Under embargo until May 11, 2009 at 2 p.m. EST

Transcription

1 Under embargo until May 11, 2009 at 2 p.m. EST

2 This report has been prepared by CNA to provide information on a particular topic or topics. The views and opinions expressed in this report do not necessarily reflect the views of the CNA Board of Directors. All rights reserved. No part of this document may be reproduced, stored in a retrieval system, or transcribed, in any form or by any means, electronic, mechanical, photocopying, recording, or otherwise, without written permission of the publisher. Canadian Nurses Association 50 Driveway Ottawa, ON K2P 1E2 Tel.: or Fax: Website: May 2009 ISBN

3 Authors Gail Tomblin Murphy Professor School of Nursing /Faculty of Health Professions and Director WHO/PAHO Collaborating Centre in Health Workforce Planning and Research Dalhousie University, Halifax Nova Scotia Stephen Birch Professor Centre for Health Economics and Policy Analysis McMaster University, Hamilton, Ontario, Canada Rob Alder Associate Professor of Epidemiology Faculty of Medicine and Dentistry University of Western Ontario Adrian MacKenzie Analyst School of Nursing/Faculty of Health Professions WHO Collaborating Centre on Health Workforce Planning and Research Dalhousie University, Halifax Nova Scotia Lynn Lethbridge Analyst School of Nursing/Faculty of Health Professions WHO/PAHO Collaborating Centre in Health Workforce Planning and Research Dalhousie University, Halifax Nova Scotia Lisa Little Director Public Policy Canadian Nurses Association Amanda Cook Analyst TMCI Toronto Ontario

4

5 FOREWORD Dear Colleagues, The Canadian Nurses Association (CNA) is dedicated to advancing the registered nursing profession, to ensuring that Canadians have timely access to quality registered nursing care, and to promoting health for all. CNA recognizes that health human resources planning is complex. Finding solutions to the problem of nursing shortages is a responsibility shared between federal, provincial and territorial governments, educators, professional associations and/or colleges, unions, employers and other stakeholders. As a policy-focused organization, CNA has a long history of working with governments and other partners to track and analyze issues surrounding the health workforce. CNA contributes research findings, policy documents, statistics, quantitative analysis and other resources to support policy and planning decisions that affect the nursing workforce. Tested Solutions for Eliminating Canada s Registered Nurses Shortage the third in a series of reports on the RN shortage presents a picture of the supply of and requirement for RNs in direct/clinical care (excluding nurse practitioners) in Canada over 15 years. Most importantly it shows how several viable policy options were tested, and highlights their ability to address the projected shortfall. It is these results that are the basis for the report recommendations. CNA is committed to sharing information and collaborating with nursing and other health-care stakeholders to ensure Canada has a sustainable, self-sufficient supply of health human resources. We hope that this report will bring us closer to achieving this. Kaaren Neufeld, RN, MN President Tested Solutions for Eliminating Canada s Registered Nurse Shortage i

6

7 EXECUTIVE SUMMARY There are a quarter-million registered nurses (RNs) in Canada. Representing the largest profession in the health-care workforce, the RN labour market must be carefully managed to meet the diverse and changing health-care needs of Canadians now and in the future. To this end, the Canadian Nurses Association (CNA) has created a national planning model, specific to RNs who provide direct clinical care, which estimates the supply of and requirement for RNs (excluding nurse practitioners) in Canada for each year over 15 years. Unfortunately, due to data limitations, the health needs of the Aboriginal population are not specifically accounted for in this model instead the model is based on the needs of the entire Canadian population, including Aboriginals. This model is aligned with federal, provincial and territorial policy, which calls for population health needs-based health human resource planning. Based on the model, this report estimates: 1. The future requirements for RNs services based on the size, distribution and levels of health-care needs of the population; and 2. The future availability of RNs services based on the size and characteristics of the current workforce as well as trends in entries to and exits from the workforce. The model goes one step further by enabling planners to gauge the effects of a policy or combination of policies on the supply of RNs and requirements for them in the short, medium and long term. As a result, this report is able to estimate the extent to which the implementation of certain policies would alleviate the projected RN shortage in Canada. Based on the best available data and a number of planning assumptions, the simulation model suggests that: There was a shortage of nearly 11,000 full-time equivalent (FTE) RNs in Canada in If the health needs of Canadians continue to change according to past trends, and if no policy interventions are implemented, the shortage of RNs in Canada will increase to almost 60,000 FTEs by Six policy scenarios rooted in these and other findings, and validated by nursing stakeholders were evaluated to determine their impact on the projected shortage. The results show that, on their own, each of the policy scenarios would yield the following results: Short-term: Increasing RN productivity 1% per year would have a dramatic effect on the gap, cutting the shortage by about 47% over 15 years. The effects of this policy would be seen within the first year of implementation. Reducing RN absenteeism (which currently averages 14 days/year) by half over three years would be equivalent to 7,000 new RN FTEs entering the workforce between 2007 and Tested Solutions for Eliminating Canada s Registered Nurse Shortage iii

8 Long-term: Reducing RN exit rates to 2% for RNs under the age of 60 and to 10% for those 60 and over would reduce the RN shortage by about half, or 30,000 FTEs, by Reducing attrition rates in RN entry-to-practice education programs (from 28% attrition to 15%) could cut the gap by about 24% to roughly 45,000. This scenario has a slightly more pronounced effect earlier than an enrolment increase. Increasing enrolment in RN entry-to-practice education programs by 1,000 per year from 2009 to 2011 would result in a substantial reduction in the 15-year RN gap (a shortage of 45,000 vs. 60,000 RNs). However, this scenario would have no noticeable effect on the gap until about Reducing international in-migration by 50% would result in a larger shortage of RNs; however, the effect of this change is not at all substantial (less than 10%), even in the long term. This is due to the fact that internationally educated RNs still represent a relatively small fraction of the national RN supply. Most interestingly, the simulation model showed that the combined effects of these six policy scenarios would be sufficient to eliminate the RN shortage within 15 years. In fact, the results suggest that should all the necessary policies be implemented, Canada could eliminate its shortage of RNs while halving existing levels of in-migration and recruitment from other countries. Canada s policy-makers, decision-makers, educational organizations, professional associations and/or colleges, employers and others are in a position to start addressing the RN shortage immediately. With these results, and future investments in data, stakeholders and governments will be able to make the right policy decisions for Canada s health-care workforce and for the health of Canadians. Recommendations: 1. Governments, employers, unions, professional associations and/or colleges, RNs and other health providers should work together to consider how they can enhance the productivity of the RN workforce. For example, removing non-nursing tasks and providing support staff, appropriate technology and equipment, interprofessional practice and/or effective organization of services would allow RNs to remain as focused as possible on the provision of quality RN patient care. 2. Governments, employers, unions, professional associations and/or colleges, RNs and other health providers should collaborate to focus workplace improvement efforts on strategies to improve the health and well-being of RNs. For example, addressing high role overload, acquiring technologies and equipment that help reduce injuries, and addressing workplace morale would all contribute to reducing the injury and absenteeism of RNs. 3. Governments, employers, unions, professional associations and/or colleges, and RNs should collaborate to improve the retention of RNs in the workforce. Although retention issues may be generation-specific, they generally include having control over one s work (autonomy), reducing high role overload, feeling valued and respected by one s employer, being included in decision-making, and having opportunities for continuing education and professional development. iv Tested Solutions for Eliminating Canada s Registered Nurse Shortage

9 4. Educational organizations, professional associations and/or colleges, student associations and governments should partner to examine opportunities to improve the retention of nursing students. Factors to consider include pre-admission requirements, guidance and campus counselors, remediation, availability of faculty, student financial support and teaching methods. 5. Governments, educational organizations and professional associations and/or colleges should collaborate at a pan-canadian level to increase enrolment in RN education programs by considering a variety of delivery models, availability of faculty, location of programs and opportunities for interprofessional education. 6. HHR planning should employ a continuous, comprehensive, multifaceted approach considering a variety of policy options (such as those tested in this report) to achieve greater self-sufficiency. Investments in one policy area, such as improving work environments, may simultanelously affect many issues, such as retention, health of nurses and student attrition. 7. Governments, employers and professional associations and/or colleges should invest in data including coordinating and linking data currently collected with a particular focus on: the amount and type of services RNs provide according to the health needs of the patient/population; any aspect of the work done by RNs working outside acute care (e.g., long-term care, home care, in the community), the level of service they provide, activity rates, participation rates and productivity; level of retention of nurses, both practicing and newly educated; and rates of attrition among RN education programs. 8. Governments, employers, unions and professional associations and/or colleges should invest in a national health provider unique identifier to provide more accurate and reliable HHR data. Tested Solutions for Eliminating Canada s Registered Nurse Shortage v

10

11 CONTENTS FOREWORD...i EXECUTIVE SUMMARY...iii INTRODUCTION...1 METHODS...3 Conceptual Framework...3 Analytical Framework...4 Simulation Model...5 Data Elements...7 The Training Module...7 The Supply Module...8 The Work and Productivity Module...8 The Needs Module...10 RESULTS...14 Initial Data Analysis...14 Training Module...14 Supply Module...14 Work and Productivity Module...14 Needs Module...15 Simulation Modelling Results...25 Development and Testing of Policy Scenarios...29 CONCLUSIONS...37 RECOMMENDATIONS...38 GLOSSARY...40 REFERENCES...41 Tested Solutions for Eliminating Canada s Registered Nurse Shortage vii

12 Appendix I: Developing an Aboriginal-Specific HHR Planning Model...45 Health Needs...45 Age Groups...46 Service Levels...46 Appendix II: List of Data Sources...47 Appendix III: Data Limitations and Associated Assumptions...49 Appendix IV: Measurement of Low Income...51 Appendix V: Forecasting Future Needs Based on Past Trends...53 Population...53 Health Needs...53 Trending Methodology...55 Health Status and Income...56 viii Tested Solutions for Eliminating Canada s Registered Nurse Shortage

13 INTRODUCTION The quarter-million registered nurses (RNs) in Canada constitute the largest profession in the health-care workforce. They play a key role in illness prevention and health promotion, as well as treating illness and helping individuals, families and communities throughout the life cycle. Given the importance of this profession to the health and well-being of those it serves, it is crucial that Canada carefully manage its RN workforce so it can meet the diverse and changing health-care needs of the Canadian population. Health human resources (HHR) planning is a priority for many stakeholders, including governments, professional associations and/or colleges, unions, employers, and research funding agencies. HHR accounts for about 70% of health-care operating budgets (Advisory Committee on Health Delivery and Human Resources, 2004). The strategic value of meaningful HHR planning is clear provincial, territorial and federal (jurisdictional) health-care systems need enough health-care providers to meet the requirements for service. While this objective is quite straightforward, identifying the best strategies for ensuring an adequate supply of HHR can be challenging. To project future service requirements, traditional approaches to HHR planning have relied primarily on age- and gender-standardized provider-to-population ratios. However, it is widely acknowledged that provider-to-population ratios are of limited value (Birch et al., 1994; Lavis and Birch, 1997; Birch, 2002; Birch et al., 2003). First, these approaches tend to assume that the future demand for health care is a function of the current supply of health-care providers (this is typically referred to as supply-based planning). Where the workplace deployment and use of RNs, licensed practical nurses (LPNs) and registered psychiatric nurses (RPNs) is changing consider, for example, the rapidly evolving role of the nurse practitioner (NP) in Canada existing measures of service provision (i.e., current provider-topopulation ratios) are of no relevance to future planning. Second, provider-to-population ratio approaches to HHR planning do not take into account the fact that health-care needs of populations will vary between communities and over time, despite the fact these variations are well documented (Evans, 1994; Roos et al., 2001; O Brien-Pallas et al., 2007). Third, supply-based approaches to HHR planning do not account for variations in the productivity of providers between health-care settings and over time. New technologies and new models of service delivery can have a marked effect on provider productivity (Gray, 1982). However, recent reviews have found little evidence of these considerations being factored in by planners or researchers (Tomblin Murphy et al., 2007; Tomblin Murphy et al., 2004). CNA and members of the current project team have been involved in efforts to develop and implement more comprehensive and effective HHR planning processes. Recent work in this field includes the Atlantic Health Human Resources Planning Study (Birch et al., 2005) and work on the Canadian Nurse Practitioner Initiative (Tomblin Murphy et al., 2006). In both these projects, the authors developed models for needs-based HHR planning. These models estimate the future requirements for the services of a given group of providers based on the size, distribution, and levels of health-care needs of the population. In addition, they estimate the future supply of services, based on the size and characteristics of the current workforce as well as trends in entries to and exits from the workforce. With these models, planners can gauge the effects of various policy scenarios, as well as combinations of policies, on the balance between supply of and requirements for health-care providers over short-, medium- and long-term planning periods. Tested Solutions for Eliminating Canada s Registered Nurse Shortage 1

