Background
The Institute of Medicine (IOM) estimated in 2012 that 30% of annual health care spending is wasteful, citing unnecessary services and inefficient care as primary drivers of cost [
1]. Months before the IOM report, Choosing Wisely (CW) launched in the United States [
2]. CW is a clinician-led campaign with the intention of encouraging clinicians and patients to engage in conversations about potentially unnecessary services – i.e. tests, treatments, and procedures where care is unlikely to be beneficial and/or may cause avoidable harm to the patient [
2‐
4]. Since the launch of CW in 2012, the CW campaign has been adopted internationally with over 20 countries that have either started their own versions of the campaign or have local CW campaigns in development [
4]. As interest in tackling the problem of unnecessary care grows worldwide, so does the need for quantifying the extent of the issue at a national level [
4]. Initial estimates of the prevalence of unnecessary care are predominantly derived from U.S. data [
5]; however, prior studies from the U.S. have largely been limited to Medicare beneficiaries or select commercial health plan members [
6‐
8]. While some population-based studies have assessed overuse in Australia and Canada, these works have often been limited to a single region [
9‐
13]. To our knowledge, there has not been a prior study that quantifies the use of unnecessary care across multiple services and Canadian jurisdictions.
Based on a published framework for measuring unnecessary or “low-value” care, we aim to develop common methodological approaches to estimating overuse at a national level by utilizing existing, routinely collected administrative and survey data in Canada [
14]. To facilitate measurement across multiple Canadian jurisdictions, the Choosing Wisely Canada (CWC) campaign partnered with the Canadian Institute for Health Information (CIHI) to provide large-scale estimates of unnecessary care across provinces and territories. By piloting a practical application of this measurement framework, we hope to develop standardized methods of measuring unnecessary care that might be applied across Canada and, in doing so, may serve as a model for how other countries approach quantifying unnecessary care at a national level [
14].
In this paper, we focus on three CWC recommendations relevant to primary care that advise against the following potentially unnecessary tests: 1) imaging for lower back pain in the absence of red flags [
15,
16]; 2) routine performance of preoperative testing for patients undergoing low-risk surgeries [
17‐
19]; and 3) routine screening mammography for average-risk women aged 40–49 [
16,
20]. These three CWC recommendations, highlighted in CIHI’s recent
Unnecessary Care in Canada report [
21], were identified by multiple Canadian specialty societies to be of concern and were suspected to be common practice despite evidence-based recommendations against their use [
15‐
20]. The objectives of this study are to quantify the frequency of these unnecessary screening and diagnostic imaging services across multiple Canadian provinces, and to investigate factors that may affect their likelihood of being ordered.
Discussion
In our study, the first to measure unnecessary care across multiple Canadian provinces and territories, we found evidence of substantial overuse for each service we examined – lower back pain imaging in the absence of red flags, preoperative cardiac testing for low-risk surgeries, and routine screening mammography for women 40–49 years old – in scenarios where CWC recommendations indicate care is unnecessary or low-value [
15‐
19]. Across these three non-recommended services, overuse ranged from 17.9 to 38.7% depending on the test and jurisdiction considered [
21]. Moreover, our test-specific findings are consistent with previous smaller-scale studies in Canada [
11,
12,
21]. As Kirkham et al. found in Ontario, our findings suggest preoperative cardiac testing is fairly common in Alberta and Saskatchewan (17.9 and 21.8% respectively) [
10,
21]. Observed rates of CT/MRI scans for lower back pain in Alberta are similar to those reported by Pendrith and colleagues for Ontario [
11]; however, the latter study investigated three-month rates [
11,
21]. The national rate of screening mammography among women 40–49 was 22%, with significant variation among provinces. Our investigation revealed multiple factors associated with receipt of an unnecessary service. Older age, male sex, and living in a rural health zone were all patient characteristics associated with increased odds of lower back pain imaging. Patients with higher versus lower income had increased odds of having back imaging and early screening mammograms. Older, male patients have been previously found to be more likely to receive unnecessary testing have findings regarding older age and male previous studies that show that older age and male sex are patient characteristics are risk factors for potentially unnecessary electrocardiograms (ECGs) and chest x-rays [
12,
25]. Contrary to our study, rural patients were less likely to receive a potentially unnecessary ECG, which differs from our findings for imaging for lower back pain and demonstrates the importance of understanding the specific drivers for individual unnecessary services [
12]. All estimates were based on data that predates the 2014 launch of the CWC campaign and subsequent release of these three recommendations against low-value care; therefore, this study provides additional support to the existence of these recommendations [
15‐
19,
21].
