Every time I log onto social media, I am reminded that people have not really matured in their consumer skepticism. You can laugh about how silly it is that anyone fell for “a shake for breakfast, a shake for lunch and a sensible dinner” as a diet solution or lacing candy with an appetite suppressant.1 But we still fall for this, as 80% of the ads on social media can attest. Today these ads often use AI-generated video to convince you that one tablet will create a foam volcano and return all your pipes to as-new condition, or that this simple comb will generate bags-full of dog hair that was, apparently, never visible on the dog.2 This is not an essay on deceptive advertising practices. It is simply calling out that we are not as savvy as we might think we are.
In PX, there are salesmen who will say that there is an “easy button” to fixing patient perceptions, if you just use their technological gewgaw or mnemonic phrase. These claims are harmful not just because they are selling a pig-in-a-poke3 but because they imply that improvement in the PX space is quick and dramatic. Not only is PX work hard, tedious, and often detail-oriented, it rarely responds immediately like a rocket to the moon with the introduction of “one simple trick that the big corporations don’t want you to know about.” PX requires consistent execution that will, in the long-run, drive scores in the right direction. That, though, is not helped by imagining that these scores will continually climb with unflagging incremental improvement.
Like last time, I need to address some out-of-scope questions or concerns.
- When looking at the trendline of PX scores, one needs to remember that the data is the product of a sample and not the entire population of discharged patients. This essay will not discuss the reality of measurement error, sample variation, or possible participation bias. I personally feel that I am overdue for an essay on math, but that will not be today.
- Associated with the first point, I will not be addressing confounding variables, either. There can be staffing changes, leadership changes, environmental changes, population changes, expectation changes, or external changes that can affect scores independently of the PX work done. Anyone not sure what this means can examine the impact of COVID on patient experience scores for an array of examples that illustrate this.
- This is NOT an essay on how PX teams should not be held accountable. All employees should be asked to explain their positive contribution to the organization. Scores that dip or fail to improve should be discussed and if the change-agents are deemed an obstacle, they should have to answer for this. I will address accountability for improvement in this essay series. Just not right now.
Improvement is not incremental
All improvements are incremental. An organization is not likely to go from the 1st percentile to the 91st percentile in a day. When I use the word “incremental” here, I mean that the relationship between action and outcome is less hard-nosed mathematical certainty and more a broad identifiable pattern.
For example, I have worked with a number of hospitals that used their surveys to capture patient-recollection on specific behaviors. A system in North Carolina, for example, included two Yes/No questions asking patients if they remembered (a) a leader rounding on them during their stay, and (b) a post-discharge phone call. The group of patients that recalled ONE of those events had Overall Rating top-box scores that were about 10%-15% higher than those who recalled neither event. The patient group that recalled BOTH events had scores were about 20%-25% higher than the base group.4 So, clearly executing on these behaviors in a memorable way led to higher scores.
The problem was that the senior leaders wanted to push the data further and create an arithmetic equation, where an X% improvement in leader rounding would generate a Y% improvement in overall scores. They also wanted to know why Hospital B did not show this trend as their scores were actually slightly worse if the patient recalled both. Both requests are reasonable. Considering the first, as a math nerd, I am always fascinated with logistic regression outputs and the incremental increase in odds. The problem is that these are still an approximation. They are real numbers, but they also carry a level of error as well as a host of unexplained variance.5 Considering the second, it is valuable to figure out why the pattern does not hold at certain locations. Early on, the speculation was that the leaders at Hospital B only rounded on patients who were identified as having problems or concerns. The leader-round was, then, mostly an attempt to bail water rather than improve overall care. So, perhaps, the scores being flat was evidence of success, because they were not lower. (Our old friend confounding variables.) It turned out that this was not likely the reason, but that is a much longer story for another time.
While these are both reasonable follow-up questions, the problem with my North Carolina client was that these questions fixated attention. This focus on the granular and the exceptions was getting in the way of the conversation on how to generate better performance on the behaviors and better recall in the patient. They ended up wanting to put weight and rigor on the pattern and, by extension, on the individual patient’s perceptions that neither could easily support. In an effort to buttress against hypothetical questions from hypothetical staff, they created enough smoke to confuse the message and limit the acceptance of the work by creating actual questions from actual staff.
I see this all the time in hospitals. The care-experience is complex, so often organizations focus on one piece at a time. They might say, “Oh the doctors are such delicate geniuses who wilt under any criticism, so we will ignore them and focus on the nursing staff.” Or, maybe, “The negotiations for the new nursing union contract are coming up, so let us not give them any ammunition by requiring them to add new tasks to their list. Instead, we will focus on our non-union staff.” While it is true that one needs to tailor messaging to physicians, clinical-nursing, clinical-non-nursing, and non-clinical audiences, breaking these things into distinct independent pieces misses the real point of patient experience. Namely, that all of these interactions are collectively important.
