The early retiree divests the health workforce: a quantitative analysis of early retirement among Canadian Registered Nurses and allied health professionals

We developed a conceptual model of early retirement among RNs and AHPs—see Hewko et al. [9] to view the complete model. The contents of the model were derived from the retirement literature (broad and profession-specific). We established its face validity through a series of purposive interviews with Canadian RNs and AHPs between 45 and 85 years of age. Elder’s [10] Life Course Perspective guided our conceptualization of retirement; notably, this perspective encourages interdisciplinarity and recognizes the value of micro-, meso-, and macro-level factors in the study of individual life courses [10]. This perspective is commonly applied in studies of retirement [11–15]. All identified factors were incorporated into one of eight categories of predictors of early retirement among RNs and AHPs. Categories included lifestyle and health, attitudes and beliefs, work-related, organizational factors, sociodemographics, broader context, and family and workplace characteristics [10]. In the analysis reported below, we were unable to empirically test all factors included in the conceptual model.

Attitude and belief factors linked to early retirement include the sense of being “tired of work” [16] and the desire for leisure time [16, 17]. Among the organizational factors connected with early employee retirement are as follows: (i) access to programs incentivizing retirement [16], and (ii) organizational restructuring [17, 18]. Many sociodemographic factors are identified predictors of early retirement including household income [19] and eligibility for retirement benefits [16] such as pension. Age [20–22], often considered a sociodemographic variable, is an obvious predictor of retirement; we have classified this under broader context in recognition that year of birth has implications unrelated to biological age (i.e., generational effects [23]). Last, family considerations impact upon timing of retirement—specifically, an agreement with one’s spouse [17] and caregiving responsibilities [18] positively predict early retirement.

The RN and AHP workforces are, in many ways, very similar. Both require a minimum baccalaureate level of education and both are female-dominated. Both groups are employed in diverse settings and both commonly provide their services to patients or clients in a shared workspace. Neither is frequently the “most responsible practitioner” in the care of their patients [24]. In general, the RN and AHP professions grant practitioners a comfortable, middle-class lifestyle. Yet, there are important differences between RNs and AHPs and their work-life, differences that may lead to systematic differences in their approaches to retirement. Most tangibly, the content and scheduling of RN and AHP work can differ significantly. RNs more often work elongated, rotating shifts outside of banking hours [25] and more frequently perform physically demanding tasks.

Professions included in the allied health workforce are diverse but do share characteristics relevant to the employee-employer relationship. For example, AHPs often provide ancillary services within multidisciplinary teams and may span work boundaries (e.g., across work units or across facilities) [26]. RNs are, in general, more likely to be associated with a single work unit or team than are AHPs. We suspect that there are sufficient differences in the work experiences of members of these professional groups to result in divergent approaches to retirement decision-making.

Gaining a better understanding of retirement decision-making among Canadian, publicly employed RNs and AHPs would facilitate the development and implementation of effective workforce policy (whether governmental or institution-specific) by health care leaders and policy-makers. Even a minor increase in average age of RN/AHP retirement could lead to significant benefits at the workforce level and, correspondingly, to improvements in availability and quality of health care for Canadians.

In the RN model, each incremental increase (one increment = $50 000) in household income was associated with a 1.61 times greater odds of retiring early. Among the factors reported by RNs as contributing to their retirement decision, financial possibility and caregiving responsibilities both significantly increased odds of early retirement. Alternately, those who indicated that being “tired of work” contributed to their retirement decision had significantly lower odds of having retired early.

Early retirement (before age 65) is the norm among publicly employed Canadian RNs and AHPs. We were able to test a pared-down version of a literature-derived conceptual model of early retirement: overall our tested model explained between 19 and 25% of variance in RN and AHP early retirement. As a country with existing shortages in health professionals (e.g., audiology and speech language pathology [33], pharmacy [34], occupational therapy and physiotherapy [33]) and an aging population, it is in Canada’s best interest to try and extend the work lives of RNs and AHPs.

