Search

• DCNeT, a comorbidity network-based temporal deep learning framework is presented. • Two attention mechanisms are proposed to represent patients' hospitalizations. • Time intervals between hospitalizations are modeled to fully use time information. • DCNeT outperforms the state-of-the-art methods on real-world datasets. • Illustrate an important use of administrative health data. Patients with mental disorders (MDs) are at higher subsequent risk of developing cardiovascular diseases (CVDs) than the general population. Early identification of cardiovascular risk in patients with MDs is beneficial for timely intervention and reducing disease burden. Recently, deep learning approaches have been increasingly applied in CVDs risk prediction. However, these methods have three major issues: 1) mostly relying on multiple types of clinical data, 2) not sufficiently mining and utilizing comorbidity patterns hidden in complex correlations among various diseases, and 3) not fully leveraging the time information, including the irregular intervals. To address these issues, we propose a disease comorbidity network-based temporal deep learning framework (DCNeT) to predict the subsequent CVDs risk for patients with MDs based on routinely collected administrative health data. Firstly, to identify the comorbidity patterns of MDs, we construct a disease comorbidity network (DCN) for MDs and apply graph embedding method to generate disease embeddings for each disease in the DCN. Then, a code attention mechanism is proposed to obtain the weight of each disease which is embedded into dense vectors based on the structure of DCN. We present a view attention mechanism to compute the attention weights of different types of features including disease embeddings, basic features, and disease indicators for generating the final representations of patients' hospitalizations. Furthermore, to fully utilize the information on the irregular time intervals between hospitalizations, a time encoding module is designed, and the time-aware LSTM is adopted to model the irregular time intervals and capture the temporal patterns of patients' hospitalizations. The experimental results show that DCNeT outperforms the state-of-the-art methods, with the area under the receiver operating characteristic curve of 0.7658, 0.8143, 0.8110, and 0.7839 on four datasets, respectively. The ablation experiments further demonstrate that each module of DCNeT, including the code attention, view attention, and time encoding module, contributes to its superior performance, with average improvements of 1.20 %, 1.65 %, and 1.13 % in accuracy, respectively. Our DCNeT could be utilized as an efficient framework for identifying high-risk groups of CVDs among patients with MDs that may benefit from screening and preventive strategies. [ABSTRACT FROM AUTHOR]

To examine changes in the prescribing of end-of-life symptom management medications in long-term care (LTC) homes during the COVID-19 pandemic. Retrospective cohort study using routinely collected health administrative data in Ontario, Canada. We included all individuals who died in LTC homes between January 1, 2017, and March 31, 2021. We separated the study into 2 periods: before COVID-19 (January 1, 2017, to March 17, 2020) and during COVID-19 (March 18, 2020, to March 31, 2021). For each LTC home, we measured the percentage of residents who died before and during COVID-19 who had a subcutaneous symptom management medication prescription in their last 14 days of life. We grouped LTC homes into quintiles based on their mean prescribing rates before COVID-19, and examined changes in prescribing during COVID-19 and COVID-19 outcomes across quintiles. We captured 75,438 LTC residents who died in Ontario's 626 LTC homes during the entire study period, with 19,522 (25.9%) dying during COVID-19. The mean prescribing rate during COVID-19 ranged from 46.9% to 79.4% between the lowest and highest prescribing quintiles. During COVID-19, the mean prescribing rate in the lowest prescribing quintile increased by 9.6% compared to before COVID-19. Compared to LTC homes in the highest prescribing quintile, homes in the lowest prescribing quintile experienced the highest proportion of COVID-19 outbreaks (73.4% vs 50.0%), the largest mean outbreak intensity (0.27 vs 0.09 cases/bed), the highest mean total days with a COVID-19 outbreak (72.7 vs 24.2 days), and the greatest proportion of decedents who were transferred and died outside of LTC (22.1% vs 8.6%). LTC homes in Ontario had wide variations in the prescribing rates of end-of-life symptom management medications before and during COVID-19. Homes in the lower prescribing quintiles had more COVID-19 cases per bed and days spent in an outbreak. [ABSTRACT FROM AUTHOR]

