Patient Surveys and Reviews in an ENT Practice

Patient surveys and reviews are the first thing an ENT practice drops when the phones get loud. The intent survives every staff meeting. The send does not, because the person who was going to do it spent the afternoon rebooking a surgical block and the survey window closed.

What makes this worse than a missed task is that the practice still gets reviewed. Patients who had a bad experience find the review form without any prompting from you. Patients who had a good one mostly do not. A practice with no survey program is not a practice with no feedback loop. It has one, and it is running entirely on the people most motivated to complain.

The survey is not hard, the timing is

Sending a survey is a solved problem. Sending it about the correct visit, to a patient who is in a position to answer, is where ENT practices come apart.

The standard instruments are public and well specified. The CAHPS Clinician and Group Survey defines measures for access, provider communication, and office staff courtesy, which covers most of what a practice actually wants to know and saves a lot of arguing about question wording. Adopting an existing measure set also means the results compare to something outside your own four walls.

The operational failure is upstream of the questions. It is that nobody knows which visit to survey, on which day, through which channel, for a patient who was in the building twice this month for related reasons.

A patient with two appointments gets surveyed about the wrong one

ENT runs a high share of linked visits: an audiology evaluation paired with a physician visit, a procedure followed by a post-op check, an allergy testing appointment that leads straight into the immunotherapy build-up series.

Native EHR reminders fire on the chronologically first appointment only. A patient with a 9:30 audiology slot and a 10:00 provider visit gets reminded about one of them. Survey logic built on the same event inherits the same defect, and the practice ends up asking a patient about their 9:30 hearing test when what they want to talk about is the conversation at 10:00.

The answers that come back are not wrong exactly. They are unusable, because the respondent and the practice are discussing different appointments. Scores drift, nobody can explain the drift, and the program loses credibility with the physicians within two quarters.

The fix is to treat the linked set as the survey unit rather than the appointment. Read the day’s appointments for that patient, identify the visit that the survey is actually about under a rule the practice writes, and send one survey referencing it specifically. This is the same read-the-whole-day logic that makes multi-appointment reminders work, applied to a different outbound message, which is why it is worth doing once rather than per channel.

Route the unhappy answer inside before it becomes a public one

This is the section administrators care about, and it is a routing problem rather than a marketing one.

The useful design separates two outbound actions that most practices collapse into one. First the survey goes out. Then, and only for responses that came back positive, a review request follows. A blanket post-visit review request sent to everyone is an efficient way to convert private dissatisfaction into public dissatisfaction.

Responses that come back negative go the other direction. They route to a named internal bucket in athenaOne, to the office manager rather than the provider, with the visit, the department, and the verbatim comment attached. The value is the speed: a complaint that reaches a manager within an hour of the visit is frequently still fixable, and a patient who gets a call back about it often does not write the review at all.

None of this involves judging the substance of the complaint. The automation sorts on the response value the patient selected and delivers the text. A human reads it and decides what it means and what to do, which is exactly where that decision belongs.

Channel and language decide your response rate

A survey program lives or dies on response rate, and response rate is mostly a distribution question.

Running the same instrument across two-way SMS, the patient portal, and an outbound voice call for patients who use neither raises the denominator substantially, particularly in an ENT practice where about 15% of American adults aged 18 and over report some trouble hearing and a phone survey is the wrong instrument for exactly the patients most engaged with the practice. Offering the survey in the patient’s preferred language rather than defaulting to English does the same thing.

The practical constraint is that each channel has its own rules. Portal delivery depends on the patient being enrolled, so portal enrollment nudging becomes part of the same workflow rather than a separate initiative. SMS depends on a current mobile number and a consent record. Voice depends on a reasonable call window. Running all three from one rule set, rather than three vendors with three schedules, is the difference between a program and a set of overlapping messages that annoy people.

What the practice keeps doing

Automation here replaces the sending and the sorting, not the reading.

The office manager still reads every negative comment. The physicians still see their own results. Someone still decides whether a recurring complaint about wait time in the audiology suite is a staffing problem or a template problem, and that decision needs somebody who knows the building.

What changes is that the work reaches those people at all. The survey program stops depending on whether Tuesday was busy, the negative response arrives while the visit is still fresh rather than in a monthly export, and the review requests go to the patients who said they were happy rather than to everyone with an email address.

That is a small operational change with a compounding effect. Practices that run this consistently do not get better feedback than practices that do not. They get feedback that is representative, which is a different and more useful thing.

Key Takeaways

Use an existing measure set rather than writing your own questions. The CAHPS Clinician and Group measures cover access, communication, and staff courtesy, and they let you compare outside your own practice.
Make the linked appointment set the survey unit, not the appointment. ENT patients are frequently in twice in one day, and native reminder logic fires on the first appointment only.
Separate the survey from the review request. Send the review request only after a positive response, because a blanket send converts private dissatisfaction into public dissatisfaction.
Route negative responses to the office manager within the hour, with visit, department, and verbatim comment attached. Fixable complaints have a short shelf life.
Run one rule set across SMS, portal, and voice. In an ENT panel a phone-only survey systematically excludes the patients with hearing difficulty, who are the ones you most need to hear from.
Treat portal enrollment as part of the survey workflow. Delivery channel availability is the ceiling on response rate, and it is a fixable ceiling.

A survey program that runs only when the front desk has a quiet afternoon is not a program. The instrument is public, the routing rules are things your office manager can write in an hour, and the reason it has not been running is that somebody had to press send several hundred times a month. That part is worth handing to an AI team that already knows which visit the patient is being asked about.

Sources