14 The objectives of this study were to estimate, for each year over 15 years, the requirements for and supply of direct/clinical care RNs in Canada (excluding those who are also licensed as NPs), and to develop a simulation model that permits the testing of various HHR policy scenarios. To this end, CNA and the project team have collaborated in building a planning model specific to RNs at the national level (though the model can be adapted to the provincial/territorial level as well). The present study is directly informed by the Atlantic and Canadian Nurse Practitioner Initiative projects, and builds on that work. The conceptual and analytical frameworks used in this study and the earlier ones are discussed in the following chapter. One of the initial objectives of this study was to develop a planning model specific to the health-care needs of Canada s Aboriginal population. Unfortunately, after substantial exploration and consultation with the National Aboriginal Health Organization and the First Nations and Inuit Health Branch, it was determined that the data presently available on the health-care needs and health-care utilization of Aboriginals in Canada are not sufficient to support needs-based-planning for this population, and as such this model has not been developed. However, substantial progress was made in this study in HHR planning for the Aboriginal population, particularly in the identification and assessment of data sources. For a detailed discussion of the effort to develop an Aboriginal-centred HHR planning model, see Appendix I. 2 Tested Solutions for Eliminating Canada s Registered Nurse Shortage

15 METHODS Conceptual Framework The methodological approach in this study is informed by a conceptual framework developed by O Brien- Pallas et al. (2005). This framework has been adopted as a guiding framework for use in HHR planning by Canada s Advisory Committee on Health Delivery and Human Resources (ACHDR, 2005). As seen in Figure 1, the outer oval represents the context of social, political, geographical, technological and economic factors in which HHR planning takes place. Fundamentally, however, HHR planning starts with the health-care needs of the country, province/territory or region in question. Figure 1: Conceptual Framework. Tomblin Murphy & O Brien-Pallas, 2006 Adapted from O Brien-Pallas, Tomblin Murphy & Birch (2005), O Brien-Pallas, Tomblin Murphy, Birch & Baumann (2001) and O Brien-Pallas & Baumann (1997) Tested Solutions for Eliminating Canada s Registered Nurse Shortage 3

16 Analytical Framework The conceptual framework is the foundation for the analytical model, which can then generate simulations and recommendations. The analytical model (Birch et al., 2007) consists of two broad elements: provider supply and provider requirements. Provider supply is, in essence, the answer to the question How many providers are available to deliver health-care services to the population? Supply can be seen as the outcome of two broad determinants: 1. The stock of individuals, namely the number of providers in each age and sex group who are potentially available to provide health-care services 2. The flow of provider time from the stock, influencing the quantity of service output in short, time spent in the production of services. This time depends on: a. The proportion of the current stock participating in providing health care, or the participation rate b. The quantity of time devoted to service provision by those who do participate in the provision of health care, or the activity rate Participation and activity rates represent policy levers for HHR policy-makers and, hence, alternative or complementary approaches for changing provider supply. In addition to changes in the flow of provider time, the size of the stock changes with new entrants (inflows of health-care providers from other countries together with new graduates from within Canada) and departures from the stock (outflows of providers to other countries, and retirements and deaths among providers). In terms of policy responsibilities, education and training (i.e., the production of new providers) are generally separate from the management and regulation of providers (the use of existing providers). Thus provider supply can be seen as the combination of two components: training of new providers and management of existing providers. The second element of the analytical model, provider requirements, has four distinct components: 1. Demography: the number of people by age and gender group in the population. 2. Epidemiology: the rate of health and illness as well as risk factors for future illness across the population subgroups. 3. Level of service: the amount of health-care services to be provided for individuals at different levels of illness or risk of illness. 4. Productivity: the amount of health-care services a full-time equivalent (FTE) provider performs per unit of time. Because each of these components varies across age and gender groups in a population, the analytical model is applied to each age-gender group to come up with the provider requirements for each group. These results are then added together to provide an estimate of total provider requirements. Combining the first three components of the framework demography, epidemiology and level of service yields an estimate of the number of health-care services required by a population, given its size, demographic mix, levels and distribution of health and illness, and levels of service. The fourth component productivity 4 Tested Solutions for Eliminating Canada s Registered Nurse Shortage

17 translates the number of services required into the number of health-care providers required to perform them. The framework for this needs-based approach can be written mathematically as follows: (1), where R t is the number of providers required to meet the service requirements of a population at time t; N i,j,t is the number of providers required to perform each service to patients of age group i and sex j at time t (i.e., the inverse of productivity); Q i,j,t is the number of services required by level of need per person of age group i and sex j at time t; H i,j,t is the proportion of the population by level of need (in the simplest case this would be the proportion of the population who are sick ) for age group i and sex j at time t; P i,j,t is the size of the population of age group i and sex j at time t. Traditional HHR planning methods have limited attention to demographic change, by applying current levels of service use by age and gender to the estimated changes in the size and mix of the future population. This assumes that health-care needs of the population, the types and quantity of health-care services delivered to meet those needs, and the ways in which such services are delivered are constant over time. In contrast, the needs-based model allows policy-makers to explicitly consider the separate effects of each of these factors as they plan for their health-care workforces. Simulation Model Building on the analytical framework, a simulation model (Kephart et al., 2005) has been adapted that simultaneously estimates present and future HHR requirements and present and future HHR supply. The model was designed using a system dynamics approach (Forrester, 1968; Richardson, 1991; Sternman, 2000) and implemented using Vensim (2002) simulation software. Tested Solutions for Eliminating Canada s Registered Nurse Shortage 5

18 Figure 2: Simulation model Adapted from Kephart et al., 2005 The components of the model are grouped into four distinct modules. The training module estimates the flow of new graduates out of education/training programs. The supply module estimates the size of the stock of providers available to deliver health-care services. The needs module estimates the number of health-care services required by a population. The work and productivity module translates service requirements and counts of available providers into standard FTE providers required and available, respectively. The model is not necessarily designed to predict the future, but rather to integrate knowledge of different components of the health-care system to better understand how various factors affect the supply of and/or requirements for health-care providers. The model enables policy-makers to rehearse potential policy changes; by altering variables in the model, they can see the effects of each change on the supply of and requirements for a given type of health-care provider. Moreover, because many of the variables in the model lie beyond the scope of HHR policy (e.g., needs for care, population size), it supports consideration of the effects of HHR policies in different future contexts. In this way, it gives policy-makers a means of testing and evaluating policy options to determine the most efficient and effective ways to manage HHR under different future scenarios. Often, a central concern for policy-makers is the gap, or difference, between the number of providers available and the number required hence the prominence of this component in Figure 2; the model has been designed to calculate this gap automatically. 6 Tested Solutions for Eliminating Canada s Registered Nurse Shortage

19 The accuracy of any model s simulations depends greatly on the quality of the data used to run it. Unfortunately, timely access to reliable data is a challenge for HHR planners and researchers alike. This model, therefore, has been designed so that users can easily update it as newer and/or better data become available. The model was originally designed to be able to simulate HHR gaps over a 40-year period. In this study, however, it was used to consider the supply of and requirements for direct/clinical care RNs (with the exception of NPs) in Canada over a 15-year period. This time frame lines up with policy-making in Canada, which uses short- (1- to 2-year), medium- (2- to 5-year) and long-term (5- to 15-year) time frames. The individual elements of the simulation model follow, presented according to the module to which they belong. A list of the sources for each data element is provided in Appendix II, and a tabulated list of any associated assumptions appears in Appendix III. Data Elements The Training Module Domestic nursing education programs account for the bulk of new entrants into the stock of RNs in Canada, and as such are essential to maintaining Canada s supply of RNs. The training module estimates the future number of graduates from Canadian RN entry-to-practice (ETP) education programs using data on enrolment, length of programs, attrition rates and graduate retention rates. All figures for the training module are from the CNA/Canadian Association of Schools of Nursing (CASN) National Student and Faculty Survey of Canadian Schools of Nursing. Enrolments: The number of first-time entrants in all RN ETP education programs in Canada. Program length: The duration of RN ETP education programs in Canada. The value used in the simulations is the average (weighted by program size) of all programs. Program attrition: The proportion of entrants to all Canadian RN ETP education programs who leave the programs before completion. This can also be seen as the opposite of the graduation rate. Seventeen programs at 22 schools/school sites spanning seven provinces provided data as part of the National Student and Faculty Survey of Canadian Schools of Nursing, which supported calculation of attrition rates. Graduate out-migration: The proportion of new graduates from Canadian RN ETP education programs who do not enter practice as RNs in Canada. This includes graduates leaving Canada to practice in other countries. Estimates of this proportion were available from British Columbia, Ontario and Saskatchewan, and the average of these three provincial rates was used to approximate the national rate. The Ontario report, however, does not specify whether out-migration from Ontario is to other parts of Canada or to other countries; thus this report may overestimate the rate of national outmigration of Ontario graduates. As previously noted, the model can be easily updated should a better estimate become available. New graduates: The number of new graduates from Canadian RN ETP education programs each year is determined by enrolment, program length, attrition and out-migration rates. The age distribution of these new graduates is based on the average age distribution of Canadian-educated first-time writers of the Canadian Registered Nurse Examination over the past seven years (provided by CNA). Tested Solutions for Eliminating Canada s Registered Nurse Shortage 7

20 The Supply Module The supply module estimates the future size of the RN stock in Canada. This is a function of its current size and the flow of RNs into and out of the stock. This estimate is based on the following factors: Existing RN stock: The number of RNs in Canada who are potentially available to provide nursing services. This includes all individuals with the appropriate RN licensing, whether or not they are currently practising direct nursing care as an RN. This does not include RNs who are also licensed as nurse practitioners. Data on the current size of the stock of RNs in Canada by age were obtained from the Canadian Institute for Health Information (CIHI) s national Regulated Nursing Database. In-migration: The number of RNs entering practice in Canada from other countries. This is an estimate based on the number of internationally educated writers of the Canadian Registered Nurse Examination in In-migration is combined with new entrants from the training module (minus out-migrating graduates) to arrive at the total flow of new RNs into the stock. Exit rates: Every RN, at some point, will cease to practice, as is true of any professional. They do so for a variety of reasons, such as relocation to another country, burnout, retirement or death. Exit rates are the age-specific rates at which RNs in Canada cease renewing their professional registration and hence are no longer licensed to provide RN services. Ideally, national exit rates would be measured using a national RN database with a unique identifier for each nurse. In the absence of such a system, CNA obtained data on age-specific provincial/territorial-level non-renewal rates from provincial/ territorial RN regulatory bodies. These were then adjusted for interprovincial/territorial migration, since these relocations do not represent a loss of RNs to Canada. Based on these factors, the supply module provides estimates of the number of RNs available to provide care over time. As we have seen, however, the quantity of services provided by the RN stock depends on factors associated with working practices, such as the productivity and activity rates of the RNs in the stock. These are accounted for in the work and productivity module. The Work and Productivity Module The degree to which the supply of RNs (i.e., the RN stock) is sufficient to meet requirements depends on the amount of time RNs contribute to service delivery and their productivity during that time. Hence, the work and productivity module translates the size of the RN stock into the supply of RN hours and the rate at which RNs can provide services over time. Thus the work and productivity module allows for the comparison of the supply of and requirements for RN services. The module consists of the following components: Participation rate: The proportion of all licensed RNs in Canada who are employed in direct patient/clinical care. This does not include licensed RNs who are working in other sectors (such as research, education, policy or administration), who are not currently in the labour market (e.g., on parental or educational leave) or who have become licensed as nurse practitioners. Activity rate: The proportion of an FTE s hours that the average RN employed in direct/clinical care provides. Ideally, this would be estimated using Canada-wide administrative data (such as CIHI Management Information System data) on the worked hours of RNs. However, figures were available only for acute-care RNs and only for three provinces New Brunswick, Nova Scotia and Ontario. 8 Tested Solutions for Eliminating Canada s Registered Nurse Shortage