Medications are often needed to manage distressing end-of-life symptoms (eg, pain, agitation). In this study, we describe the variation in prescribing rates of symptom relief medications at the end of life among long-term care (LTC) decedents. We evaluate the extent these medications are prescribed in LTC homes and whether prescribing rates of end-of-life symptom management can be used as an indicator of quality end-of-life care. Retrospective cohort study using administrative health data. LTC decedents in all 626 publicly funded LTC homes in Ontario, Canada, between January 1, 2017, and March 17, 2020. For each LTC home, we measured the percent of decedents who received 1+ prescription(s) for a subcutaneous end-of-life symptom management medication ("end-of-life medication") in their last 14 days of life. We then ranked LTC homes into quintiles based on prescribing rates. We identified 55,916 LTC residents who died in LTC. On average, two-thirds of decedents (64.7%) in LTC homes were prescribed at least 1 subcutaneous end-of-life medication in the last 2 weeks of life. Opioids were the most common prescribed medication (overall average prescribing rate of 62.7%). LTC homes in the lowest prescribing quintile had a mean of 37.3% of decedents prescribed an end-of-life medication, and the highest quintile mean was 82.5%. In addition, across these quintiles, the lowest prescribing quintile had a high average (30.3%) of LTC residents transferred out of LTC in the 14 days compared with the highest prescribing quintile (12.7%). Across Ontario's LTC homes, there are large differences in prescribing rates for subcutaneous end-of-life symptom relief medications. Although future work may elucidate why the variability exists, this study provides evidence that administrative data can provide valuable insight into the systemic delivery of end-of-life care. [ABSTRACT FROM AUTHOR]

Background: The lack of epidemiological and clinical data is a major obstacle in health service planning for rare diseases. Patient registries are examples of real-world data that may fill the information gap.Objective: We describe the Rare Disease Registry of the Friuli Venezia Giulia region of Italy and its potential for research and health planning.Methods: The Rare Disease Registry data were linked with information on mortality, hospital discharges, ambulatory care, and drug prescriptions contained in administrative databases. All information is anonymous, and data linkage was based on a stochastic key univocal for each patient. Average annual costs owing to hospitalizations, outpatient care, and medications were estimated.Results: Implementation of the Registry started in 2010, and 4250 participants were registered up to 2017. A total of 2696 patients were living in the region as of January 1, 2017. The overall raw prevalence of rare diseases was 22 per 10,000 inhabitants, with higher prevalence in the pediatric population. The most common disease groups were congenital malformations, chromosomal and genetic syndromes, and circulatory and nervous diseases. In 2017, 30 patients died, 648 were hospitalized, and 2355 received some type of ambulatory care. The total annual estimated cost was approximately €6.5 million, with great variability in the average patient cost across diseases.Conclusions: The possibility of following the detailed real-world care experience of patients with each specific rare disease and assessing the costs related to each step in their care path represents a unique opportunity to identify inefficiencies, optimize care, and reduce waste of resources. [ABSTRACT FROM AUTHOR]

