Two reports can show different counts and both be internally consistent. In hospital data work, that is often the point where I stop looking at the arithmetic and ask what each number is meant to represent.
Start with the thing counted
A person, an appointment, an admission and an encounter are not interchangeable units. One person might have several encounters; an encounter might also be recorded across more than one system or workflow. Before comparing totals, I want to know which of these the report is counting.
The distinction can be easy to miss when a field has a familiar name such as “patient count”. The name alone does not tell me whether it means unique people, records, or activity within a chosen period. That definition belongs with the data, not just in someone’s memory.
Time is part of the definition
A date can refer to when something happened, when it was entered, or when it became available to another system. Those timings can fall on different days, particularly when records are updated after the original event. A midnight boundary can therefore change a daily figure without changing the underlying activity.
When I work through a mismatch, I check the date field and the period rule before assuming one source is wrong. That means asking practical questions: which timestamp is used, what time zone applies, and whether late updates are included. A clear answer makes the number easier to interpret and the logic easier to maintain.
Write down the meaning
A short definition beside a data model or report can prevent a long chain of guesswork later. It should state the unit being counted, the relevant dates and any important exclusions in plain language. If two teams use the same label for different things, recording that difference is more useful than trying to make the labels look consistent.
I have learned to treat questions about a surprising number as questions about meaning first and code second. The next time two hospital reports disagree, start by describing what each one counts. That is often where the useful investigation begins.