How Many Tests?
Laboratories revise a test menu based on clinical need or cost. The common “bread and butter” tests I blogged about last time are easy to decide on and often performed on analyzers with volume discounts. But in many cases estimating the volume of tests to be performed can make the decision for us. In other words, is it cheaper to make or buy? Do we bring a test in house or send it out?
Ideally, a test should be brought in if it fills a clinical need that changes patient treatment faster or cheaper than sending it out. Many of these fall under the “soft dollar” umbrella of reducing length of stay -- notoriously hard to sell to bean counters.
Better, a cost per test can be calculated based on an estimated volume and compared to the cost of sending it out. If these “hard dollars” (variable cost) save money, that’s an easy sell. Bean counters love saving beans.
But it’s tricky, especially with new technology or services, pushing us again into that area where the beans are slippery. Here are a few ideas:
- Look at sendout volumes (easiest) - referral labs provide usage reports or numbers can be pulled from your information system. It’s a good idea to compare your top five sendout tests against your instrumentation menus to see if it’s worth performing these tests in house. Sometimes you can save money!
- Ask the docs who will order the test (harder) - having a physician advocate for a new test can’t hurt, and docs can have an expert opinion of how many tests are likely to be ordered.
- Make an educated guess (hardest) - an educated guess can be made easier if you already have a protocol in place; for example, a sepsis protocol which includes lactates can suggest how many procalcitonins may be ordered. If you don’t have a protocol you can ask local medical centers or do research online to guess how many tests might be ordered. The more data-based your estimate, the better.
Note that this doesn’t hinge on keeping techs busy. One way or another we always have plenty to do.
NEXT: Market Your Computer Skills