatory Care Reporting System, Pharmaceutical Information Network database, Laboratory, and Electronic Medical Records. Hemoglobin A1c and diabetes medication details were used to classify diabetes groups into absent, prediabetes, meeting glycemic targets, and not meeting glycemic targets. The performance of ICD codes among these groups was compared to this standard. Within each group, the total numbers of true positives, false positives, false negatives, and true negatives were calculated. Descriptive statistics and bivariate analysis were conducted on identified covariates, including demographics and types of interacted physicians. RESULTS: A total of 12,012 NAFLD patients were registered through the Calgary NAFLD Pathway Database and 100% were successfully linked to the administrative databases. Overall, diabetes coding showed a sensitivity of 0.81 and a positive predictive value of 0.87. False negative rates in the absent and not meeting glycemic control groups were 4.5% and 6.4%, respectively, whereas the meeting glycemic control group had a 42.2% coding error. Visits to primary and outpatient services were associated with most encounters. CONCLUSION: Diabetes ICD coding in administrative databases can accurately detect true diabetic cases. However, patients with diabetes who meets glycemic control targets are less likely to be coded in administrative databases. A detailed understanding of the clinical context will require additional data linkage from primary care settings.

Automated Paper Screening for Clinical Reviews Using Large Language Models: Data Analysis Study.

BACKGROUND: The systematic review of clinical research papers is a labor-intensive and time-consuming process that often involves the screening of thousands of titles and abstracts. The accuracy and efficiency of this process are critical for the quality of the review and subsequent health care decisions. Traditional methods rely heavily on human reviewers, often requiring a significant investment of time and resources. OBJECTIVE: This study aims to assess the performance of the OpenAI generative pretrained transformer (GPT) and GPT-4 application programming interfaces (APIs) in accurately and efficiently identifying relevant titles and abstracts from real-world clinical review data sets and comparing their performance against ground truth labeling by 2 independent human reviewers. METHODS: We introduce a novel workflow using the Chat GPT and GPT-4 APIs for screening titles and abstracts in clinical reviews. A Python script was created to make calls to the API with the screening criteria in natural language and a corpus of title and abstract data sets filtered by a minimum of 2 human reviewers. We compared the performance of our model against human-reviewed papers across 6 review papers, screening over 24,000 titles and abstracts. RESULTS: Our results show an accuracy of 0.91, a macro F1-score of 0.60, a sensitivity of excluded papers of 0.91, and a sensitivity of included papers of 0.76. The interrater variability between 2 independent human screeners was κ=0.46, and the prevalence and bias-adjusted κ between our proposed methods and the consensus-based human decisions was κ=0.96. On a randomly selected subset of papers, the GPT models demonstrated the ability to provide reasoning for their decisions and corrected their initial decisions upon being asked to explain their reasoning for incorrect classifications. CONCLUSIONS: Large language models have the potential to streamline the clinical review process, save valuable time and effort for researchers, and contribute to the overall quality of clinical reviews. By prioritizing the workflow and acting as an aid rather than a replacement for researchers and reviewers, models such as GPT-4 can enhance efficiency and lead to more accurate and reliable conclusions in medical research.

Distribution of videos demonstrating best practices in preventing hemolysis is associated with reduced hemolysis among nurse-collected specimens in hospitals.

BACKGROUND: Minimizing hemolysis during phlebotomy ensures accurate chemistry results and reduces test cancellations and specimen recollections. We developed videos demonstrating best practices to reduce hemolysis and tested whether distribution to clinical nurse educators (CNEs) for provision to nursing staff affected the degree of specimen hemolysis in hospital inpatient units and outpatient clinics. METHODS: Videos of common blood collections demonstrating best practices to reduce hemolysis were filmed and then distributed via email link to all hospital-based CNEs in Calgary, Alberta, Canada. (https://vimeo.com/user18866730/review/159869683/a0cec9827f). Roche Cobas hemolysis index (H-index) results from specimens collected +/- 12 months from the date of video distribution were extracted from Roche Cobas IT middleware (cITM) and linked to collection location. An interrupted time series (ITS) analysis with collection location as the unit of anlaysis was used to quantify impact of video distribution on the trajectory of weekly mean log-H-index weighted by inverse variance. RESULTS: In +/- 3 months of data flanking video distribution (n = 137 241 collections), where overall impact was strongest, H-index trajectory (change in units per week) decreased immediately following video distribution (-5.7% / week, p < 0.01). This was accompanied by a 22% drop in overall H-index from the week before to the week after video distribution (y-intercept change, or gap). There was also a small but significant overall decrease in the proportion of hemolyzed specimens (-0.3%, p < 0.01). These changes were not observed at all collection locations, and in fact increases occured at some locations. CONCLUSIONS: We developed a novel and convenient educational aid that, when distributed, was associated with beneficial changes in specimen hemolysis at hospital inpatient units and outpatient clinics. Including it in ongoing nursing education will fill a knowledge gap that may help to reduce specimen hemolysis.