Highlights • Little is known about the use of the new oral MS drugs in clinical practice. • Uptake for an oral MS drug was 16.6/1000 person-years over the five-year study period. • Uptake and use of the new oral DMTs increased substantially from 2011–2015. • Uptake and use of the oral MS drugs were similar for men and women. • However, subjects with a higher (vs. lower) socioeconomic status had a higher uptake. Abstract Background: Little is known about the use of the new oral disease-modifying therapies (DMTs: fingolimod, dimethyl fumarate, teriflunomide) for multiple sclerosis (MS) in clinical practice. We describe their rate of uptake, and their use as compared to the established first-generation (beta-interferon and glatiramer acetate) and second-generation (natalizumab and alemtuzumab) parenteral DMTs. Methods: Persons with MS (PwMS) were identified using health administrative data from British Columbia, Canada (2011-15). We described the uptake and use of the DMTs with Poisson and Log-Binomial regression, adjusting for sex, age, and socioeconomic status (SES). Results: Among 14,148 PwMS, 2,821 were exposed to any DMT over the study period. In total, 1,019 initiated an oral DMT, a rate of 16.6/1,000 person-year, and 83/1,000 PwMS filled at least one prescription for an oral DMT during the study period. The rate of uptake for oral DMTs, and the annual proportion of PwMS using an oral DMT or second-generation parenteral DMT increased over the 5 years, whereas use of the first-generation DMTs decreased by 30%. PwMS in the higher (vs. lower) SES group had a 19-44% higher rate of uptake for teriflunomide and fingolimod. Conclusion: The uptake and use of the oral DMTs increased substantially over the first 2 to 5 years after their introduction. These recent changes highlight the importance of monitoring the risks and benefits in the real world. [ABSTRACT FROM AUTHOR]

There is a critical need for high-quality population-level data related to the health of individuals with intellectual disabilities. For more than 15 years Special Olympics International has been conducting free Healthy Athletes screenings at local, national and international events. The Healthy Athletes database is the largest known international database specifically on the health of people with intellectual disabilities; however, it is relatively under-utilized by the research community. A consensus meeting with two dozen North American researchers, stakeholders, clinicians and policymakers took place in Toronto, Canada. The purpose of the meeting was to: 1) establish the perceived utility of the database, and 2) to identify and prioritize 3-5 specific priorities related to using the Healthy Athletes database to promote the health of individuals with intellectual disabilities. There was unanimous agreement from the meeting participants that this database represents an immense opportunity both from the data already collected, and data that will be collected in the future. The 3 top priorities for the database were deemed to be: 1) establish the representativeness of data collected on Special Olympics athletes compared to the general population with intellectual disabilities, 2) create a scientific advisory group for Special Olympics International, and 3) use the data to improve Special Olympics programs around the world. The Special Olympics Healthy Athletes database includes data not found in any other source and should be used, in partnership with Special Olympics International, by researchers to significantly increase our knowledge and understanding of the health of individuals with intellectual disabilities. [ABSTRACT FROM AUTHOR]

Background Large administrative databases may prove useful to assess epilepsy-related comorbidity and mortality. Despite their increased use, their validity as data source in epilepsy is yet under-ascertained. Methods Achmea is a large Dutch health insurance company covering about 25% of the population. We performed a retrospective cohort study using data from the Achmea Health Insurance Database (AHID) over the period 2006–2009. To assess the validity of epilepsy codes in the AHID, we randomly invited 1000 individuals (age 18–75 years insured by Achmea), attending an epilepsy centre or a district hospital during 2006–2009, to participate. Informed consent was provided and 293 were eligible for inclusion. We compared the diagnostic codes for epilepsy in AHID with the diagnosis in their case-notes (reference standard). As additional measure of validity, we compared prevalence of epilepsy codes in AHID (based on anonymized data of all 26.297 subjects with this code in AHID) with epilepsy prevalence rates in the general Dutch population to estimate an age-specific standardized prevalence ratio. Results We identified 293 participants with an epilepsy code in AHID. The majority (278) of them had a definite or possible diagnosis of epilepsy in the case-notes; i.e. a positive predictive value of 0.95 (95% CI 0.92–0.97). The overall prevalence of epilepsy codes in the AHID was slightly higher than the putative prevalence in the general Dutch population (7.4/1.000 vs. 6.8/1.000) with a Standardized Prevalence Ratio of 1.08 (95% CI: 1.08–1.09). Conclusions Our findings demonstrate the validity of AHID data for a diagnosis of epilepsy and confirm previous work on using administrative data for epilepsy research. [ABSTRACT FROM AUTHOR]

