I’m working my tail off but not getting any work done (Part 3 of 3)

Positive feedback loops and hacking into the Productivity Formula for good

In my previous two posts I sought to break down the reasons we, as radiologists, feel so busy yet, paradoxically, unproductive.

In the first part, I broked down the more “obvious” causes:

1. The normal ebb and flow of patient volume.
2. Fires.
3. The Dump.
4. Complex cases.
5. Not being “present.”
6. Team not being “present.”
7. Broken systems.

In the second part, I defined the terms Quantum Unit of Work and the Time Tax and I further introduced the productivity formula, a schematic that allows us to approach the process of supercharging our productivity in a systematic fashion.

P = ((60 — ((Ti1 + Ti2 + Tf) * N)) * e ) / (Q)

P = Productivity = cases per hour

Tt = Time Tax (in minutes) = Ti1 + Ti2 + Tf

Ti1 = Time spent dealing with any random interruption that is part of the work of being a radiologist.

Ti2 = Time spent dealing with interruptions that could have been dealt with by someone else.

Tf = Time spent to disengage/regain focus when you are interrupted in the middle of reading a study.

N = Number of studies on your worklist.

e = the elasticity factor for how fast you are reading (0–1). 1.0 is reading at your fastest sustainable pace when you are mentally focused and in a flow state…Of course, e is usually well below 1.0 (at least for those of us who are not robots).

Q = Quantum unit of work (minutes).

I covered the variables P, Ti1, and Ti2 in the previous post. Onward with Tf!

Tf — Time to focus.

The time it takes to close your thoughts and then reopen them (let alone actually opening and closing an exam on the PACS system) is a real and documented phenomenon in the setting of multitasking. All the available research points to the inefficiency of multitasking…not to mention how error-prone it is. However, the reality of our jobs (indeed, the job of all physicians) is that it is constructed on a scaffold of multi-tasking. The most obvious way to decrease Tf would be to have less interruptions in general. Each individual interruption accounts for a fraction of the total Tf. No matter what the interruption is, if you are in the middle of a study, each individual interruption will have its own associated Tf cost. In other words, interruptions occurring between studies, right after you have closed out a study, will not have the same Tf as one occurring while you are in the middle of reading a study. One way to minimize this is to batch the interruptions together. Instead of having 5 interruptions with 5 separate attempts to close and reopen your focus, you could deal with all 5 interruptions at once and have one round of closing and reopening your focus. This results in a huge decrease in Tf.

Here is where you can utilize your knowledge of Q to your advantage. If you know you have to deal with an urgent interruption, it is much easier to have them wait 30 seconds until you finish reading a chest X-ray then having them wait until you finish reading a CT for pulmonary embolism. We all do this instinctively — for example, when we have people wait till we finish dictating a sentence before attending to them. By pushing things to the “in-between” times we are instinctively trying to minimize Tf. Therefore, if you are expecting interruptions, it makes sense to transition to reading X-rays or other exams with a small Q so that you can push expected interruptions more comfortably to the end of a Q cycle, minimizing the cost on your focus and minimizing the wait for the interrupting party.

Finally, you can try to figure out a way to deal with your interruptions completely outside the stream of reading…for example, when we protocol exams all at once at the end of the day.

Some other examples I have seen of attempts to minimize Tf costs:

- Dedicated office hours in which clinicians can come and consult with a specific radiologist who will probably be reading plain films or similar small Q studies.

- Radiologist of the Day — a single rotating radiologist in a group who is dedicated to consultations, answering questions, doing procedures, etc.

- Having a PACS system that allows the radiologist to have more than one exam open at a time. (I’ve encountered more than one PACS system that only allows the radiologist to have only one exam open at a time. This results in the painful process of having to close an exam you are in the middle of when a clinician arrives with a question about a different case, and then having to search for, reopen, and rehang that case before reengaging with it.)

e — Elasticity factor

This is a function of the mental focus of the radiologist and represents an appropriate level of stress/anxiety/desire to keep reading at a fast pace through an entire shift. It is decidedly unrealistic to expect anyone to function at 100% capacity all day, day in and day out, but there are certainly things that can be done to improve mental efficiency without breaking out the whip. I encourage anyone looking at this to take a holistic view. Radiologists are not robots and burnout is real. The pressure to maximize e can arise both internally and externally, and like all stress, is only productive to a point and is different for different individuals. Here are some factors that can affect e in no particular order and without judgement.

- Adequate sleep

- Adequate exercise

- Adequate time away from the reading room

- Copious amounts of caffeine

- Minimizing distractions, such as mobile apps and messaging in this social media steeped world

- Limiting access to distracting websites through hospital VPN

- A healthy level of competition between members of the group who are working together.

- Incentives for radiologists who best maximize productivity.