How much is enough? Proposing achievement thresholds for core EPAs of graduating medical students in Canada.

PURPOSE: The transition towards Competency-Based Medical Education at the Cumming School of Medicine was accelerated by the reduced clinical time caused by the COVID-19 pandemic. The purpose of this study was to define a standard protocol for setting Entrustable Professional Activity (EPA) achievement thresholds and examine their feasibility within the clinical clerkship. METHODS: Achievement thresholds for each of the 12 AFMC EPAs for graduating Canadian medical students were set by using sequential rounds of revision by three consecutive groups of stakeholders and evaluation experts. Structured communication was guided by a modified Delphi technique. The feasibility/consequence models of these EPAs were then assessed by tracking their completion by the graduating class of 2021. RESULTS: The threshold-setting process resulted in set EPA achievement levels ranging from 1 to 8 across the 12 AFMC EPAs. Estimates were stable after the first round for 9 of 12 EPAs. 96.27% of EPAs were successfully completed by clerkship students despite the shortened clinical period. Feasibility was predicted by the slowing rate of EPA accumulation overtime during the clerkship. CONCLUSION: The process described led to consensus on EPA achievement thresholds. Successful completion of the assigned thresholds was feasible within the shortened clerkship.[Box: see text].

Risk Factors and Outcomes of Bloodstream Infections Among People With Human Immunodeficiency Virus: A Longitudinal Cohort Study From 2000 to 2017.

Background: Bloodstream infections (BSIs) among people with human immunodeficiency virus (PWH) remain a poorly studied source of morbidity and mortality. We characterize the epidemiology, microbiology, and clinical outcomes including reinfection, hospitalization, and mortality rates of both community-acquired and hospital-acquired BSI in PWH. Methods: We identified all BSI, between January 1, 2000 and December 31, 2017 in PWH in care at Southern Alberta Clinic, by linking data from laboratory and clinical databases. Crude incidence rates per 1000 person-years for BSI and death were calculated. Cox proportional hazards models estimated crude and adjusted hazard ratios (HRs) and 95% confidence intervals (CIs) to conduct a risk factor analysis of BSI in PWH. Logistic regression models with generalized estimating equations estimated crude and adjusted odds ratios (aORs) to identify characteristics associated with 1-year mortality after BSI. Results: Among 2895 PWH, 396 BSI episodes occurred in 228 (8%) PWH. There were 278 (72%) Gram-positive and 109 (28%) Gram-negative BSI. People with human immunodeficiency virus with lower CD4 nadirs, higher Charlson comorbidity indices, and hepatitis C virus were at highest risk for BSI. Long-term all-cause mortality was greater in those experiencing BSI (HR, 5.25; 95% CI, 4.21-6.55). CD4 count <200 cells/mm3 measured closest to the time of BSI was associated with 1-year mortality after BSI (aOR, 3.88; 95% CI, 1.78-8.46). Repeat episodes (42%) and polymicrobial BSI (19%) were common. Conclusions: Bloodstream infections continue to occur at an elevated rate among PWH with high reoccurrence rates and associated morbidity and mortality. To risk stratify and develop targeted interventions, we identified PWH at greatest risk for BSI. People with human immunodeficiency virus with low immunity at the time of BSI are at highest risk of poor outcomes.

Variation in Patient-Described Barriers to and Facilitators of Diabetes Management by Individual-Level Characteristics: A Cross-Sectional, Open-Ended Survey.

