Neftzger, Amy L. MA; Walker, Shannon MA
* Become familiar with the available tools for measuring health-related productivity loss, including the reasons why current approaches may not be appropriate for all jobs.
* Outline the authors' new Minimum Acceptance Performance methodology and its appropriate use, as part of a Multi-Method Approach, in measuring health-related productivity loss.
* Review the implications of the new methodology for measuring health-related productivity loss, including the balance between increased accuracy and increased costs.
Measuring health-related productivity loss (HRPL) has been the focus of employers, wellness, or health service providers, as well as researchers who strive to fully understand the return on investment of employee health programs. Recognizing the need to aggregate HRPL across different jobs and industries, several tools have been created to measure this construct on a global level in the form of absenteeism and presenteeism. However, while attempting to quantify HRPL, these measures may be producing biased estimates of productivity loss for certain types of jobs. It has long been recognized that employees do not equally contribute to organizational success, nor does each task an individual employee performs contribute equally to organizational success.1 In addition, there are some key sources of error that may be encountered during the process of measuring human performance2 that can be mitigated through the use of different measurement approaches. Therefore, in certain cases, it may be beneficial to reduce the potential for these types of measurement errors by accounting for the type of job, the type of tasks, and the repercussive effects of diminished performance.
In this article, we are introducing a modification to the current methodology that involves using a decision-based approach that takes into account the type of job and type of task to identify jobs for which a minimum acceptable performance strategy should be used to improve productivity loss estimates. The result of this adjustment to measurement is a more precise and valid quantification of HRPL for certain jobs. As part of our recommendations, we are proposing that the current tools continue to be used to measure productivity loss for situations in which they are appropriate, while also incorporating what we refer to as the minimum acceptable performance methodology for other types of jobs. In addition, we will explain why different methods are necessary and when each is appropriate. These two approaches to measurement (current tools and proposed methodology), when used appropriately in combination for different jobs as determined through a decision framework (also discussed later in this article), are what we refer to as the multi-method approach. We will also demonstrate through examples how the validity of HRPL measurement may be improved through utilization of this modified approach. However, before outlining the proposed methodology and decision framework, we will begin with a brief review of the current tools and why these may not be appropriate for all jobs.
Current Measurement Instruments
There are a number of tools designed to assess an individual worker's level of absenteeism (days missed) or presenteeism (being at work, but performing at reduced capacity) due to health. For the most part, the current tools have followed the general health research model of measuring productivity loss through self-reported survey items concerning the number of days absent and using some form of a Likert rating scale to assess the impact of presenteeism. Although some of the measures ask about the degree, frequency, or percentage of work time during which health has adversely impacted overall job performance, other instruments ask the individual to focus specifically on the impact of health on the quality, quantity, or timeliness of work. A comparison of several commonly used tools is listed in Table 1.
Regardless of which aspect of performance is being measured, the general approaches are fairly consistent in using a Likert scale by asking an individual to estimate the impact of health on work performance. Without delving too deeply in the argument as to whether or not Likert scales, themselves, are interval or ordinal in nature,10 we do know that the scales are designed to measure the magnitude of a construct. Therefore, the scales are constructed so that a rating of 9 indicates more productivity loss than a rating of 8, a rating of 8 indicates more productivity loss than a rating of 7, and so forth. However, whether or not the distance between each of the scale points is equal (interval) or simply ordered in terms of directional magnitude (ranked or ordinal) is important to the validity of cost estimates of productivity loss. Treating all individuals' productivity as equal and measured in equal units during the calculation of productivity loss may not be consistent with the actual productivity units and performance criteria for each job.2 In some cases, the nature of the work, and thus the impact of health on productivity, may be ordinal rather than interval. Therefore, treating all units and aspects of work equally may produce biased estimates of productivity loss for some jobs. Some jobs may even reach a point of diminishing returns in terms of added productivity gains once the employee reaches a certain performance level. Therefore, it may be possible to improve the validity of cost estimates for these types of jobs by using a threshold approach to HRPL.
