Original Research
Prevalence and Epidemiology of Diabetes in Canadian Primary Care Practices: A Report from the Canadian Primary Care Sentinel Surveillance Network

https://doi.org/10.1016/j.jcjd.2014.02.030Get rights and content

Abstract

Objective

The Canadian Primary Care Sentinel Surveillance Network (CPCSSN) is a large, validated national primary care Electronic Medical Records (EMR)-based database. Our objective was to describe the epidemiology of diabetes in this Canadian sample.

Methods

We analyzed the records of 272 469 patients10 years of age and older, with at least 1 primary care clinical encounter between January 1, 2011, and December 31, 2012. We calculated the age-gender standardized prevalence of diabetes. We compared health care utilization and comorbidities for 7 selected chronic conditions in patients with and without diabetes. We also examined patterns of medication usage.

Results

The estimated population prevalence of diabetes was 7.6%. Specifically, we studied 25 425 people with diabetes who had at least 1 primary care encounter in 2 years. On average, patients with diabetes had 1.42 times as many practice encounters as patients without diabetes (95% CI 1.42 to 1.43, p<0.0001). Patients with diabetes had 1.29 times as many other comorbid conditions as those without diabetes (95% CI 1.27 to 1.31, p<0.0001). We found that 85.2% of patients taking hypoglycemic medications were taking metformin, and 51.8% were taking 2 or more classes of medications.

Conclusions

This study is the first national Canadian report describing the epidemiology of diabetes using primary care EMR-based data. We found significantly higher rates of primary care use, and greater numbers of comorbidities in patients with diabetes. Most patients were on first-line hypoglycemic medications. Data routinely recorded in EMRs can be used for surveillance of chronic diseases such as diabetes in Canada. These results can enable comparisons with other national EMR-based datasets.

Résumé

Objectif

Le Réseau canadien de surveillance sentinelle en soins primaires (RCSSSP) qui maintient une importante base de données nationales recueille et valide des données de soins primaires à partir de dossiers médicaux électroniques (DMÉ). Notre objectif était de décrire l’épidémiologie du diabète de cet échantillon canadien.

Méthodes

Nous avons analysé les dossiers de 272 469 patients de 10 ans et plus ayant eu au moins 1 intervention clinique en soins primaires entre le 1er janvier 2011 et le 31 décembre 2012. Nous avons calculé la prévalence standardisée du diabète selon l’âge et le sexe. Nous avons comparé l’utilisation des soins de santé et les comorbidités de 7 affections chroniques sélectionnées chez les patients souffrant ou non de diabète. Nous avons également examiné les profils d’utilisation des médicaments.

Résultats

L’estimation de la prévalence du diabète dans la population était de 7,6 %. Notamment, nous avons étudié 25 425 personnes souffrant de diabète qui avaient eu au moins 1 intervention en soins primaires en 2 ans. En moyenne, les patients souffrant de diabète avaient 1,42 fois plus d’interventions que les patients ne souffrant pas de diabète (IC à 95 % 1,42-1,43, p < 0,0001). Les patients souffrant de diabète avaient 1,29 fois plus d’autres comorbidités que ceux ne souffrant pas diabète (IC à 95 % 1,27-1,31, p < 0,0001). Nous avons constaté que 85,2 % des patients prenant des hypoglycémiants prenaient de la metformine, et que 51,8 % prenaient 2 classes ou plus de médicaments.

Conclusions

Cette étude constitue le premier rapport national canadien qui décrit l’épidémiologie du diabète en utilisant les données de soins primaires des DMÉ. Nous avons constaté des taux significativement plus élevés d’utilisation des soins primaires et un plus grand nombre de comorbidités chez les patients souffrant de diabète. La plupart des patients prenaient des hypoglycémiants de première intention. Au Canada, les données systématiquement enregistrées aux DMÉ peuvent être utilisées pour la surveillance des maladies chroniques comme le diabète. Ces résultats peuvent permettre de faire des comparaisons sur le plan national avec d’autres ensembles de données issues des DMÉ.

Introduction

Diabetes is an important and prevalent chronic health condition. It is a significant contributor to cardiovascular morbidity and renal disease. The management of diabetes and its complications is associated with substantial and increasing healthcare costs 1, 2. Estimates of the prevalence and reports on the epidemiology of diabetes are needed to help plan and guide the management of this condition.

Large-scale reports on the epidemiology of diabetes in Canada have usually relied on administrative datasets such as the National Diabetes Surveillance System (NDSS, a collaborative network supported by the Public Health Agency of Canada) (3) or on survey data such as reports from the Canadian Community Health Survey (4). Similar administrative and survey-based data are available provincially 5, 6, 7. These datasets have consistently reported increases in the prevalence of diabetes (5). The most recent NDSS estimate of the prevalence of diagnosed diabetes in Canada was 6.8% in 2009 (3).