21 Therefore, the ratio of worked hours to actual RNs in these three provinces was used as a proxy for the ratio for all of Canada. As noted, these values are easily replaced should actual data become available. The National Survey on the Health and Work of Nurses (CIHI, 2006) included nurses own estimates of the hours they worked in an average week. However, when extrapolated to yearly estimates, these self-reported data differed substantially from all available administrative data, perhaps indicating that self-reported weekly estimates are not an accurate basis for estimating yearly worked hours. Productivity: The number of services performed per FTE RN per year. Because of the significant differences in the type of care provided across various sectors, this value was estimated separately for RNs in acute care, long-term care, home care and community care (which includes primary health care and public health nursing). Estimating productivity was complicated, however, by a general absence of administrative data on the actual amounts of care provided by RNs, except in the acute-care sector. In each sector, productivity was estimated by taking the best available measure of the services performed per unit of time and dividing that by the number of FTE RNs working in that sector during that time. In acute care, the best available measure of RN services was acuity-adjusted episodes of care. These data were provided by CIHI and broken down by in-patient, day surgery and emergency episodes of care. No figures were available from Quebec, day surgery data were not available from Alberta, and emergency data were available only from Ontario. In an attempt to estimate the total number of acuity-adjusted episodes of care in Canada, the model relied on provinces for which data were available. The ratio of the population to the number of acuity-adjusted episodes of care by age and sex in those provinces for which data were available was used as an adjustment factor to amplify the provincial data to a national estimate. As noted, CIHI data on productivity of RNs working in acute care were limited to RNs providing in-patient, day surgery and emergency care. Without information on the contribution of RNs to out-patient care in acute-care organizations, the model underestimates the total productivity of acute-care RNs. However, the impact of this on the model is more or less negated by corresponding underestimates in the levels of service they provide (see the level of service discussion that follows). As noted, the model can easily be updated should more relevant or complete data become available. In long-term care, the only available measure of services provided by RNs was the total patient days of care provided in long-term care facilities. This information was obtained from Statistics Canada s Residential Care Facilities Survey (RCFS Statistics Canada, 2008a). In home care, the only measure of RN services was information from patients about the number of home-care visits they received. These self-reported data came from Statistics Canada s Canadian Community Health Survey (CCHS Statistics Canada, 2005a). No information was available on the services performed during visits or their duration. Similarly, in the community sector, the only available measure of the services provided by RNs was reports from patients on their use of community nursing, again from the CCHS. Information was limited to the number of consultations with a nurse, with nothing on the services performed or the duration of the consultations. Tested Solutions for Eliminating Canada s Registered Nurse Shortage 9

22 In this way, the work and productivity module translates the number of available RNs (estimated by the training and supply modules) into the number of FTEs available based on the proportion of RNs participating in patient care and the hours they work. The rate at which an RN FTE can provide patient care services translates the needs-based service requirements (estimated by the needs module) into the number of RN FTEs required. The Needs Module Estimating the service requirements for RNs in Canada based on the health-care needs of the population uses the following components: Population: The size of the Canadian population by age and sex, at present and projected into the future. Statistics Canada s (2005) medium-growth population projections were used to estimate future population size. Need: An estimate of the level of need for RN services. Because of the significant differences in the services provided by RNs across sectors, different measures for needs were used for each sector; see Table 1. The aim was to identify indicators of population health needs that were appropriate to the particular sector, and that were independent of service use. However, choice of need measures was limited by availability of data over several years, which is required to consider trends over time. For acute-care RN services, the measure of need was a combination of rates of injury and chronic conditions, and self-assessed unmet requirements for health-care services. Utilizing injuries and chronic conditions as need indicators was based on the need for acute care RN services arising from either acute episodes of need brought on by chronic conditions (e.g., asthma attacks, hypertensionrelated cardiac incidents) or injury (e.g., fractures, lacerations). Unmet need was included since it suggests a level of service is required beyond that reflected in current utilization patterns. A potential problem with using chronic conditions and injury as needs indicators is their dependence on utilization. Because individuals are identified as having a chronic condition or an injury based on a health professional s diagnosis, as access to health care and diagnostic criteria expand, rates of diagnoses will increase even if the prevalence in the population does not change; the population may appear to be becoming less healthy when in fact it is not. Combined rates of injury and chronic conditions were estimated based on self-reported data from the CCHS. The proportion of the population within each level of need reporting unmet needs for hospital services (emergency, overnight and outpatient) was calculated from the CCHS. Categories are not mutually exclusive; that is, individuals reporting unmet need in more than one service are counted more than once. To quantify a level of service for this group, the total number of services for each need level was inflated by the percentage reporting unmet need. The total inflation amount can be considered the level of service for those with unmet need 1. 1 This is used as a possible scenario using the assumption that the required proportionate increase in level of service to accommodate current self-reported unmet need for care is equal to the proportion of the population within each need level reporting unmet need for care. However, there is no objective basis for this assumption and alternative scenarios for the levels of unmet need for care can easily be accommodated in the model should information become available. 10 Tested Solutions for Eliminating Canada s Registered Nurse Shortage

23 For long-term care, the proportion of the Canadian population requiring services was estimated by combining: (1) individuals already in long-term care facilities (on the assumption that they have met explicit needs criteria for admission); (2) individuals occupying acute-care beds while waiting for beds in long-term care facilities (i.e., alternate level of care, or ALC, patients); and (3) individuals living alone and being unable to perform personal care and/or mobilize in their homes without help. The population in long-term care was estimated using the Residential Care Facilities Survey. The number of individuals living alone and requiring assistance with personal care and/or to move about their homes was estimated using data from the CCHS. The number of ALC patients was not available at a national level. Corresponding data for Ontario were provided by the Ontario Hospital Association. These were used to generate a rate of ALC patient population. This rate was then used as a proxy for the average rate of ALC patients in Canada. For home-care RN services, a combination of met and unmet need was used. The proportion of the Canadian population requiring home-care RN services includes both those who already receive such services (having met explicit criteria of need used in the assessment procedures of home-care programs) and those who do not but identify themselves as requiring such services. The assumptions of this approach are that Canadians do not receive home-care nursing services unless they meet objective criteria of need and that self-reports of unmet need for home-care services are valid. Data on both measures were obtained from the CCHS. For community RNs, the distribution of self-assessed general health status by income level was used. This is because community RNs provide primary health-care and/or public health services, with much of the public health work aimed at low-income families or individuals. Self-assessment prompts people to visit a primary health-care provider, and such providers respond to general health issues. Measures of self-assessed general health status have also been found to correlate with a wide range of health and socioeconomic variables, both at the population and individual level (Birch et al., 1996). Data on self-assessed health by income level were obtained from the CCHS. For a detailed discussion of how income levels were measured, see Appendix IV. As with the acute-care sector, unmet needs were estimated using self-assessed reports of health-care services perceived to be required, but not received. The percentage of respondents in the CCHS who report trying unsuccessfully to receive services from a doctor s office, walk-in clinic, appointment clinic or a community health centre was calculated within each level of need for community nursing care, with multiple selections counted more than once. Note, however, that only a fraction of the respondents reporting unmet care from a doctor s office was included as not all offices employ an RN. A lack of data on the proportion of services performed at a doctor s office that could be administered by an RN meant expert consultation was required to estimate this value. Twenty-five percent of those reporting unmet care from a doctor s office was chosen, but this should only be viewed as a possible scenario until more detailed data become available. Finally, as was done in the acute-care sector, the level of service for each need category was inflated by the percentage with unmet needs 2. 2 As in Note 1, this is used as a possible scenario using the described assumptions, and alternative scenarios for the levels of unmet need for care can easily be accommodated in the model should information become available. Tested Solutions for Eliminating Canada s Registered Nurse Shortage 11

24 Table 1: Needs Indicators and Data Sources by Sector Sector Needs Indicators Data Source Acute care Long-term care Home care Community care Injury Number of chronic conditions Living alone and requiring assistance with activities of daily living (ADLs) Use of alternative level of care (ALC) beds in hospitals Residence in long-term care facilities Receipt of home-care nursing services (publicly or privately funded) Self-reported unmet need for home-care nursing services Self-assessed health status by income level Self-assessed unmet need CCHS CCHS CCHS OHA RCFS CCHS CCHS CCHS CCHS When planning to allocate resources based on health-care needs, it is necessary to estimate what those future needs will be. In this study, population demographics population and health-care needs were projected over a 15-year time horizon and fed into the planning model. No one can predict with certainty how the health-care needs of Canadians will change in the future. Still, if these needs continuing to change as they have in the recent past, it is possible to simulate the effects of these changes by fitting statistical models to the historical data and projecting forward. For a detailed discussion of the methodology used to project past trends in health needs into the future, see Appendix V. Level of service: This measures the amount of care or services a person requires by level of need. In the absence of gold standards, existing levels of service (where available) were used as a baseline, keeping in mind that the model can be changed or updated as desired. Moreover, unlike the other elements of the model, the level of service will be influenced by non-clinical factors such as the resources available for the delivery of care and the other demands on those resources. The same unit of service volume was used in the calculation of both level of service and productivity, before these measures were multiplied to estimate FTE RN requirements. Because of significant differences in the type of care provided across sectors, level of service was estimated separately for acute care, long-term care, home care and community care. In acute care, data on acuity-adjusted episodes of care were not available by level of need. Instead, estimates were made by combining data from various sources. Specifically, the distribution of actual hospitalization by chronic conditions and injuries was assumed to be the same as the self-reported distribution in the CCHS data. Institutional-level data (provided by CIHI, as described previously) on actual hospitalizations by age and sex were combined with self-reported population-based data on hospitalizations by age and sex for the same year to calculate the ratio of actual utilization to self-reported use by age and gender. This ratio was then applied to self-reported use by level of need (i.e., chronic conditions and injuries). 12 Tested Solutions for Eliminating Canada s Registered Nurse Shortage

25 As no administrative data were available on the amount of outpatient services provided by RNs, this value could not be included in simulations; thus the level of service provided by RNs is underestimated in these models. However, the impact of this on the accuracy of the model is more or less negated by the fact that productivity is similarly underestimated due to this same data gap. In long-term care, the amount of care measured by days of care in long-term care per patient per year used was from work by Tomblin Murphy et al. (2008) for Ontario. The level of service for long-term care patients in Ontario was used as a proxy for the average level of care for long-term care for Canada. Because the level-of-service measure was not estimated by patient age and gender, the same rate was applied to all age and gender groups admitted to long-term care. The report by Tomblin Murphy et al. (2008) also provided an estimate of level of care for the homecare sector, but the same limitations applied to using this source of data as those listed previously for long-term care. For the community sector, the number of community nursing consultations per person by income level and self-assessed health status was estimated from self-reported data in the CCHS. For each year of the projection period, the needs module estimates the future requirements for the healthcare services of RNs by combining the projected size of the population and its health status rates with estimates of the amounts of RN health-care services it receives by health status. Using RN productivity data (number of services delivered per FTE RN per year), the requirement for services is converted to the requirement for FTE RNs in each year. This value is then compared with the estimate of the future availability of FTE RNs in each of those years to determine whether there will be a surplus or shortage of RNs in Canada. The results produced by the simulation model (see the following section) are dependent on the range and quality of data available and should be interpreted with caution; still, they are based on the best data available. The qualitative findings, in particular, provide valuable insight to planners. Further, this approach will become more valuable as new and better-quality data become available to support efficient, effective HHR planning. Tested Solutions for Eliminating Canada s Registered Nurse Shortage 13

26 RESULTS Initial Data Analysis Before the model was populated and simulations run, substantial analysis of the data from various sources was required to extract the necessary information. The results of this analysis are organized by the module to which the data pertain. Training Module According to data from the Nursing Education in Canada Statistics, report, in 2007 there were an estimated 13,000 first-time entrants in all ETP programs. Recently announced funding increases in several provinces will increase this number by about 900 by the end of the academic year. Based on figures from the 17 programs that provided sufficient data to calculate graduation rates, approximately 28% of students admitted to RN ETP education programs do not complete them. In addition, recent estimates from three provinces put the proportion of new graduates staying in the province to practise at 95% (CRNBC, 2005), 98% (Insightrix, 2007) and 96% (OMHLTC, 2008) in British Columbia, Saskatchewan and Ontario, respectively. Data from the Nursing Education in Canada Statistics, report indicate that the number of RN graduates in Canada nearly doubled between 1999 and 2007, increasing from 4,833 to 9,447. The average age of these graduates was approximately 26. Supply Module Data from the Regulated Nursing Database show that as of 2007 there were approximately 270,000 licensed RNs in Canada. Of these, about 217,000 or 81% are employed in the delivery of direct/clinical care. The average age of these RNs is 44. There is an annual influx of approximately 1,000 RNs from other countries entering practice in Canada, with an average age of 35. Provincial/territorial registration data provided to CNA, adjusted for inter-jurisdictional migration, indicate that about 3% of RNs in Canada cease to renew their registration each year (varying from less than 2% for 35-to-49-yearolds to more than 10% for those over 60). Stated another way, these data indicate that Canada retains about 97% of its potential RN workforce from year to year. Work and Productivity Module CNA (2008) has defined an FTE RN as one who accumulates 1,950 earned hours in a year. This is based on RN collective agreements in Canada. However, since RNs do not deliver patient care during vacations or sick time, the model uses actual worked hours to measure the activity of RNs. According to CIHI Management Information System data on nursing hours in hospitals in Nova Scotia, New Brunswick and Ontario, about 83% of earned hours are actually worked; thus an FTE RN is defined here as one who works 83% of 1,950 hours, or 1,618.5 hours, per year. Other administrative data sources indicate that the 54,000 hospital-based unit-producing RNs in these provinces worked a total of about 75 million hours in 2005, or over 1,100 hours per RN. This quantity is about 86% of the 1,618.5 worked hours per year an FTE accumulates, meaning the average acute-care RN in these provinces represented approximately 86% of an FTE. As this was the best available estimate of the activity level of any type of RN, 86% was used as the activity level of all RNs. 14 Tested Solutions for Eliminating Canada s Registered Nurse Shortage