Validation Efforts of Current Tools
In keeping with psychometric methodologies, the work surrounding validation of the current instruments has included concurrent, discriminant, face, and criterion-related validation efforts. Most relevant to employers is how well the instruments measure differences in actual work performance, or criterion-related validity. Considerable variation exists in the different forms of validity across tools. Validity coefficients range from approximately 0.10 to above 0.60, depending on the tool and the type of validation effort.11 These validation efforts represent a large body of work designed to demonstrate the link between employee health and productivity and have been significant in advancing the field. Although this is a relatively young science, several studies have already been published establishing the importance of HRPL. A number of the early validation studies have demonstrated the discriminant validity of the tools.8,12 As the validation efforts have progressed, researchers have shown the association of higher levels of presenteeism and absenteeism with increased medical costs through the linking of the HRPL survey data and medical or disability claims data.13–17 Still other studies have shown the association between HRPL and physical functioning or other health status measures.18,19 Yet another study examined the relationship between health status, presenteeism, and absenteeism and found that presenteeism was a better predictor of future absenteeism than health status.20 Still other studies have demonstrated that HRPL is associated with specific medical conditions to varying degrees.21–24 As this brief overview shows, there have been significant contributions to the field in establishing the validity and justification for pursuing this type of measurement in health outcomes research. As part of this body of validation work, researchers have also outlined some of the issues and challenges associated with creating a valid measure of job performance.25
Costing Productivity Loss
Although validity expressed as a coefficient has great meaning for researchers, it is often viewed as irrelevant to business professionals. This may be due to the fact that the language of business is dollars and not correlation or validity coefficients.26 Knowing this, efforts have been made to monetize HRPL to aid in proving the value or return on investment of health improvement efforts. A number of studies have provided estimates of the costs associated with HRPL, with the majority of these studies following a human resource cost accounting methodology and a salary conversion approach. In these techniques, the presenteeism measure is used to estimate reduced performance whereas the absenteeism measure is used to estimate work time lost. In these instances, productivity loss is expressed as salary paid (or salary plus benefits load) per unit of unproductive time. Because most presenteeism and absenteeism measures are global and not job specific, this method of cost estimation allows researchers to easily calculate the cost of productivity loss across all jobs within an organization.
Advancement on this methodology is the acknowledgment that interdependency may exist between jobs, thereby complicating the HRPL measurement to include an impact on the productivity of other workers or team members within an organization. One particular study examined the impact of HRPL on teams27 and demonstrated that the cost impact of absenteeism or presenteeism can be much greater than the dollar amount of the impaired individual's salary. Measuring this type of repercussive effect is one way to more accurately evaluate the cost of poor health to an organization. Most current practitioners in the business of measuring HRPL acknowledge the benefits of assessing the impact of productivity loss on teams and accordingly address the limitations of not doing so in studies where this type of measurement has been difficult.
Regardless of which measurement or cost estimation approach is used, all of them may be useful at different times or for different purposes. The key to successful measurement is often in finding a technique that provides both the employer and researcher with the desired specificity of information at a reasonable cost. A measurement system that is perfectly accurate may bring joy to the heart of the researcher, but cost more than the employer's health improvement programs or health care savings from the program, thereby rendering the measurement system accurate but impractical. In other words, to be cost effective and useful, the outcomes measurement system and intervention should not exceed the savings of those programs.
Although the trade-off between accuracy and practicality may appear to be a small matter, this issue is important to the validation efforts of the current tools. This is because the field of performance measurement has evolved over the years to consistently demonstrate that the most accurate tools for measuring job performance are job specific.28–30 Despite the advancement of this knowledge, the field of HRPL has chosen a more general approach so that estimates of productivity loss may be easily aggregated across jobs and even organizations. Although these general measures are easier to aggregate, the validity of this type of approach becomes more important as many of these measures are the basis of outcomes for health and wellness programs. In their 1979 capstone article in the field of human performance, Schmidt et al28 demonstrated a need for utility analysis for the use of assessment tools, given that precision in measurement is often coupled with a higher cost of measuring performance. The authors contend that not all organizations will seek the highest level of precision due to the increased cost and that there is a point at which the additional precision ceases to add value. Applying this knowledge to the current tools designed to measure HRPL may help to strike a balance between providing more accuracy in measurement and the ability to easily aggregate productivity loss across jobs. Rather than replacing current HRPL instruments with job specific tools, in this article, we are proposing an enhancement to the methodology that improves the accuracy in estimating true cost. In addition, we propose an approach that incorporates a decision framework to more accurately measure the impact of health on knowledge-based work, and thus provide more valid measurement for health and wellness outcomes.
There have been a number of previous efforts to estimate the cost of productivity loss due to health. However, because we can not possibly know or measure all possible costs associated with productivity loss and measure it perfectly, any estimate provided will always contain some element of error. Stated this way, we can express productivity loss cost estimates in the following manner:
The equation above is a modified version of the assessment theory outlined by Lord and Novick.31,32 Depending on the type of error, the measured cost may be grossly overestimated or underestimated. One major source of error is due to the job specific nature of productivity loss. Because of this, the same methodology used within one organization may result in overestimation of costs for some jobs and underestimation of costs for others. Once these costs are aggregated to the organizational level, the errors may cancel one another out or skew the total estimated cost in one direction or another. Without an ability to determine the sources or types of errors, we would never be able to know the accuracy of estimates. However, there are several common sources of potential error that may be minimized to produce a more accurate cost estimate. Three of these potential sources of error include the type of job, the type of tasks, and the repercussive effects of diminished performance. In the next few sections of the article, we will describe these three potential sources of error and how these may contribute to reduced accuracy in the measurement of HRPL. Once the potential for error has been defined, we will then describe our proposed solution to these issues in the form of the minimum acceptable performance methodology, the multi-method approach, and a decision framework.
