APHL-CDC Influenza Cost Estimation Models Introduction An optimal influenza surveillance system requires adequate resources to support all elements of the system. Sustaining the national, state and local components of the system is becoming increasingly threatened by the decline in funding. While funding is often cited as a key limiting factor, the true costs of virologic surveillance are not well defined, likely because of the complexity of the system. Optimizing resources and justifying costs associated with influenza surveillance at the national, state and local level is becoming increasingly important. In response to the need for a better understanding of the true cost of influenza virologic surveillance, and in association with the Influenza Virologic Surveillance Right Size Roadmap, the Association of Public Health Laboratories (APHL) and the Centers for Disease Control and Prevention (CDC), in collaboration with Booz Allen Hamilton, conducted the Influenza Cost Estimation Models Project. The model is a cost-per-specimen metric based on operational and inventory-related costs, testing methodology and overall laboratory processes. Public health laboratories (PHLs) consistently have to decide between cost efficiency and increased capacity (i.e. increasing the number of specimens tested by the laboratory). And there are times when one is more effective than the other depending on the testing situation. This costing model serves as a metric to allow individual laboratories to determine the true cost for influenza testing, identify costsaving opportunities for future specimen testing and to make an informed decision between cost efficiency and capacity. Methods The costing model built upon the APHL-CDC Influenza Laboratory Resource and Process Modeling Project, conducted in 2011 and 2012, which focused on laboratory capacity for influenza testing. The resource and process modeling project collected and evaluated laboratory-specific data in order to analyze current PHL testing capacity and determine the optimal processes and resource allocations in PHLs for influenza testing. To build on the existing modeling project, APHL, CDC, and Booz Allen Hamilton worked to: (1) identify additional key inputs that had not already been collected from the previous resource and process modeling project and that needed to be collected for the cost model; (2) determine key value-added outputs to analyze; and (3) design an Excel based model and dashboard to utilize inputs and simulation results, perform calculations and present the findings on influenza laboratory costs. Six public health laboratories participated in both projects.
Three different testing algorithms were used in the costing model based on the CDC Human Influenza Virus Real-Time RT-PCR Diagnostic Panel: Results 1) Influenza A/B Typing Assay with full influenza A subtyping using baseline resources (9 specimens per plate), 2) Influenza A/B Typing Assay with reflex influenza A subtyping using baseline resources (19 specimens per plate for typing; 9 specimens per plate for subtyping), 3) Influenza A/B Typing using surge resources (19 specimens per plate). **costing differed for each algorithm due to different numbers of specimens per plate Detailed data defining labor, consumable materials, equipment and overhead and shipping were collected from each laboratory utilizing Excel based input worksheets and the data was directly imported into a discrete event simulation model using Arena simulation software. Analysis was performed on the data for the six laboratories, individually and at an aggregate level. For laboratory one, the total cost per specimen was more than one standard deviation above the average testing cost across all six laboratories; therefore, Laboratory 1 was labeled as an outlier and excluded from further analysis. After analysis, each of the participating laboratories received an interactive dashboard allowing the users to easily alter inputs (such as costs) and correspondingly see the results change. The results presented are the aggregate results of the six participating PHLs. Baseline Resources: the algorithm used and resources [i.e. staff and equipment] available to the laboratory on a daily basis Surge Resources: the algorithm and resources that are expected to be in place during emergency periods of high testing demand The results of the costing model showed that, at baseline influenza testing costs vary widely among laboratories, ranging from $81 to $398 per specimen for the Influenza A/B Typing Assay with full subtyping algorithm, and from $83 to $620 for the Influenza A/B Typing Assay with reflex influenza A subtyping algorithm (table 1). When labor costs were excluded, the difference in the cost per specimen between the highest and lowest cost per specimen decreased from $537 to $175 in both testing algorithms. 2 P age
