Clinical Quality & Patient Safety

Reports land in a queue. A nurse files one about a medication given late, a tech files one about a patient who slid to the floor, somebody files one that is really a complaint about another department. Most arrive as a short free-text narrative typed at the end of a shift by someone who wanted to go home, filed under a category picked from a dropdown that did not quite fit. You read them, assign a harm level, separate the near misses from the events that reached the patient, and route each one to the manager who owns it. A few have to move today: something that may meet the sentinel event definition, something the state has to be told about, something risk management and counsel need before anyone else has it. The rest wait while you work out whether the several reports about the same infusion pump from different units are separate problems or one. And you know the queue is not the world. What gets filed depends on who had time to file it, and the units that report most are often the ones paying most attention.

You know the incident reports are not the whole picture, so you go looking for harm in the chart instead. A sample of records is screened against a list of triggers: a reversal agent given, an unplanned transfer to a higher level of care, an unexpected return to the operating room, a medication stopped abruptly, a transfusion nobody anticipated, a positive culture that appeared days into the stay. A trigger is not a finding. It is a reason to read the record. Reviewers read the flagged charts, a second confirms, and together they decide whether harm actually occurred and how severe it was. It is slow, so you sample rather than screen everything, which means your harm rate is an estimate drawn from a sampled subset of admissions and can move for reasons that are really sampling. Alongside it you carry the harm that gets counted whether or not anyone reports it: infections submitted through national surveillance, the safety indicators that fall out of coded claims, the conditions flagged as not present on admission.

The scored questions tell you the number moved. The comments tell you why, and nobody has time to read them all. The standardized survey items are closed-ended, so the free text arrives through the supplemental questions your organization added to the instrument its survey vendor fields, plus the letters, the compliments, the complaints handed to a front desk, and whatever patients wrote on consumer review and physician-rating sites where everyone can read it and you cannot take it down. You skim for the ones that have to become something: a complaint that is legally a grievance and starts a formal clock, a comment describing what sounds like a safety event, a comment naming a specific member of staff. The rest get coded by theme so a service line director can be shown what patients on their floor are saying, usually well after they said it. And the sample is not the population. The patients who answer surveys are not the patients you treated.