Need to estimate costs? We frequently receive questions from researchers asking us how to estimate costs for their study. There are many methods by which one could use to estimate costs. Costs are most often thought of in accounting terms because billing data are more frequently available. However, the economic cost and the accounting cost do not always agree. Economic costs are based on the concept of opportunity costs (Garber 2000).
Learn more about opportunity costs »
There is a spectrum of cost determination methods (see Figure 1). Accounting and billing systems use direct measurement methods, whereby very detailed estimates of time and products (inputs) are combined with unit costs to estimate total costs. Sometimes these systems are called activity-based cost (ABC) systems. The highly precise methods are extremely challenging to implement because a single inpatient stay or outpatient procedure might have hundreds or thousands of inputs. Even when there is just a single input, such as a pill or medication, the cost can vary by location or day. Researchers can use less precise methods, such as an average cost per day. These methods are easier to implement than the more precise methods, but their ease of use comes at a cost of decreased precision. Researchers need to identify the level of precision necessary for their study.
Figure 1
Blending two or more methods in a study is frequently needed because a researcher might need to estimate the cost of an intervention and the cost of subsequent health care. Different methods may be ideal for each of these goals. After we describe the methods, we discuss which method is best.
On precision and accuracy: Direct measurement is more precise than an average cost per day. This added precision might not yield greater levels of accuracy. Researchers should choose a method that offers a sufficient amount of precision and then work to validate the accuracy of the data.
Direct Measurement / Activity Based Costing (ABC)
Most researchers are not able to directly measure the cost of health care. Such a task would require considerable time, resources, and very detailed production data. Researchers sometimes use direct measurement to estimate the cost of an intervention. For example, a study that compared outreach workers and usual care to improve smoking cessation would need to use direct measurement to estimate the cost of the outreach workers. To directly measure the costs, the research would need to identify the quantity and prices of all inputs, such as labor, space, supplies, contracts, training and quality assurance.
A related method is activity-based costing. This approach was used to develop the VA Managerial Cost Accounting (MCA) System workload and cost estimates. Researchers have access to many MCA data extracts.
Learn more about micro costing »
Learn more about the Managerial Cost Accounting (MCA) »
Learn more about measuring staff activities »
Learn more about the cost of VA staff & labor »
Learn more about determining VA capital costs »
Learn more about pseudo-bills »
Learn more about the HERC Average Cost Outpatient Dataset »
Pseudo-Bill
The pseudo-bill method combines VA utilization data with unit costs from non-VA sources to estimate the cost of patient care. This is commonly referred to as the pseudo-bill method because the itemized list of costs is analogous to a fee-for-service hospital bill. The unit cost of each item may be estimated by using Medicare reimbursement rates, the charge schedule of an affiliated university medical center, or some other non-VA source. We use this method to estimate outpatient costs in the HERC Average Cost Dataset.
Learn more about Pseudo-Bill »
Learn more about the HERC Average Cost Outpatient Dataset »
Reduced Cost List
The reduced cost list is similar to the pseudo-bill method, except that only a limited itemized list is used. Ideally, the list is limited to those items that account for the majority of the costs.
Cost Regression
The cost regression method requires detailed cost and utilization data for a specific, non-VA service to simulate the cost of a comparable VA service. If suitable non-VA data are available such as cost-adjusted charges, a regression can be estimated using cost-adjusted charges as the dependent variable and information about the encounter as the independent variable. VA costs are simulated using VA utilization data and the function’s parameters. The chief advantage of this method is that it requires less data than is needed to prepare a pseudo-bill, making it a more economical way of micro-costing. We used this method to calculate the inpatient costs of the HERC Average Cost Dataset.
Learn more about cost regressions and analyzing cost data »
Learn more about the HERC Average Cost Inpatient Dataset »
Estimated Medicare Payments
Medicare uses DRGs to assign inpatient reimbursement rates and CPT codes for outpatient reimbursement rates to providers. Researchers can use this information to calculate an expected Medicare payment. The process of estimating Medicare payments is complex because payments are adjusted to account for regional wage differences and individual hospital factors.
Learn more about estimating Medicare payments »
Average Daily Rate
This method assumes that the cost of an inpatient stay is directly proportionate to the length of stay. We use this method for rehabilitation, mental health and long term care in the inpatient HERC Average Cost Dataset. This method is also frequently used to estimate costs when the utilization data come from participant self-report.
