A lab director tells a budget committee her DNA unit closed 4,000 cases last year. A different lab director, same investigative area, also reports 4,000 cases. The committees walk away thinking they have comparable numbers.
They might not. A case is a request from a customer: a police agency, a prosecutor, a medical examiner, an attorney. But a case can arrive with one item of evidence or with forty. It can require one sample per item or several. The case count alone says nothing about how much work sits behind it. Two labs can report identical case numbers and be doing very different amounts of underlying work. A case is not a case is not a case.
This is why Project FORESIGHT built a common vocabulary before it built a benchmark. Comparing labs without first agreeing on what’s being counted produces numbers that look precise and mean nothing.
Case
A case is a request from a crime laboratory customer that includes forensic investigation in one or more areas. LabRAT reports a case for each investigative area a request touches, which is what’s meant whenever a Project FORESIGHT table says “cases per FTE” or “cost per case” for a given discipline.
Item
An item is a single physical object submitted for examination: a swab, a firearm, a bag of pills, a hard drive. One item can be relevant to more than one investigative area. A firearm recovered at a scene might generate work in firearms examination and in DNA casework from the same object.
Items per case varies enormously by discipline, and that variance is the point, not noise. A drug case might arrive with one item. A trace evidence case from a vehicle fire can arrive with dozens. Reporting cases without items obscures how much physical material backs up the case count.
Sample
A sample is an item, or a portion of an item, that produces a reportable result. One item frequently yields more than one sample. A single piece of clothing might be sampled at four locations for four separate DNA results.
This is where case-level counts and bench-level reality start to diverge. The relationship compounds: cases contain items, items contain samples, and the multiplier at each step is discipline-specific. DNA casework runs close to 1.5 samples per item at the median; serology and biology can run several times higher, depending on how a lab partitions its sampling strategy.
Test
A test is an analytical process: instrumental analysis, microscopic examination, a presumptive screen, an extraction, a comparison. It does not include technical or administrative review. Review confirms a test was done correctly; it isn’t itself a test.
Tests per sample is where the analytical intensity of a discipline shows up most clearly. Toxicology screens for a long panel of compounds; trace evidence comparisons can involve several instrumental passes on a single sample. The count reflects method, not effort, but the two are easy to conflate if the terms aren’t pinned down first.
Report
A report is the formal statement of results, the thing that actually leaves the laboratory and enters the case file. Reports per case sits close to 1.0 for most disciplines, for the obvious reason that one case usually generates one report. It is not always 1.0. A lab that issues supplemental or amended reports, or that reports interim results separately from final results, will show a ratio above 1.0. A lab that bundles results from multiple cases into a single combined report will show a ratio below it.
Why the terms matter
Five terms, four ratios connecting them: items per case, samples per item, tests per sample, reports per case.1 None of these ratios is good or bad on its own. They describe how a discipline does its work, not how well.
Where they earn their keep is in comparison. Two labs reporting the same cost per case can be doing completely different amounts of underlying work if their items-per-case or samples-per-item ratios differ. A discipline with a high test-to-sample ratio is not necessarily inefficient; it may simply require more analytical steps per unit of evidence to produce a defensible result.2 Reading cost or productivity metrics without the unit-of-work context behind them is reading half the picture.
This is also why Project FORESIGHT keeps the definitions fixed across years even when individual labs count things differently in their day-to-day operations. A lab’s internal definition might be more convenient for its own case management system but it would not be comparable to anyone else’s, which defeats the purpose of a benchmark.
What would change in your own reporting if you had to state, every time, which of these five units you were counting?
Of course, these are not the only ratios that can be used. For example, tests per case may be used to evaluate methods used by different labs for the same type of analysis.
A FORESIGHT toxicology lab had a good example of this. They were testing well above the average tests per sample but at an average cost per case and turnaround time. When it was suggested they could reduce their testing intensity to save money, the lab said they were intentionally over-testing because they served an aging population in a jurisdiction with multiple research hospitals (meaning a massive medical presence). The lab knew the medical information systems didn’t communicate well, meaning a patient with multiple doctors might be prescribed confounding medications. The lab deliberately over-tested to look for those interactions that might impinge on the patient’s health (and death).