Predicting the target specialty of referral notes to estimate per-specialty wait times with machine learning.

BACKGROUND: Currently, in Canada, existing health administrative data and hospital-inputted portal systems are used to measure the wait times to receiving a procedure or therapy after a specialist visit. However, due to missing and inconsistent labelling, estimating the wait time prior to seeing a specialist physician requires costly manual coding to label primary care referral notes. METHODS: In this work, we represent the notes using word-count vectors and develop a logistic regression machine learning model to automatically label the target specialist physician from a primary care referral note. These labels are not available in the administrative system. We also study the effects of note length (measured in number of tokens) and dataset size (measured in number of notes per target specialty) on model performance to help other researchers determine if such an approach may be feasible for them. We then calculate the wait time by linking the specialist type from a primary care referral to a full consultation visit held in Ontario, Canada health administrative data. RESULTS: For many target specialties, we can reliably (F1Score ≥ 0.70) predict the target specialist type. Doing so enables the automated measurement of wait time from family physician referral to specialist physician visit. Of the six specialties with wait times estimated using both 2008 and 2015 data, two had a substantial increase (defined as a change such that the original value lay outside the 95% confidence interval) in both median and 75th percentile wait times, one had a substantial decrease in both median and 75th percentile wait times, and three has non-substantial increases. CONCLUSIONS: Automating these wait time measurements, which had previously been too time consuming and costly to evaluate at a population level, can be useful for health policy researchers studying the effects of policy decisions on patient access to care.

The receipt of information by family physicians about their patient's emergency department visits: a record linkage study of electronic medical records to health administrative data.

BACKGROUND: Canadians are known to be frequent users of emergency department (ED) care. However, the exchange of information from ED visits to family physicians (FPs) is not well known. Our objectives were to determine whether Canadian FPs received information about their patient's ED visit and the patient characteristics related to the receipt of ED information. METHODS: This study was a descriptive record linkage study of FP Electronic Medical Record (EMR) data linked to health administrative data. Our study cohort included patients who had at least one ED visit in 2010 or 2015 in Ontario, Canada. An ED visit could include a transfer to or from another ED. The receipt of information from an ED note was examined in relation to patient age, sex, neighbourhood income quintiles, rurality and comorbidity. RESULTS: There were 26,609 patients in 2010 and 50,541 patients in 2015 with at least one ED visit. In 2010, 53.3% of FPs received an ED note for patients having a single ED visit compared to 41.0% in 2015. For patients with multiple ED visits, 58.2% of FPs received an ED note in 2010 compared to 45.7% in 2015. FPs were more likely to receive an ED note for patients not living in low income neighbourhoods, older patients, patients living in small urban areas and for patients having moderate comorbidity. FPs were less likely to receive a note for patients living in rural areas. CONCLUSIONS: Community-based FPs are more likely to get information after an ED visit for their older and sicker patients. However, FPs do not receive any information from EDs for over half their patients. Electronic health record technologies and their adoption by ED providers need to improve the seamless transfer of information about the care provided in EDs to FPs in the community.

The quality of diabetes care among cancer survivors: a retrospective cohort study.

BACKGROUND: As cancer survivorship continues to improve, management of co-morbid diabetes has become an increasingly important determinant of health outcomes for people with cancer. This study aimed to compare indicators of diabetes quality of care between people with diabetes and without a history of cancer. METHODS: We used the Electronic Medical Record Administrative data Linked Database (EMRALD), a database of Ontario primary care EMR charts linked to administrative data, to identify people with diabetes and at least 1 year follow-up. Persons with a history of cancer were matched 1:2 on age, sex and diabetes duration to those without cancer. We compared recommended diabetes quality of care indicators between persons with and without cancer using a matched cohort analysis. RESULTS: Among 229,627 people with diabetes, we identified 2275 people with cancer and 4550 matched controls; 86.5% had diabetes diagnosed after cancer. Compared to controls, cancer people with diabetes were significantly less likely to receive ACE inhibitors or angiotensin receptor blockers (OR 0.75 [95% CI 0.64-0.89]), receive statin therapy if age 50-80 years (OR 0.79 [95% CI 0.68-0.92]) and achieve an LDL cholesterol level <2.0 mmol/L (OR 0.82 [95% CI 0.74-0.91]). There were no differences in recommended clinical testing or achieving A1C and blood pressure targets between groups. CONCLUSION: Cancer survivors with diabetes are less likely to receive recommended cardiovascular risk-reducing therapies compared to people with diabetes without cancer of similar age, sex and diabetes duration. Further studies are warranted to determine if these associations are linked to worse survival, cardiovascular outcomes and quality of life.