Reach Us +44 1285300058

Referral Bias in Coronary Micro vascular Dysfunction: a Retrospective Study

Mary McCarthy*, Olga Toleva, Karen Elliott, Chris Ardern, Steven Miner

Southlake Regional Health Center, Newmarket, ON, Canada

*Corresponding Author:
Mary McCarthy
Southlake Regional Health Centre
596 Davis Dr, Newmarket, ON, L3Y 2P9
Tel: +19058302605
Email: [email protected]

Received Date: July 10, 2020; Accepted Date: August 18, 2020; Published Date: August 25, 2020

DOI: 10.36648/2049-5471.17.3.211

Visit for more related articles at Diversity & Equality in Health and Care


Obstructive coronary artery disease (CAD) is a well-defned condition and a signifcant cause of myocardial ischemia. Patients with angina typically undergo non-invasive cardiac investigations initially as a means of risk-stratifying. Coronary angiography is the diagnostic test of choice for patients with chest pain (angina) and signs of myocardial ischemia on noninvasive testing. Typically CAD is diagnosed based on the patency of the epicardial coronary arteries [1]. Approximately 30% of patients with angina have non-obstructive arteries on coronary angiogram and area diagnostic and management challenge for clinicians, contributing to signifcant social and economic burden [2]. They undergo multiple angiograms, experience a hospitalization rate 80% greater than patients with single vessel CAD and up to 50% have persistent symptoms leading to functional disability [3]. Historically, this population was given reassurance as long-term prognosis was believed to be benign [4]. We now know that a subset of these patients suï¬Â€er from coronary micro vascular dysfunction (MVD) [1] and are at increased risk for adverse cardiac events including congestive heart failure (CHF), myocardial infarction (MI), left ventricular (LV) dysfunction, and death [1,5]. The Cardiovascular Integrated Physiology (CVIP) program at Southlake Regional Health Centre (SRHC) investigates and manages patients with MVD. CVIP has received referrals for only a small fraction of patients with angina and non-obstructive angiograms at SRHC. It is unclear what factors promote referral. We hypothesized that referrals were not random and that the clinical characteristics of the referred patients diï¬Â€ered from the larger population of patients with angina and nonobstructive CAD. The primary objective of this study was to determine whether the patients referred to CVIP are an accurate representation of the overall population of patients with angina and non-obstructive coronary arteries.


Obstructive coronary artery disease; Patients; Population

Methods and Procedures

The study was approved by the Research Ethics Board at SRHC.

Patient population

Two sub groups of patients were analyzed. The first included all patients referred for angiography with high-risk non-invasive testing to rule out CAD but with non-obstructive coronary angiograms at SRHC from 2016-2018 (n=1,816). The second included patients referred to CVIP for the assessment of possible MVD due to the presence of angina and non-obstructive CAD from 2017-2018 (n=281).

Data sources

Data was retrospectively obtained from two databases. The Cor Health database is a provincial administrative database that gathers patient information on various cardiac procedures for the measurement of quality and outcomes. Only data relating to SRHC was abstracted for this study. The TEAMWORKS© database gathers patient information relating to their clinic or procedural visit. It serves as both a research database and a reporting tool.

Cor Health data was collected from the patient’s angiogram referral form and angiogram report and entered into the Cor Health database by data clerks. This information included age, sex, ethnicity, co-existing co morbidities, coronary risk factors, significant cardiac history, electrocardiogram, left ventricular (LV) function, non-invasive cardiac investigations and angiogram results.

Data from TEAMWORKS© was obtained during the patient’s CVIP clinic visit. Patients completed a questionnaire consisting of questions about their race/ethnicity, social history, and medical history. Patients were assessed by a nurse practitioner who gathered the patient’s past medical history, previous cardiac investigations and symptom history. Physical examination included blood pressure, heart rate, height, weight, and body mass index. Where appropriate, patients had a transthoracic echocardiogram and a symptom-limiting graded exercise stress test (GXT). Patients were also assessed by an interventional cardiologist who determined whether the patient was an appropriate candidate for an invasive coronary physiology study. Patient data collected from CVIP was electronically stored into TEAMWORKS©. Patient information was extracted from the database for the purpose of analysis.

Statistical analysis

Results of continuous (mean ± SD) and categorical (%) variables were compared using chi square and two-tailed independent t-tests, respectively. A P value of <0.05 was considered statistically significant.


From 2016-2018, 21% of patients undergoing angiography at SRH Crafter having a high-risk non-invasive cardiac test were found to have non-obstructive CAD (n=1,816). From 2017-2018, 281 patients were referred to CVIP for suspected MVD. The clinical characteristics of these patient populations were compared and are presented in Table 1. Comparing CVIP MVD patients with Cor Health non-obstructive patients revealed that patients referred to CVIP tended to be younger (58 ± 11.8 vs 63 ± 11, p<0.0001), more likely to be female (63% vs 43%, p<0.0001), of white ethnicity (83% vs 79%, p<0.0001), weighed less (80 ± 20 kg vs 88 ± 21 kg, p<0.0001), with less hypertension (51% vs 64%, p=0.0001), hyper lipidemia (48% vs 63%, p<0.0001), and no difference in diabetes (20% vs 25%, p=0.07). There were significant differences in smoking status: never smoked (51% vs 48%); former smoker (37% vs 36%); current smoker (12% vs 12%); unknown smoking history (0% vs 4%), p=0.0001. There were no significant differences in the prevalence of CHF (2% vs 3%, p=0.13), however there were differences between LV function: normal function (72% vs 79%); mild dysfunction (23% vs 10%); moderate dysfunction (0.5% vs 3%); severe dysfunction (0.5% vs 8%); unknown LV function (4% vs 8%), p<0.0001.