27 Table 2: Training, Supply, and Work and Productivity Data Data Element Value ETP enrolment 13,000 ETP program attrition 28%* ETP program length 3.6 years* ETP new graduates 9,447 ETP Graduate out-migration 5% In-migration 1,023 Existing provider stock 270,485 Exit rates All ages: 3% Ages 25 34: 6% Ages 35 49: 2% Ages 50 59: 3% Ages 60+: 11% Participation rate 81% Activity rate 86% * Average weighted according to size of program enrolment Needs Module One advancement of the needs-based approach over traditional HHR modelling is the ability to account for estimated future changes in population health needs. This section opens with population projections from Statistics Canada. It then presents, by sector and for each indicator, the age/gender distribution for the most recent year in which data are available. The age/gender distribution data are followed by estimated population health needs over 15 years. For reasons of brevity, the estimates are for all ages and both genders combined, although the separate age and gender specific estimates are used in the modelling. POPULATION Population projections from Statistics Canada indicate that the Canadian population is aging, and will continue to age through to 2022 (see Figure 3.) Tested Solutions for Eliminating Canada s Registered Nurse Shortage 15

28 Figure 3: Age Distribution of Canadian Population, 2007 and 2022 Source: Statistics Canada Population Projections By 2022, the proportion of Canadians age 65 and over will increase from approximately 13% to 19%, so that in 15 years almost one Canadian in five will be a senior citizen. This is an important consideration for health-care planners, as older people tend to require more health-care services, other things being equal (Eyles et al., 1991). NEEDS INDICATORS Acute-Care Sector Injuries and chronic conditions were the key needs indicators in this sector. Prevalence of injuries is highest among youth, and within this age group it is highest in young males (Figure 4). Since these data relate to injury requiring the attention of a health-care professional, the prevalence of such cases is particularly relevant for HHR planning. Figure 4: Prevalence of Injury by Age and Sex, Canada, 2005 Source: 2005 CCHS 16 Tested Solutions for Eliminating Canada s Registered Nurse Shortage

29 There is a clear gradient in the prevalence of chronic conditions with age; older individuals tend to have the most chronic conditions (Figures 5a and 5b). In addition, older females tend to have slightly more chronic conditions than older males. Figure 5a: Prevalence of Chronic Conditions by Frequency and Age Males, % 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% Proportion of Canadian Population by # of Chronic Conditions and Age Males Source: 2005 CCHS Figure 5b: Prevalence of Chronic Conditions by Frequency and Age Females, % 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% Proportion of Canadian Population by # of Chronic Conditions and Age Females Source: 2005 CCHS Tested Solutions for Eliminating Canada s Registered Nurse Shortage 17

30 As illustrated in Figure 6, individuals with more chronic conditions tend to report more days spent in hospital. Further, this relationship appears more pronounced among those who reported a significant injury within the past year. Data of this nature are directly relevant to plans for the provision of health care based on population health needs. Figure 6: Self-Reported Hospital Days per Person per Year in Canada by Injury Status and Number of Chronic Conditions, CCs No Injuries Average # of Hospital Days per Person per Year by Injury Status and # of Chronic Conditions, Canada, CC 2 CCs 3+ CCs 0 CCs Injured 1 CC 2 CCs 3+ CCs Source: 2005 CCHS Included in this section are figures that show both the point values and estimated change in future need for the various needs indicators. As mentioned, the trends illustrated are not used directly in the model. The calculated proportions in the figures are derived from data that combine males and females as well as all age groups, whereas those used in the model are specific to age and sex groups. While the graphs help convey the methods and give an overall sense of the estimated average change in the population for each need indicator, results for each age/sex group within the population may show different estimated changes in need. For example, younger age groups may demonstrate a falling trend while older individuals an increasing trend for any particular indicator. Rather than weigh this report down with figures, those included here reflect the full population (all age and gender groups). As well, for the sake of brevity, figures show the presence of any chronic conditions rather than stratifying by the total number of conditions as utilized in the model. Finally, chronic conditions and injuries are separated for clarity. The left-hand panels in Figures 7 and 8 show the individual point estimates used to calculate the trend function or equation. The number of data points used depends on data availability. Examining these prevalence rates across survey years gives a sense of whether the overall trend for each indicator is rising, falling or more or less flat. The right-hand panel shows the plot of the predicted future values over a 15-year time horizon using the power trend function as described in Appendix V. Note that the value for 2005 in the trend plot is the same as the 2005 value in the left-hand panel. 18 Tested Solutions for Eliminating Canada s Registered Nurse Shortage

31 As indicated in Figures 7 and 8, the all-ages/both-genders projection data estimate a slight increase in the prevalence of chronic conditions and somewhat smaller increase in the incidence of injuries over 10 to 15 years. If all else remained equal, this would translate into an increase in the level of future health-care services required in the acute-care sector. Figure 7: Prevalence of Any Chronic Conditions in Canada, 2001, 2003, 2005 percentage 45.0% 40.0% 35.0% 30.0% 25.0% 20.0% 15.0% 10.0% 5.0% 0.0% Point Estimates Any Chronic Condition* Males and Females, All Ages 33.8% 35.3% 35.3% * Diagnosed with high blood pressure, diabetes, cancer, arthritis, heart disease or COPD. percentage 16.0% 14.0% 12.0% 10.0% 8.0% 6.0% 4.0% 2.0% 0.0% percentage 45.0% 40.0% 35.0% 30.0% 25.0% 20.0% 15.0% 10.0% 5.0% 0.0% Estimated Change in Needs Indicator* Any Chronic Condition Males and Females, All Ages * Power function based on point estimates from 2001, 2003, Figure 8: Incidence of Injury in Canada, 2001, 2003, 2005 Point Estimates Any Injuries in Past 12 Months* Males and Females, All Ages 13.4% 13.2% 13.7% As described previously, self-assessed unmet need for acute-care services was also included as an indicator for this sector. The proportion of the population within each level of need reporting unmet needs for hospital services (emergency, overnight and outpatient) was calculated from the CCHS. Categories are not mutually exclusive; that is, individuals reporting unmet need in more than one service are counted more than once. To quantify a level of service for this group, the total number of services for each need level was inflated by the percentage reporting unmet need. The total inflation amount can be considered the level of service for those with unmet need 3. percentage 16.0% 14.0% 12.0% 10.0% 8.0% 6.0% 4.0% 2.0% 0.0% Estimated Change in Needs Indicator* Any Injuries in Past 12 Months Males and Females, All Ages * Power function based on point estimates from 2001, 2003, This is used as a possible scenario using the assumption that the required proportionate increase in level of service to accommodate current self-reported unmet needs for care is equal to the proportion of the population within each need level reporting unmet need for care. However, there is no objective basis for this assumption; alternative scenarios for the levels of unmet need for care can easily be accommodated in the model should information become available. Tested Solutions for Eliminating Canada s Registered Nurse Shortage 19

32 NEEDS INDICATORS Community Care Sector Self-assessed general health status was the key need indicator in the community sector. The data presented in Figure 9 show that self-reported fair or poor health is more common among older individuals, both male and female, low- and non-low income. However, these lower levels of health are more common among low-income individuals, particularly at younger ages. Figure 9: Prevalence of Fair/Poor Health by Age, Sex and Income Status in Canada, % Prevalence of Fair/Poor Health by Age, Sex, and Income Level 15% 10% 5% 0% Non Low Income Low Income Non Low Income Low Income Female Male Source: 2005 CCHS The data in Figure 10 suggest that individuals with low income levels tend to consult nurses in the community substantially less than those with non-low income levels, regardless of their health status. Among those with non-low income, those who report their health status as good or better report twice as many consultations with community nurses than do those who report their health status as fair or poor. 20 Tested Solutions for Eliminating Canada s Registered Nurse Shortage

33 Figure 10: Community Nursing Consultations per Person per Year by Income Status and Self-Assessed Health Status, 2005 Average Number of Community Nursing Consultations per Person per Year by Income Status and Self-Assessed Health Excellent/Very Good/Good Low Income Fair/Poor Excellent/Very Good/Good Non Low Income Fair/Poor Source: 2005 CCHS Figure 11 provides all-ages/both-genders estimates on future self-assessed health status. For low-income Canadians the prevalence of fair/poor health status is projected to increase over 10 to 15 years. The corresponding prevalence among those not in low income is expected to be unchanged. Note that the trend is estimated within income categories, meaning that predictions are made for levels of health status, but not for the proportion of those results in the low-income category. For an explanation of how lowincome individuals were identified, see Appendix IV. Tested Solutions for Eliminating Canada s Registered Nurse Shortage 21

34 Figure 11: Prevalence of Fair/Poor Health in Canada by Income Status, 1994, 1996, 2003, 2005 percentage percentage 40.0% 35.0% 30.0% 25.0% 20.0% 15.0% 10.0% 5.0% 0.0% 40.0% 35.0% 30.0% 25.0% 20.0% 15.0% 10.0% 5.0% 0.0% Point Estimates Fair/Poor Health Low Income Males and Females, All Ages 17.4% 14.4% 23.1% 23.6% Point Estimates Fair/Poor Health Not in Low Income Males and Females, All Ages 8.7% 6.5% 8.4% 8.4% Estimated Change in Needs Indicator* Fair/Poor Health Not in Low Income Males and Females, All Ages As described, self-assessed unmet need for community nursing care was included as a needs indicator for this sector, as well as self-assessed health status and income. As one possible estimate of unmet community nursing needs, the percentage of respondents in the CCHS who report trying unsuccessfully to receive services from a doctor s office, walk-in clinic, appointment clinic or a community health centre was calculated within each level of need for community nursing care, with reported unmet need in multiple service areas counted more than once. Because not all doctor s offices have an RN, only a fraction of the respondents reporting unmet care in this setting were included in this estimate. A lack of data on the proportion of services performed at a doctor s office that could be administered by an RN meant that expert consultation was required to estimate this value. Twenty-five per cent of those reporting unmet care from a doctor s office was chosen, but this should only be viewed as a possible scenario until more detailed data become available. Finally, as was done in the acute-care sector, the level of service for each need category was inflated by the percentage with unmet needs 4. percentage percentage 40.0% 35.0% 30.0% 25.0% 20.0% 15.0% 10.0% 5.0% 0.0% 40.0% 35.0% 30.0% 25.0% 20.0% 15.0% 10.0% 5.0% 0.0% Estimated Change in Needs Indicator* Fair/Poor Health Low Income Males and Females, All Ages * Power function based on point estimates from 1994, 1996, 2003, * Power function based on point estimates from 1994, 1996, 2003, As in Note 1, this is used as a possible scenario using the described assumptions, and alternative scenarios for the levels of unmet need for care can easily be accommodated in the model should information become available. 22 Tested Solutions for Eliminating Canada s Registered Nurse Shortage

35 NEEDS Long-Term Care The data in Figure 12 show that the vast majority of the need for long-term care is among older individuals (age 75 and over) and that this need is greater among females. Further, as need for long-term care is defined here, these data indicate that the amount of unmet need is relatively small in comparison to the need being met, and this unmet need is most prevalent among those age 75 and older, particularly females. Figure 12: Breakdown of Need for Long-Term Care in Canada, 2005 Proportion of Canadian Population Requiring Long-Term Care Services, % 9% 8% 7% 6% 5% 4% 3% 2% 1% 0% Lives Alone, Needs Help with ADLs Residing in Acute Care Facility, Awaiting LTC Bed Females Males Already Residing in LTC Facility Sources: 2005 CCHS, 2005 RCFS, Ontario Hospital Association (OHA) As seen in Figure 13, five years of data were used to estimate the future change in need represented by the proportion of those requiring help with daily personal care and mobility (referred to activities of daily living, or ADL) and who live alone. When all ages are included, the trend line suggests there will be an increase in need over 15 years, but the authors note there is wide variation within age and gender categories (data not shown). Tested Solutions for Eliminating Canada s Registered Nurse Shortage 23