Imagine trying to institute leader rounding at a hospital and getting met with, “We can try leader rounding, but if we cannot deliver on that, can we instead do two nurse-manager rounds, an extra hourly round, a food-service round, and maybe a visit from the pharmacist?” Trying to compartmentalize each of these events fails to understand the gestalt. This is what I mean when I say that improvement is not incremental. We know that consistent execution on key behaviors will likely improve scores, but expecting hard numbers both fails to understand the ecological fallacy and distracts staff from the essential purpose of an hourly-round, a post-discharge phone call, or a leader-round. It fails to see patient experience “forest” for all of these discrete behavior “trees.”
Improvement is not linear
Associated with the fact that improvement is not incremental is the fact that improvement is not linear. Again, it is a reasonable expectation that effort appropriately applied will yield results. What is not reasonable is to expect those results to be a stairstep rise. Any improvement activity comes with fits and starts. Anyone beginning any new hobby or sport understands this. You may not demonstrate any improvement early-on, as you struggle to master the essential mechanics and start the learning curve. From here, you may demonstrate great improvement over a short period of practice. Once you have harvested the gains from the initial improvement, you might plateau. Constant work on mechanics and mindset yield modest improvements, as your form and muscle-memory improve. But in order to get another windfall of improvement, you likely need to add, tweak, or change some other part of your performance. So you will get better, but it will not be a linear improvement.
Further, as you change, you might initially get worse before you get better. I bowled on a company team in a league. I enjoyed it, but I was not, um, what’s the phrase?, oh, yeah, GOOD at it. The team was focused on having fun rather than crushing our opponents, but my inconsistent performance limited the “fun” I was having. A teammate could see I was frustrated and suggested that I move from a 3-step approach to a 4-step approach. I did as he suggested and immediately got worse. I got worse because I was trying to fit this new thing into my approach, which meant also having to adjust my swing and my release point and my foot placement. But, as I worked with this new approach and its impact on my form, I did get better. In fact, I significantly improved my average and my consistency. And then I plateaued. That one change was not going to turn me from sad-sack to professional bowler. It did, though, prove to me that bowling is not random, and improved form can produce better results. It also inspired an interest both in the game and in getting better. So, after a year or so, my teammate could give me a second thing to work on.
Any change in behaviors designed to improve PX has the same learning curve. Even something as straightforward as sitting down or introducing oneself needs to be incorporated into the full picture. In the short-run this can seem clunky or distracting from the rest of the work and cause dips in scores, even if we know that it will pay dividends in the long-run. Thinking that changing one PX process today will start producing positive results tomorrow is unreasonable. It is also unreasonable that one change will continue to make results incrementally better. PX work, like finance, quality, or maintaining a loving and supportive relationship with a significant other will improve with effort and plateau without additional effort.
Wondering why one nursing unit improves faster or with a higher ceiling is like wondering why one rookie quarterback improves faster than another. Context, coaching, innate ability, personal confidence, and individual effort all play a part. This is not meant to excuse differences in performance. It is meant, though, to encourage people to NOT see PX processes like hourly rounding or multidisciplinary rounding or post-discharge phone calls as producing identical results across all spaces. This is why we want to celebrate high-performers and help mentor middle-performers. These tactics are not coins used to buy gumballs. They are investments in long-term performance. By demanding an ROI from every line-item fails to understand the nature of improvement.
1Yes, you read that correctly: weight-loss candy. Those out there who are drinking vodka-and-RedBull are encouraged to not judge other people’s oxymorons.
2If you have experienced miraculous outcomes using one of these products, well, clearly, I am talking about their competitors.
3Geez. A paragraph that references Ayds, “gewgaw,” and a pig-in-a-poke? I am verging very close to “you kids, get off my lawn” territory. My apologies. For those who are under the age of 100, a “gewgaw” is a bright shiny object that has no real use or value. A pig-in-a-poke is something you buy, sight-unseen without validation or verification, meaning that you don’t know if it really has value or quality. Ayds were a weight-loss candy with what turned out to be an INCREDIBLY UNFORTUNATE name.
4My fellow number-nerds likely have a bunch of questions, comments and follow-ups to this data. Feel free to ask clarifying questions in the comments, if you want. I just didn’t want to go full math-modeling here.
5I know, I know. I just cannot help myself with math-speak. I will stop. For now…
PX Matters with Joseph Snipp 
Leave a comment