This study analyzed patient-described barriers and facilitators related to diabetes management, focusing on how these differ by glycemia and across individual characteristics. A cross-sectional telephone survey was conducted with adult patients with diabetes in Alberta, Canada, asking two open-ended questions to describe the most helpful and difficult components of their diabetes management. Responses were analyzed using directed content analysis using the Theoretical Domains Framework as a template. The most frequently cited facilitator was care context and information, and the most frequently cited barriers were cognitive challenges and structural barriers, with patient-perceived barriers and facilitators varying by individual-level factors.

Using a systematic approach to strategic innovation in laboratory medicine to bring about change.

There is a growing mismatch with regard to demand, supply, and affordability in healthcare systems in developed countries. Innovation is required to address this, but roadmaps for innovation in laboratory medicine are largely lacking. Advances in process and instrument digitization are driving a revolution in medical laboratory practice but changes are not strategically focused on improved patient care. Laboratory services therefore largely remain transactional so that customer access and experience are suboptimal, especially for vulnerable populations. Laboratory medicine must be integrated back into clinical care pathways, thereby transforming services to be more responsive to end-user needs. Healthcare trends show that patients, physicians, and allied healthcare professionals will increasingly dictate what and how services are provided. Laboratories will be pressed to restructure to address these healthcare trends. Since the primary goal of ambulatory practice is to prevent expensive hospital admissions for patients with complex chronic diseases, specific services (e.g. ambulatory clinics, surgeries, deliveries, procedures) that could be safely provided in the community are moving out of acute care hospitals. This review addresses the existing barriers to innovation faced by medical/scientific and managerial services as well as outlines a systematic approach used by other industries to bring about transformative change. Enabling disruptive innovation that improves the clinical and economic effectiveness of laboratory practice is critical to sustain clinically relevant services as an essential cornerstone of patient care within the healthcare systems of developed countries.

Serum 25-hydroxyvitamin D and risk of cancer in a large community population under investigation for cardiovascular disease: a retrospective cohort study.

OBJECTIVES: It remains unclear whether vitamin D status is related to cancer risk. We examined this relationship using laboratory, administrative and survey data. DESIGN: Retrospective cohort study. SETTING: All care settings within Calgary, Alberta, Canada and surrounding rural communities. PARTICIPANTS: Patients tested for serum 25-hydroxyvitamin D from 2009 to 2013 without a past cancer diagnosis but with an ECG and body mass index ±3 months from testing were included. Age, sex, mean hours of daylight during month of testing were linked to census dissemination area-level indicators of socioeconomic status measured in 2011. PRIMARY AND SECONDARY OUTCOME MEASURES: Hospital discharge diagnoses for any cancer, major cancer (colorectal, breast, lung, prostate, skin) and other cancers >3 months from testing from 2009 to 2016. Cox proportional hazard models were used to examine associations with incident cancer after adjusting for potential confounders. Interactions were tested using multiplicative terms. RESULTS: Among 72 171 patients, there were 3439 cancer diagnoses over a median of 5.9 years. After adjustment, increasing quartile of serum 25-OH vitamin D was significantly associated with an increased risk of any cancer and major cancer, however this was completely driven by an increased risk of skin cancer (Q4 vs Q1: HR=2.56, 95% CI 1.70 to 3.86, p for linear trend <0.01). This association was strengthened among individuals residing in communities with higher proportions of non-citizens, recent immigrants, visible (non-white) minorities and those not speaking an official Canadian language (English or French) at home. CONCLUSIONS: Higher vitamin D status was associated with a greater risk of skin cancer in a large community population under investigation for cardiovascular disease. This association was likely due to sun exposure and may be modified by community variation in vitamin D supplementation.

A longitudinal population-based study of predictors of mortality from bloodstream infections in Calgary, Alberta, Canada.

Aim: To study the predictors of mortality from nine major pathogens causing approximately 70% of cases over a 7-year period. Materials & methods: A population-based surveillance cohort of all adult and pediatric patients in the Calgary Zone with an initial episode of bloodstream infections (BSI). Results: The 1-year mortality was 29.2% among 9524 patients (5164 males [54%]). Incidence rates for BSI increased annually to 119.7/100,000 persons by 2016. Distinct survival curves were found for each specific pathogen. Age, comorbidity burden and infecting organism were significantly associated with increased hazard of death. No relationship occurred between the time to positivity for blood cultures and overall mortality. Conclusion: BSI has a high mortality, but overall survival depends on underlying host health and the type of pathogen acquired.