The accuracy of identification of chronic diseases depends on the data sources and methods used to ascertain the presence of conditions 8, 9. There are many potential issues underlying the quality of data in large datasets (10). Health administrative databases use validated algorithms to ascertain disease prevalence and incidence. The assumptions in these algorithms depend on properties of the data (10) such as physician billing behaviour; this can change over time as health systems evolve and billing incentives are modified. Ontario, as an example, has recently switched from a largely fee-for-service billing system to a system in which a majority of family physicians practising comprehensive care are remunerated primarily through capitation for their enrolled patients; physicians receive partial payment (currently 15%) for most fees submitted for services provided (11). The percentage of Canadian family physicians reporting being paid mostly through fee-for-service has decreased from 51% in 2004 to 38% in 2013 (12). The switch to non–fee-for-service payments could be impacting the accuracy of diabetes detection when using administrative data (13). A commonly used Canadian algorithm using administrative data to ascertain the presence of diabetes employs a combination of hospital discharge coding and physician service claims. This had a sensitivity of 86% and a specificity of 97% when validated against primary care chart audits (7); the lower sensitivity leads to an underestimation of true prevalence. A recent systematic review of this algorithm found a pooled sensitivity of 82.3% and noted that periodic revalidations may be needed if trends in the prevalence of diabetes change (14).

The sensitivity of survey-based self-reports may be even lower than that of administrative data. A recent study noted that the prevalence of diabetes in Ontario in 2005 was 7.2% using administrative data and 5.4% using self-reported survey data from the Canadian Community Health Survey; sensitivity was 73% and positive agreement was 82% for self-report as compared to administrative data (8).

Administrative datasets can be limited by the lack of clinical data, such as laboratory data, medication prescriptions for patients not covered by provincial plans, routinely collected vital signs or presence of risk factors such as tobacco use. New methods of data capture could provide rich clinical information, augmenting currently available data sources and improving the accuracy and completeness of information available concerning diabetes and other chronic diseases 15, 16. Because of the increasing uptake of Electronic Medical Records (EMRs) in Canada (64% of family physicians reported using EMRs in 2013) (17), the collection and analysis of routinely recorded clinical data have recently become feasible (18). In this study, we used data extracted from primary care EMRs across Canada. We provide EMR-based information to complement and extend knowledge about the epidemiology of diabetes in Canadian primary care.

Data extracted from EMRs have been used in international settings to study chronic diseases at national levels 19, 20, 21. In addition to improving national surveillance data, the availability of national primary care clinical databases can enable comparisons across different countries using a similar approach.

Our objective was to describe the epidemiology of diabetes primary care, including utilization, comorbidities and patterns of medication use for this condition.

Section snippets

Data sources and study population

We extracted data from the EMRs of primary care providers participating in the Canadian Primary Care Sentinel Surveillance Network (CPCSSN). CPCSSN is Canada's first multidisease EMR-based surveillance system (18). It includes 10 primary care practice-based research networks in 8 provinces across Canada. Consenting family physicians and other primary care providers participating in CPCSSN contribute anonymized EMR data to a regional CPCSSN repository; data from all participating networks are

Results

The population consisted of 272 469 patients and 380 participating primary care providers. Of the patients, 25 425 with diabetes had at least 1 encounter with the practice in 2 years. The age- and gender-adjusted prevalence of diabetes for patients seen in this sample of Canadian primary care practices was 8.2%. The population-based estimate of prevalence, using a corrected yearly contact group practice denominator, was 7.6%. Table 2 provides the age- and gender-specific observed prevalence.

Discussion

We present diabetes prevalence and cross-sectional descriptive data for a new Canadian initiative to study the epidemiology of chronic diseases as they are managed in primary care. This study uses EMRs as a new clinical data source in Canada to provide information on epidemiologic trends and patterns. This is similar to recent approaches using primary care EMR clinical databases for diabetes research and epidemiologic studies in Spain (31), the United Kingdom (32) and the United States 33, 34.

Acknowledgements

Grant provided by the Public Health Agency of Canada under a contribution agreement with the College of Family Physicians of Canada, fund # 6271-15-2009-10010002. The views expressed herein do not necessarily represent the views of the Public Health Agency of Canada. No potential conflicts of interest relevant to this article were reported.

References (53)

  • L.L. Lipscombe et al.