In 1998, the Institute of Medicine defined all healthcare quality problems as fitting into three categories: underuse, misuse, and overuse [
34,
35]. Since then, much of the focus of research and quality improvement initiatives has been on underuse of appropriate medical services [
36]. However, as concern has recently grown about the overprovision of medical services in situations where evidence does not support their use, so too has an interest in understanding how frequently inappropriate care is delivered [
4,
5,
12]. The body of literature on overuse continues to grow, but to date, most of the research into overuse has been conducted in the U.S., with limited studies having been conducted in other countries, such as Sweden and Australia [
5].
Overall estimates of overuse in Canada appear to be highly prevalent and remarkably similar to those in other countries, despite differences in healthcare systems and underlying populations [
13,
37,
38]. However, comparing our test-specific findings with prior U.S. research demonstrates the need for country-specific data to understand which unnecessary services are potential drivers of overuse domestically [
1]. For example, rates of imaging for lower back pain from the US ranged from 4.5 to 12.4% compared to our finding of 30.7% [
6]. Based on data from two U.S. federally-sponsored health care programs (i.e. Veterans’ Affairs and Medicare), Kerr et al. concluded preoperative cardiac testing is a low priority issue after observing its low prevalence (0.62 to 4.34%) [
6,
8]. Conversely, our Canadian study indicated low-value preoperative tests are relatively pervasive (17.9 to 35.5%) and worthy of further attention. These contrary findings show that we cannot simply generalize results from U.S. studies to other countries without careful consideration of the differences in the underlying patient cohorts and health care systems. Furthermore, this discrepancy emphasizes the importance of developing standardized definitions and measures of unnecessary care specific to a given country, and the need for international collaboration [
8,
10,
14].
International disparities in the frequency of unnecessary care are unsurprising given the substantial variation we observed between and within levels of the Canadian health care system. For example, the rate of routine screening mammography among average-risk women aged 40–49 was nearly triple in Nova Scotia versus Quebec [
21]; the rate of CT/MRI imaging was approximately two-fold in rural versus urban Albertan health regions [
21]; and Albertan facilities ranged from almost never to nearly always ordering a cardiac test prior to a low-risk surgery [
21]. This ordering variation across the Canadian health system is consistent with existing research and represents opportunities for provincial and local quality improvement initiatives to reduce unnecessary care by targeting those facilities and providers with above average ordering rates first [
10‐
12,
14,
21].
This study represents the first attempt at measurement of unnecessary care at a national level both in Canada and internationally. However, several limitations with the study design are worth noting. While claims and survey data can help approximate the occurrence of unnecessary tests, these data sources lack clinical information that ultimately informs the care provider’s decision to recommend, order, or withhold a service [
14]. We captured imaging at six months to achieve comparability with prior research while also considering wait times in Alberta (e.g. MRI 90th percentile wait time, 235 days) [
11,
21,
23,
24]. Consequently, some imaging may have been indicated for cases of chronic back pain or other clinical conditions with lumbar pain symptoms. Responses to the CCHS are voluntary and self-reported, and as such, may be subject to volunteer and social desirability biases [
28,
29]. Furthermore, cancer screening rates derived from CCHS data are consistently higher than rates found using administrative data [
39]. Additionally, there are several limitations that highlight the challenges faced when estimating overuse as a national population level and the importance of common methodological approaches to quantifying low-value care. Jurisdictional variation may be partially attributed to the lack of standardized billing codes and public health programs, as well as differential access to diagnostic services [
21]. To achieve comparisons between Alberta and Ontario, we identified lower back pain diagnoses using three-digit ICD-9 codes, which may have overestimated the number of patients compared to using more specific four- or five-digit codes [
21]. However, among Albertan visits where fourth and fifth digits were available, we found that 80% of the visits selected using three-digit codes were for lower-back pain [
21]. Finally, the regression analyses were not standardized across the three measures studied. While the variables examined were informed by previous studies, due to inconsistencies in the data sets used, we were unable to use a standard set of variables for each regression model.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.