The life of a laboratory requisition form: Patient compliance with clinical laboratory testing in a Canadian primary care health region.

OBJECTIVE: To examine patient laboratory testing compliance by tracking time to submission of laboratory requisitions in Southern Alberta, Canada as part of a quality improvement initiative. METHODS: Data was collected retrospectively from patients from the Chinook Primary Care Network in Alberta, Canada, who received a laboratory requisition consisting of a complete blood count (CBC) test order between September 1, 2016 and August 31, 2017. To allow for all laboratory requisitions created to be submitted within one year, the study collection period was from September 1, 2016 to August 31, 2018. Patient age, sex, and dates of laboratory requisition creation and submission were collected. The days-to-test-submission served as a marker of compliance. Association of age, sex, and clinic location with time to laboratory requisition completion was determined using Cox regression analysis. RESULTS: During the study period, 70.4% (n = 1607) of laboratory requisitions created were completed within one year, and over half (50.5%) of the laboratory requisitions ordered were completed within two weeks. There were no significant associations between time to laboratory requisition submission and sex or clinic locations (P > 0.05), but there were significant associations between patients who were 20-49 or 70-79 and increased laboratory requisition compliance (P < 0.05). However, 26.0% of the laboratory requisitions created were not submitted at all. CONCLUSIONS: This was the first study that quantified the proportion and timing of laboratory requisitions that were submitted by patients in a primary care setting. Community patients should be engaged and educated regarding the importance of complying with their physician-ordered laboratory requests in a timely manner.

The mean abnormal result rates of laboratory tests ordered in the emergency department: shooting percentage insights from a multi-centre study.

OBJECTIVE: While there is concern about excessive laboratory test ordering in the ED, it is difficult to quantify the problem. One solution involves the Mean Abnormal Result Rate (MARR), which is the proportion of tests ordered that return abnormal results. The primary objective of this study was to calculate MARR scores, and factors associated with MARR scores, for tests ordered between April 2014 and March 2019 at adult EDs in Calgary. METHODS: Administrative data were obtained for 40 laboratory tests that met selection criteria. One possible MARR correlate, physician experience, was quantified for 209 ED physicians as number of years since licensure. Analyses were descriptive where appropriate for whole-population data. RESULTS: The condensed dataset comprised 3,395,312 test results on 415,653 unique patients. The aggregate 5-year MARR score was 40.1%. The highest per-test score was for BNP (80.5%), while the lowest was for glucose (7.9%). MARR scores were higher for nurse-initiated orders than for physician-initiated orders (44.7% vs. 38.1%). The MARR score correlated inversely with number of tests per order (r = - 0.90; 95% confidence interval [CI] - 0.65 to - 0.94; p < 0.001) and directly with physician experience (r = 0.28, 95% CI 0.20 to 0.27; p < 0.001). CONCLUSION: This is the first study to measure MARR scores in an ED setting. While lower scores (close to 5%) are less optimal in principle, ideal scores will depend on the clinical context in which tests are used. However, once departmental benchmarks are established, MARR score-monitoring allows efficient tracking of ordering practices across millions of tests.