Learn more about the HERC Average Cost Inpatient Dataset »
Which method is best?
The best method to use depends on the level of precision required, and the levels of resources available (jump to Table B1.1 listing the advantages and disadvantages of each method). Micro-costing methods are precise, but expensive to employ. Average cost methods are easier to undertake, but the cost estimate may not precisely reflect how the intervention affects the resources used in providing care. In fact, it is often appropriate to use mixed methodologies in the same study.
The analyst must consider whether the assumptions used to create average cost estimates are appropriate to all utilization data within the study; for example, whether the intervention might affect the cost of hospitals stays in a way that will not be captured by the DRG or length of stay, or whether it will effect the cost of ambulatory visits in a way that will not be captured by the relative value units associated with CPT codes.
Estimates of outpatient costs based on average cost methods do not reflect the cost of prescription drugs. The cost of prescriptions may be found in the MCA national data extracts, or they may be compiled from the Pharmacy Benefits Management database.
It is uncertain if the VA national outpatient utilization databases include all outpatient care. It is possible that they understate laboratory tests and prosthetic supplies. If this is true, then analysts who need an estimate that reflects this type of utilization must turn to micro-costing. Orders for laboratory tests must be extracted from the VISTA system. Prosthetics data are kept in a national prosthetics database.
Table B1.1 - Overview of Cost Methods
| Method | Source of Data | Assumptions | Advantages/Disadvantages |
|---|---|---|---|
| Pseudo-bill | Detailed utilization data Schedule of charges adjusted for cost |
Schedule of charges reflects relative resource use Cost-adjusted charges reflect VA costs |
Pro: Captures effect of intervention on pattern of care within an encounter. Con: Expense of obtaining detailed utilization data Charge schedule may not represent VA costs. |
| Cost function based on non-VA data | Previous study with cost-adjusted charges and detailed utilization Reduced list of utilization measures previously identified as important |
Same as for pseudo-bill The relationship between cost and utilization is the same in the current study as in the previous study |
Pro: Less effort to obtain reduced list of utilization measures than to prepare a pseudo-bill. Con: Must have detailed data from a prior study, may result in error or bias. |
| Direct measurement | Staff activity analysis Payroll data on labor cost Estimate of supply costs |
May assume all utilization uses the same amount of resources | Pro: Useful to determine cost of a program that is unique to VA. Con: Limited to small number of programs, can't find indirect costs, can't find total healthcare cost. |
| HERC average cost per inpatient day for psychiatric, rehabilitation, and long-term care | Patient Treatment File and aggregate cost from MCA. | All inpatient days have equal cost | Pro: Simple, may be accurate for psychiatric and rehabilitation stays. Con: Doesn't capture case-mix variation in long-term care. |
| HERC average cost of acute medical and surgical stays | Patient Treatment File and aggregate cost from MCA Relative Values from Analysis of Cost of Veterans' Medicare Stays |
VA use of resources for different diagnoses and lengths of stay are the same as for non-VA hospitals | Pro: Avoids bias of assuming all days of equal cost, can estimate cost from administrative data. Con: Only appropriate for acute medical and surgical stays. Not sensitive to all sources of variation in resource use cost. |
| Average cost per clinic visit | Outpatient Care file and aggregate cost from MCA | All visits have the same cost | Pro: Can estimate cost from administrative data Con: Does not capture variation in ambulatory care cost |
| HERC outpatient average cost method | Outpatient Care File and aggregate cost from MCA | All visits with the same CPT codes have the same cost | Pro: Can estimate cost from administrative data Con: Assumes that VA characterizes care with appropriate CPT codes, that non-VA charge schedules represent VA relative cost of production, and that Medicare level of facility costs are incurred in all visits. |
| Decision Support System | MCA national extract or MCA production data | MCA accurately assigns costs, finds relative value units, and identifies utilization | Pro: Staff at each facility develop estimates of department costs, products and encounters. Con: Needs to be validated, especially in earliest years. |
References
Garber, A. M. 2000. Advances in cost-effectiveness analysis of health interventions. In Handbook of Health Economics, edited by A. J. Culyer and J. P. Newhouse. Amsterdam: Elsevier Science.
Last updated: April 24, 2023