    Trends in diabetes prevalence, incidence, and mortality in Ontario, Canada 1995-2005: A population-based study

    Lancet

    (2007)
  • Standards of Medical Care in Diabetes 2010

    Diabetes Care

    (2010)
  • Canadian Diabetes Association 2013 Clinical Practice Guidelines for the Prevention and Management of Diabetes in Canada

    Can J Diabetes

    (2013)
  • Report from the National Diabetes Surveillance System: Diabetes in Canada, 2009

    (2009)
  • Canadian community health survey

    (2008)
  • J.F. Blanchard et al.

    Incidence and prevalence of diabetes in Manitoba, 1986-1991

    Diabetes Care

    (1996)
  • J.E. Hux et al.

    Diabetes in Ontario: Determination of prevalence and incidence using a validated administrative data algorithm

    Diabetes Care

    (2002)
  • E. Muggah et al.

    Ascertainment of chronic diseases using population health data: A comparison of health administrative data and patient self-report

    BMC Pub Health

    (2013)
  • L.M. Lix et al.

    Population-based data sources for chronic disease surveillance

    Chronic Dis Can

    (2008)
  • J. Hirdes et al.

    An evaluation of data quality in Canada's Continuing Care Reporting System (CCRS): secondary analyses of Ontario data submitted between 1996 and 2011

    BMC Med Informat Decis Making

    (2013)
  • R.H. Glazier et al.

    Capitation and enhanced fee-for-service models for primary care reform: A population-based evaluation

    CMAJ

    (2009)
  • National Physician Survey, results for family physicians, remuneration. National Physician Survey, 2013....
  • A.M. Alshammari et al.

    The impact of non-fee-for-service reimbursement on chronic disease surveillance using administrative data

    Can J Pub Health

    (2009)
  • A. Leong et al.

    Systematic review and meta-analysis of validation studies on a diabetes case definition from health administrative records

    PLoS ONE

    (2013)
  • S.W. Pascoe et al.

    Identifying patients with a cancer diagnosis using general practice medical records and Cancer Registry data

    Fam Pract

    (2008)
  • M.C. Gulliford et al.

    Selection of medical diagnostic codes for analysis of electronic patient records. application to stroke in a primary care database

    PLoS ONE

    (2009)
  • National Physician Survey, result on use of electronic records. National Physician Survey, 2013....
  • R. Birtwhistle et al.

    Building a pan-Canadian primary care sentinel surveillance network: initial development and moving forward

    J Am Board Fam Med

    (2009)
  • L.A. Garcia Rodriguez et al.

    Use of the UK General Practice Research Database for pharmacoepidemiology

    Br J Clin Pharmacol

    (1998)
  • M.C. Biermans et al.

    Striking trends in the incidence of health problems in The Netherlands (2002-05): findings from a new strategy for surveillance in general practice

    Eur J Pub Health

    (2009)
  • J. Hall et al.

    A general practice minimum data set for New Zealand

    N Z Med J

    (2002)
  • R.V. Birtwhistle

    Canadian Primary Care Sentinel Surveillance Network: a developing resource for family medicine and public health

    Can Fam Phys

    (2011)
  • Statistics Canada 2011 census. Government of Canada, 2013. http://www12.statcan.gc.ca/census-recensement/index-eng.cfm....
  • Primary health care intelligence: 2013 progress report of the Canadian Primary Care Sentinel Surveillance Network (CPCSSN)

    (2013)
  • M. Greiver et al.

    Who are your patients with diabetes?: EMR case definitions in the Canadian primary care setting

    Can Fam Phys

    (2012)
  • T. Williamson et al.

    Validating the CPCSSN algorithms for eight chronic conditions

    (2013)
  • Cited by (47)

    • New Diabetes Guidelines: Impact on Eligibility for Sodium-Glucose Cotransporter-2 Inhibitors and Glucagon-like Peptide-1 Receptor Agonists in Canada

      2022, Canadian Journal of Diabetes
      Citation Excerpt :

      NDR includes patients’ demographics, health conditions, physical parameters, encounters, medications, lab data, smoking status and billing codes. These data have been used in previous research (17,18). We reviewed data from July 1, 2018, until June 30, 2020, with additional health conditions data from July 1, 2010.

    • Management of diabetes and hypertension in people with multiple sclerosis

      2020, Multiple Sclerosis and Related Disorders
      Citation Excerpt :

      This case definition has a PPV of 87.0% and NPV of 99.1% (Williamson et al., 2014). We assessed disease control and the intensity of treatment for hypertension and diabetes using approaches already published by CPCSSN investigators (Greiver et al., 2014; Godwin et al., 2015). We assessed blood pressure (BP) control based on the average of the last 3 measures in the 2 years before data extraction.

    • Primary care electronic health data: Good to the last byte

      2018, Health Efficiency: How Can Engineering be a Player in Health Organization?
    View all citing articles on Scopus
    View full text