Speed Data

Patient journeys consecutively and rapidly

One of the questions that keeps me awake at night is this: "When you've got a problem, and you don't know what the solution to that problem is—but you do know that information will help in the understanding and solving of the problem—how do you decide what information to look at?"

Mario Testino and Anna Wintour looking at photographs in London's Ritz Hotel. This is a scene from The September Issue, whose relevance to this article will only become clear if you read on.

When Nick Tordoff and I teach our Data Conversations course, we get participants to discuss our hypothesis that there are several distinct decision-making scenarios that data analysts commonly find themselves in. What usually emerges is that NHS analysts spend most of their time in "Monitor" scenarios, where managers (apparently) know what the cause-and-effect dynamics are in a system, and we just monitor the key performance indicators of what we hope is a stable, steady state.

But "Monitor" scenarios are at one end (the comfortable end) of the decision scenario "spectrum"; at the other end we find another scenario. This—the fuzziest, the muddiest, the untidiest—is what Nick and I call the "Explore" scenario). "Explore" scenarios are those in which decision-makers don't have all the answers, and they need more information in order to understand things better so that they might eventually arrive at answers.

I'm currently operationalising a method called Anaximaps which is specifically about trying to identify the data that'll help us get out of the "Explore" phase. It's a method whereby managers and professionals draw maps (or diagrams) of the systems they inhabit so that analysts can get a clearer sense of what managers' and clinicians' mental models of those systems look like. Once we are equipped with that clear sense, so my argument runs, we'll be in a better position to analyse and present data that's genuinely meaningful.

I've been through this mapping process with about 100 individuals so far. And what nearly all of the maps have in common is that the patient or service user is drawn in the middle of the page. The maps are all therefore self-consciously person-centric. So if we are using them as a guide to what data to analyse, and how to analyse it, then these maps usually point clearly in one direction: start with the patients (as opposed to—for example—departments) and look at individual patient journeys as a way of making sense of complex systems.

Indeed, when I look at maps drawn like this my own thinking starts to become person-centric, too. I find myself wanting to see each map somehow projected as an exhibit on a screen while someone is simultaneously recounting an individual case study. And as they refer to each staging post of the patient journey, the relevant box or circle or arrow on the map lights up, so that people can see how the factual case study (often told as a chronologically-correct narrative with calendar dates and times) fits together with the map. An architect acquaintance of mine has told me that what I am visualizing here is a juxtaposition of a spatial map and a temporal map.

And it occurs to me that this technique is quite possibly the best way of progressing if we want to try and make sense of the system as whole (in other words if we are trying to get out of the "Explore" decision-making scenario), as well as just trying to make sense of one patient's journey.

This idea of person-centric data versus aggregated and summarised data is one that's familiar to information analysts. We are used to making a distinction between—on the one hand—patient-based data and—on the other—summarised data. We often think of it as a distinction between operational information (information about an individual patient that is used at the coalface to help clinicians make decisions about what happens next) and management information (information that is used mainly retrospectively to look back at what happened in the past to see if it can help with decisions about—for example—resource allocation in the future.

But management information—the sort of information that's made up of summaries—can only be useful once the thought process has already been gone through to decide which summaries, which aggregations, which counts matter the most, or which are most useful to help us make sense of a complicated reality. If we don't yet know how to make sense of that reality, then our management information is going to be nothing more than a shot in the dark.

We need to do something first so that we can make our shots in the dark a bit more accurate.

We have to describe individual patient experiences, pathways and journeys. But we have to do so in the context of it being management information as opposed to operational information.

But how? How are we to analyse and present data so that it retains its "individual person-ness" and yet also somehow allows us to get a sense of any patterns there might be? How can we analyse and present this person-centric data so that it helps managers and clinicians generate hypotheses about how systems work?

To which one answer is: repetition. In fact, not just any old repetition but rapid repetition.

What if you presented data on individual patient journeys to a meeting, but you did it repeatedly? One after another. And rapidly. Here's the first one, here's the second one, here's the third one. Maybe allowing people a couple of minutes per patient before moving on to the next one. And what if the data about each journey has been designed so that it is quick and easy to digest, in a way that enables people to see any patterns there might be in the sequence?

One of my favourite scenes in The September Issue is the one where Anna Wintour and Mario Testino meet up in a red room in London's Ritz Hotel to discuss Testino's thoughts about the Sienna Miller cover shoot Wintour wants him to do in Rome. Amidst all of Testino's melodramatic hyperbole about how he wants there to be lots of statues and white horses, what we witness is these two powerhouses of the fashion industry flicking rapidly and consecutively through a stack of images the photographer has brought with him. Dozens and dozens of images get glanced at, each one receiving maybe a millisecond of attention before the next one is uncovered. And yet the trained eyes of these two fashion moguls still manage to spot what they want.

My imagination was captured by this scene because of the velocity with which two experts can rattle through a series of exhibits or stories. It got me wondering: could we do something similar in a healthcare setting? I'm always looking for opportunities to join together the clinician world of individual patient stories with the analysts' world of counted, summarised and averaged data. And this scene in the Ritz seemed to offer a way into it. A high velocity sequence of individual patient stories-if we get through the sequence at sufficient speed, maybe clinicians would see any patterns that might be revealed by the consecutiveness, the "one-thing-after-another-ness" of it.

A year ago, the Chief Nurse of Edinburgh's Community Health Partnership gave me the opportunity to put this technique to the test. We took a sample of 30 patients who crossed the boundary from healthcare to social care during a two-month period. We got together with a healthcare analyst from NHS Lothian and a social care data analyst from the City of Edinburgh Council and we compiled as much information as we could about the primary care, social care community nursing and acute hospital interventions that had taken place prior to and during the delayed discharge episode, then we took these assembled case studies to a multi-disciplinary group of staff (occupational therapists, social workers, district nurses, geriatricians, GPs) to see if this technique had legs.

I'll leave the story of what happened next for another day…

[20 January 2017]