Equation (Uncited)Image Tools
Type of Job as a Potential Source of Measurement Error
Role of Absenteeism
In the contemporary workforce, being absent from work does not always equate to productivity loss, and the cost/benefit of absenteeism depends on the job specific requirements of the work. In some instances, attendance is mandatory to effectively perform the job. Examples of mandatory attendance are physicians who physically examine patients, janitorial work, assembly line workers in a manufacturing setting, and restaurant servers. In each of these examples, it would be difficult or impossible for individuals to do the job from a remote location, because the physician needs to be where the patients expect to be examined and janitorial workers need to go to the locations to be cleaned, and so forth. However, in some jobs, especially knowledge-based workers, it is not necessary to be physically present to perform the work, such as in telephone sales or computer programmers. Therefore, before estimating the cost of absenteeism it is important to define the job as being one in whih mandatory attendance is required for effective job performance. For those jobs falling outside of this category, it is reasonable to focus only on measuring productivity loss due to presenteeism, because absenteeism would not be relevant for those jobs. Including those jobs in a cost estimate would introduce error into the measurement by assuming productivity loss without sufficient evidence that such productivity loss exists. In other words, it is an invalid assumption based on the type of job.
Although it may seem counter intuitive not to ask all individuals if poor health has kept them from attending work, it is important to determine whether or not attendance is relevant to achieving performance goals. In many cases, individuals may be completely unable to physically attend work but still perform 100% of the essential job functions. In these types of jobs, the productivity loss will be manifested in the form of failure to meet performance thresholds (presenteeism) during the same time period. Therefore, it is the individual's job performance and not attendance that determines the amount of productivity loss in these situations. An exception to this would be any formal work leave such as short-term disability, long-term disability, or worker's compensation claims. In these cases, individuals should still be counted as absent for the length of the claim regardless of the type of job. This is because the formal leave serves as an indication that the individual has stopped working for a period of time and is therefore not productive.
Pivotal Versus Important Jobs
When examining the relationship between the type of job and presenteeism, it is important to distinguish how much a particular job impacts overall organizational performance. In the general field of performance measurement some jobs may be classified as pivotal whereas others are classified as important.33 Whether or not a job is pivotal is a reflection of how differences between high or low performance on the same job impact overall organizational performance. A pivotal job is one in which large improvements in performance translate into greater strategic success for the organization. In jobs that are considered important, but not pivotal, there is simply a minimum level of performance that must be maintained. Any increases in performance above this minimum threshold do not continue to add value to the strategic success of the organization and therefore these incremental improvements in performance yield diminishing returns. What this means in terms of HRPL is that for “important” jobs, a decrease in performance may not be creating HRPL if performance stays above the minimum threshold. Conversely, actual productivity loss may be much higher than the estimates provided by the current methods, because falling below the performance threshold means that the employee is essentially not contributing to strategic success at all. In other words, in cases where performance falls below the threshold the true estimate of productivity loss would be 100% of salary and benefits load for the time period, rather than a fraction.
Let's consider the example of a mail room clerk for a manufacturing plant. Suppose that the organization defines strategic success as the ability to supply more widgets to market than their competitors. Improving the quality of mail delivery does not add incrementally to the organization's strategic success. Meeting specific performance standards is necessary, such as delivering the mail to the correct person within a specified time frame (daily delivery schedule). However, increasing the accuracy or timeliness from 85% to 90% accuracy and 12 rather than 24-hour delivery cycles would not add substantially to achieving the organizational goals in the marketplace.
To determine if a job is pivotal or important, a series of questions may be used to classify the job. The questions are as follows:
1. Do incremental increases in quality of performance add incremental (additional intervals of) value to the organization's strategic success?
2. Is there a point at which increases in the quality of the individual's performance would cease to add incremental value to the organizations strategic success?
Continuing with the mail room clerk example, answering no to the first question and yes to the second would classify the job as important, but not pivotal. This is because the job is a support position with indirect impact on the organization's strategic goals. On the other hand, an example of a pivotal job would be a front line Sales Manager. If we look at the two questions above, we can see that the response to the first question would be yes and the response to the second item would be no. This is because an increase in the number of widgets sold by the manager's team would add to the organizations strategic success, and there is no point at which adding more sales stops adding value, because each widget sale adds another increment of revenue. Jobs that are neither important nor pivotal are likely to be eliminated as unnecessary positions.
Although it is unrealistic to answer these questions for all jobs, most mid to large organizations have grouped jobs into families where the performance criteria for the jobs are closely related. In this way, similar jobs within the organization may easily be identified and categorized rather than treated individually. Working with categories of jobs, rather than with each individual job is one way to make this approach scalable. However, assuming that all jobs add equal intervals of value when estimating costs of productivity loss may introduce a source of error. The best way to minimize the potential of this error is to identify the type of jobs as being pivotal or non-pivotal and apply the appropriate performance measurement methodology accordingly.