RéSUMé: OBJECTIFS: Bien que l'on s'inquiète d'une commande excessive de tests de laboratoire dans le service d'urgence, il est difficile de quantifier le problème. Une solution implique le taux de résultat anormal moyen (MARR), qui est la proportion de tests commandés qui renvoient des résultats anormaux. L'objectif principal de cette étude était de calculer les scores MARR, et les facteurs associés aux scores MARR, pour les tests commandés entre avril 2014 et mars 2019 dans les services d'urgence pour adultes à Calgary. LES MéTHODES: Des données administratives ont été obtenues pour 40 tests de laboratoire répondant aux critères de sélection. Une corrélation possible du MARR, soit l'expérience des médecins, a été quantifié pour 209 médecins urgentistes en nombre d'années depuis l'autorisation d'exercer. Les analyses étaient descriptives, le cas échéant, pour les données sur l'ensemble de la population. RéSULTATS: L'ensemble de données condensé comprenait 3,395,312 résultats de tests sur 415 653 patients uniques. Le score MARR global à 5 ans était de 40,1%. Le score le plus élevé par test était pour le peptide cérébral natriurétique BNP (80,5%), tandis que le plus bas était pour le glucose (7,9%). Les scores de MARR sont plus élevés pour les ordonnances initiées par les infirmières que pour les ordonnances initiées par les médecins (44,7% contre 38,1%). Le score MARR s'est corrélé inversement avec le nombre de tests par ordre (r = − 0,90; intervalle de confiance de 95 % [IC] − 0.65 to − 0,94; p < 0.001) et directement avec l'expérience du médecin (r = 0.28, IC à 95 % 0.20­0.27; p < 0.001). CONCLUSIONS: Il s'agit de la première étude à mesurer les scores du MARR dans un contexte de service d'urgence. Bien que des scores inférieurs (près de 5%) sont moins optimaux en principe, les scores idéaux dépendront du contexte clinique dans lequel les tests sont utilisés. Cependant, une fois les points de repère ministériels établis, la surveillance des scores du MARR permet de suivre efficacement les pratiques de commande sur des millions de tests.

Age and sex-specific incidence rates of group A streptococcal pharyngitis between 2010 and 2018: a population-based study.

Aim: Group A streptococcus (GAS) pharyngitis is a common clinical infection with significant morbidity but remains understudied. Materials & methods: We sought to assess the rates of testing and incidence of GAS pharyngitis in Calgary, Alberta based on age and sex. Results: A total of 1,074,154 tests were analyzed (58.8% female, mean age 24.8 years) of which 16.6% were positive. Age-standardized testing and positivity was greatest in the 5-14 years age group and lowest in persons over 75 years. Females had greater rates of testing and positivity throughout. Testing rates (incidence rate ratios: 1.40, 95% CI: 1.39-1.41) and case rates (incidence rate ratios: 1.36, 95% CI: 1.33-1.39) increased over time. Conclusion: Future studies should focus on evaluating disparities in testing and treatment outcomes to optimize the approach to this infection.

A multicenter study investigating SARS-CoV-2 in tertiary-care hospital wastewater. viral burden correlates with increasing hospitalized cases as well as hospital-associated transmissions and outbreaks.

SARS-CoV-2 has been detected in wastewater and its abundance correlated with community COVID-19 cases, hospitalizations and deaths. We sought to use wastewater-based detection of SARS-CoV-2 to assess the epidemiology of SARS-CoV-2 in hospitals. Between August and December 2020, twice-weekly wastewater samples from three tertiary-care hospitals (totaling > 2100 dedicated inpatient beds) were collected. Hospital-1 and Hospital-2 could be captured with a single sampling point whereas Hospital-3 required three separate monitoring sites. Wastewater samples were concentrated and cleaned using the 4S-silica column method and assessed for SARS-CoV-2 gene-targets (N1, N2 and E) and controls using RT-qPCR. Wastewater SARS-CoV-2 as measured by quantification cycle (Cq), genome copies and genomes normalized to the fecal biomarker PMMoV were compared to the total daily number of patients hospitalized with active COVID-19, confirmed cases of hospital-acquired infection, and the occurrence of unit-specific outbreaks. Of 165 wastewater samples collected, 159 (96%) were assayable. The N1-gene from SARS-CoV-2 was detected in 64.1% of samples, N2 in 49.7% and E in 10%. N1 and N2 in wastewater increased over time both in terms of the amount of detectable virus and the proportion of samples that were positive, consistent with increasing hospitalizations at those sites with single monitoring points (Pearson's r = 0.679, P < 0.0001, Pearson's r = 0.799, P < 0.0001, respectively). Despite increasing hospitalizations through the study period, nosocomial-acquired cases of COVID-19 (Pearson's r = 0.389, P < 0.001) and unit-specific outbreaks were discernable with significant increases in detectable SARS-CoV-2 N1-RNA (median 112 copies/ml) versus outbreak-free periods (0 copies/ml; P < 0.0001). Wastewater-based monitoring of SARS-CoV-2 represents a promising tool for SARS-CoV-2 passive surveillance and case identification, containment, and mitigation in acute- care medical facilities.

Laboratory reporting of framingham risk score increases statin prescriptions in at-risk patients.