It is unsurprising that much of the variation in early retirement remains unexplained by our model; we were unable to test associations of many meso-level variables (i.e., those identified in the categories of work-related, organizational factors and workplace characteristics in the complete conceptual model [9]) as these factors were not measured in the CLSA. The Life Course Perspective, used to guide our conceptualization of retirement, emphasizes the importance of micro-, meso-, and macro-level factors as influences of life course decisions.

Unfortunately, nearly all factors significantly associated with early retirement in our analyses are not easily mitigated or altered by health administrators or policy-makers. Although only 12% of RNs and 9% of AHPs reported that organizational restructuring contributed to their retirement decision, it was a significant predictor of early retirement among both RNs and AHPs. Previous research has demonstrated that the older workforce is at disproportionate risk throughout the process of organizational restructuring. There is a tendency to introduce and/or expand early retirement programs during restructuring in order to reduce the size of the workforce as the optics are better than those associated with sizeable layoffs [35]. Additionally, as restructuring is frequently triggered by financial constraints, reducing the number of older, more “expensive” [36] workers is seen as a way of reducing organizational expenses. In a unionized system, such that exists across Canada, administrators may seek to vacate positions by offering older workers incentivized early retirement packages.

Across Canada and the world, organizational restructuring is a common response to resource scarcity (actual or expected); as health care costs make up a significant part of many government’s budgets, the need for cost-cutting in health systems is unlikely to diminish. For nurses in particular, organizational restructuring commonly results in an increase in workload [37]. Early retirements triggered by organizational restructuring may exacerbate resource scarcity. Burke, Ng, and Wolpin [38] suggest that collaboration during implementation of change between hospital management and unions (such as nursing unions) may mitigate negative impacts on the employees in the organization.

Desire to stop working or being “tired of work,” which in this analysis was associated with significantly lower odds of early retirement among RNs, is a “catch-all” response worthy of further study. Respondents were specifically asked if wanting to stop working contributed to their decision to retire. They were given an opportunity to offer “other” reasons contributing to retirement but only two respondents in our analytic sample elected to do so. Without a better understanding of what it is specifically that made them want to stop working, it is difficult to develop strategies to alter this desire. It is possible that many of the meso-level factors not measured in the CLSA (but included in our conceptual model) such as frequent/pervasive change in the workplace and opportunities for flexible hours of work (or lack of) may have contributed to respondents’ desire to stop working.

In our analysis, caregiving responsibilities were associated with significantly higher odds of early retirement among RNs. The health professions are female-dominated, and according to US data, 60% of unpaid care providers are female [39]. In addition, female care providers are more likely than male care providers to provide caregiving as a reason for retirement [1]. Glenn [40] argues that for many women—whose roles often include mother, wife, and daughter—the duty to care is a role obligation.

Humble, Keefe, and Auton [41] noted that caregivers reported a willingness to remain in the paid workforce should circumstances be altered to facilitate their doing so. Strategies to subsidize caregiving support, expand leave policies to accommodate employees’ needs to provide care [42], and/or facilitate work flexibility [40] may be effective deterrents to early retirement. Glenn [40] also argues that, due to existing inequalities in the labor market that perpetuate gendered caring, affirmative-action policies and application of anti-discrimination laws could serve to equalize the cost of engagement in caring work for women.

It is clear, both from the number of literature-derived variables appearing in our model that were not measured in the CLSA and from the fact that our tested model left the majority of variance in the outcome of early retirement unexplained, that our understanding of retirement decision-making among publicly employed RNs and AHPs is far from complete. Future studies with a focus on measurement of workplace characteristics, attitudes, and beliefs and work-related factors would facilitate testing the influence of the remaining factors in our conceptual model on early retirement.

The diversity of AHPs in our sample likely contributed to the limited number of statistically significant relationships identified in the AHP model of early retirement. Although adequately powered to detect predictors of early retirement among AHPs in general, there were insufficient numbers of individual groups of professionals to test for differences across allied health professions. It is possible that a factor that contributed to higher odds of early retirement in one sub-group (e.g., pharmacists) contributed to lower odds of early retirement in another (e.g., social work) and thus sub-group analyses should be undertaken with a sufficient sample.