36 Figure 13: Prevalence of Requiring Assistance with ADL and Living Alone in Canada, 1994, 1996, 2001, 2003, 2005 percentage 0.20% 0.15% 0.10% 0.05% 0.00% Point Estimates Assisted Daily Living* Males and Females, All Ages 0.06% 0.18% NEEDS Home Care 0.17% 0.10% 0.09% * Defined as those needing assistance with mobility and personal care and who live alone. As is the case with long-term care, the data on home-care nursing services in Figure 14 show that most of the need is among older individuals, particularly females. In this sector, though, the difference in need between those age 75 and older and younger individuals is not as pronounced. And again, as with the long-term care sector, it appears that the bulk of the need (as it is defined here) for home-care nursing services is being met, mostly by publicly funded home-care programs. Figure 14: Breakdown of Need for Home-Care Nursing Services in Canada, 2005 percentage 0.20% 0.15% 0.10% 0.05% 0.00% Estimated Change in Needs Indicator* Assisted Daily Living* Males and Females, All Ages * Power function based on point estimates from 1994, 1996, 2001, 2003, % 9% 8% 7% 6% 5% 4% 3% 2% 1% 0% Proportion of Canadian Population Requiring Home-Care Nursing Services, Females Males Source: 2005 CCHS Received Publicly Funded Home Care Nursing Only Received Privately Funded Home Care Nursing Only Received Both Reported Unmet Home Care Nursing Needs Figure 15 suggests there will be a slight fall in the future need for nursing services in the home-care sector. As with ADL, however, there is variation within age groups and between genders (data not shown). Also, data were only available for two years, meaning few data points were available with which to calculate a future trend. 24 Tested Solutions for Eliminating Canada s Registered Nurse Shortage

37 Figure 15: Prevalence of Need for Nursing Services in Home Care in Canada, 2003, % 2.00% 1.50% 1.00% 0.50% 0.00% Point Estimates Nursing Services, Home-Care Males and Females, All Ages 2.03% 1.94% Estimated Change in Needs Indicator Nursing Services, Home-Care Males and Females, All Ages Making HHR policy decisions based on the health needs of the population is the foundation of needsbased HHR planning. But even though past and current levels of population health needs may be understood, it is difficult to predict how these needs will change in the future. For example, even though the prevalence of chronic conditions has been fairly well documented over time, no one can predict with certainty how this prevalence will change; it may double over the next decade, drop by half in the next year, remain constant for the forseeable future only time will tell. However, using statistical models fitted to recent historical data and projecting forward, one can estimate how many health-care providers would be required should recent trends in health status continue over the next few years. One benefit of this approach is that policy-makers can at least have some sense of likely fluctuations in the number of providers required, based on reasonably expected changes in the population s health status. Another merit of these models is that they can illustrate the potential benefits like improvements in population health of such things as public health initiatives. percentage 2.50% 2.00% 1.50% 1.00% 0.50% 0.00% * Power function based on point estimates from 2003 and Simulation Modelling Results All the previous data presented were used in the simulation model to estimate the present and future supply of, and requirements for, RNs in direct/clinical care (less NPs) in Canada. The results of these simulations follow. To serve as a baseline when interpreting these results, a status quo scenario was simulated that is, one in which no changes were made to RN HHR policies. The reader must bear in mind that these results are based on the best data available at the time of the study, as well as on a number of assumptions (see Appendix III) made necessary by limitation in these data, as detailed in the previous chapter. Development of more comprehensive and reliable data sources and subsequent updating of the model will strengthen the validity of these projections. The two immediate determinants of the gap in RN FTEs are the number of RN FTEs required and the number of RN FTEs available. Figure 16 depicts the number of RN FTEs required through to 2022 under two needs scenarios: one in which the health needs of Canadians do not change, and another in which these needs continue to follow trends observed over the past 10 years. While the trends in the individual needs indicators take a variety of directions, overall it appears that the needs of Canadians for RN healthcare services are increasing. Tested Solutions for Eliminating Canada s Registered Nurse Shortage 25

38 Figure 16: FTE RNs Required Under Two Needs Scenarios These data indicate that as of 2007, Canadians required the services of approximately 198,000 FTE RNs. If the health needs of Canadians remain unchanged for 15 years, the number of RN FTEs required will increase by roughly 27%, from about 198,000 to 253,000, over that period. If needs follow observed trends, the number of RN FTEs required will increase by about 30%, from 198,000 to around 259,000, over that period. Thus a failure to consider the changing health needs of Canadians could result in an underestimation of the number of required RNs by about 6,000 FTEs over 15 years. The number of RN FTEs available over time is the same under both needs scenarios, since population health needs do not influence the availability of RNs. As of 2007, there were approximately 188,000 FTE RNs available to provide direct clinical nursing care in Canada. If RN management policies remain the same, this number will be relatively constant through to 2022, increasing about 6% from 188,000 to 199,000. Next, the difference between RN FTEs available and required the gap over time under these two needs scenarios is simulated. As was just noted, as of 2007 there were approximately 188,000 RN FTEs available in Canada to provide direct care compared with roughly 198,000 RN FTEs required, so the shortage as of 2007 was just under 11,000 RN FTEs. 26 Tested Solutions for Eliminating Canada s Registered Nurse Shortage

39 Figure 17: Comparison of Canadian RN Gap Under Two Needs Scenarios Figure 17 indicates that if the health needs of Canadians continue to change according to past trends, the shortage of RNs in Canada will increase to almost 60,000 FTEs by If the health needs of Canadians remain as they are today, the gap will be smaller, at approximately 54,000 FTEs. Thus a failure to account for the changing health needs of Canadians could result in an underestimation of the future shortage of RNs by approximately 6,000 FTEs. The results suggest that the gap will grow under both scenarios. This is because while the supply of RNs in Canada will increase over 15 years, the number of RNs required will increase at a higher rate over the same period. This is due at least in part to a growing and aging population and an aging RN workforce. Since the care provided by RNs varies significantly across sectors, it seems worthwhile to examine how the number of required RNs will change in each over time. Tested Solutions for Eliminating Canada s Registered Nurse Shortage 27

40 Figure 18: FTE RN Requirements by Sector Over Time Of all the sectors, acute care has the largest RN workforce, both in number available and number required. In this sector, the number of RN FTEs will increase from about 131,000 to 179,000, or about 37%, by In long-term care, the number of FTE RNs required will increase about 36%, from 27,000 to 36,000, over the same period, while the number of community RNs required will increase 12%, from about 25,000 to 28,000. Finally, the requirement for home-care RNs will increase by roughly 29%, from about 6,900 to 8,900 FTEs, by Without data on the rates at which RNs enter and exit the practice of direct/clinical care by sector, it is not possible to estimate the future supply of RNs and thus the RN gap by sector. However, with data on the current distribution of RNs by sector, it is possible to estimate the initial RN gap by sector, shown in Figure Tested Solutions for Eliminating Canada s Registered Nurse Shortage

41 Figure 19: Simulated RN FTE Gap by Sector at Baseline Based on the assumptions described previously, the shortage would be largest in acute care and long-term care, at roughly 4,500 FTEs, and smaller in home care and community care, at about 700 and 900, respectively. In the next section, the possibility of reducing or even eliminating the shortage of RNs is explored by simulating the effects of a number of policy scenarios. These scenarios were developed in consultation with key stakeholders in the nursing community. Development and Testing of Policy Scenarios To make the policy scenarios tested in this project as valid as possible, CNA sought input from a number of nursing stakeholders. National nursing stakeholders, as well as CNA board members and principal nurse advisors from provincial and territorial governments, were given a list of potential policy scenarios and asked to pick the top five, ranked in order of priority, and/or suggest others. Each of the final scenarios is presented here in comparison to the baseline scenario, which assumes no changes to RN management policies but assumes that the health needs of Canadians follow observed past trends (as detailed in the previous section). Scenario 1: Increased RN Enrolment A number of jurisdictions have recently funded substantial increases to the capacity of their RN ETP education programs. These will result in an increase in seats by approximately 900 from 2007 to The effect on the RN gap of an additional 1,000 students per year from 2009 to 2011 is simulated in Figure 20. Tested Solutions for Eliminating Canada s Registered Nurse Shortage 29

42 Figure 20: Effect of Enrolment Increase on RN Gap 0 10,000 RN Gap After Enrolment Increase vs. Baseline ,000 30,000 40,000 50,000 60,000 70,000 Increase Enrolment 1,000 Per Year from Baseline Increasing RN first-year ETP enrolment by 1,000 per year for three years would result in a substantial reduction in the 15-year RN gap: the shortage would be reduced from about 60,000 to approximately 45,000, a difference of about 25%. However, it is important to note that this policy would have no noticeable effect on the gap until about There are several reasons for the delayed impact of an enrolment increase on the gap. For one, RN ETP education programs are at least two years long (and most are twice that long). If an enrolment increase were implemented in 2009, no additional RNs would be available to the workforce until 2011, and the full effects of the increase would not be felt until Another reason for the delayed impact is that the pool of extra graduates is initially small relative to the existing stock of providers (which is currently about 270,000). In short, although increasing enrolment has been a popular choice of policy-makers seeking to address RN shortages, it is best thought of as a long-term rather than a short-term solution. Scenario 2: Improved Retention of Student Nurses The best available data from RN ETP education programs shows that about 28% of students, or more than one in four, do not complete the program. The effects of reducing this loss rate to 15% over three years are simulated in Figure Tested Solutions for Eliminating Canada s Registered Nurse Shortage

43 Figure 21: Effect on RN Gap of Reducing Student RN Attrition 0 10,000 RN Gap After Reduction in Program Attrition vs. Baseline ,000 30,000 40,000 50,000 60,000 70,000 Reduce Program Attrition to 15% over three years Baseline As Figure 21 indicates, a reduction in attrition rates in education programs has a substantial long-term effect on the RN shortage, reducing the gap by about 24% to roughly 45,000. It is worth noting that this is almost identical to the long-term effect of the substantial enrolment increase simulated previously, and that a reduction in program attrition has slightly more pronounced effects earlier than the enrolment increase. Scenario 3: Improved Retention of Practising RNs Present estimates of the rates at which Canadian RNs do not renew their licences (referred to here as exit rates) range from roughly 2% per year for 35-to-49-year-olds to over 11% for RNs age 60 and over. The simulation depicted in Figure 22 represents the effects on the RN gap of a gradual reduction in these exit rates to 2% for all RNs except those 60 and over; the rates for this group are instead reduced to 10%. One example of a strategy to achieve such retention rates is the 80/20 model of staffing, introduced in Ontario to help retain older nurses. In this model, a nurse spends 80% of his or her time in direct patient care and 20% on professional development. Research at the University Health Network in Toronto found that the initial investment was recovered within two years from lower rates of turnover, absenteeism and overtime (Bournes and Paré, 2007). Another example is a study of components that contribute to positive work environments and job satisfaction of nurses in Ontario working in community and public health (Bookey-Bassett, S., et al, 2008). It found that these include having control over one s work (autonomy); feeling respected and valued by one s employers; being included in decision-making; being able to make a difference in the lives of patients; and having opportunities for continuing education and professional development. These findings are consistent with those from previous studies and across settings. Many other examples of evidence based leading practices for retaining RNs can be found in Within Our Grasp: Healthy Workplace Action Strategy for Success and Sustainability in Canada s Healthcare System (2007). Tested Solutions for Eliminating Canada s Registered Nurse Shortage 31

44 Figure 22: Effect of Reduced RN Exit Rates on RN Gap Reducing RN exit rates appears to have substantial short- and long-term effects on the gap. At 15 years, this strategy reduces the RN shortage by about half, or 30,000 FTEs. Scenario 4: Reduced RN Absenteeism A recent national survey of the health and work of nurses in Canada (CIHI, 2006) indicates that RNs miss an average of about 14 days of work per year. This is almost twice the average of seven days per year for all working people in Canada. Figure 23 shows the effect on the gap of reducing absenteeism among RNs over a three year period to that seven-days-per-year average. Reducing job strain and role overload, improving support from supervisors and co-workers, and reducing the physical demands of work can improve the health of nurses and reduce absenteeism (Shields and Wilkins, 2006). An example of an initiative to reduce on-the-job injury is the best practices musculoskeletal injury prevention program in nursing homes. Best practices like investment in mechanical lifts and reposition aids, a zero-lift policy, and training for employees on the use of equipment reduce injuries as well as violence by residents toward caregivers (Collins et al., 2004). This program recovered its equipment costs within three years through reduced workers compensation costs. 32 Tested Solutions for Eliminating Canada s Registered Nurse Shortage

45 Figure 23: Effect of Reducing RN Absenteeism on RN Gap RN Gap After Reducing RN Absenteeism vs. Baseline 0 10,000 20,000 30,000 40,000 50,000 60,000 70, Reduce RN Absenteeism Baseline The extra work time yielded by reducing RN absenteeism by half over three years is equivalent to an additional 7,000 RN FTEs entering the workforce over this period, substantially reducing the RN shortage in both the short and long term. Scenario 5: Increased RN Productivity RN productivity is the rate at which RNs perform nursing services; that is, the amount of services they perform per unit of time. Productivity is influenced by a variety of factors, including (but not limited to) the safety of the work environment, the availability of support staff and services, the efficiency with which RNs are deployed and managed, and, of course, the competencies (skills, knowledge and judgement) of the RN her/himself. Increasing the productivity of RNs is not about nurses working harder. Rather, it involves changing working practices and conditions to enable nurses to be more productive in other words, to work smarter as opposed to harder. This means providing resources, such as support staff and appropriate technology, and organizing services effectively so nurses can focus on providing quality patient care. Productivity among RNs has been improved, for example, by removing non-nursing tasks and adding support staff at Sisters of Charity of Ottawa Health Service (Mantha and Stewart, 2007). Another example is the Releasing Time to Care project being piloted across Saskatchewan. It is based on a U.K. initiative, highlighted by the NHS Institute for Improvement, that is also known as lean methodology. To assess the potential benefits of facilitating increased RN productivity, the project team considered a modest increase in productivity of 1% 5 per year. This would mean that, for example, an RN who treats 100 patients one year would be able to treat 101 patients the following year (other things being equal), through things such as removal of non-nursing tasks, working to full scope of practice (or optimizing competencies), better technology, appropriate staffing, and equipment such as patient lifts to help with burdensome care. 5 This increase is non-cumulative. Tested Solutions for Eliminating Canada s Registered Nurse Shortage 33