- Fulfillment. Work is more than just a paycheck. We all have deep needs that must be met for us to feel truly motivated. If you really want the most out of yourself or your coworkers, the work we do should be be structured to address our deeper needs in a healthy way. This necessarily involves individual insight, an understanding that each radiologist within a group is an individual and making space for that, and understanding the concept of “moral injury” in the workplace.

Q — Quantum unit of work

As radiologists, we instinctively try to improve Q by “reading as fast as we can”. The more you decrease Q, the less of a “block effect” you will have. If your Q for an exam was 60 minutes, even a single minute of interruptions would result in your not reading a single case for that entire first hour — and there are many potential interruptions that could happen in that hour, which would exact further costs on your focus, etc. Exams with a large Q make you a proverbial “sitting duck”. Exams with a Q of 1 minute would more easily accommodate multiple interruptions without grinding your productivity to a halt.

There are more ways to decrease the quantum unit of work than we can go into here but I’ll throw out a few interesting examples. The first is increasing the efficiency of the workflow. An outpatient imaging center I’ve worked at had an extremely primitive PACS system (admittedly back in 2014). I disparagingly referred to it as the Fisher-Price PACS. A chest X-ray would take 10 seconds from clicking on the exam to displaying the images and being able to dictate it. So if you have read 60 X-rays, you would have spent at least 10 minutes just waiting for the exams and dictation windows to open. Even more frustrating is that there is also a significant delay when closing the exam!

Any time you can shave off, from case opening to case closing, by optimizing your PACS and RIS system, is an obvious way to decrease Q and to decrease frustration.

All of those micro-moments when you are not actively interpreting or reporting your studies are opportunities to decrease your Q. Like literally — if you are not actively looking at anatomy, clinical history, or generating the report, then that is something that can be tweaked. This includes the time spent on loading images and studies, manually hanging and rehanging images, having to repeatedly adjust window/level, scrolling through a worklist, choosing appropriate dictation templates, choosing appropriate priors, trying to find the relevant information in a verbose history, etc.

Q can also be decreased by optimizing the voice recognition workflow and the templates used. The goal of decreasing Q must of course be balanced against providing an appropriate level of quality in your diagnosis and reporting. This isn’t about popping out a report of dubious quality. Spending the appropriate amount of time to come up with a proper impression, complete with a specific differential and helpful management recommendations is part and parcel of the service that we provide. The smallest possible Q is not the end goal…rather you want the Q that accommodates all of your best practices — and nothing more. Fortunately, this is becoming easier than ever with the current leaps in healthcare technology. No longer do you need to pull a heavy textbook off the shelf and thumb through an index to find answers. There are innumerable tools that can help us improve or maintain quality while minimizing the impact on our time. A few that come to mind are StatDx, Radiopaedia, Radiologyassistant.nl, and of course, RadsBest. Having a team member who can sift through all of these offerings and find the one that best fits the needs of your practice might be a worthwhile endeavor. Moreover, dedicating a team member to work “on” your practice and improving it’s systems, rather than only “in” it may pay dividends to the whole group.

N

If you understood my points about the time tax, you will understand that the number of studies on the worklist is a powerful factor in determining our ability to be productive.

Crucially, the closer N is to zero, the less you will be affected by Tt, since they are multipliers. In an ideal setting, your starting worklist would be short…theoretically N = 0 would be ideal since you would mostly only deal with studies that you are actively reading or have just read. It would synchronize the “non-productive” work to the work you are actually elbow deep in. Of course, you don’t want your worklist to be zero all the time…if your worklist is not replenishing and you don’t have enough cases to read, that would result in a different kind of decreased productivity…i.e. starvation! In an ideal world, studies would come in at the exact same pace at which you are reading the exams…no faster and no slower….like the finely tuned intake and exhaust of a performance car. Any extra studies sitting on the list increase the probability of having an associated interruption.

Of course, this is all a simplification. Having an empty worklist does not mean zero interruptions. As I previously stated, the Time Tax has a nose and a tail; I more accurately account for this component in the appendix to this post.

Importantly, the interplay between the size of the worklist and the Time Tax goes beyond that. There is actually a positive feedback loop that can dramatically contribute to our stress levels.

Indulge me for a minute, this is major. As the number of studies on the worklist (N) increases, with all else being equal, the productivity (P) will begin to drop, based on the aforementioned formula. This is one of the fundamental points that I’m trying to make — but there’s more. For simplicity’s sake, I have been assuming a constant rate of imaging studies being fed to the worklist, the inflow. In that case, the number of studies on the worklist (N) is really just the inflow minus the rate at which studies are being read — the productivity (P).

N = Inflow — P

Therefore, as P drops (assuming a stable inflow), N will increase. As N increases, the effect of the Time Tax is multiplied (from our original formula), which further contributes to P dropping!…a positive feedback loop.