Type of Task as a Potential Source of Measurement Error
Just as different types of jobs add value to an organization's strategic success differently, the type of tasks performed will determine the best way to gauge effective performance, as well. Converting a Likert scale of global performance to equally measured units of productivity loss has the underlying assumption that all tasks for the job are incremental or cumulative in nature. Experts in workgroup performance refer to tasks that can be broken down into equal, independent units as “additive.” However, not all jobs are comprised of additive tasks and this assumption is another potential source of error in measuring productivity loss.
If we examine the history of performance measurement, the science began with quantifying goods produced through piece-rate work and evolved into time and motion studies.34 These types of measures were designed for simple, additive tasks, which were the core job functions of most workers at the beginning of the industrial age. Today, organizations rely more on knowledge-based workers and the value of intellectual capitol, thus making the measurement more difficult than counting the pieces of work completed per hour.35 In these more complex jobs, measuring effective performance then becomes an issue of assessing a latent trait, rather than a concrete, observable, and easily quantifiable behavior. Which latent traits are relevant, how these contribute to job performance, and how much the behavior or trait contributes to organizational effectiveness is determined through a process called job analysis, which we briefly discuss shortly. Similar to the manner in which jobs that are not pivotal do not always contribute additional organizational value, non-additive tasks simply need to meet a minimum performance threshold to be considered successful.
Non-additive tasks include those done by most managerial or knowledge workers in which decision-making or intellectual faculty require a specific level. Making more decisions does not necessarily make these workers more effective but making a poor quality decision may be detrimental to the organization.
When determining the nature of the task as additive or non-additive, the following question may be used:
1. Does an increase in quantity of the outcome of the task result in incremental (additive interval) value to the organization?
An example would be a physician who is paid per patient examined. The more patients examined the more revenue the practice receives. Another example would be a lawn mowing service that is paid per lawn mowed. Adding more mowers would increase profits because more lawns mowed is directly related to revenue. An example in which the nature of a task is not additive involves the decision-making task in managerial work. Making more decisions does not necessarily improve profitability for the organization and may even inhibit it if the decisions contradict one another. Similarly, hiring more managers to make more decisions will not necessarily increase revenue. Therefore, the nature of the decision-making task is non-additive.
As a general rule, piece-rate and assembly work tend to be pivotal jobs with additive tasks. This is why replacing these types of workers with robots is often attractive to organizations: because having a machine working 24/7 to produce more output adds value to the organization. On the other hand, most managerial or knowledge jobs tend to be either important (non-pivotal) jobs or pivotal jobs with non-additive tasks. Although these classifications are too general to be implemented in this simplistic manner, these examples may be helpful for those who are new to these concepts.
Before leaving this subject it is important to note that one job may involve performing multiple tasks. Determining the nature of all tasks would be a substantial body of work and may yield low return on value. Therefore, we recommend identifying the key performance tasks or those tasks which are essential to the job. Because these are the most critical to job performance, classifying these tasks will yield the largest decrease in error of cost estimation. Other tasks may also be classified, but once again, the utility of the precision should be considered to maintain a feasible approach.
The Role of Job Analysis in Defining Pivotal, Important, and Additive
The process of gathering the information to determine whether or not a job is pivotal versus important, the nature of the tasks performed (additive vs non additive), and which tasks are critical to successful performance is a field known as job analysis. Most mid sized and large organizations will already have this information as part of the formal human resource system. This is because job analysis is the foundation of job descriptions, selection systems, compensation, and performance appraisal systems. If an organization does not have this information available or up to date, the organization may either choose to have their own human resource professionals conduct the job analysis or hire consultants for this purpose. There are quite a number of fully developed systems already well established that can be applied to obtain this information. Examples of two of these systems are the Fleishman Job Analysis Survey and the Functional Job Analysis System Scales.36 However, because there are additional legal concerns that may be encountered during or resulting from job analysis, it is not recommended that health care researchers attempt to conduct one without proper training. It is preferred to reference any data already collected through a previous job analysis or partner with the employer's human resource professionals to obtain new information.
Repercussive Effects as a Potential Source of Measurement Error
The third source of potential measurement error we are going to discuss is the presence of repercussive effects. Repercussive effects include any other processes or individuals impacted by the worker's job performance. These effects may include impacts on team members, production lines or processes, sales volume, changes in customer satisfaction or loyalty, sales volume, safety violations or workplace injuries, regulatory violations and fines, or any other outcome.
In some cases, the repercussive effects may be easy to locate and measure, whereas in others, the task may be far more difficult. Furthermore, if processes and measurement systems are not already in place, repercussive effects may not be identified simply because there was no previously established metric designed to capture them. For example, a repercussive impact of presenteeism or absenteeism may primarily impact the immediate work team, but the decreased performance of the work team may impact other teams or processes within the organization. In such instances, we may fail to see evidence simply because we were not looking for it outside of the immediate proximity of the work situation and therefore did not measure the appropriate or complete outcomes.
Repercussive effects are much like a drop of rain falling into a pool. Although the effects may continue to reach farther and father outward, the intensity may diminish as the effect travels farther away from the source. In addition, the act of measuring all processes and all impacts may be cost prohibitive. Because of this and other difficulties related to job specific measurement, some researchers have opted to take a broader approach to quantifying the cost associated with productivity loss. These approaches use organization level metrics, similar to the methodology used in measuring the impact of employee engagement on overall organizational performance.
