A baseline starts by defining what was measured
4 min read
Lesson objectives
- Build a bounded baseline of patient-experience signals from heterogeneous reports without manufacturing a rate, clinical outcome or causal conclusion.
Please sign in to use all the tools
Once you sign in, the reading tools are yours: larger or smaller text, typeface, line spacing, focus mode, highlighting and bold, and notes saved to your account and gathered in My study.
A baseline starts by defining what was measured
Before recommending an improvement, write a baseline stating what is known and how it was learned. A phone account, staff observation, booking log and survey response answer different questions. A booking record may establish when an appointment was created, but not whether a patient understood the directions. One comment may expose an important failure point but cannot estimate its prevalence among all visitors. A defensible foundation-level baseline is an evidence register connecting the experience domain, source, period, group and unknowns. Do not combine counts from unlike sources in one denominator unless the unit, definition and time period match. Name the source before calculating a percentage so the reader knows exactly whom the figure describes.
Define an experience indicator with a question about something the patient can report, such as whether the next step after the visit was explained. Suppose 20 people were invited to a local survey, 12 responded and 9 respondents said the next step was clear. Nine divided by twelve is 75% of respondents; twelve divided by twenty is a 60% response proportion. It would be wrong to state that 75% of all patients understood. The eight nonrespondents are unknown, and an answer about clarity alone does not prove comprehension or clinical correctness of the plan. Record collection mode and date and consider whether response patterns differ among groups. Standardized survey instruments have their own guidance. This invented question is not a CAHPS item, and its number is not comparable with published CAHPS measures.
For narratives, preserve meaning and use a limited descriptive code: insufficient location information, unknown follow-up contact, or understandable explanation. A pattern can justify a check, not an automatic causal verdict. A single complaint may flag a serious safety concern even when its frequency is unknown; its escalation should not wait for a prevalence estimate. Ten delay complaints could reflect higher service volume or easier collection as well as a worse process. Examine context before attributing the change to one intervention. Specify what the next measurement would repeat: the same event definition, a sensible time window and a balancing observation such as staff burden or access for people who do not use digital channels.
Lesson summary
Link counts to an observation unit, period and source. Build a baseline showing denominators, unknowns and the next verification step.