Limits...
Marginal costing methods highlight the contributing cost of comorbid conditions in Medicare patients: a quasi-experimental case-control study of ischemic stroke costs.

Simpson AN, Bonilha HS, Kazley AS, Zoller JS, Ellis C - Cost Eff Resour Alloc (2013)

Bottom Line: Cost of illness studies are needed to provide estimates for input into cost-effectiveness studies and as information drivers to resource allocation.However, these studies often do not differentiate costs associated with the disease of interest and costs of co-morbidities.Additionally, correct cost estimates are needed as inputs for valid cost-effectiveness studies.

View Article: PubMed Central - HTML - PubMed

Affiliation: Department of Healthcare Leadership and Management, Medical University of South Carolina, 151 Rutledge Avenue, MSC 962, Charleston, SC 29425, USA. simpsona@musc.edu.

ABSTRACT

Background: Cost of illness studies are needed to provide estimates for input into cost-effectiveness studies and as information drivers to resource allocation. However, these studies often do not differentiate costs associated with the disease of interest and costs of co-morbidities. The goal of this study was to identify the 1-year cost of ischemic stroke compared to the annual cost of care for a comparable non-stroke group of South Carolina (SC) Medicare beneficiaries resulting in a marginal cost estimate.

Methods: SC data for 2004 and 2005 were used to estimate the mean 12 month cost of stroke for 2,976 Medicare beneficiaries hospitalized for Ischemic Stroke in 2004. Using nearest neighbor propensity score matching, a control group of non-stroke beneficiaries were matched on age, gender, race, risk factors, and Charlson comorbidity index and their costs were calculated. Marginal cost attributable to ischemic stroke was calculated as the difference between these two adjusted cost estimates.

Results: The total cost estimated for SC stroke patients for 1 year (2004) was $81.3 million. The cost for the matched comparison group without stroke was $54.4 million. Thus, the 2004 marginal costs to Medicare due to Ischemic stroke in SC are estimated to be $26.9 million.

Conclusions: Accurate estimates of cost of care for conditions, such as stroke, that are common in older patients with a high rate of comorbid conditions require the use of a marginal costing approach. Over estimation of cost of care for stroke may lead to prediction of larger savings than realizable from important stroke treatment and prevention programs, which may damage the credibility of program advocates, and jeopardize long term funding support. Additionally, correct cost estimates are needed as inputs for valid cost-effectiveness studies. Thus, it is important to use marginal costing for stroke, especially with the increasing public focus on evidence-based economic decision making to be expected with healthcare reform.

No MeSH data available.


Related in: MedlinePlus

Standardized Difference of Means of PS Covariates Before and After Matching.
© Copyright Policy - open-access
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC3842820&req=5

Figure 1: Standardized Difference of Means of PS Covariates Before and After Matching.

Mentions: A plot of the absolute standardized differences in means (Figure 1) offers a good representation of whether selection bias of known factors has been reduced by matching on propensity score [10]. Balance (i.e. reduction in standardized mean differences between stroke cases and selected controls) for each covariate that was used in the logistic regression model of propensity is improved when differences are reduced after matching to no greater than 0.25 [11]. Figure 1 illustrates that the standardized differences in means for each of the fourteen covariates included in the propensity model is reduced to less than the 0.25 recommended level after matching, indicating good control of potential selection bias of known confounders.


Marginal costing methods highlight the contributing cost of comorbid conditions in Medicare patients: a quasi-experimental case-control study of ischemic stroke costs.

Simpson AN, Bonilha HS, Kazley AS, Zoller JS, Ellis C - Cost Eff Resour Alloc (2013)

Standardized Difference of Means of PS Covariates Before and After Matching.
© Copyright Policy - open-access
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC3842820&req=5

Figure 1: Standardized Difference of Means of PS Covariates Before and After Matching.
Mentions: A plot of the absolute standardized differences in means (Figure 1) offers a good representation of whether selection bias of known factors has been reduced by matching on propensity score [10]. Balance (i.e. reduction in standardized mean differences between stroke cases and selected controls) for each covariate that was used in the logistic regression model of propensity is improved when differences are reduced after matching to no greater than 0.25 [11]. Figure 1 illustrates that the standardized differences in means for each of the fourteen covariates included in the propensity model is reduced to less than the 0.25 recommended level after matching, indicating good control of potential selection bias of known confounders.

Bottom Line: Cost of illness studies are needed to provide estimates for input into cost-effectiveness studies and as information drivers to resource allocation.However, these studies often do not differentiate costs associated with the disease of interest and costs of co-morbidities.Additionally, correct cost estimates are needed as inputs for valid cost-effectiveness studies.

View Article: PubMed Central - HTML - PubMed

Affiliation: Department of Healthcare Leadership and Management, Medical University of South Carolina, 151 Rutledge Avenue, MSC 962, Charleston, SC 29425, USA. simpsona@musc.edu.

ABSTRACT

Background: Cost of illness studies are needed to provide estimates for input into cost-effectiveness studies and as information drivers to resource allocation. However, these studies often do not differentiate costs associated with the disease of interest and costs of co-morbidities. The goal of this study was to identify the 1-year cost of ischemic stroke compared to the annual cost of care for a comparable non-stroke group of South Carolina (SC) Medicare beneficiaries resulting in a marginal cost estimate.

Methods: SC data for 2004 and 2005 were used to estimate the mean 12 month cost of stroke for 2,976 Medicare beneficiaries hospitalized for Ischemic Stroke in 2004. Using nearest neighbor propensity score matching, a control group of non-stroke beneficiaries were matched on age, gender, race, risk factors, and Charlson comorbidity index and their costs were calculated. Marginal cost attributable to ischemic stroke was calculated as the difference between these two adjusted cost estimates.

Results: The total cost estimated for SC stroke patients for 1 year (2004) was $81.3 million. The cost for the matched comparison group without stroke was $54.4 million. Thus, the 2004 marginal costs to Medicare due to Ischemic stroke in SC are estimated to be $26.9 million.

Conclusions: Accurate estimates of cost of care for conditions, such as stroke, that are common in older patients with a high rate of comorbid conditions require the use of a marginal costing approach. Over estimation of cost of care for stroke may lead to prediction of larger savings than realizable from important stroke treatment and prevention programs, which may damage the credibility of program advocates, and jeopardize long term funding support. Additionally, correct cost estimates are needed as inputs for valid cost-effectiveness studies. Thus, it is important to use marginal costing for stroke, especially with the increasing public focus on evidence-based economic decision making to be expected with healthcare reform.

No MeSH data available.


Related in: MedlinePlus