This is important because while these are difficult to measure, repercussive effects may account for the larger share of costs in productivity loss than wage and salary amounts. Keep in mind that the use of fully loaded wage and benefits is merely “an approximation and that the assumption that total pay is equal to the value of employee time is not generally valid.”37 The work that an employee accomplishes may be actually worth more or less than the amount of salary and benefits paid to the individual for the work.
It should also be noted that repercussive effects are often job and industry specific. Therefore, it is advisable to work closely with human resource professionals on any study to accurately identify potential repercussive effects. It may also be helpful to use an established human factors taxonomy in this process, because these tools have been designed to identify potential sources and outcomes of poor performance.38 Repercussive effects may include such measures as average cost of scrap per piece or average loss in sales revenue per poor decision. More detailed examples on quantifying repercussive effects can be found in the published literature.39
Decision Framework and Utility
It has often been noted that measuring presenteeism loss due to health is a difficult task. However, given the drive to demonstrate value for health improvement efforts this task is not likely to go away in the near future. The current methodologies have provided us with some rough cost estimates, but the general and less precise nature of these estimates has left some individuals skeptical of the value. As mentioned previously, the measurement of effective job performance can be extremely precise, but this precision comes at a higher price. Conducting job analysis for every position and then conducting health research using these very specific measures can also become expensive in terms of hours spent by human resource professionals and researchers to monitor, record, and then analyze the productivity loss. Although all this is possible, it is not always practical. Therefore, a decision should be made to determine what level of accuracy and precision is required and how much an organization is willing to invest financially in this measurement. At some point, the cost of the measurement may exceed the organization's investment in the health programs or any savings generated by the program. With this in mind, the following adaptation of the current methodologies is offered as a way to improve precision and minimize the amount of error in measurement.
In this technique, we recommend asking a short series of questions to determine whether to use the current methodology or a minimum acceptable performance (threshold based) approach. The list of questions is as follows:
1. What are the key job performance criteria?
2. How critical is each facet to overall total job performance?
3. Is the job pivotal or important?
a. How much incremental organizational value is created by differences in performance quality? Note: large differences indicate that the job is more pivotal. Conversely, small differences may indicate that a job is important, but not pivotal.
b. Is the performance yield curve (slope) steep? If you have a performance yield curve for the job, look at the slope. A steeper slope or curve indicates the job is more pivotal.12
4. Is the primary job function an additive task?
5. What are the repercussions of poor performance?
A summary of this process can be seen in Fig. 1, which provides an overview of the decision framework and application. However, once the appropriate process has been identified for determining productivity loss, repercussive effects need to be considered and added into the equation.
Applying a Multi-Method Approach
Based on the new decision framework, we can move forward with producing more accurate estimates of productivity loss due to health. The first step is to determine in which jobs attendance is mandatory for effective job performance. The cost of productivity loss due to absenteeism will only be included for those jobs. As previously stated, the exception to this exclusion is any formal disability or worker's compensation leave. The second step is to determine which jobs are pivotal and have additive tasks. For these jobs, the traditional measures of presenteeism are appropriate. For jobs that are non-pivotal and/or have non-additive tasks, the minimum performance methodology should be used. This process is outlined in Fig. 2. Once key performance criteria are defined and performance thresholds set, we can then begin to calculate the number of units, such as hours, an individual did not meet minimum performance thresholds due to health. The number of units of productivity loss due to health are then combined from both the traditional, minimum performance methodology, and absenteeism measures and then multiplied by the cost per unit (wage and benefits load). Finally, we add any repercussive costs.
Stated in the form of an equation, the calculation would appear as follows:
Equation (Uncited)Image Tools
CostPH = total cost of productivity loss due to health.
ΣUPH = sum of units of productivity loss caused by a decrease in health.
Absenteeism is included here only for jobs in which attendance is required to achieve effective job performance. For pivotal jobs or jobs with non-additive tasks, this is the total number of units below the minimum acceptable performance threshold. For non-pivotal jobs with additive tasks, this is the estimate of units of productivity loss obtained from traditional measures of presenteeism.
CostPU = cost of one productivity loss unit. This may be calculated though the cost of wage and benefits load per unit (such as workday or hour) or through other units of measure.
CostR = repercussive costs. These may include the cost of scrap, rework, customer loss, increases in turnover, lost time other workers expend to remedy errors, safety violation fines, decreases in sales, lost productivity of team members who are dependent on the work, stalls in an assembly line when product is not available to continue work etc. Repercussive costs will vary by job type and are most likely to be larger in situations where core job components are based on intellectual capitol.
U = Unit of measure, such as workday, hour, minute, or widget. The unit is required to be consistent throughout the equation.
UP = Units of Productivity loss may be expressed in time units, such as hours or minutes or through other quantitative productivity measures, such as number of widgets produced below acceptable standards.
