Abstract
Objective To explore whether participation in a series of cardiology continuing medical education (CME) activities affects primary care providers’ (PCPs’) lipid management for their patients.
Design This retrospective cohort study used a database of participation in cardiology CME activities (2011 to 2017) linked to electronic medical records. Statistical analyses were completed using logistic regression with generalized estimating equations.
Setting Manitoba.
Participants Patients receiving care from 225 PCPs participating in the Manitoba Primary Care Research Network.
Main outcome measures Recommended lipid management was defined as prescription of statins (yes or no) among patients diagnosed with cardiovascular disease (CVD), patients diagnosed with diabetes mellitus (DM; 40 years or older), and patients diagnosed with chronic kidney disease (CKD; 50 years and older) in 2017. Treatment was identified using the ATC (Anatomical Therapeutic Chemical) system (ATC code C10AA or C10B).
Results After adjusting for relevant confounders, the odds of prescribing statins to patients with CVD, DM, or CKD among PCPs who did not participate in the cardiology CME activities were 50%, 55%, and 67% lower, respectively, than among PCPs who participated in 2 or more activities. The odds of prescribing statins to patients with CVD and DM among PCPs who participated in only 1 cardiology CME activity were also 67% and 63% lower, respectively, than among PCPs who participated in 2 or more activities.
Conclusion Results suggested that PCPs who participated in 2 or more cardiology CME activities were more likely to prescribe recommended lipid management (statins) for adults with CVD, DM, or CKD.
Cardiovascular disease (CVD) is the primary cause of death globally, accounting for 31% (17.9 million persons) of all deaths in 2016.1 In 2015, heart and cerebrovascular diseases were the second and third leading causes of deaths in Canada.2 Researchers have demonstrated a reduction of CVD mortality in patients with high-risk conditions who are treated with statin medication.3-5 Consequently, clinical practice guidelines recommend statin (HMG-CoA [β-hydroxy-β-methylglutaryl coenzyme A] reductase inhibitor) medications for patients diagnosed with CVD, patients 40 years or older who have diabetes mellitus (DM; type 1 or type 2), and patients 50 years or older diagnosed with chronic kidney disease (CKD).6-9 Evidence-based clinical practice guidelines are commonly incorporated into continuing medical education (CME) activities.
Continuing medical education aims to enhance health care professionals’ knowledge, teach new skills, and build on existing skills to provide the best possible care, improve patient outcomes, and protect patient safety.10,11 Summative assessments (ie, self-reported questionnaires, knowledge tests, and commitment-to-change exercises) used to measure CME effectiveness demonstrate knowledge gains.11 However, the lack of metrics and objective data has hindered measurement of clinical performance and patient outcomes in relation to CME activities. Rayburn and Davis12 suggest that patient health records and administrative data contained in databases could be used to overcome this knowledge gap. To the best of our knowledge, however, few have attempted to use these sources of data to assess the effectiveness of CME.13 Therefore, this study used electronic medical records (EMRs) to explore associations between recommended statin prescribing to patients with CVD, DM, or CKD and primary care providers’ (PCPs’) participation in a series of cardiology CME activities.
METHODS
This retrospective study, approved by the health research ethics board of the University of Manitoba in Winnipeg, linked information about PCPs’ participation in cardiology CME activities (2011 to 2017) to patients’ EMRs. All PCPs in the study participated in the Manitoba Primary Care Research Network (MaPCReN) in 2017. A unique provider identification number was used to link participation in CME activities with de-identified records from EMRs. The MaPCReN database contains information extracted from 48 primary care clinics and more than 255 PCPs (family physicians and nurse practitioners), representing more than 288 000 patients in Manitoba. The Manitoba Primary Care Research Network is part of the Canadian Primary Care Sentinel Surveillance Network database. The Canadian Primary Care Sentinel Surveillance Network is a pan-Canadian network that extracts de-identified EMR data from consenting PCPs.
It has developed and validated methods for capturing information about patients with several chronic diseases, including CVD, DM, and CKD.
Study setting
Manitoba is 1 of the 3 Prairie Provinces of Canada, with a population of 1 356 961 (57% in Winnipeg) in 2017.14 Canada maintains a universal, publicly funded health care system that provides Manitobans access to health care services. The provincial pharmacare program covers the cost of some prescribed medications for Manitoba residents who spend a large amount of their income on medications, once the income-based deductible is paid. In 2017, approximately 1648 PCPs (178 [11%] nurse practitioners and 1470 [89%] family physicians) were registered in Manitoba.