46 Figure 24: Effect of Improved RN Productivity on RN Gap As seen in Figure 24, even this modest, 1% increase in RN productivity per year has dramatic short- and long-term effects, cutting the shortage by about 47% by It appears that investments in improving RNs productivity is a very effective strategy for addressing RN shortages. Scenario 6: Reducing In-migration of Foreign-Trained RNs Recently there has been growing concern over the ethics of recruiting health-care professionals from disadvantaged countries. To assess the impact of reducing dependence on this practice, a scenario in which the rate of in-migration of RNs from countries outside Canada is reduced by 50% is simulated in Figure 25. Figure 25: Effect of Reduced In-migration on RN Gap 34 Tested Solutions for Eliminating Canada s Registered Nurse Shortage

47 While reducing in-migration by 50% would result in a larger shortage of RNs, the effect of this change is not all that substantial (less than 10%), even in the long term. This is because foreign-trained RNs still represent a relatively small fraction of the national RN supply. Figure 26 compares the effects of each of the six policy scenarios. Figure 26: Individual Effects of Various Policy Scenarios on RN Gap Individual Effects of Various Policy Scenarios on RN Gap 0 10, ,000 30,000 40,000 50,000 60,000 70,000 Baseline Reduced Program Attrition Increased RN Productivity Reduced RN In Migration Increased RN Enrolment Improved RN Retention Reduced RN Absenteeism Tested Solutions for Eliminating Canada s Registered Nurse Shortage 35

48 Combined Policy Scenarios The results above indicate that none of the policy scenarios simulated would be sufficient on its own to eliminate the shortage of RNs consistently over 15 years. However, since each of them showed substantial effects in reducing the gap (albeit only in the long-term for some), combining several policy scenarios could be used to further reduce the estimated gap. Figure 27: Combined Effects of Various Policy Scenarios on the RN Gap 60,000 Cumulative Effects of Various Policy Scenarios on the RN Gap 40,000 20, , ,000 60,000 80,000 AND Reduce In Migration AND Reduce Absenteeism AND Reduce Program Attrition Baseline AND Increase Productivity AND Reduce Exit Rates Increase Enrolment Figure 27 shows that if all the policy scenarios as previously discussed were implemented, their combined effects would be sufficient to eliminate the RN shortage by In particular, these results highlight the potential to improve the flow of services from Canada s RN workforce. Note that these results suggest that Canada s RN shortage could be eliminated without additional recruitment from other countries beyond recent levels of in-migration and recruitment, as demonstrated by the scenario in which the number of RNs educated in other countries beginning practice in Canada is reduced by 50%. It is important to note that the set of policy scenarios described is only one potential combination of policies that may eliminate the RN shortage during the next 15 years. Considering there is likely to be a practical limit to each of the interventions described, comprehensive, multi-faceted approaches will likely be required to eliminate the RN shortage. When comparing multiple-policy initiatives, it is important to bear in mind that an intervention aimed at one variable in the system modelled here may simultaneously affect other variables. For example, policies aimed at increasing RN productivity (such as increased job security or reduced safety concerns, extra opportunities for skill development or career advancement, and better technological resources) may also improve RN retention rates (O Brien-Pallas et al., 2008). 36 Tested Solutions for Eliminating Canada s Registered Nurse Shortage

49 CONCLUSIONS A needs-based simulation model that estimates both the supply of and requirements for direct/clinical care RNs (excluding NPs) in Canada has been developed. The model is based on a number of planning assumptions and is designed to accommodate new and more reliable data on an ongoing basis. Using assumptions about levels of unmet need for care and the extra RN activity required to meet those needs, a baseline scenario of a shortage of just under 11,000 RN FTEs in 2007 for Canada as a whole was adopted and used to estimate the changes in the shortage. This shortage represents the difference between an estimated 188,000 FTE RNs available and an estimated 198,000 FTEs required. If the health needs of Canadians continue to change according to recent trends and there are no policy interventions to alter the supply of or requirement for RNs, the shortage of RNs will increase to almost 60,000 FTEs by Although single policy changes, such as simply increasing enrolment in nursing ETP education programs, may reduce the gap in the long term, such initiatives alone (of the magnitude being contemplated) are unlikely to be enough to eliminate the gap. More comprehensive, multifaceted approaches aimed at, for example, improving the retention and productivity of existing RNs have the potential to eliminate the projected shortage, particularly where policies in one area may result in improvements in other areas. Tested Solutions for Eliminating Canada s Registered Nurse Shortage 37

50 RECOMMENDATIONS The simulation models provide a good base for projecting the depth of the current and future nursing shortage, and how it can be addressed. These findings provide evidence to inform HHR planning. With this information in hand, the Canadian Nurses Association and the authors recommend the following: 1. Governments, employers, unions, professional associations and/or colleges, RNs and other health providers should work together to consider how they can enhance the productivity of the RN workforce. For example, removing non-nursing tasks and providing support staff, appropriate technology and equipment, interprofessional practice and/or effective organization of services would allow RNs to remain as focused as possible on the provision of quality RN patient care. 2. Governments, employers, unions, professional associations and/or colleges, RNs and other health providers should collaborate to focus workplace improvement efforts on strategies to improve the health and well-being of RNs. For example, addressing high role overload, acquiring technologies and equipment that help reduce injuries, and addressing workplace morale would all contribute to reducing the injury and absenteeism of RNs. 3. Governments, employers, unions, professional associations and/or colleges, and RNs should collaborate to improve the retention of RNs in the workforce. Although retention issues may be generation-specific, they generally include having control over one s work (autonomy), reducing high role overload, feeling valued and respected by one s employer, being included in decision-making, and having opportunities for continuing education and professional development. 4. Educational organizations, professional associations and/or colleges, student associations and governments should partner to examine opportunities to improve the retention of nursing students. Factors to consider include pre-admission requirements, guidance and campus counselors, remediation, availability of faculty, student financial support and teaching methods. 5. Governments, educational organizations and professional associations and/or colleges should collaborate at a pan-canadian level to increase enrolment in RN education programs by considering a variety of delivery models, availability of faculty, location of programs and opportunities for interprofessional education. 6. HHR planning should employ a continuous, comprehensive, multifaceted approach considering a variety of policy options (such as those tested in this report) to achieve greater self-sufficiency. Investments in one policy area, such as improving work environments, may simultanelously affect many issues, such as retention, health of nurses and student attrition. 7. Governments, employers and professional associations and/or colleges should invest in data including coordinating and linking data currently collected with a particular focus on: the amount and type of services RNs provide according to the health needs of the patient/population; 38 Tested Solutions for Eliminating Canada s Registered Nurse Shortage

51 any aspect of the work done by RNs working outside acute care (e.g., long-term care, home care, in the community), the level of service they provide, activity rates, participation rates and productivity; level of retention of nurses, both practicing and newly educated; and rates of attrition among RN education programs. 8. Governments, employers, unions and professional associations and/or colleges should invest in a national health provider unique identifier to provide more accurate and reliable HHR data. Tested Solutions for Eliminating Canada s Registered Nurse Shortage 39

52 GLOSSARY Activity rate: The proportion of one full-time equivalent (FTE) that the average RN represents. Enrolment: The number of first time entrants in RN entry-to-practice education programs in Canada. Exit rate: The rate at which RNs do not renew their licences in a given year. Graduate out-migration: Proportion of new graduates who do not practise nursing in Canada after graduation. In-migration: Internationally educated RNs who obtain licences to practise nursing in Canada. Level of service: The amount and type of RN health-care services delivered according to an individual s level of need. Needs: The distribution of the population by level of health status (i.e., need for RN health-care services). Participation rate: The proportion of the stock of RNs involved in the delivery of patient care. Population: The size and distribution of the Canadian population by age and sex. Productivity: Average number of services RNs provide per unit of time. Program attrition: The proportion of students who leave RN entry-to-practice education programs without graduating. Program length: The average length of RN entry-to-practice education programs in Canada, weighted according to enrolment size. RN stock: Number of RNs in Canada who are licensed to provide nursing care. 40 Tested Solutions for Eliminating Canada s Registered Nurse Shortage

53 REFERENCES Advisory Committee on Health Delivery and Human Resources (ACHDHR). (2005). A Framework for Collaborative Pan-Canadian Health Human Resources Planning. Available at: sc.gc.ca/ahc-asc/alt_formats/ccs-scm/pdf/public-consult/col/hhr-rhs/pancanhhr_framework_sept- 05_e.pdf. Ottawa: Health Canada. Bernard, A., Li, C. (2006). Death of a Spouse: The Impact on Income for Senior Men and Women. Ottawa: Statistics Canada. Birch, S. (2002). Health human resource planning for the new millennium: Inputs in the production of health, illness and recovery in populations. Canadian Journal of Nursing Research, 33, Birch, S., Eyles, J., Newbold, B. (1996). Proxies for healthcare need among populations: Validation of alternatives a study in Quebec. Journal of Epidemiology and Community Health, 50, Birch, S., Kephart, G., O Brien-Pallas, L., Tomblin Murphy, G., Alder, R. et al. (2005). Atlantic Health Human Resources Planning Study. Mississauga: Med-Emerg. Birch, S., Kephart, G., Tomblin Murphy, G., O Brien-Pallas, L., Alder, R., MacKenzie, A. (January 2007). Human resources planning and the production of health: A needs-based analytical framework. Canadian Public Policy, XXXIII, supplement. Birch, S., Lavis, J., Markham, B., Woodward, C., O Brien-Pallas, L. (1994). Nursing requirements for Ontario over the next twenty years: Development and application of estimation methods. Report commissioned by the Ontario Nursing Human Resources Data Centre. Hamilton: McMaster University. Birch, S., O Brien-Pallas, L., Alksnis, C., Tomblin Murphy, G., Thomson, D. (2003). Beyond demographic change in human resources planning: an extended framework and application to nursing. Journal of Health Services Research and Policy, 8, Bookey-Bassett, S., Laporte, D., Hayes, L., Etowa, J., Hiroz, J., Raqueno, M., O Brien-Pallas, L. (July 2008). Sector Specific Components that Contribute to Positive Work Environments and Job Satisfaction for Nurses (SSC) Interim Report. Ontario: Ministry of Health and Long-term Care. Bournes, D., Ferguson-Paré, M. (2007). Human Becoming and 80/20: An Innovative Professional Development Model for Nurses. Nursing Science Quarterly, 20(3), Brown, R. and Prus, S. (2003). Social transfers and Income Inequality in Old Age: A Multi-National Perspective. Luxembourg Income Study Working Paper 355. Canadian Institute for Health Information (2006). National Survey on the Health and Work of Nurses. Ottawa: CIHI. Canadian Nurses Association (2008). HHR Policy Brief #4: January Ottawa: CNA. Tested Solutions for Eliminating Canada s Registered Nurse Shortage 41

