Report 3 - Exploring the Link Between Work-Life Conflict and Demands on Canada's Health Care System

Appendix E - Calculating the Costs to the Health Care System of Work-Life Conflict

The first step in obtaining data for an economic estimate of work-life conflict is to select the stressor and the specific consequences of interest from the wide range of potential consequences that exist. In the case of work-life conflict, we defined the stressor as the perception of role overload, work to family interference, family to work interference and caregiver strain.

This leaves the question of which consequences to explore, given the wide range of health care outcomes available from this research. This task is made somewhat easier by the requirement that the economic and health cost data be available on a national level from secondary sources. As expressed by Cooper et al. (1996), "The problem is...not the lack of economic calculation methods, but the lack of factual material on which to base calculations" (p. 78). After much searching,³⁵ we could find reliable national data only on the following three health care outcomes: physician visits, inpatient hospital stays and emergency room visits.

The following are the basic components of the socioeconomic assessment model (Levi & Lunde-Jensen, 1996), and the Canadian data sources used to meet these needs:

Health care system data and basic economic indicators are needed from which one can calculate the total use of this component of the health care system and to assign a value to the average cost per use (in our case, physician visits, inpatient hospital stays and emergency room visits). Information on the health care system data used in this report is included in Boxes E1 (Physician visits), E2 (Inpatient hospital stays) and E3 (Emergency visits).

An estimate of the proportion of the use of this component of the health care system related to work-life conflict. The data set from the 2001 "National Study on Balancing Work, Family and Lifestyle" was used to estimate prevalence, relative risk and the etiologic fraction. The questions to be answered from this dataset are:

1. What proportion of workers are exposed to the risk factor (in our case, work-life conflict)? This is the prevalence question.
2. What proportion of visits to the physician, inpatient hospital stays and emergency room visits can be associated with the risk factor? This answers the question of relative risk.
3. With data on both the exposed population and their excess risk, how much of the total health care system use would not have occurred had the risk factor been absent? This expression is referred to as the etiologic fraction.

Tables E1 through E3 provide the data used to answer these questions with respect to physician visits, inpatient hospital stays and emergency room visits.

Table E1
Costs of Work-Life Conflict: Physician Visits

Table E2
Costs of Work-Life Conflict: Inpatient Hospital Stays

Table E3
Costs of Work-Life Conflict: Emergency Room Visits

Question 1 (prevalence) was answered by calculating the proportion of the sample who reported high work-life conflict (i.e. operationalized as high role overload, high work to family interference, high family to work interference, high caregiver strain).

Question 2 (relative risk - RR) was obtained as follows:

• RR: Visits to physician was obtained by calculating the mean number of visits for those with high levels of work-life conflict and dividing this number by the number of visits for a control group of individuals with low levels of work-life conflict. Relative risk was then obtained to determine the excess number of visits associated with the risk factor (in this case, role overload, work to family interference, family to work interference and caregiver strain).
• RR: Inpatient hospital stays was obtained by calculating the mean number of inpatient hospital stays for those with high levels of work-life conflict and dividing this number by the number of stays for a control group of individuals with low levels of work-life conflict. Relative risk was then obtained to determine the excess number of stays associated with the risk factor.
• RR: Visits to emergency room was obtained by calculating the mean number of visits for those with high levels of work-life conflict and dividing
  this number of visits by the number of visits for a control group of individuals with low levels of work-life conflict. Relative risk was then obtained to determine the excess number of visits associated with the risk factor.

Once the population at risk is quantified, and their excess risk identified, we can calculate how much of the total use of the health care system would not have occurred had the risk factor not been present in the population (Question 3: the etiologic fraction). This is calculated from prevalence (P) and relative risk (RR) with the formula:

Etiologic fraction = (RR - 1) * P / ((RR - 1) * P + 1)

So, for example, 58.0% of the sample report high role overload and 12.3% report low role overload. Employees with high role overload made an average of 1.74 visits to a physician in a six-month period compared to only 1.11 visits for those with low role overload. This yields a RR of 1.58 (i.e. visits to physician among employees with high role overload was 1.58 times the number of visits for workers with low role overload).