Cardiology CME
The Office of Continuing Competency and Assessment in the Rady Faculty of Health Sciences at the University of Manitoba hosted a yearly CME activity in the form of “Cardiology Day” from 2011 to 2017. We offered a series of 20- to 30-minute lectures followed by about 10 minutes of questions and answers that fulfilled the interactive component of educational standards for accredited CME. (Accredited programs must dedicate at least 25% of instruction time to interactive learning.) These 1-day (8:30 am to 2:45 pm) accredited programs were planned by a committee composed of family physicians and specialists in cardiology, who defined the most suitable themes and speakers based on their experience, informed by literature and local needs assessment surveys. The cardiology CME activities aimed to provide an overview of advances made in diagnostics, prevention, and medical and surgical treatment of CVD. The target audience comprised family physicians, medical and surgical specialists, nurse practitioners, nurses, physician assistants, and other allied health professionals who wished to update their general knowledge of cardiology. Apart from 2016, a session on recommended lipid management was incorporated into each cardiology CME program, reflecting its important role in decreasing CVD mortality.
Data
Data from PCPs and patients were extracted from the MaPCReN repository (2017, quarter 2) for the year 2017. Patients’ EMRs were reviewed for diagnoses of CVD, CKD, or DM (ie, ICD-9 diagnostic codes):
the CVD ICD-9 Clinical Modification (ICD-9-CM) code (429.2), the ICD-9 codes for atrial fibrillation or flutter (427.3, 427.31, 427.32), or the ICD-9 code for valvular disease (394.x, 395.x, 396.x, 424.0 or V43.3);
the CKD ICD-9 code (585.9); and
the DM (types 1 and 2) ICD-9-CM code (250), Anatomical Therapeutic Chemical (ATC) codes for diabetes medicine (A10), or hemoglobin A1c level of 7 or more. Patients were excluded if they had a diabetes medication and an ICD-9-CM code for gestational diabetes (648.8), chemical-induced (secular) diabetes (249), neonatal diabetes (775.1), polycystic ovary syndrome (256.4), or nitric oxide synthesis hyperglycemia (790.29).15,16
How many times PCPs attended cardiology CME activities from 2011 to 2017 was linked to PCPs’ EMRs from MaPCReN in 2017 and grouped in 3 categories according to participation in 2 or more activities, 1 activity, or no activities. The 2016 cardiac CME activity did not include a session on lipid management and therefore was excluded from analyses.
Outcome measures
Among patients diagnosed with CVD, DM (≥ 40 years), and CKD (≥ 50 years) in 2017 were identified using the ATC system (ATC code C10AA or C10B).
Providers’ and patients’ characteristics
Characteristics of PCPs and patients listed in MaPCReN and included in the study are age (in years) and sex (female vs male). Additional provider characteristics included type of provider (family physician vs nurse practitioner), location of practice (rural vs urban), type of funding (salary vs fee-for-service), large practice (837 or more patients; 837 is the median practice size of the cohort), and graduate of a Canadian medical school (no vs yes). Prescribing data within the EMR can be linked to individual patients and providers by unique identification numbers.
Analysis
We used the conceptual framework of Moore et al,17 which assesses the learner’s change in performance after participation in a CME activity, to explore the effect of educational activities on clinical practice. This theoretical model is based on Kirkpatrick and Kirkpatrick’s framework,18 a benchmark for assessing desired learning outcomes, which also incorporates theories specific to medical education.
Descriptive statistics were used to present the characteristics of PCPs in relation to their participation in cardiology CME activities. Percentages were used for categorical variables and means and SDs for continuous variables.
Crude and adjusted logistic regression models with generalized estimating equations clustered at the PCP level were used to analyze statin prescription among patients diagnosed with CVD, patients 40 years or older with DM, or patients 50 years or older with CKD in 2017 in relation to PCPs’ participation in cardiology CME activities. The category of participation in 2 or more cardiology CME activities was used as a reference.
Statistical significance was accepted at P values below .05. All analyses were performed using SPSS software, version 25.0.
RESULTS
Two hundred twenty-five PCPs had in their practices patients with CVD, patients 40 years or older with DM, or patients 50 years or older with CKD in 2017. The mean (SD) age of PCPs was 47.6 (10.0) years, 40% were male, and 84% were family physicians. Four percent of PCPs participated in 2 or more cardiology CME activities at the University of Manitoba, 16% participated in 1 activity, and 80% participated in none. There were significant differences in sex (P = .01), type of provider (P < .01), and practice size (P = .04) in relation with participation in cardiology CME activities (Table 1). However, no other characteristics of PCPs, nor proportions of patients with CVD, DM, and CKD in their practices were statistically different across groups (Table 1). The PCPs who participated in 2 or more cardiology CME events prescribed statins to a larger percentage of their patients with CVD, DM, and CKD in 2017 than PCPs who participated in 1 event or no events (Table 2).