BACKGROUND: The under-utilization of cardiovascular preventative therapy with statins warrants novel interventions to optimize prescriptions in at-risk patients. We investigated the role of a laboratory generated Framingham Risk Score (FRS) provided to primary care clinicians in changing statin use in a primary care setting. METHODS: Data was acquired from the electronic medical records of 1573 anonymized patients undergoing routine lipid testing. Follow-up statin use and low-density lipoprotein cholesterol levels were obtained for 2 years post intervention. FRS parameters were entered into a laboratory information system, and provided to ordering physicians along with the cholesterol profile and the appropriate current Canadian Dyslipidemia treatment recommendation in a single report. Statin prescription rates following the intervention were compared with historical use 6 months prior to the study. RESULTS: A total of 1283 participants (mean age of 60 ± 11 years) had an FRS report and were considered for analysis. Two hundred individuals filled a statin prescription in the 6 months prior to their index lipid test, and an additional 84 filled a statin prescription following the intervention (42% increase). The relative and absolute increase in statin prescription was 47.3% and 13.6% in the high-risk group p < 0.001, 53.3% and 8.1% in the intermediate-risk group p < 0.001, and 17.0% and 1.42% in the low-risk group p = 0.008, respectively. CONCLUSION: The use of the laboratory reported FRS was associated with a significant increase in the rate of statin prescription across all risk groups. The expansion of FRS reporting across other health regions would improve cardiovascular risk prevention.

Factors Associated with Hyponatremia in Patients Newly Prescribed Citalopram: A Retrospective Observational Study.

BACKGROUND: Hyponatremia is a common and under-recognized adverse drug reaction of selective serotonin re-uptake inhibitor (SSRI) antidepressants. Despite its clinical importance, there are few large-scale studies on the factors associated with hyponatremia. OBJECTIVE: The aim of this study was to determine the incidence of hyponatremia and to identify patient factors associated with hyponatremia in a large, population-based cohort initiating new prescriptions for citalopram. METHODS: We included all patients with a new prescription for citalopram during 2010-2017, inclusive, with baseline and post-initiation serum sodium values available. Data were obtained from an Alberta Health Pharmacy database to identify new citalopram prescriptions. Laboratory values for patients with new prescriptions were obtained from linked Calgary Laboratory Services data. Incident hyponatremia was defined as serum sodium level < 135 mmol/L, following prescription initiation. Associations were determined by performing Cox regression with time-varying covariate analysis, with the development of hyponatremia as the dependent variable. RESULTS: A total of 19,679 patients with new prescriptions were identified; 12,842 females and 6837 males. The mean age was 55.48 years (SD 21.35). Of these patients, 3250 (16.5%) developed hyponatremia, 1996 (15.5% of) females and 1254 (18.3% of) males (p = 0.002). Cox regression showed significant associations of hyponatremia with lower baseline sodium (HR 0.788), older age (HR 1.029), thiazide diuretic use (HR 1.141), and male sex (HR 1.168). Pharmaceutical manufacturer or strength of citalopram did not have significant effects on the development of hyponatremia. CONCLUSION: This study provides additional data on the predictors of hyponatremia among patients initiating citalopram therapy. We report a 16.5% incidence of hyponatremia after starting citalopram treatment, and significant new findings include a higher incidence in males. This is the first published incidence of hyponatremia following the initiation of citalopram treatment across all ages in Canada.

Serum 25-hydroxyvitamin D, cardiovascular risk markers, and incident cardiovascular disease in a high risk community population.