The limited AHP sample size bounded our capacity to test more factors associated with early retirement among RNs and AHPs. A more comprehensive survey, completed by a larger number of AHPs, in particular, would be required to test the complete conceptual model [9]. Additionally, our inability to confidently isolate all RNs in the sample limited the RN sample size. While the use of nationally representative samples in analysis can maximize the generalizability of results, it should be noted that the CLSA sampling frame was not designed to specifically ensure representativeness according to profession or setting of employment. For this reason, the AHP and RN samples may not be representative of the broader publicly employed AHP and RN population. There is, however, no indication that our sample is not representative—gender representation in the sample is in-line with proportions reported by the Canadian Institute for Health Information. In 2006, 94.5% of nurses were female [43] and, because the proportion of males in the profession is gradually increasing over time [44], it is likely that the proportion of those who are female that are over the age of 45 is greater (i.e., closer to our observed proportion of 97%). Pharmacists, however, a group with significant representation in our sample, were 59.8% female as of 2012 [45]; thus, it is expected that the AHP sample would have a larger proportion of male respondents. A future survey, designed explicitly to explore retirement decision-making in the RN and AHP workforce, should be administered using methods that maximize the representativeness of the sample with consideration for profession, setting of employment, province/region of residence, age, and gender.

The reference point (time-wise) for many of the variables was a limitation. For many questions, particularly those related to sociodemographics, participant responses represented their status at the time of data collection. It is possible that many respondents’ marital status, province of residence, etc. had changed since the time of retirement. Interestingly, many variables predictive of early retirement among RNs and AHPs, including financial possibility, “tired of work,” organizational restructuring, and caregiving responsibilities, were measured using questions that called on participants to think back to what had contributed to their retirement. Although there was potential for recall bias [31] when responding to these questions, these variables were more temporally connected to our outcome variable. Additionally, as we have analyzed only the first wave of data (cross-sectional) collected in the CLSA, we can only confirm associations (not causality) between independent variables and early retirement. In future, we hope to re-test the model using multiple waves of CLSA data.

We relied on free-text variables to assign participants to both an occupational category and a work setting; our goal was to ensure that only RNs and AHPs employed in the Canadian public health sector were included in the analytic sample. It is possible that participants were misallocated (either in or out of the analytic sample). An error may have occurred at either or both the data entry stage or during the selection process (conducted by SH). Due to space constraints, we were unable to discuss varied theories and conceptualizations of retirement and how the process may differ when retirement is early vs. delayed. We recommend Feldman [46], Shultz and Wang [47], and Fisher et al. [48] to those seeking in-depth theorization of retirement and timing of retirement.

The analytic method we have selected is not designed to test mediation effects. We have considered the magnitude and significance of correlations and tested for multicollinearity in our model. The literature review conducted to inform development of our conceptual model [9] did not identify specific factors as mediators in the relationships between predictors of retirement and early retirement. Theoretical or conceptual mediation effects would be better detected via other analytic methods, such as structural equation modeling.

The majority of publicly employed, Canadian RNs and AHPs retire before age 65. Organizational restructuring, although cited by fewer than 15% of RNs and AHPs as a factor contributing to their retirement, increased odds of early retirement in this population. Health administrators may want to re-consider use of early retirement as a tool to achieve cost reduction during organizational restructuring. Although this strategy can help to improve the optics of system-wide change, it is a blunt tool that can exacerbate existing shortages in the health professions. Among RNs, both financial possibility and caregiving responsibilities predicted early retirement. The maximum explained variance in early retirement was 25% (among RNs) indicating that there is much to be learned about both RN and AHP pathways to early retirement. Further research, ideally exploring the role of workplace characteristics, attitudes, and beliefs and work-related factors (e.g., organizational tenure) in retirement decision-making among Canadian, publicly employed RNs and AHPs, would deepen our understanding of this phenomenon.