Measuring Reachability on Post-ED Callbacks

Measuring reachability is the first thing a post-emergency callback program should do and usually the last thing it gets around to. Most groups report calls made. Calls made is an activity number that rises when the team works harder and tells you nothing about whether anybody was actually reached, which is the only thing the program exists to achieve.

The reporting gap is easy to explain. Calls made comes free from the phone system. Reachability has to be defined, and defining it means making decisions that expose how the program is really doing.

What counts as reached. Does a voicemail count. Does reaching a family member count. If the number was disconnected, is that a failed attempt or a data problem. How many attempts before the account is closed, and closed as what.

Until those are settled, two programs reporting the same number are measuring different things, and neither can tell whether a change helped. Groups end up arguing about whether the callback program works based on numbers that were never comparable.

The good news is that these are administrative definitions, they can be written down in an afternoon, and once they exist the reporting becomes genuinely useful.

Start with an honest denominator

Reachability is a fraction, and most of the dishonesty in follow-up reporting happens in the bottom half of it.

The denominator should be every discharged patient the program was supposed to contact, not every patient someone got around to calling. A program that quietly drops patients with no usable phone number reports a wonderful contact rate while missing the people who most need the outreach.

Emergency medicine makes that denominator large. CDC data from the National Hospital Ambulatory Medical Care Survey shows that 11.5% of emergency department visits result in hospital admission, which means the great majority of encounters end with the patient going home and becoming eligible for follow-up.

So the first report a group should build is coverage. How many eligible discharges were there, how many entered the outreach queue, and where did the difference go. Patients lost between those two columns are usually lost to a data rule nobody remembers writing, and finding them is the cheapest improvement available.

Only after that is settled does contact rate mean anything.

Define reached before you count it

Write down the outcome categories, keep them few, and make them mutually exclusive.

A workable set is: spoke with the patient, spoke with an authorized person other than the patient, left a message on a permitted channel, attempted and no answer, and could not attempt because no valid contact route existed. Everything that happens on an outreach attempt should land in exactly one of those.

The distinction that matters most is between no answer and no valid route. The first is a reachability problem the program can work on with better timing and channels. The second is a data problem that belongs to registration, and lumping them together hides a fixable defect behind a metric that looks like patient behavior.

AHRQ’s re-engineered discharge toolkit sets out how to conduct a postdischarge follow-up phone call, including planning who makes the calls, when they happen, and what to do when the patient cannot be reached. Borrowing that structure is easier than inventing one, and it gives the definitions a defensible source.

One more rule keeps the numbers honest. A left message is not a contact. It is worth recording because message-then-callback is a real pattern, but folding it into contact rate turns a metric into a comfort.

Report contact rate by attempt, not overall

A single blended contact rate hides the operational question a manager actually needs answered, which is how much effort is worth spending.

Break the number down by attempt. What share of patients were reached on the first attempt, what share on the second, what share on the third. That curve flattens somewhere, and where it flattens tells the group how many attempts to fund and when to switch approaches instead of repeating one.

Break it down by timing too. Attempts land differently at different hours and on different days, and this is measurable rather than a matter of opinion. A program calling exclusively during business hours is measuring how many of its patients are at work.

Channel belongs in the same view. A text that asks the patient to call back, a portal message, and a phone call have different response profiles, and a group that only counts phone calls cannot see that one of the others is doing the work.

The payoff is a program that can be tuned with evidence. Three attempts across two channels at two different times of day is a defensible policy. Six calls to the same number between nine and five is a habit.

Close reasons are the taxonomy the whole report rests on

Every outreach task should end in a recorded reason, and those reasons should come from a short controlled list rather than free text.

Inside athenaOne, patient case documents carry close reasons, and using that structure means the reporting comes out of the same system the work happened in. No separate spreadsheet, no reconciliation, and no argument about which system is right.

A short list works better than a thorough one. Contacted and resolved. Contacted and escalated to clinical staff. Unreachable after the defined attempts. No valid contact route. Patient declined. Five categories that everyone applies the same way produce better reporting than fifteen that get used inconsistently.

The escalation category deserves separate visibility. Any follow-up program will produce contacts where the patient says something that needs clinical attention, and the front-office layer’s only job at that point is to route it quickly and record that it was routed. Counting those separately is how a group shows the boundary is being respected.

What this produces is a report that reconciles. Eligible discharges, contacted, escalated, unreachable, and no valid route should add up, and when they do not the gap is a real process defect rather than a reporting artifact.

What not to claim from this data

The last discipline is about restraint, and it is what keeps the reporting credible with clinical colleagues.

Contact rate measures the front office. It does not measure whether patients recovered, whether they followed instructions, or whether the follow-up prevented anything. Those are clinical questions requiring clinical study design, and a callback program that starts implying them will lose the trust of the people whose cooperation it needs.

Similarly, the program should not rank patients by how urgently they seem to need a call. Working the list by discharge time is administrative. Reordering it by who sounds worse is a clinical call wearing an operational hat, and it belongs to clinicians.

What the group can claim is real and worth having. A defined share of discharged patients were contacted within a defined window. A defined share were escalated to clinical staff, with the time from contact to acknowledgment recorded. A defined share could not be reached, with the reasons broken out and a plan for the data-quality portion.

That is a scorecard a practice administrator can take to a partners meeting without anyone challenging the definitions, which is the actual test of a measurement program.

Key Takeaways

Make the denominator every eligible discharge, not the patients someone got around to calling, or the contact rate flatters itself.
Separate no answer from no valid contact route, because the first is a reachability problem and the second is a registration data defect.
Do not count a voicemail as a contact, even though message-then-callback is worth recording on its own.
Report contact rate by attempt so the curve shows how many attempts are worth funding.
Vary time of day and channel deliberately, then measure the difference instead of debating it.
Close every outreach task with a reason from a short controlled list so the report reconciles inside athenaOne.
Count clinical escalations separately, including time to acknowledgment, since that is how the boundary shows up in the numbers.
Claim only front-office outcomes from this data and leave recovery and adherence to clinical study.

A callback program is judged on whether it reached people, and reaching people is measurable the moment someone writes down what reached means. Fix the denominator, define the outcomes, break the rate down by attempt and channel, and the report stops being a defense of the program and starts being the thing that improves it.

Sources