Excluding both Laboratory 1 and labor costs, the average cost per specimen ranged from $81 to $140 for the Influenza A/B Typing Assay with full subtyping algorithm and $83 to $152 for the Table 1: Capacity per Day and Cost per Specimen *AB screen analysis includes surge resources where the analysis for the Full Subtyping Panel and Reflex Assay do not Testing Algorithm Average Capacity per Day Range of Cost per Specimen Excluding IRR Materials Range of Cost per Specimen Including IRR Materials Full Subtyping Panel 52 specimens $ 42-358 $ 81-398 Reflex Assay 65 specimens $ 39-573 $ 83-620 AB Screen* 148 specimens $ 34-137 $ 60-156 Influenza A/B Typing Assay with reflex influenza A subtyping algorithm. The large variation in cost per specimen was likely due to varying labor and equipment costs among the participating PHLs (Figure 1). Labor costs most likely varied due to differences in cost of living and wages among the six laboratories and equipment cost variations were most likely due to the differing extraction platforms utilized by each of the six laboratories. Varying costs between the Algorithms During baseline, the Influenza A/B Typing Assay with full subtyping algorithm was recommended for cost efficiency. The higher cost in the Influenza A/B Typing Assay with reflex influenza A subtyping algorithm may be attributable, at least in part, to the additional labor costs associated with the algorithm (i.e. plate prep, adding specimen and controls to RT-PCR). Analysis was performed on the cost of labor for each processing step of the three algorithms (outlined in Figure 2). For the Influenza A/B Typing Assay with full subtyping algorithm and Influenza A/B Typing 3 P age
Assay with reflex influenza A subtyping algorithm results showed that the reporting and supervisory review steps accounted for 37% and 23%, respectively, of the total labor cost (figure 2). The results reporting, supervisory review and specimen accessioning steps accounted for approximately 40% of the total cost for the Influenza A/B Typing Assay with reflex influenza A subtyping algorithm. The addition of a supervisory review step increased the cost to 21% of the total labor cost. The results reporting, supervisory review and specimen accessioning steps accounted for approximately 40% of the total cost for the Influenza A/B Typing Assay with reflex influenza A subtyping algorithm. Accounting for Reagents The analysis demonstrated that the Influenza Reagent Resource (IRR) absorbs an average of 26% to 53% of the overall testing cost for each specimen tested across the three protocols (excluding Laboratory 1) (Figure 3 and 4). Further analysis determined which algorithm would be the most cost efficient method with the IRR absorbing the cost of reagents. During baseline, it was determined that the Influenza A/B Typing Assay with full influenza A subtyping algorithm was the more cost efficient method (cost range of $81 to $398) for laboratories. However, the algorithm that provides the laboratory with the highest capacity (but Figure 5: Cost Per Specimen By Extraction Platform 4 P age
not necessary the most cost effective) was the Influenza A/B Typing Assay with reflex influenza A subtyping algorithm (cost range of $83-$620/specimen). IRR subsidized costs were further broken down into costs associated with extraction methodologies as seen in Figure 5. Reagent costs used in the analysis were based on the 2012 pricing of IRR reagents and do not include overhead and shipping. Major Findings The major findings that resulted from the data analysis include: The Influenza A/B Typing Assay with full subtyping algorithm was recommended for cost efficiency, while the Influenza A/B Typing Assay with reflex influenza A subtyping algorithm is recommended when increased capacity is needed. The IRR absorbs an average of 26% to 53% of the overall testing costs for a specimen across the three algorithms (analysis excludes Laboratory 1). The results reporting, supervisory review and specimen accessioning steps accounted for approximately 40% of the total cost for the Influenza A/B Typing Assay with reflex influenza A subtyping algorithm. The analysis of the costing model further showed that each laboratory s viral culture testing is different. Therefore, conclusions were difficult to draw based on the data set. For example, although the labor cost type was a prominent cost in some laboratories for PCR testing, there was not one consistent cost type that provided the largest portion of total cost for viral culture. Additionally, the cost per specimen model allows PHLs to identify cost-saving opportunities for future specimen testing as well as resource needs to help improve pandemic and outbreak response planning. The aggregate data generated will also help policy and decision makers understand the current cost of influenza testing for PHLs in order to support an effective influenza virologic surveillance system. Acknowledgements APHL would like to the public health laboratories who participated in this project. APHL would also like to thank CDC s influenza division for their continued support and Booz Allen Hamilton for their contributions related to the success of this project. This project was fully funded by Cooperative Agreement # U60HM000803 from CDC and/or Assistant Secretary for Preparedness and Response. Its contents are solely the responsibility of the authors and do not necessarily represent the official views of CDC and/or Assistant Secretary for Preparedness and Response. Copyright 2012, Association of Public Health Laboratories. All Rights Reserved. 5 P age