Primary care providers by participation in cardiology CME activities from 2011 to 2017: Four percent of PCPs participated in ≥ 2 activities, 16% participated in 1 activity, and 80% participated in none.
Primary care providers’ participation in cardiology CME activities from 2011 to 2017 and prescription of statins in 2017: N = 225 PCPs.
Statins for patients with CVD
Adjusted odds ratios (aORs) of prescribing statins to patients with CVD among PCPs who participated in 1 (aOR = 0.31, 95% CI 0.19 to 0.50, P < .001) or no (aOR = 0.50, 95% CI 0.33 to 0.77, P = .002) cardiology CME activities were lower compared with PCPs who participated in 2 or more cardiology CME activities. The fully adjusted model also identified lower odds of prescribing statins to patients with CVD associated with female patients (aOR = 0.58, 95% CI 0.50 to 0.69, P < .001), younger patients (aOR = 1.006, 95% CI 1.002 to 1.011, P = .01), salaried providers (aOR = 0.71, 95% CI 0.56 to 0.89, P = .003), and providers in rural practices (aOR = 0.81, 95% CI 0.68 to 0.97, P = .02) (Table 3).
Primary care providers’ prescription of statins to patients in 2017 according to recommended clinical practice: A) Crude model, B) adjusted model. Bolded values indicate statistical significance.
Statins for patients with DM
Adjusted odds of prescribing statins to patients 40 years or older with DM among PCPs who participated in 1 (aOR = 0.37, 95% CI 0.29 to 0.47, P < .001) or no (aOR = 0.45, 95% CI 0.36 to 0.55, P < .001) cardiology CME activities were lower than among PCPs who participated in 2 or more cardiology CME activities. The fully adjusted model also showed that lower odds of prescribing statins in this group of patients were associated with female patients (aOR = 0.76, 95% CI 0.71 to 0.82, P < .001), younger patients (aOR = 1.008, 95% CI 1.006 to 1.011, P < .001), salaried providers (aOR = 0.75, 95% CI 0.68 to 0.84, P < .001), providers in rural practices (aOR = 0.69, 95% CI 0.63 to 0.75, P < .001), and providers trained in Canada (aOR = 0.77, 95% CI 0.70 to 0.85, P < .001) (Table 3).
Statins for patients with CKD
Adjusted odds of prescribing statins to patients 50 years or older with CKD among PCPs who did not participate in cardiology CME events were lower (aOR = 0.33, 95% CI 0.15 to 0.75, P = .01) compared with PCPs who participated in 2 or more events. No other variables were significantly associated with prescribing statins to this group of patients (Table 3).
DISCUSSION
We found that PCPs who participated in 2 or more cardiology CME activities were more likely to provide recommended statin treatment to adults with CVD, DM, or CKD. This suggests that participation in CME activities improves clinical practice (Moore et al’s level 5 performance17). Evidence shows that CME activities increase physicians’ knowledge and competence.11 However, studies that assess the effect of CME in clinical practice and patient health outcomes are scarce. To the best of our knowledge, this is the first study to use medical records to evaluate the association between participation in recurrent CME activities and recommended clinical practice. These findings support targeted medical education and recurrent CME programs for health care providers.
Besides PCPs’ participation in cardiology CME activities, results also showed that some characteristics of patients and providers were associated with prescription of recommended statin treatment to adults with CVD or DM. Female and younger patients had lower odds of receiving prescriptions for statins, which aligns with previous data showing that men 60 years and older reported the highest percentage of lipid-lowering medication intake in the United States.19,20 These findings could be influenced by previous evidence that associated prescribing statin therapy with larger reductions in mortality rates in men than in women21 and that showed positive effects of statin therapy among elderly patients.22 Cardiovascular risk increases with age; thus the prospective risk reduction is greater in older patients, meaning this finding of our study supports the practice pattern expected if clinical decision making is patient centred.
Salaried providers, providers in urban practices, and Canada-trained providers had lower odds of prescribing statins. The impact of physician payment method on quality of care is controversial. Previous studies suggest that salaried PCPs prescribe less often23,24; however, it is possible that less prescribing might better reflect the needs of patients because salaried physicians are able to expend more time analyzing each case. Several health care challenges have been documented in rural areas, usually expressed in terms of access, use, cost, and geographic distribution of providers and services.25 These challenges (together with absence of communities of practice,26 barriers to accessing CME, and unique CME needs27) might have affected the practice of evidence-based care in rural settings. It is possible that foreign-trained PCPs felt more pressure to comply with local standards of care28 and were therefore more inclined to incorporate Canadian guidelines into their clinical practice.
