The Discharge Disconnect

A while back I was working Process Improvement for a hospital group when I received a request from my boss. I was to compile discharge data for two hospital campuses I was working with. Evidently, Hospital Leadership wanted to increase room turnover to improve patient throughput. Just like restaurants, the more customers, the more money. So they established a goal of getting a patient out within two hours of a provider giving discharge orders. And the first order of business was to find out how long it was taking to discharge patients (which, when you think about it, was backwards…setting a goal before knowing the process time. But more about that later).

Unfortunately, the metric was not so simple to capture. Doctors do not order simple “this patient is now ready to go home” orders. Nope. Often discharge orders are dependent upon some variable. Orders might read “Discharge when patient can keep down food” or “Patient may go after X-ray clears them.” It then becomes a matter of monitoring the patient for hours (or longer) or waiting for radiology or the lab to be able to clear the patient.

The immediate solution to open-ended discharge orders would be not to allow them. But that would require additional time and work from the doctor. They’d have to write an order that the patient would need to keep food down. Then write an additional order after that was confirmed stating “Discharge Patient.” And Providers’ time is more valuable than early discharge.

Even when a discharge order was clear and simple “The Patient is Ready to go” there were still a host of other issues that would cause delays. Family members who were to pick them up are not available. Patients with nobody to drive them needed transportation, and if they needed specialized transportation, such as wheelchair access, such services were scarce. Often special needs transportation needed a day’s notice to be scheduled.

Also, there was the issue of where the patient was going. The majority were simply going home. But many needed equipment or extra care to be arranged for their home; which required time, and did not often align with the discharge time. Others were not going home but to a skilled nursing facility, or hospice, or another healthcare facility. Often there were waitlists for these other places. So there may be a discharge order, but it could take a week for a space to become available.

Lastly, there were “special circumstances.” Some were noble, such as the person who lived on the street having their stay extended until after dinner so they could have one more hot meal. Others were a bit more absurd, like the guy who refused to leave until after he was done watching a football game.

With all the variables involved, the discharge metric was not accurate, nor did truly capture the process…it was not a valid metric. To top it off, if the goal was to improve overall patient throughput, focusing on discharge time was perhaps the least meaningful process step to look at.

The average patient stay was just under four days (3.92 days if I remember correctly). A four-hour discharge amounted to about 4% of total process time. Well over 80 percent of the total process time was a combination of procedures, tests, recovery, and waiting. There were so many variables with those steps, it was daunting to address them. Discharge appeared to be a standard process that could be tackled for a quick win. But that was not the case, the discharge process was not that simple, and the approach taken was wrong.

Setting an arbitrary goal for a process that one does not even have basic data on is not the way to go about it. The process should have been observed, defined, and then addressed. The Hospital Leadership did it backward. Had they done it the proper way, they could have created control charts and realized the wide variation they were dealing with, and moved forward from there. But, as many organizations do, they just wanted immediate answers and actions.

While we collected the data, we were also asked to note causes. Again, not the structured 8-step approach I used, but it was not my project, I was just being asked to provide observations. With no standard approach to Root Cause Analysis the team leading the project jumped to conclusions and began several tests of experiments.

A major cause of delay noted was waiting for transportation (which was often late). To address this issue, they created a “Discharge Waiting Area” in the front lobby. Patients discharged could be moved out of their rooms to the designated spot in the lobby to await their rides. Unfortunately, they didn’t fully think the process through. The waiting area had to be staffed to keep track of who was coming and going, as well as monitoring the patients…they could not be left unattended until they were off hospital grounds. This required multiple staff members to ensure coverage for the day. This staffing also limited the hours of the waiting area…it was only open 9 to 5 during the weekdays. This led to confusion on the floors as to when they could send a patient there…could they go at 8:50 am? At 4:45? When in doubt, the staff kept patients in the rooms. Also, as it should have been expected, rides got delayed past 5pm. This caused major issues…either the staff had to wait with them past 5, or the patient had to be returned to the floor, where their room had been given to another. The experiment didn’t last a month.

About this time, because of different improvement event, I was attending a weekly “Special Cases” meeting held by the Family Liaison staff. This was the team that was to help patients transition to other facilities. Most were patients who were going to require longer-term care, such as skilled nursing facilities (SNF) like nursing homes or assisted living. The cases were a bit startling. Many of them were patients with stays over 30 days. Many of these were eligible for discharge but did not have a place to go.

One patient was in a room for over 100 days. They were to go to a SNF, but had no money. SNF’s have limited spaces for those requiring financial assistance. When a space would open up, the patient, and his family, would object. The quality of care was nowhere near that of the hospital, so they did not want to leave. So they would challenge the move, and when told no, they filed an appeal. The appeal to the hospital would be denied, but, by the time the bureaucratic machine had spun its wheels, the spot would be taken by another, and the patient would remain at the hospital. This happened three times, resulting in a stay of 112 days.

This was the extreme case, but there were many like it. As I learned more about these cases and the team of Family Liaisons that worked with them, I realized they could have a greater impact on patient throughput than anything else. If we could cut these special case stays in half, the entire patient stay metric could drop by 20%. It was easily doable.

I sent a note to the Discharge Team with my notes and recommendation. I also told the leadership of the two campuses I supported. Unfortunately, it turned into a game of passing the buck. The Discharge team said their charter had them focused on one thing. The Leaders I reported to told me the Senior Leadership was driving the throughput initiative. The Senior Leadership did not want to meet me, even with support from the Quality Director.

There were other projects I had… so I went back to them. I dutifully continued to collect the discharge data requested of me. But a year and a half later, as I moved to another position, there had been no downward movement in discharge times, and patient throughput remained about the same.

But at least a story came out of it…

#quality #lean #leansixsigma #operationalexcellence #processimprovement #totalqualitymanagement #storytelling innovation? #lean #leantraining? #leanthinking?

?