Imbalances in the oral health workforce: a Canadian population-based study | BMC Health Services Research

Study setting

As in many countries, issues of spatial, gender, and other imbalances in the dentistry and allied health professional workforces in Canada are understudied [35]. Under Canada’s multi-stakeholder health system governance model, public funding for basic dental services is organized at the level of the provinces and territories, yielding inconstant coverage across jurisdictions and sociodemographic groups [33]. Most Canadians access oral health care through private dental offices, with services paid for out-of-pocket and/or through voluntary private insurance [8, 36]. Dentists in private practice set their patient fees by type of procedure or service, although provincial and territorial dental associations suggest annually updated fee guides. The organization of the dentistry workforce thus differs starkly from the physician workforce, for which financing is ensured across the country through a single-payer universal medical coverage system. The education and licensing of dental practitioners are also regulated at the level of provinces and territories. Some reports point to a growing pool of registered dentists in large urban centres, while rural and remote communities remain underserved [37]. The dental hygiene, dental therapy, and dental assisting workforces are characterized with variations in training requirements and scopes of practice across jurisdictions and over time [37].

A new federally-funded public dental insurance plan, being phased in over the period 2022–2025, is intended to accelerate universal access to the treatment of oral disease, notably by complementing provincial and territorial dental plans and reducing the financial burden among patients lacking private insurance [8, 36, 38, 39]. While the federal plan recognizes the crucial role of HROH in service delivery [39], it does not explicitly address potential impacts of labour supply, mix, or distribution to fill existing coverage gaps. Like in other countries, concerns have been raised whether dental offices are prepared to anchor their practices in person-centredness and have enough hygienists and assistants to respond to increasing numbers of patients seeking professional oral healthcare services [38, 40].

Study design

We sourced microdata from the latest 2021 Canadian population census, conducted by Statistics Canada. The census includes a long-form questionnaire, which captures detailed sociodemographic and labour market information from a 25% sample of the household population. The 2021 individual response rate to the long-form was 95.7% [41]. The target population for this study consisted of all respondents who self-reported their main occupation in selected dental care professions, in the core working ages of 25–54 years, having postsecondary educational attainment, and who earned professional income in the previous two years.

We analyzed a set of key indicators of occupational distribution, geographic distribution, service sector, and gender representation. We included three occupational groups aligned with the primary tasks performed in jobs, based on the systematic taxonomy of the 2021 National Occupational Classification: dentists; dental hygienists and therapists; and dental assistants and laboratory assistants (Table 1) [42].

To facilitate geographic analyses across Canada’s vast physical landscape, we linked the census data deterministically by respondents’ place of residence to the Index of Remoteness (IR), a geocoded measure of accessibility and connectivity for all inhabited communities (5,161 census subdivisions) [43]. Developed by the national statistical agency, the continuous index is based on a spatial gravity model for gauging communities in terms of population size, proximity to population centres, and accessibility to services and transportation infrastructures [44]. The index is considered useful to help nuance heterogeneous population needs across the country’s rural and remote areas, given that the most urbanized and accessible areas represent only 6.1% of the total landmass [45]. To enhance meaningful comparisons, we ranked communities by IR deciles into four categories: highly urbanized and accessible areas (decile 1), accessible areas (decile 2), moderately accessible areas (decile 3), and more rural and remote areas (deciles 4–10).

Places of work for dental practitioners were delineated according to the North American Industry Classification System, specifically: (i) ambulatory healthcare services, such as dentist offices, diagnostic laboratories, and other outpatient facilities (NAICS code 621); or (ii) any other subsector (e.g., hospitals and other service settings) [46]. Other tracers of potential institutional imbalances included work status (full-time versus part-time) and worker class (self-employed versus employee).

Gender was based on self-identification as a woman or man (with a very small number of non‑binary persons being distributed in the other two gender categories) [47].

Statistical analyses

Following common approaches for HROH planning applications [20], we started with descriptive analyses of dental workforce size and distribution, including calculating workforce-to-population densities by occupation and geographies. Second, we conducted bivariate analyses of the interplays between gender, occupation, and earnings. Person-level annual earnings data were captured in the census from integrated administrative income tax and benefits records, including all wages from paid employment and net self-employment income (before income taxes and deductions) in the preceding calendar year [47]. The earnings data were logged to address skewness.

Third, we assessed gender inclusion in the dental workforce by means of multivariate econometric decomposition of earnings differentials between women and men for each of the three occupational groups. The wage equations for the counterfactual decomposition technique, known as the Blinder-Oaxaca method [48, 49], are based on two single-gender regressions as follows:

where \(\:{X}_{i}\) denotes predictors of differed wages for both males and females, with \(\:\beta\:\) representing its estimated parameter, and \(\:\epsilon\:\) indicating the associated error. Differences in mean (logged) earnings between men and women, in relation to the predicted mean earnings specific to each gender, distinguish between the “explained” moderators of wage differences (i.e., those attributable to the observed characteristics of men and women as included in the models) and to estimate any residual or “unexplained” component. A significant unexplained portion of the gender wage gap is often termed in the literature as the effect of discrimination and other (unmeasured and unmeasurable) structural problems in the labour market [27, 50]. In addition to adjusting for geographic and institutional factors, we controlled for various professional and personal characteristics including age group, advanced university qualifications, household status (whether living with a marital partner and/or children), ethnic/ancestral minority status (whether the person identified as Indigenous or a visible minority versus Caucasian or white), and adult migrant status (whether the person immigrated to Canada in adulthood, i.e., at ages 18 or over).

The de-identified census microdata used in the analyses were accessed in the secure computing facilities of the Statistics Canada Research Data Centre at the University of New Brunswick (Fredericton, Canada). Individual-level sampling weights were applied to ensure population representation of the descriptives and robust 95% confidence intervals (CIs) for results of the bivariate and multivariate analyses. Population counts were rounded and all statistical outputs were subject to risk-based confidentiality vetting in respect of Statistics Canada data privacy guidelines. The decomposition methods were implemented using the Stata v16 statistical software [51].