It is important to note that the (ΣUPH × CostPU) may need to be calculated separately in any cases where the wage and benefits loads differ significantly between jobs. If the CostPU is the same for all jobs, then aggregating all ΣUPH before multiplying by the CostPU would be appropriate. However, this is rarely the case in practice. In addition, while the examples below demonstrate the calculation for an individual worker, it may also be acceptable to group jobs into categories or use an average CostPU for all jobs within the organization to save time and resources on the calculation of these estimates. Again, the utility of the precision should be weighed against the cost of obtaining the estimate.
Simple Managerial Example
For the sake of example, we will consider a case where 1 hour of productivity loss due to a migraine which caused the manager's performance on the “decision-making skills” or “judgment” performance dimension to fall below the minimum performance threshold. The result of this is that the manager agreed to a contract that would result in decreased sales of 5% over the next year for a small customer. The small customer brings in $45,000 in revenue annually. With a wage and benefits load of 125k per year, the manager is worth $60 per hour. Therefore, the total productivity loss due to health would be:
Equation (Uncited)Image Tools
Total cost = (1 hour × wage and benefits load per hour) + repercussive costs of losing 5% of revenue from a small client
Total cost = (1 hour × $60) + $2250
Total cost = $2310
Assembly Line Worker Example
In our second example, we will examine the case of a line worker who, because of a physical ailment, performs below the minimum performance threshold for 1 full workday (8 hours). The line worker paints the side of a TV cabinet and is paid $60,000 in wages and benefits. The cost of rework exceeds the cost of errors to cabinet walls that are not painted to meet minimum quality standards, thus resulting in scrap for each piece that is below the standard. We will assume for the sake of this example that other line segments have enough cabinet walls on hand that meet quality specifications so that the work of others is not inhibited by this particular employee. This particular employee has a rate of production of painting four cabinet walls per hour. So, with these figures, we can calculate the total cost of productivity loss as such:
Equation (Uncited)Image Tools
Total cost = (8 hours × wage and benefits load per hour) + repercussive costs of units of scrap
Total cost = (8 hours × $28.85) + ($3 × 32)
Total cost = ($230 + $96) = $326
Limitations of the New Approach
Although much of the information contained in this article is not new, it is new to the health care field. Part of our goal in writing this article was to help bridge the differences between how employers measure general productivity loss and how the health care industry is measuring HRPL. As such, we hope that this article will bring greater understanding and insight even if the methodology is not adopted by everyone. In fact, we acknowledge that there are instances when researchers should continue to use the current methodologies. One of the reasons we included a decision framework in this article is to aid researchers in determining how much precision is desirable and whether or not a minimum acceptable performance strategy would be more appropriate. We also acknowledge that the methodology outlined here in the multi-method approach is considerably more work than the current tools, which have the benefit of being less cumbersome to implement. This more complicated measurement design translates into higher costs for outcomes evaluation, whether it is part of the health care providers' outcome services or whether this evaluation is done by a third party (Table 2).
In addition, using the same assessment tool for all employees, regardless of job type (as in the current methodologies), has the advantage of being able to easily aggregate responses across all job types and industries. In some instances, this ease of aggregation may be preferred over the potentially higher validity coefficients and greater precision offered by the multi-method approach. Although some employers are interested in knowing accurate cost estimates of HRPL, others may simply be interested in knowing if any change has occurred in how much employees perceive their health to be impacting their work. For this second group, the current tools and methodologies would be preferable. As stated earlier in this article, the researcher, health care provider, and employer should work together to determine the best methodology for evaluating HRPL based on the employer's needs and goals for the assessment.
Given that there are some situations in which one method may be preferable over another and to help clearly demonstrate the differences between the current tools and the two approaches outlined in this article (the minimum acceptable performance methodology and the multi-method approach), we have constructed a table that provides a broad comparison of the three approaches. Note that the classification of measurement expertise needed to implement these is relative to the other methods proposed. We have included these ratings as a proxy for the comparative cost and difficulty required to implement each of the measurement techniques.
The issues surrounding the tradeoff between higher validity with a job specific assessment tool and the ability to aggregate results (in some cases resulting in reduced validity) are not new. In fact, the difficulty has been previously outlined in the health care literature when it was suggested that job specific measures for each individual job would be too cumbersome.25 Our efforts in creating the minimum acceptable performance methodology were focused on suggesting an additional measure that is more appropriate for a large class of jobs, namely intellectual capital and knowledge workers, thereby increasing the validity and cost estimates for the productivity loss measurement of these jobs as a partial solution to this measurement issue. We view this effort as building on the significant work that has already been done.
The current methodologies and tools for costing productivity loss due to health are most accurate when the jobs are pivotal and the critical performance tasks are additive. Knowing which jobs are pivotal, the pivot points, and the performance thresholds for non-pivotal jobs can help an organization to make a more informed decision about what types of health programs or wellness efforts are likely to yield the highest ROI for the organization. It will also aid in more accurate calculations of ROI.