Previous research found that, despite strong evidence that statin treatment is effective in primary prevention,29 PCPs still underprescribe recommended statin treatment for people at high risk of CVD.15,30-32 In addition to the variable captured in this study, other factors could increase reluctance to adopt recommended care (clinical guidelines), such as disagreement with the underlying evidence, individual patient factors (eg, adverse effects, nonadherence), lack of knowledge of guidelines, and unclear or ambiguous recommendations.27
Limitations and strengths
Some limitations of this study should be considered when interpreting the results. First, we were unable to measure whether PCPs received similar CME from other sources outside our institution. However, this research provides an overview of the effect of a recurrent (annual) and wellattended (about 120 to 180 health care providers yearly) CME activity led by the University of Manitoba, which is one of the main CME providers in the province. Second, although data included in this study represent a large sample of PCPs and patients in Manitoba, the MaPCReN database includes only consenting PCPs in Manitoba (approximately 16% of Manitoba providers). Third, it is possible that prescriptions that did not align with the guidelines were influenced by PCPs’ disagreement or personal interpretations, which are impossible to account for in the EMR data. Fourth, it was impossible to capture PCPs’ interest in and knowledge of the recommendations before the CME activity. The PCPs who attended a cardiology-focused CME activity could have had previous interest in CVD and could have been more likely to be aware of relevant recommendations and to prescribe statins accordingly. In this scenario, CME providers should explore strategies to reach PCPs who are not participating in current CME events. Fifth, our study used only structured data files available within the EMR and not clinic encounter notes, which might have provided more detail about treatment decisions (eg, clinical presentation, discussion with patients). Finally, this study was based on EMRs that could have had incomplete data.33,34 Nevertheless, using clinical data from EMRs has been shown to be valid for identifying diagnoses.35,36
Strengths of this study include the use of good-quality EMR data from the MaPCReN database and longitudinal information about participation in a recurrent CME activity, together with the ability to link the 2 data sets. Further, this is one of the few studies objectively exploring the potential association between PCPs’ participation in CME and clinical practice.
Conclusion
This study found that PCPs who participated in 2 or more cardiology CME activities had higher odds of providing recommended statin treatment to adults with CVD, DM, or CKD. While several factors might influence clinical decision making, our results still suggest that participation in CME activities has a positive effect on clinical practice. This valuable information can serve to support targeted medical education and recurrent CME activities for health care providers.
Notes
Editor’s key points
▸ Primary care providers who participated in 2 or more cardiology continuing medical education (CME) activities were more likely to prescribe recommended lipid management (statins) for adults with cardiovascular disease, diabetes mellitus, or chronic kidney disease.
▸ Results also showed that some characteristics of patients and providers were associated with prescription of recommended statin treatment to adults with cardiovascular disease or diabetes mellitus. Female and younger patients had lower odds of receiving prescriptions for statins. Salaried providers, providers in urban practices, and Canada-trained providers had lower odds of prescribing statins.
▸ To the best of our knowledge, this is the first study to use medical records to evaluate the association between participation in recurrent CME activities and recommended clinical practice. These findings support targeted medical education and recurrent CME programs for health care providers.
Points de repère du rédacteur
▸ Les professionnels des soins primaires qui avaient participé à 2 activités ou plus de formation médicale continue (FMC) en cardiologie étaient plus susceptibles de prescrire le contrôle des lipides (statines) recommandé aux adultes atteints de maladies cardiovasculaires, de diabète sucré ou de néphropathie chronique.
▸ Les résultats ont aussi démontré que certaines caractéristiques des patients et des professionnels étaient associées à la prescription de statines recommandée aux adultes atteints de maladies cardiovasculaires ou de diabète sucré. Il était moins probable que les femmes et les patients plus jeunes reçoivent une prescription de statines. Les professionnels salariés, les professionnels dans les pratiques urbaines et les professionnels formés au Canada étaient moins enclins à prescrire des statines.
▸ Au meilleur de nos connaissances, il s’agit de la première étude dans laquelle les dossiers médicaux ont servi à évaluer l’association entre la participation à des activités de FMC récurrentes et la pratique clinique recommandée. Ces constatations étayent l’importance d’une formation médicale ciblée et de programmes de FMC récurrents à l’intention des professionnels de la santé.
Footnotes
Contributors
Dr Diana C. Sanchez-Ramirez contributed to the concept, design, and data analysis of the study. Leanne Kosowan assisted with the data gathering. All authors contributed to the interpretation of the data and the preparation of the manuscript.
Competing interests
Dr Alexander G. Singer received a research grant administered by the Canadian Institute for Military and Veteran Health Research funded by IBM and Calian not related to the findings of this study.
This article has been peer reviewed.
Cet article a fait l’objet d’une révision par des pairs.
- Copyright © 2022 the College of Family Physicians of Canada