Incidence and prevalence estimates in myasthenia gravis have varied widely. Recent studies based on administrative health data have large sample sizes but lack rigorous validation of MG cases, and have not examined the North American population. Our aim was to explore trends in MG incidence and prevalence for the years 1996–2013 in the province of Ontario, Canada (population 13.5 million). We employed a previously validated algorithm to identify MG cases. Linking with census data allowed for the calculation of crude- and age/sex-standardized incidence and prevalence rates for the years 1996–2013. The regional distribution of MG cases throughout the province was examined. Mean age at the first myasthenia gravis encounter was 60.2 ± 17.1 years. In 2013, there were 3611 prevalent cases in Ontario, and the crude prevalence rate was 32.0/100,000 population. Age- and sex-standardized prevalence rates rose consistently over time from 16.3/100,000 (15.4–17.1) in 1996 to 26.3/100,000 (25.4–27.3) in 2013. Standardized incidence rates remained stable between 1996 (2.7/100,000; 95% CL 2.3–3.0) and 2013 (2.3/100,000; 2.1–2.6). Incidence was highest in younger women and older men, and geographic variation was evident throughout the province. In conclusion, this large epidemiological study shows rising myasthenia gravis prevalence with stable incidence over time, which is likely reflective of patients living longer, possibly due to improved disease treatment. Our findings provide accurate information on the Canadian epidemiology of myasthenia gravis and burden for health care resources planning for the province, respectively. [ABSTRACT FROM AUTHOR]

Background: Early childhood immunizations have been viewed as promoters of asthma development by stimulating a TH2-type immune response or decreasing microbial pressure, which shifts the balance between TH1 and TH2 immunity. Objective: Differing time schedules for childhood immunizations may explain the discrepant findings of an association with asthma reported in observational studies. This research was undertaken to determine whether timing of diphtheria, pertussis, tetanus (DPT) immunization has an effect on the development of childhood asthma by age 7 years. Methods: This was a retrospective longitudinal study of a cohort of children born in Manitoba in 1995. The complete immunization and health care records of cohort children from birth until age 7 years were available for analysis. The adjusted odds ratio for asthma at age 7 years according to timing of DPT immunization was computed from multivariable logistic regression. Results: Among 11, 531 children who received at least 4 doses of DPT, the risk of asthma was reduced to ½ in children whose first dose of DPT was delayed by more than 2 months. The likelihood of asthma in children with delays in all 3 doses was 0.39 (95% CI, 0.18-0.86). Conclusion: We found a negative association between delay in administration of the first dose of whole-cell DPT immunization in childhood and the development of asthma; the association was greater with delays in all of the first 3 doses. The mechanism for this phenomenon requires further research. [Copyright &y& Elsevier]

Abstract: Objective: To derive and validate an International Classification of Diseases-10 (ICD-10) version of the Ontario Acute Myocardial Infarction (AMI) mortality prediction rules, used to adjust for case-mix differences in studies of AMI patients using administrative data. Study Design and Setting: We linked the records of all Ontario patients admitted with AMI (2002–2004) with mortality data. The original ICD-9 codes were mapped to ICD-10-CA (Canada) codes using both a translation produced by coding experts and a manual search of codes; the final codes were determined by consensus. Comorbidity prevalence and mortality rates were calculated. Multivariable logistic regression models were used to predict 30-day and 1-year mortality and the C-statistic was used to evaluate the discrimination of the models. Results: We identified 37,271 AMI patients. The most common comorbidities were heart failure and dysrhythmias; 30-day and 1-year mortality rates were 12.3% and 21.8%, respectively; and mortality rates were highest among patients with shock, cancer, and acute renal failure. The C-statistics were 0.77 and 0.80, compared with 0.78 and 0.79 in the ICD-9 version, for 30-day and 1-year mortality, respectively. Conclusion: An ICD-10 version of the AMI mortality prediction rules predicted 30-day and 1-year mortality as well as the original ICD-9 version. [Copyright &y& Elsevier]