In jobs that are not pivotal, the cost of productivity loss may be 0, as long as the individual meets minimum performance thresholds. Nevertheless, productivity loss may be closer to 100% of the worker's salary and benefits load, plus additional repercussive costs. This may be true even in cases where there is only a slight decrease in performance.
We acknowledge that the cost of performing a thorough job analysis to obtain some of this additional information may be substantial. However, organizations interested in optimizing human capitol will need this information to calculate ROI in other employee programs, such as training or selection tools. In these cases, knowing how much the job contributes to organizational success aids in making decisions on where to invest in employee performance to get the greatest benefit in overall organizational performance. Measuring the cost of productivity loss due to health is an incomplete estimate and fraught with error if we do not know how much the job contributes to overall strategic success. In the absence of the detailed information outlined, using a marginal approach provides a generic estimate of the impact of costs that may be useful until more specific sophisticated and precise methodology may be employed.
Finally, the measurement of the impact of health on employee productivity is a relatively young science and one that will continue to develop over time. Those of us who are researchers should persist in our investigations concerning the validity of the current tools and refine the instruments if new insights or evidence indicate a need to do so. In addition, we should also continue to develop new tools and methodologies to advance this science and improve the measurement in a manner acceptable to business leaders. If not, our efforts to quantify productivity loss due to health will have no applied value if we do not also find a way to express it in a way in which executives clearly understand.
The methodology presented in this article is intended as a first step toward finding an appropriate balance between the desire for more accurate cost estimates and the cumbersome precision required to obtain these metrics. Although many of the concepts discussed in this article have already been proven and established in the field of general productivity measurement,30,40 these concepts have not yet been integrated into the field of health-related productivity measurement. We do not expect this integration and evolution of the field to take place immediately.
The next step in this process would be to implement the methodology outlined in this article along with the current methodologies and begin examining the differences between the cost estimates obtained in an employer setting. If there is no significant difference between the estimates obtained, there is little justification to continue adding complexity to the HRPL measurement process. This initial comparison of estimates should then be followed by studies demonstrating the validity of the methodology using an organizational level criterion, such as revenue per employee or another global metric appropriate to the employer and industry.
Additionally, if the validation efforts follow the path that it has in general productivity measurement and the multi-method approach gains acceptance by employers and researchers, it would be prudent to build a user friendly system for more easily classifying jobs based on the job analysis data. Such a system may consist of general categories of jobs for quick reference and comparison. A large number of job analysis classification systems already exist and adding an extra component for measuring HRPL may save time or money in the long term.
Finally, locating organizations who strongly believe in improving employee health and partnering with them to implement these measurement techniques is important. The monetization of productivity loss estimates is an excellent way to communicate the value of health improvement programs. Most health care professionals intrinsically understand the value of good health but not every employer does unless we can demonstrate it in a way in which they understand.
1. Motowidlo SJ, Borman WC, Schmit MJ. A theory of individual differences in task and contextual performance. Hum Perf. 1997;10:71–83.
2. Spector P. Industrial and Organizational Psychology. Hoboken, NJ: John Wiley and Sons; 2008:81–87.
3. Reilly MC, Zbrozek AS, Dukes EM. The validity and reproducibility of a work productivity and activity impairment instrument. Pharmacoeconomics. 1993;4:353–356.
4. Lerner D, Amick BC, Rogers WH, Malspeis S, Bungay K, Cynn D. The Work Limitations Questionnaire. Med Care. 2001;39:72–85.
5. Koopman C, Pelletier KR, Murray JF, et al. Stanford presenteeism scale: health status and employee productivity. J Occup Environ Med. 2002;44:14–20.
6. Endicott J, Nee J. Assessment measures for clinical studies: Endicott Work Productivity Scale (EWPS): a new measure to assess treatment effects. Psychopharmacol Bull. 1997;33:13–16.
7. Stewart WF, Ricci JA, Chee E, Morganstein D. Lost productive work time costs from health conditions in the United States: results from the American Productivity Audit. J Occup Environ Med. 2003;45:1234–1246.
8. Kessler RC, Barber C, Beck A, et al. The World Health Organization Work Performance Questionnaire (HPQ). J Occup Environ Med. 2003;45:156–174.
9. Goetzel RZ, Ozminkowski RJ, Long SR. Development and reliability analysis of the Work Productivity Short Inventory (WPSI) instrument measuring employee health and productivity. J Occup Environ Med. 2003;45:743–762.
10. Jameison S. Likert scales: how to (ab)use them. Med Educ. 2004;38:1217–1218.
11. Lofland JH, Pizzi L, Frick KD. A review of health-related workplace productivity loss instruments. Pharmacoeconomics. 2004;22:165–184.
12. Turpin RS, Ozminkowski RJ, Sharda CE, et al. Reliability and validity of the Stanford Presenteeism Scale. J Occup Environ Med. 2004;46:1123–1133.
13. Goetzel RZ, Long SR, Ozminkowski RJ, Hawkins K, Wang S, Lynch W. Health, absence, disability, and presenteeism cost estimate of certain physical and mental health conditions affecting U. S. employers. J Occup Environ Med. 2004;46:398–412.