Table 1: Comparison of patients referred to CVIP for suspected MVD versus those referred for angiography to rule out CAD but with non-obstructive angiograms (Cor Health).

Variable CVIP MVD (n= 281) CorHealth Non-obstructive, high risk non-invasive (n=1,816) P Value
Sex (% female) 63 43 <0.0001
Age (years) 58 ± 11.8 63±11 <0.0001
White Ethnicity (%) 83 79 <0.0001
Weight (kg) 80 ± 20 88±21.6 <0.0001
Hypertension (%) 51 64 0.0001
Hyperlipidemia (%) 48 63 <0.0001
Smoking (%)      
Never 51 48 0.0001
Former 37 36  
Current 12 12  
Unknown 0 4  
Diabetes (%) 20 25 0.07
CHF (%) 2 3 0.13
LV Function (%)      
Normal 72 79 <0.0001
Mild Dysfunction 23 10  
Moderate Dysfunction 0.5 3  
Severe Dysfunction 0.5 0  
Unknown 4 8  


This study revealed that patients referred for the investigation of possible MVD do not accurately reflect the larger population of patients with angina and non-obstructive CAD. A referral bias was demonstrated that favours younger, thinner, white females with fewer traditional CV risk factors, and better LV function.

An overarching theme in the current literature surrounding MVD is that a sex difference exists in those who are affected by MVD. As a result men have been underrepresented in studies [6]. One study found that MVD was common in both sexes and was associated with adverse events [7]. We found that patients referred to CVIP for suspected MVD were more likely to be female, yet the majority (57%) of patients with non-obstructive CAD found at the time of angiography were male. This referral bias suggests that clinicians believe that MVD typically affects females. Underrepresentation of men in research of MVD is problematic as it disregards the potential impact of MVD in the male population.

Ethnic differences in CV disease have also been extensively researched. Factors aside from geographic location may contribute to the development of CV disease such as genetic predisposition to various disease states [8]. South Asians living in the United States and African Americans are at a higher risk for developing ischemic heart disease due to many causes including higher prevalence of traditional CV risk factors [8,9]. Patients referred to CVIP were disproportionately of white ethnicity. It is unclear whether this is a result of the geographic homogeneity or the referring clinician’s bias. This is of concern because the WISE study found that women of non-white race had higher rates of mortality [10]. Non-white ethnic groups have higher rates of CV risk factors which have been associated with the development of MVD [11]. There may be other ethnic factors relating to the development of MVD and this hypothesis will be difficult to determine if non-white patients are disproportionately excluded from advanced testing.

The referral patterns to CVIP also revealed a bias away from most traditional CV risk factors. Patients referred to CVIP had less hypertension and hyper lipidemia compared to Cor Health patients with non-obstructive CAD. Traditional CV risk factors have been associated with the development of MVD [11] and were independent predictors of all-cause mortality in the WISE study [12]. The current data suggests a referral bias in the opposite direction for reasons which are unclear.

Numerous studies have found a link between aging and MVD, thought to be from the effects of vascular remodelling that comes with the aging process [1,5]. Older women with MVD had an increased risk of CV mortality compared to younger women [12]. CVIP receives referrals for women who are younger than the overall population of patients with angina and non-obstructive coronary arteries. The reason for this bias is unclear and would tend to exclude those at highest risk of adverse cardiovascular events.

Progressive LV dysfunction has been implicated in MVD [5]. A growing body of research suggests a link between heart failure with preserved ejection fraction (HFpEF) and MVD [13]. Patients referred to CVIP for suspected MVD had better LV function and lower incidence of CHF which is particularly counterintuitive given the association between MVD and HFpEF [13-20].

This study has multiple limitations. The Cor Health data does not give a thorough understanding of the patient’s symptom burden, which may be a driver for referral. TEAMWORKS© is a clinical database and as such captures more detailed data than Cor Health. This may lead to overrepresentation of some disease states in the CVIP data. Since our concern relates to lower prevalence of CV risk factors, this limitation does not seem to affect our conclusions. Comparing body mass index would have been superior to using weight, especially given the differences in sex, however the Cor Health database did not collect BMI. This dataset does not include definitive diagnoses for the CVIP patients, thus we cannot conclude that patients were referred inappropriately [21-27].


MVD is a poorly understood, but emerging condition that plays a significant role in a subset of patients with angina and non-obstructive coronary arteries. Our data suggests a significant referral bias that may not reflect our current understanding of MVD and potentially limits future research. Efforts to educate clinicians on potential risk factors for development of MVD should be utilized in order to reduce bias in the referrals and capture a broader population of patients who may be affected by MVD.

Conflict of Interest

The authors declare that there is no conflict of interest.


Mary McCarthy received funding for this research from the Registered Nurses Association of Ontario Advanced Clinical Nursing Fellowship.


Select your language of interest to view the total content in your interested language

Viewing options

Recommended Conferences
Flyer image
journal indexing image

Share This Article