It should now be abundantly clear why busy times result in an almost exponential drop in productivity, and how I believe that a large worklist is a disproportionate contributor to that busy-ness. I am sure that we have all walked into a busy hospital with a large worklist and felt the anxiety that comes with it. That anxiety is the visceral manifestation of this phenomenon. It’s the reason we work so hard to chop the worklist down to a manageable level before we take that overdue bathroom break.

In view of all of this, decreasing the number of studies on the worklist should be a top priority. Of course it must be balanced against real life fluctuations in work volume. But assuming you have enough cases, there are many ways to optimize N. I know practices that spend a lot of creative energy on this issue and many of the solutions are things that we see being done in our own practices. Some example include having the overnight and duskhawk radiologists clear the worklists, spreading the work out over multiple radiologists during times in the day when the list is large, having an early morning radiologist clear the list before normal hospital hours, or having cross-coverage across multiple networked sites so the list is never unattended and always being read down. As an individual, I used to do this by reading down all of the fastest (lowest Q) exams first, i.e. chest xrays, screener head CTs, stone protocol CTs. It would be insane to spend 20 minutes on a complex ENT case when there are 100 studies on the list. The goal was always to decrease the number of buzzards circling overhead so you could make time to dig into the tougher pieces of meat. Finally, when all else fails, there is always the last ditch effort of surreptitiously cutting the cord on the CT scanner.

In summary, I’ve spent a lot of time explaining the ideas of the Quantum Unit of Work and the Time Tax and how they interact with other variables in the formula for productivity that I presented earlier. Here it is again:

P = [(60 — ((Ti1 + Ti2 + Tf) * N)] * e ) / (Q)

Admittedly this is a ranging and multifaceted topic. In addition to defining the components of the productivity formula and going over some cursory examples of how each variable can be tweaked, I really wanted to emphasize the point that the existence of a time tax for every study on our worklist and that the size of our worklist together create a positive feedback loop that is further exacerbated by the quantized nature of our work output.

I hope that this helps to schematize the craziness that we can feel one day and the relative chillness we feel the next…and why, when things are crazy, we have a hard time getting our “real work” done. It really shines a light on the outsize impact of the size of our active worklists. By breaking out, categorizing, and understanding the relationship between the components of productivity, we are afforded a systematic way to approach productivity hacking in the workplace. At its most basic level, this could be utilized to directly increase output; or alternatively, on a more holistic level, these variables could be tweaked to help smooth out stress levels, increase workplace satisfaction, create space for improving quality of reporting, and create more time for radiologists to add indirect value to the practice, such as by consulting directly with referrers or taking on active roles in hospital leadership/decision-making.

I welcome further discussion on these ideas — where I’m right, where I’m wrong, and where the ideas are muddled. See the appendix to this post below for a more detailed and nuanced productivity formula.

*** APPENDIX

Here I’ll go into a more nuanced version of the Productivity Formula that I presented earlier. It basically unpacks the Time Tax, pulling out the nose and tail.

We are assuming constant inflow volume

P = [ { 60 — ((Tt nose * V) + (Tt tail * RR) + (Tt active * N)) } * e ] / Q

Tt nose = Time tax exacted in the now by exams that have not been performed yet.

Tt active = Time tax exacted in the now by active exams on worklist that are awaiting a read or being read. This period ends when the exam is closed and report crosses over.

Tt tail = Time tax exacted in the now by studies that have been reported (i.e. reviewing exams for tumor board). All exams have a “curing” period over which the likelihood that that exam will require further attention from the radiologist will fade over time. You are more likely to spend time talking to a physician about an exam performed last month than an exam performed 3 years ago. Tt tail is just the sum of that time averaged out on a per exam basis.

Helpful to think of each of the three categories of time tax as consisting of the three separate components mentioned in first formula. Therefore:

Tt nose = Ti1 nose + Ti2 nose + Tf nose

Tt active = Ti1 active + Ti2 active + Tf active

Tt tail = Ti1 tail + Ti2 tail + Tf tail

V = total active imaging studies ordered (all the studies on deck)

I = Inflow = V / Scan rate

N = I — P

RR = recently read volume. This is just a measure of how many exams have been read in the recent past. It is just a reflection of the fact that a radiologist who is in a busy practice and has read twice as many cases in the past week is twice as likely to have to reattend to some of those cases than a radiologist who read half as much.

e = has a dynamic relationship to N. So as the stress levels go up, our e will adjust. At most simple level, as the N goes up, stress goes up, and we adapt by reading faster…but the reality is that sometimes it can be the opposite, where we maladaptively read less productively (the throwing up of hands). The other thing that sometimes happens with a busy worklist is that our per exam read time can drop below Q, which is not safe.