14. Keyes CLM, Grzywacz JG. Health as a complete state: the added value in work performance and healthcare costs. J Occup Environ Med. 2005;47:523–532.
15. Goetzel RZ, Carls GS, Wang S, et al. The relationship between modifiable health risk factors and medical expenditures, absenteeism, short term disability, and presenteeism, among employees at Novartis. J Occup Environ Med. 2009;51:487–499.
16. Loeppke R, Taitel M, Richling D, et al. Health and productivity as a business strategy. J Occup Environ Med. 2007;49:712–721.
17. Johnston K, Westerfield W, Momin S, Phillippi R, Naidoo A. The direct and indirect costs of employee depression, anxiety, and emotional disorders—an employer case study. J Occup Environ Med. 2009;51:564–577.
18. Tsutsumi A, Nagami M, Yoshikawa T, Kogi K, Kawakami N. Participatory intervention for workplace improvements on mental health and job performance among blue-collar workers: a cluster randomized control trial. J Occup Environ Med. 2009;51:554–563.
19. Wynne-Jones G, Buck R, Varnava A, Phillips C, Main CJ. Impacts on work absence and performance: what really matters? Occup Med. 2009;59:556–562.
20. Bergstrom G, Bodin L, Hagberg J, Aronsson G, Josephson M. Sickness presenteeism today, sickness absenteeism tomorrow? A prospective study on sickness presenteeism and future sickness absenteeism. J Occup Environ Med. 2009;51:629–638.
21. Burton WN, Pransky G, Conti DJ, Chen C, Edington DW. The association of medical conditions and presenteeism. J Occup Environ Med. 2004;46:S38–S45.
22. Schultz AB, Edington DW. Employee health and presenteeism: a systematic review. J Occup Rehabil. 2007;17:547–579.
23. Loeppke R. Are we underestimating the costs of poor employee health? J Compensat Benefit. 2007;23:15–19.
24. Aronsson G, Gustafsson K. Sickness presenteeism: prevalence, attendance-pressure factors, and a outline of a model for research. J Occup Environ Med. 2005;47:958–966.
25. Kessler RC, Ames M, Hymel PA, et al. Using the world health organization health and work performance questionnaire (HPQ) to evaluate the indirect workplace costs of illness. J Occup Environ Med. 2004;46(suppl):S23–S37.
26. Cascio W. Preface. In: Janke RA, Tibbets KM, eds. Costing Human Resources: The Financial Impact of Behavior in Organizations. Boston, MA: PWS-Kent Publishing Company; 1992:vii–ix.
27. Pauly MV, Nicholson S, Xu J, et al. A general model of the impact of absenteeism on employers and employees. Health Econ. 2002;11:221–231.
28. Schmidt FL, Hunter JE, McKenzie RC, Muldrow TW. Impact of valid selection procedures on work-force productivity. J Appl Psychol. 1979;64:609–626.
29. Catano VM, Darr W, Campbell CA. Performance appraisal of behavior-based competencies: a reliable and valid procedure. Pers Psychol. 2007;60:201–203.
30. Kline TJB, Sulsky LM. Measurement and assessment issues in performance appraisal. Can Psychol. 2009;50:161–171.
31. Lord FM, Novick MR. Statistical Theories of Mental Test Scores. Reading, MA: Addison-Wesley;1968:667–668.
32. Meyer GJ, Finn SE, Eyde D, et al. Psychological testing and psychological assessment: a review of evidence and issues. Am Psychol. 2001;56:128–165.
33. Boudreau JW, Ramstad M. Beyond HR: The New Science of Human Capitol. Boston, MA: Harvard Business School Publishing; 2007:55–59.
34. Brogdon HE, Taylor E. The dollar criterion—applying the cost accounting concept to criterion construction. Pers Psychol. 1950;3:133–154.
35. Davenport TH. Thinking for a Living. Boston, MA: Harvard Business School Press; 2005:5–10.
36. Wilson MA. A history of job analysis. In: Koppes LL, ed. Historical Perspectives in Industrial and Organizational Psychology. Mahwah, NJ: Lawrence Earlbaum Associates; 2007:219–242.
37. Cascio W, Boudreau J. Investing in People: Financial Impact of Human Resource Initiatives. Upper Saddle River, NJ: Ft Press; 2008:33–34.
38. Salvendy G. Handbook of Human Factors and Ergonomics. Hoboken, NJ: John Wiley and Sons; 2006:708–757.
39. Wiegmann DA, Shappell SA. A Human Error Approach to Aviation Accident Analysis: The Human Factors Analysis and Classification System. Hoboken, NJ: Ashgate Publishing:72–98.
40. Bernardin HJ, Beatty RW. Measuring the effectiveness of appraisal data. In: Beatty RW, ed. Performance Appraisal: Assessing Human Behavior at Work. Boston, MA: Kent Publishing Company; 1984:128–186.