Colorectal cancer (CRC) is the second leading cause of cancer death in the US, and screening to allow early detection and prevention of CRC substantially reduces CRC mortality. The importance of optimizing CRC screening in VHA has been highlighted by recent research indicating the CRC screening is only current in 32% of age-eligible veterans. At many VA facilities, options for CRC screening are substantially limited by the availability of endoscopists. For example, until recently, Ann Arbor VA patients at average CRC risk were not considered eligible for primary screening sigmoidoscopies or colonoscopies due to excessive backlog; screening was thus restricted to fecal occult-blood testing (FOBT). Recent studies by the CRC QUERI have found similar issues in other VA facilities. In a recent national directive, the Department of Veterans Affairs stated that "Given that each [screening] modality has advantages and disadvantages and that none has clearly been proven to be superior, the choice of specific screening strategy needs to be based on patient preferences." The directive has mandated that all screening options be allowed for patients, based on individual preferences. While this is an admirable goal, it is currently unclear whether a patient preference driven strategy is feasible due to staffing and resource limitations; planning to meet possible demand for endoscopy is clearly necessary.
We conducted a QUERI study to estimate endoscopy demand and staffing needs in VHA. This project will build upon several QUERI projects that will provide critical data inputs for a simulation model projecting endoscopy demand. The specific aims of the project are as follows:
1) To project demand for CRC screening related endoscopy
2) To estimate staffing needs to meet projected demands for CRC screening
3) To create a flexible tool that can be used by facility managers to project local staffing needs and to help plan to meet demand for endoscopic services
We used a previously published simulation model, along with aggregate reports of surveys of veterans, to estimate likely demand for various endoscopic procedures related to CRC screening in VHA. We then examined, based upon CRC QUERI estimates, the current and projected staffing requirements to meet the national directive towards allowing a broader range of screening modalities than is currently applied. This will guide the VHA in setting policy for screening recommendations for CRC.
We compiled a number of data sources to provide data estimates for our model. There are unfortunately few fully representative data to guide us, so that we relied on a number of sources that may not be fully representative of other regions in the VA. As a result, we put together a spreadsheet that allows local users to input some key pieces of data that may be locally available in order to get more accurate local estimates.
We used a population of approximately 242,495 veterans who were represented by a random sampling process of 24 facilities as part of the SCREEN survey put together by the CRC QUERI. This survey elicited preferences for screening, and found that approximately 27% of subjects prefer FOBT; 40% colonoscopy; 8% sigmoidoscopy; 3% no screening; and 18% don't know. Endoscopist supply was estimated from the GI CPOS survey, which found an average of ~2.5 FTE per facility, though the distribution varied widely, with many facilities having none and many of the academic affiliated facilities having 8 or more.
We then used our simulation model to project, for the population of 242,495 veterans, how many procedures would need to be done under a variety of scenarios. In our base estimate, we assume that adherence to screening is 70% overall, a number that is approximately in-line with performance standards. In our first scenario, we projected based on 100% use of either colonoscopy or FOBT. We then varied these projections based on the preference data above. We found that for a pure colonoscopy strategy, the average demand would be for 26 colonoscopies per endoscopist per month, while for an FOBT strategy the demand would be for 13.5 colonoscopies per endoscopist per month. A mixed strategy, based on patient preference from the survey, would lead to about 21 procedures per endoscopist per month.
To put this in context, we compared this to reports of procedure volume at a large VA center. They averaged a total of 29 colonoscopies per endoscopist per month, and at least half of these were not screening related, so that the typical screening volume is about 15 colonoscopies per endoscopist per month. Thus, a preference-based strategy is likely to require substantial increases in FTE, probably on the order of 30%, in order to meet demand. An FOBT-based strategy, in contrast, should keep demand approximately in-line with current screening.
We have created a simple spreadsheet tool that allows users to input several variables, including their local population size, age distribution of their population, the number of endoscopists available for that population, and a percentage of procedures done either outside the VA or by another non-GI provider (e.g., surgeons). The spreadsheet will then estimate how many procedures would be required per endoscopist per month.
We view our results as conservative for several reasons. First, adherence to screening may be rising given current performance measures, and there seems to be a trend towards allowing colonoscopy as the primary screening test. Second, we do not currently included missed appointments or poor preparation in our model, and these often require repeat testing. Third, there is documented overuse of screening in the VA, particularly terms of the use by the elderly and in overuse of surveillance (e.g., performing more frequent follow-up than guidelines suggest).
We have provided estimates of the effect of preference on likely demand for colonoscopy. Based on survey data, it appears that a preference for colonoscopy-based screening is likely to lead to an increase in demand that the current system cannot fully meet. This, however, is highly dependent on a number of important variables that are not fully identifiable at present, including the proportion of tests that are done outside of the VA system. We have created a simple, flexible spreadsheet that can help users with better local data estimate their likely demand for colonoscopy.
External Links for this Project
None at this time.