54 Canadian Nurses Association and Canadian Association of Schools of Nursing. (2008). Nursing Education in Canada Statistics, Ottawa: CNA and CASN. College of Registered Nurses of British Columbia (2005). New Graduate Registered Nurse Study Vancouver: CRNBC. Evans, R. (1994). Strained Mercy: The economics of Canadian health care. Toronto: Butterworths. Eyles, J., Birch, S., Chambers, S., Hurley, J., Hutchinson, B. (1991). A needs-based methodology for allocating health-care resources in Ontario: Development and application. Social Science and Medicine, 33, Fisher, A. (2006). Understanding Nursing Vacancies: A Key to Improved Nurse Utilization, Cost Effectiveness and Patient Outcomes. Nursing Health Services Research Unit Fact Sheet, January Hamilton: Nursing Health Services Research Unit, McMaster University. Forrester, J. W. (1968). Market Growth as Influenced by Capital Investment. Industrial Management Review, 9(2). Fries, J. (1980). Aging, natural death and the compression of morbidity. New England Journal of Medicine, 303, Gray, A. (1982). The production of dental care in the British National Health Service. Scottish Journal of Political Economy, 29, Insightrix Research Services (2007). Nursing Education Program of Saskatchewan (NEPS) 2-Year Follow-Up Survey: 2005 Graduates. Regina: College of Nursing of the University of Saskatchewan. Kephart, G., Birch, S., Tomblin Murphy, G., O Brien-Pallas, L., MacKenzie, A. (2005). Health Human Resources Planning Simulation Model. Mississauga: MedEmerg. Lavis, J., Birch, S. (1997). The answer is now what was the question? Applying alternative approaches to estimating nurse requirements. Canadian Journal of Nursing Administration, 10, Mantha, A., Stewart, N. (2007, March). Total work redesign: Leading nursing care towards the future. Presentation at the HHR Knowledge Series meeting of the Canadian Nurses Association, March Milligan, K. (2007). The Evolution of Elderly Poverty in Canada. SEDAP Research Paper No O Brien-Pallas, L., Tomblin Murphy, G., Birch, S., Kephart, G. et al. (2007). Health Human Resources Planning: Challenging the Past, Creating the Future. Ottawa: Canadian Health Services Research Foundation. O Brien-Pallas, L., Tomblin Murphy, G., Shamian, J., Li, X. et al. (2008). Understanding the Costs and Outcomes of Nurses Turnover in Canadian Hospitals. Ottawa: Canadian Institute for Health Research. Ontario Ministry of Health and Long-Term Care (2008, March 6). Nursing Graduate Guarantee 86 per cent of participants gain full time employment. Ministry of Health and Long-Term Care News. 42 Tested Solutions for Eliminating Canada s Registered Nurse Shortage

55 Osberg, L. (2000). Poverty in Canada and the United States: Measurement, Trends and Implications. Canadian Journal of Economics, 33(4), Quality Worklife Quality Healthcare Collaborative. Within Our Grasp: A Healthy Workplace Action Strategy for Success and Sustainability in Canada s Healthcare System. (2007). Ottawa: Canadian Council on Health Services Accreditation. Richardson, G. (1991). Feedback Thought in Social Science and Systems Theory. Philadelphia: University of Pennsylvania Press. Romanow, R. (2002). Building on Values: The Future of Health Care in Canada Final Report. Commission on the Future of Health Care in Canada. Roos, N., Shapiro, E., Bond, R. et al. (2001). Changes in health and health care use of Manitobans, Winnipeg: Manitoba Centre for Health Policy and Evaluation. Shields, M., Wilkins, K. (2006). Findings from the 2005 national survey of the work and health of nurses. Ottawa: Statistics Canada. Statistics Canada (1999). Information about the National Population Health Survey. Statistics Canada Catalogue 82F0068XIE. Statistics Canada (2005a). Canadian Community Health Survey Guide. Statistics Canada Catalogue 82M0013GPE. Statistics Canada (2005b). Population Projections for Canada, Provinces and Territories. Statistics Canada Catalogue XIE. Statistics Canada (2008a). Residential Care Facilities Survey. Statistics Canada Catalogue XWE. Statistics Canada (2008b). Survey of Labour and Income Dynamics: a Survey Overview. Statistics Canada Catalogue 75F0011XIE. Sterman, J. (2000). Business Dynamics, Systems Thinking and Modeling for a Complex World. Toronto: McGraw-Hill. Tomblin Murphy, G., Birch, S., MacKenzie A. (2007). Needs-Based Health Human Resources Planning: The Challenge of Linking Needs to Provider Requirements. Ottawa: Canadian Nurses Association/Canadian Medical Association. Tomblin Murphy, G., Birch, S., Wang, S., O Brien-Pallas, L., Alder, R., Betts, H. (2005). Development of Human Resource Projection Models for Primary Health Care Nurse Practitioners in Canada. Mississauga: MedEmerg. Tomblin Murphy, G. and O Brien-Pallas, L. (2006). Appendix: Example of a Conceptual Model for HHR Planning in: A Framework for Collaborative Pan-Canadian Health Human Resources Planning. (pp ). Ottawa: Federal/Provincial/Territorial Advisory Committee on Health Delivery and Human Resources (ACHDHR). Tomblin Murphy, G., O Brien-Pallas, L., Birch, S., Kephart, G., MacKenzie, A. (2004). Hospital Service Utilization: Implications for Nursing Resource Planning. Ottawa: Health Canada. Tested Solutions for Eliminating Canada s Registered Nurse Shortage 43

56 Tomblin Murphy, G., O Brien-Pallas, L., Birch, S., Wang, S., Li, X. (2008). Health Human Resources Planning: An Examination of Relationships among Nursing Service Utilization, Estimates of Population Health, and Overall Health Outcomes in the Chronic Care and Home Care Sectors of the Province of Ontario. Toronto: Ontario Ministry of Health and Long-Term Care. Veall, M. (2007). Which Canadian Seniors Are Below the Low-Income Measure? SEDAP Research Paper No Vensim. Copyright (2002). Ventana Systems, Inc. Harvard, Massachusetts. 44 Tested Solutions for Eliminating Canada s Registered Nurse Shortage

57 Appendix I: Developing an Aboriginal-Specific HHR Planning Model In any forecasting exercise, particularly one that uses data from a variety of sources, there will be assumptions and constraints in the model that should be clearly stated and rationalized in the final analysis. Some limitations can be dealt with in a reasonable and justifiable manner; others, however, are too restrictive and would compromise the integrity of the model. In the process of research, sometimes it becomes clear that certain aspects of the planned analysis cannot be implemented, at least not at that time. The Aboriginal population is an important element in any analysis of health needs and human resource planning in Canada. As Romanow (2002) notes, there are serious disparities in health outcomes and access to care for Aboriginal people in this country, and these must be addressed. While the urgency of tackling these problems is clear, data limitations meant that Aboriginal people have not been directly included in the planning model. Consultations with stakeholders in this project led to an agreement to try to build an Aboriginal component into the HHR planning model. The first step in including Aboriginals was to check for comparable data. With the Canadian Community Health Survey (CCHS) as the model s main data source for the population in general, the project team needed Aboriginal survey data with similar health indicators and health service information. An exploration of data sources turned up two Aboriginal health surveys that could potentially be used in the model. The Aboriginal People s Survey (APS), conducted in 2001, and the First Nations and Inuit Regional Health Survey (RHS), conducted in 2002/2003, represent Aboriginal peoples living off First Nation reserves and on them, respectively. A public-use version of the APS microdata allowed the team to calculate individual health indicators for use in the model. The RHS, however, involves a large sample size from a relatively small population, resulting in very strict confidentiality rules. As a result, the team did not have direct access to these data. However, tabulations calculated by an RHS analyst were requested and received. Health Needs Across the CCHS, APS and RHS, there is significant overlap in health information. Some gaps in the data were found, which were limitations, but they were not significant enough to prevent the integration of Aboriginals into the model. For example, while chronic conditions were stratified by those with and without an injury as a needs indicator from the CCHS, the injury question was not asked in the APS. Furthermore, in the RHS, confidentiality rules prevented the team from obtaining information at three levels of stratification. Data on chronic conditions were available by age and sex but not by age, sex and the presence of an injury. Even given these constraints, chronic-condition information for the Aboriginal population could still be included in the model in the community sector. In the home-care sector, data from the CCHS on met and unmet need for home-care nursing services was used for the general population. This information is not available in the APS or RHS surveys a limitation that, again, would not, on its own, prevent the inclusion of Aboriginals. The model can be structured to handle differences in needs indicators across populations while maintaining the integrity of the results. Tested Solutions for Eliminating Canada s Registered Nurse Shortage 45

58 Age Groups As noted earlier, needs indicators are stratified by sex and age whereby proportions were calculated across five age categories for both males and females. In examining the APS and RHS, it was noted that the age groupings did not align exactly with those in the CCHS. In particular, the oldest age category available in the Aboriginal data was 55 and older, whereas the oldest age category used for the CCHS was 75 and older. As with the differences in the needs indicators, this is a limitation but not a barrier. The model can be adjusted to handle variations in age groupings across populations. Service Levels The main challenge regarding the Aboriginal data is in linking the needs indicators to health service utilization levels. For the general population, various data sources, including the CCHS, can be used to get the proportion of those who have had any contact with the health-care system, as well as the number of contacts by various needs indicators. The proportion of the population with any contact together with the average number of visits can be used to calculate the level of service. In the APS and RHS, there is no information on the number of contacts with the health-care system. Furthermore, the lack of information from other sources makes it impossible to estimate the level of service for the Aboriginal population. Because Aboriginals use of health services could be very different from patterns in the general population, it did not seem valid to apply estimates from the non-aboriginal population. The National Aboriginal Health Organization was consulted, and there are no other reliable sources on Aboriginal health service utilization at this time. The APS has some information on health service utilization specific to Métis respondents; but as with the general population, there is no information to suggest that the Métis responses would align with those of the rest of the Aboriginal population. The link between health need and health service levels is a crucial component in a needs-based HHR model. At this stage, the data about Aboriginals that would allow this link to be made in the HHR model is not available. As well, it would be wrong to assume levels for Aboriginals are the same as those for the general population. These circumstances have resulted in the exclusion of Aboriginal specific health needs from the simulation model. 46 Tested Solutions for Eliminating Canada s Registered Nurse Shortage

59 Appendix II: List of Data Sources Source Canadian Community Health Survey (CCHS) Statistics Canada population projections Residential Care Facilities Survey (RCFS) Data Elements Provided Distribution of chronic conditions and prevalence of injury by age and sex Distribution self-assessed health status by income level, age and sex Prevalence of difficulty with activities of daily living among those living alone by age and sex Average number of hospital days per person by number of chronic conditions, injury status, age and sex Average number of community nurse consultations by self-assessed health status, income level, age and sex Proportion of the population using public or private nursing home-care services, or both Proportion of the population reporting unmet nursing home-care needs Proportion of the population reporting unmet acute-care needs Proportion of the population reporting unmet community care needs Projected distribution of Canadian population by age and sex to 2022 Total patient days of care in long-term care facilities Number of individuals in long-term care facilities by age and sex CIHI Discharge Abstract Database CIHI National Ambulatory Care Reporting System Database CIHI Management Information System Database Acuity-adjusted episodes of care for in-patient, day surgery and emergency (certain provinces) Acuity-adjusted episodes of care for day surgery and emergency (certain provinces) Total acute-care RN earned hours by subsector for New Brunswick, Nova Scotia and Ontario Worked, benefit and purchased hours for all acute-care nurses by subsector for New Brunswick, Nova Scotia and Ontario Tested Solutions for Eliminating Canada s Registered Nurse Shortage 47

60 CNA/CASN National Student and Faculty Survey of Schools of Nursing Number of first-year ETP admissions to RN education programs Duration of all ETP RN education programs Attrition rates for 17 ETP RN education programs CIHI Regulated Nursing Database Canadian Registered Nurse Examination (CRNE) Existing stock of RNs in Canada by single year of age Number of Canadian-trained first-time writers of national RN exam by single year of age Number of foreign-trained writers of the national RN exam by single year of age CIHI Regulated Nursing Database RN provincial/territorial regulatory bodies CIHI National Survey on the Health and Work of Nurses Survey of Labour and Income Dynamics Ontario Hospital Association Adjusted exit rates for Canadian RNs by age group Self-reported average weekly hours worked, overtime hours worked, and days missed in the past year among RNs Threshold for low-income measure Number of people occupying acute-care beds while waiting for long-term care beds 48 Tested Solutions for Eliminating Canada s Registered Nurse Shortage

61 Appendix III: Data Limitations and Associated Assumptions In conducting this study, it was found that comprehensive, national-level data on certain measures did not exist or were not available; these items are in the left-hand column. In the right-hand column are the assumptions that were made in the absence of these data. Module Data Element Data Limitation Assumption Training Program attrition Graduate outmigration Attrition/graduation rates of RN ETP education programs Rates of retention/outmigration of new RN grads Supply Exit rates Exit rates for RNs by single year of age Attrition estimates provided by 17 programs representing 22 schools/sites across seven provinces/territories are representative of all ETP programs Estimates of RN ETP graduate retention from three provincial reports are representative of Canada Exit rates provided by CIHI are constant within age groups Age distribution of RNs moving between provinces is the same as that of RNs leaving the country Work and productivity Activity rate Hours worked by RNs The ratio of worked to benefit hours is the same for acute-care RNs as it is for all nurses, and this ratio in New Brunswick, Nova Scotia and Ontario is representative of all of Canada Productivity Productivity rates or services performed by RNs The ratio of worked hours to RNs in New Brunswick, Nova Scotia and Ontario is representative of this ratio for all of Canada The data for acute-care RNs are representative of RNs working in direct/clinical care in all sectors Acute-care service rates per population by patient age and sex in the provinces where data are available (all but Quebec for in-patient care, none but Ontario for emergency care, all but Quebec and Alberta for day surgery care, no data at all for outpatient care) are representative of national rates Tested Solutions for Eliminating Canada s Registered Nurse Shortage 49