Continuing the calculations, if the risk of the exposed group is 1.58 times the risk of the control group, and the prevalence is 58%, then the formula for the etiologic fraction gives (1.58 - 1) * 0.58 / [(1.58 - 1) * 0.58 + 1] = 0.25. Excess visits to physicians among employees who worked under conditions of high work-life conflict, therefore, can be quantified as 25%. This works out to an excess of 48 million physician visits per year that can be attributed to high levels of role overload (i.e. 25% of 192 million). To estimate the costs associated with this excess number of visits, we use the national average cost of a visit to a physician. For 1999 (the most recent year available), this worked out to a cost of $38.31 per visit. We then multiplied the excess number of visits by the cost per visit to arrive at an estimated cost per year of excess visits to a physician that can be attributed to high levels of role overload: $1.84 billion per year. Similar calculations were done to calculate increased costs due to work to family interference, family to work interference and caregiver strain (see Table E1) and increased costs due to each form of work-life conflict for inpatient hospital stays (Table E2) and emergency room visits (Table E3).

Limitations of the Estimates

To our knowledge, these estimates represent the first attempt to assign a dollar value to the health care costs of work-life conflict at the national level and, as such, are not without their limitations. First, we must again stress that these estimates are conservative and likely represent only a fraction of the costs that could be attributable to work-life conflict. For the health care sector, we could get reliable data only on three aspects of use of the health care system: physician visits, inpatient hospital stays and emergency room visits. Our estimates, therefore, cannot gauge the contribution of work-life conflict to public expenditures for services such as diagnostic procedures, visits to mental health professionals, and the governments' share of the costs of drugs used to treat stress-related illnesses. Nor can we assign costs to private companies and individuals incurred from employees' use of other health care professionals and prescription drug use (e.g. employee assistance program costs, the cost of chiropractors, physiotherapists, prescription drugs which are paid for by the employer as part of the employees' benefit plan or paid for by individual employees without such benefits).

Our estimates are also limited by the sample we used as a basis for calculating our multipliers. The employees in our sample may not be representative of employees across Canada as they work for companies employing more than 500 people. Levi and Lunde-Jensen (1996), however, argue that generalizability can be enhanced by comparing estimates to those obtained in other samples. In this regard, we note that our etiologic fractions (i.e. the proportion of excess physician visits for example associated with work-life conflict) are in line with numbers obtained in the European Community studies (Levi & Lunde-Jensen, 1996).

Finally, our estimates are also limited by our national-level data sources.³⁹ Suffice it to say that finding appropriate health care information was a considerable challenge. Although both Statistics Canada and CIHI provide excellent health data, it was nearly impossible to compare "oranges and oranges." Large interprovincial differences in payment schedules and classification categories made the costing of physician services and hospital visits extremely difficult. Many data are still not collected at the national level and were available for specific provinces only. The high degree of variability of these data between provinces (e.g. the outpatient diagnostic data) meant that we were not comfortable extrapolating these data to the national level. Another difficulty can be attributed to the fact that how the data are collected, grouped and captured has changed dramatically over the past several years as Statistics Canada and CIHI have tried to improve the data collection process. This means that it is difficult to do comparisons over time. This caution is particularly important with respect to comparing the costs to the health care system of increased physician visits reported in our 1999 study (Duxbury et al., 1999) to the calculations contained in this report.⁴⁰

CIHI (2002) also discusses the difficulties of obtaining national health care data. It attributes some of these difficulties to the fact that setting up electronic health records is complex, and appropriate privacy safeguards have yet to be established. Many jurisdictions, in fact, have passed legislation protecting the privacy of health information. Hopefully, this situation will be rectified in the near future as the Government of Canada has committed $500 million to the Canada Health Infoway Inc (CHII) to establish and accelerate the development of modern health information systems (CIHI, 2002).

Overall, we believe our estimates to be fair, given the data at hand. We are further reassured by the fact that these figures represent a lower end estimate in that we examined only a few of the possible costs for individuals who reported high levels of conflict (individuals with moderate levels of conflict also likely contribute to increased costs). Should better data become available, the multipliers obtained in this study should allow recalculation with relative ease.

[Previous] [Table of Contents]