BACKGROUND: It is unclear whether vitamin D status is related to cardiovascular risk beyond that explained by conventional risk markers. We examined the relationship between serum 25-hydroxy (OH) vitamin D and incident cardiovascular disease (CVD; heart attack/stroke) after adjusting for individual- and community-level covariates from laboratory, administrative and survey data. METHODS: Patients receiving their first 25-OH vitamin D test in Calgary, Alberta from 2009 to 2013 without a past CVD diagnosis but an electrocardiogram and body mass index (BMI) +/- 3 months from testing were included. The following was merged to this data: first results for laboratory-measured CVD risk markers (lipid profile, fasting plasma glucose, and HbA1c) measured +/- 3 months from testing; Census Dissemination Area (CDA)-level indicators of socioeconomic status (SES) in 2011; and CVD diagnoses > 3 months from testing between 2009 and 2016. Linear and Poisson regression were used to examine associations between 25-OH vitamin D quartile and covariates, and Cox proportional hazard models were used to examine associations with incident CVD before and after adjusting for covariates. RESULTS: Among 72 348 patients, there were 1898 CVD events over a median of 6.0 years. Increasing quartile of 25-OH vitamin D was associated with improved lipid and glycemic profiles (p < 0.01), higher proportion of CDA-level indicators of high SES (p < 0.01), and a lower risk of CVD (Q4 vs Q1: HR: 0.72, 95% CI: 0.63-0.81, p for trend < 0.01) after adjusting for age, sex and average daily hours of sunlight during month of testing. The association with CVD was unchanged after adjusting for BMI, slightly attenuated after adjusting for SES but completely abolished after adjusting for laboratory-measured cardiovascular risk markers. CONCLUSIONS: Vitamin D status likely offers no additional information on CVD risk over conventional laboratory-measured risk markers.

Electronic Medical Record-Based Case Phenotyping for the Charlson Conditions: Scoping Review.

BACKGROUND: Electronic medical records (EMRs) contain large amounts of rich clinical information. Developing EMR-based case definitions, also known as EMR phenotyping, is an active area of research that has implications for epidemiology, clinical care, and health services research. OBJECTIVE: This review aims to describe and assess the present landscape of EMR-based case phenotyping for the Charlson conditions. METHODS: A scoping review of EMR-based algorithms for defining the Charlson comorbidity index conditions was completed. This study covered articles published between January 2000 and April 2020, both inclusive. Embase (Excerpta Medica database) and MEDLINE (Medical Literature Analysis and Retrieval System Online) were searched using keywords developed in the following 3 domains: terms related to EMR, terms related to case finding, and disease-specific terms. The manuscript follows the Preferred Reporting Items for Systematic reviews and Meta-analyses extension for Scoping Reviews (PRISMA) guidelines. RESULTS: A total of 274 articles representing 299 algorithms were assessed and summarized. Most studies were undertaken in the United States (181/299, 60.5%), followed by the United Kingdom (42/299, 14.0%) and Canada (15/299, 5.0%). These algorithms were mostly developed either in primary care (103/299, 34.4%) or inpatient (168/299, 56.2%) settings. Diabetes, congestive heart failure, myocardial infarction, and rheumatology had the highest number of developed algorithms. Data-driven and clinical rule-based approaches have been identified. EMR-based phenotype and algorithm development reflect the data access allowed by respective health systems, and algorithms vary in their performance. CONCLUSIONS: Recognizing similarities and differences in health systems, data collection strategies, extraction, data release protocols, and existing clinical pathways is critical to algorithm development strategies. Several strategies to assist with phenotype-based case definitions have been proposed.

Cervical Screening Practices and Outcomes for Young Women in Response to Changed Guidelines in Calgary, Canada, 2007-2016.

OBJECTIVE: The aim of the study was to describe temporal trends in screening and outcomes for women, after changes in guidelines in Alberta, Canada, that raised starting age to 21 years, then to 25 years of age, and reduced frequency to 3 yearly. MATERIALS AND METHODS: Calgary Laboratory Information System data were used to examine screening rates, follow-up procedures, and cancer among women 10-29 years from 2007 to 2016 in the whole population of Calgary. Interrupted time-series analyses were used to assess changes in screening and subsequent diagnostic procedures over the 10-year period. RESULTS: Annual screening rates dropped by approximately 10% at all ages older than 15 years after the 2009 Alberta cervical cancer screening guidelines, followed by a steady decrease. Further change continued subsequent to minimal apparent effect of the 2013 Canadian Task Force on Preventive Health Care guidelines. The rates of abnormal test results decreased in concert with decreased screening. No increases in cervical intraepithelial neoplasia 1, cervical intraepithelial neoplasia 2/3, or invasive cervical cancer rates were observed after reduced testing. CONCLUSIONS: The largest decrease in screening and follow-up procedures occurred in the period immediately after implementation of 2009 Alberta screening guidelines. The number of consequent procedures also decreased in proportion to decreased screening, but there was no increase in cancer rates. Starting screening at the age of 25 years and reducing intervals seem to be safe.