62 Needs Needs Health needs of and health service utilization by children under 12 Levels of need and levels of health-care services delivered by need are the same for children under 12 as they are for those age 12 to 19 Small number of CCHS respondents age with three or more chronic conditions and at least one night spent in hospital The proportion of year-olds with three or more chronic conditions is the same as the proportion with two Number of individuals occupying acute-care beds while waiting for placement in longterm facilities The prevalence of these individuals in Ontario is representative of the country as a whole Needs Level of service Data element by age or sex The distribution of these individuals by age or sex is the same as the distribution of those individuals who are already in long-term facilities Level of service for long-term care Level of service for Ontario is representative of the whole country Long-term care level of service by age or sex Level of service in long-term care is the same regardless of patient age or sex Level of service for home care Level of service for Ontario is representative of the whole country Home-care level of service by age or sex Level of service in home care is the same regardless of patient age or sex 50 Tested Solutions for Eliminating Canada s Registered Nurse Shortage

63 Appendix IV: Measurement of Low Income Although there are various measures of low income, the one used in the model is similar to the lowincome measure developed by Statistics Canada (Statistics Canada, 1999). Low-income households are those that fall below the threshold of half the median adjusted household income at the national level, where the adjustment factor is for household size. Note that this measure is a relative, rather than an absolute, low-income threshold. In adjusting for household size, total household income is divided by the square root of the number of people living in the household. This income adjustment, known as the Luxembourg Income Study scale, has been shown to be a superior method of adjusting income for household size (Brown and Prus, 2003). For example, it s clear that a four-person household requires more income to have the same standard of living as a one-person household. However, that household is not likely to need four times the level of income, since its residents share expenses, such as for appliances, heating, etc. Dividing income by the square root of the number in the house allows comparisons in the standard of living across different-sized households. To calculate the low-income threshold, the Survey of Labour and Income Dynamics (SLID) is used. SLID is a nationally representative annual survey with a focus on the income of Canadians: At the heart of the survey objectives is the economic well-being of Canadians (Statistics Canada, 2008b, p. 1). Once calculated, the low-income threshold is entered into the CCHS data set. The public-use CCHS does not contain household income at the individual observation level. For confidentiality reasons, individual household income is identified as being in one of five categories: less than $15,000; $15,000 to $29,999; $30,000 to $49,999; $50,000 to $79,999; or $80,000 and over. Since SLID contains continuous rather than grouped income information, these data were used to calculate the average income within each of these categories. Each household in the CCHS was then assigned one of these average income levels, depending on which income category it fell into. The adjusted income was then calculated and compared with the low-income threshold. If the adjusted income fell below the threshold, the household was flagged as low income. Since the main focus of SLID is income while the main focus of the CCHS is health, low-income rates from two data sources were compared. Using all ages, rates were quite comparable: 14% of males and 16% of females were low income using SLID, while the respective figures were 13% and 17% using the CCHS. However, further analysis indicated that low-income rates are overestimated in the CCHS among both males and females age 65 and older. Males 75 and older show a low-income rate of 32% using the CCHS data, but only 8% using SLID. Similarly, the rates for 65-to-74-year-olds are 28% using CCHS, versus 10% using SLID. For females, the gap is not quite as pronounced, with nearly 35% of the oldest age group in low income in the CCHS, and 20% using SLID. Numbers in the two surveys were similar for 65-to-74-year-olds. The rates using the SLID data are consistent with previous research on seniors income in Canada using various data sources (for example, see Veall, 2007; Bernard and Li, 2006; Milligan, 2007; Osberg, 2000). Tested Solutions for Eliminating Canada s Registered Nurse Shortage 51

64 There may be a variety of reasons for the apparent overestimation of low-income rates among seniors in the CCHS. There is some evidence that using grouped income in the CCHS may inflate low-income rates. Another possibility is that health measures are the main focus in the CCHS, not income as in the SLID. As noted by Osberg (2000), rates of low income among seniors are sensitive to even slight changes in income; this is because in this age group, there is a spike in the income distribution close to the lowincome threshold. Overestimating the prevalence of low income in the model could result in an underestimation of service utilization. Generally, people with lower incomes have been shown to have higher rates of health-care need. One component of the model links those below the low-income threshold to levels of health service through CCHS data. By overestimating those in low income, average service levels may drop. 52 Tested Solutions for Eliminating Canada s Registered Nurse Shortage

65 Appendix V: Forecasting Future Needs Based on Past Trends When planning to allocate resources based on health needs, it is necessary to estimate what those future needs will be. In this study, demographic population and health indicator proportions were projected over a 15-year time horizon and fed into the planning model. In any forecasting, there is a degree of uncertainty; however, reasonable estimates are possible based on available information. Population The needs-based health human resources planning model tries to predict service needs within population groups based on changes in health needs over time (Birch et al., 2007). This means that future population estimates by sex and age group are required. Statistics Canada provides various projection numbers of the Canadian population as well as annual updates. For this analysis, projections that assume a medium population growth with migration based on recent trends (Statistics Canada, 2008) are used. Health Needs The need indicators used in the model are outlined elsewhere in this report. As noted by Birch et al. (2007), needs are incorporated into the model independent of demography. As a result, the model calls for needs projections that are separate from population. A straightforward starting place is to assume health needs will remain at current baseline rates. As noted in the text of this report, the base year in the model is The first projection scenario, therefore, assumes rates for all needs indicators will remain unchanged at 2007 levels for the full 15 years of the model. Under this steady-state needs assumption, surplus or deficit service requirements are estimated using various policy scenarios but keeping needs constant. Indeed, comparisons can be made across policy options without the influence of changing health needs. Using the needs-based model, it is also possible to analyze HHR requirements assuming there are changes in health needs over the next 15 years. Rather than arbitrarily varying projected needs indicators, recent historical data can be used to forecast future trends. For many of the need indicators used in this study, baseline estimates are calculated using the 2005 CCHS, which was the third cycle of this survey data. The first cycle was conducted in 2001 and the second in The CCHS replaced the National Population Health Survey (NPHS) as the nationally representative health survey in Canada. The NPHS was administered in 1994, 1996 and Since the CCHS was a replacement data set, there is a great deal of overlap in the two surveys questionnaires meaning there are many indicators that are comparable over the full span. However, gaps do remain, and not all information is available in both surveys. In fact, changes in questionnaires have been made within surveys, meaning some information is not available in all years even within cycles of the CCHS. As a result, some indicators have fewer data points with which to predict need than others. The following table summarizes the data years available for trending purposes for each of the needs indicators. Tested Solutions for Eliminating Canada s Registered Nurse Shortage 53

66 Sector Needs Indicator Number of data points available for trending Survey and Year Community Long-term care Acute care Home care Health status by income 4 NPHS 1994, 1996 CCHS 2003, 2005 Self-assessed unmet needs 1 CCHS 2005 Total resident days 3 RCFS , , Alternate level of care (bed blockers) Requires Assistance with ADL and lives alone 1 Trended with total resident days 5 NPHS 1994, 1996 CCHS 2001, 2003, 2005 Chronic conditions and injuries 3 CCHS 2001, 2003, 2005 Self-assessed unmet needs 1 CCHS 2005 Nursing services any plus unmet need 2 CCHS 2003, 2005 Health need in the community sector is represented by self-assessed health status, stratified by lowincome status. The self-assessed health question was asked of participants in all cycles of both surveys. However, data necessary to calculate low income is only available for two cycles of the NPHS and two cycles of the CCHS. As described in Appendix IV of this report, the low-income measure used in the model requires data on income as well as the number of individuals in the household. The 1998 NPHS and 2001 CCHS public-use data files do not provide this information on household size, meaning these two years cannot be used in the trending analysis for this indicator. A trend was determined for the proportion of the Canadian population in long-term care facilities at a given time using three iterations of the Residential Care Facilities Survey (RCFS); , and Unfortunately, the only available estimate of numbers occupying acute-care beds while waiting for longterm care beds was an overall measure for Ontario; thus establishing a trend for this value over time was not possible. To further represent need in the long-term care sector, total patient days in long-term care is included in the model. To minimize the influence of actual use of services in forecasting health need, survey data on assistance required with activities of daily living (ADL) are included to forecast this care needs indicator. How an ADL person is defined depends on what is available in the data. In both the CCHS and the NPHS, respondents are asked if they require help with various aspects of daily living. To reflect needs that likely require long-term nursing care, the proportion of individuals who report needing help with personal care and moving about in the house as well as living alone was measured. Data are available for all three of these factors in 1994, 1996, 2001, 2003 and In 1998, it cannot be determined from the data whether the individual lives alone. 54 Tested Solutions for Eliminating Canada s Registered Nurse Shortage

67 The number of chronic conditions stratified by the presence of an injury represents need in the acute-care sector of the model. Sample sizes are too small in the NPHS to calculate reliable estimates for this indicator, leaving 2001, 2003 and 2005 available for projection purposes. For the home-care sector, the proportion of those who have had any home-care nursing services or report having unmet need for home-care nursing services is measured. While data on any home-care nursing services is available in all cycles of each survey, the unmet-needs question was only asked in the 2003 and 2005 CCHS. As a result, there are two data points available to calculate a future trend. Needs indicators are stratified by sex as well as five age groups: 0-19, 20-44, 45-64, and 75 and older. In the home-care sector, there is no information available for the youngest age group before 2005; since determining a trend requires more than one data point, the trend for 20-to-44-year-olds is used instead. Also, the proportion of individuals in this age group deemed ADL is zero in all survey years. The trend for the group is used for the youngest group in this sector as well. Trending Methodology There are various ways to use existing data to predict future values. Generally speaking, one can assume the trend will follow a straight line, a curve or perhaps a jagged, up-and-down pattern. A linear trend is the most straightforward assumption. Such an approach will yield either an upward or downward trend over time, with the average rate of change each year reflected in the steepness of the line. While appealing in its simplicity, a linear trend forecasted over a number of years can result in extreme or even impossible predictions. An upward-sloping linear trend can predict implausibly large values, while a downward sloping trend can result in negative values. In forecasting populations and proportions, negative values do not make sense, and proportions must remain below 100%. Applying a linear trend may not meet these criteria, particularly over a 15-year projection period. The authors assumed that a practical health needs forecast will incorporate the upward or downward trend as suggested by past data, but will also level off so as not to produce untenable results. A power trend projection takes the following form: Y=β 1 X i β 2 where Y=predicted outcome β 1 =baseline value X i =time (year) β 2 =rate of change To forecast Y, one must plug in values for β 1, X i and β 2. The variable X i will range from 1 to 15 representing the 15-year time horizon, while β 1 will assume the 2005 baseline value. The final element, β 2, can be estimated using recent data, as previously summarized. Tested Solutions for Eliminating Canada s Registered Nurse Shortage 55

68 In estimating β 2, the properties of logarithms allow for the transformation of the previous equation to: ln Y= ln β 1 + β 2 ln X i where ln=natural logarithm This transformation means it is possible to calculate a value for β 2 using standard regression techniques and use this estimate to predict future values for Y. Through this methodology, estimates of need forecasted over 15 years will be based on trends in past data. The power function is appealing as it will not predict negative values. If the past data indicate a negative trend (i.e., the β 2 element is negative), the predicted future values will also show a negative trend but will level off before reaching zero. In the case of a positive trend, in most instances, the future values will also level off so projections do not stray unrealistically from the baseline values. This power trend function, therefore, was the forecasting technique chosen to predict future values for the needs indicators in the model. There are instances when the power function predicts values that are unrealistic and not usable in a HHR projection model. Past data with a very steep increasing trend can result in the estimate for β 2 as a value that is very near, or even greater than, one. This can yield exponential increases in the projection period. In such cases, an alternative technique should be employed. A logarithmic trend line will plateau in situations where the power trend line does not. The transformation equation for a logarithmic function is as follows: Y= β 1 + β 2 ln X i When compared with the power function, less weight will be given to the extreme values further into the future, producing more realistic predictions. The logarithmic trend is not generally preferred to the power trend as the initial methodology in the model, since a negative slope can result in projected negative values. In the modelling, there were two instances in which the logarithmic function was preferred to the power function. The trends in ADL among males and lead to outlier projections using the power function. The alternative logarithmic methodology was used in both these cases. Health Status and Income In the community sector, the proportion of individuals reporting fair or poor health versus excellent, very good or good health is calculated for those in the low-income threshold and for those not in the low income threshold in the baseline year. As noted previously, those with lower income tend to be less healthy than those with higher income. When trending, however, health status is forecasted within income categories only in other words, changes in population income itself are not forecasted. The distribution of income, particularly at the lower end, is strongly influenced by government policy, making it difficult to project reliable future estimates. Certainly the trend in health status can be very different for those in low-income threshold compared with the rest of the population; therefore, trends in self-assessed health status are calculated separately for low- and high-income populations but only among those within income categories each year. 56 Tested Solutions for Eliminating Canada s Registered Nurse Shortage

69

70