Tuesday, August 23, 2016

Who'll stop the rain (Musings on SHOT 2015's use of narrative fallacy)

Updated: 24 Aug. 2016 (tweaks +added video clip)
As last year, the August 2016 blog is stimulated by the UK's Annual SHOT Report, often featured in past blogs. SHOT has long been the best hemovigilance program anywhere and is a treasure trove of educational goodies.

This year's blog is not going to focus on the key results and recommendations of the SHOT 2015 Report but instead give a selective overview. Readers are directed to the Report for all its informative details (Further Reading). Instead the blog will feature only two pages of the 190 pp. report concerning a concept that fascinates me. The case discussed is techie-oriented but the broader lessons apply to all transfusion professionals.

EXECUTIVE SUMMARY: For eager readers who want to know the bottom line before reading further: The blog examines a thesis in the 2015 SHOT report that the narratives we construct after-the-fact to explain why events happen are often wrong and, moreover, prevent us from learning from our errors. Let's face it. SHOT's annual reports document the same and similar human errors year after year. My musings attempt to explain further some of the reasons we create narrative fallacies.

The blog's title derives from a 1970 song by John Fogerty of Creedence Clearwater Revival, rated #188 on Rolling Stone's '500 Greatest Songs of All time.'

Tidbits related to metrics:
First, for perspective, in 2014 there were 2,663,488 blood components issued in the UK (74% RBC). SHOT received reports of 3668 cases or 13.8 reports per 10,000 blood components.

By comparison, in 2015, the number of blood components issued in the UK decreased by 3.2% to 2,577,276 (73.5% RBC). SHOT received reports of 3965 cases or 15.4 reports per 10,000 blood components, an 11.6% increase. Of course, besides more events occurring, the increase could also be due to more complete reporting.

The total number of reports analysed and included in SHOT's 2015 Report is 3288 compared to 3017 reports in the 2014 SHOT Report (9% increase). 

Risk of potentially infected donation entering the blood supply (2012-14)
Hepatitis B 1 in 1.6 million
Hepatitis C 1 in 26 million
Human immunodeficiency virus  1 in 6 million

Risk of death or serious harm from transfusion per components issued (imputability 1-3*) 2015
Death 1 in 100,000
Death from error  1 in 320,000
Major morbidity  1 in 15,500

* Imputability: 1=Definitely related;2=Probably related;3=Possibly related

ERRORS (2015) 
Similar to 2014, ~78% of 3288 analysed cases were caused by error.
Human error accounted for 96.7% of serious adverse events reported to the MHPRA.

Most SHOT reports involve often multiple mistakes related to human factorsThe number of adverse reactions and events related to poor communication and poor clinical decisions are a concern.

Laboratory errors have increased and many laboratories are under pressure with vacancies (some longstanding) and increased workloads.

More than a third of NHS staff reported work-related stress in the 2015 staff survey. Emergency departments struggle: 2 in 5 new consultant physician posts were not filled. Overall funding is tight. 

Patient transfusion-related death is the worst adverse event. In 2015 there were 2 deaths definitely related to transfusion, 9 probably related, and 15 possibly related.

A good figure to show deaths (n=26) correlated to type of serious adverse event is Figure 3.2 on p. 10 of the SHOT Report:

Transfusion-related deaths reported in 2015 by imputability (Copyright SHOT) [Click to enlarge]

As noted, this blog focuses on but two pages of SHOT 2015 because they interest, indeed fascinate, me. The pages appear under 'Error Reports: Human factors' (pp. 24-5) and discuss the role of narrative fallacy in explaining why people do not learn from mistakes. In particular SHOT provides a case (Case 6) that outlines how 3 narrative fallacies occurred involving only one patient's case.

Honestly, who else but the British would discuss a psychological phenomenon like narrative fallacy in a major transfusion medicine report on hemovigilance? With narrative fallacy we concoct erroneous stories to explain facts that we encounter. 

I'm pretty sure in Canada, if we had a hemovigilance system that published its finding for the public, including the transfusion community (we don't), it would deal with numbers almost exclusively, and, beyond that, propose how transfusion professionals could leverage errors into improved transfusion safety and innovate to span the translational continuum from vein-to-vein.

Narrative fallacy can be discussed on many levels (Further Reading). Political examples:

UK: Some Labour Party MPs on the losing 'remain' side in the 2016 brexit vote appear to have rationalized the 'leave' result as significantly due to the lukewarm support and effort of their leader, Jeremy Corbyn. Labour passed a no-confidence vote triggering an upcoming leadership election. It's a narrative fallacy that gives a scapegoat, always satisfying as it stops the need for more in-depth analysis and perhaps unpleasant revelations about Labour Party MPs and officials misjudging the mood of its base.

USA: Some foreign watchers of the 2016 Presidential election patronizingly see Trump supporters as mainly poorly educated bigots, a narrative fallacy that has the advantage of fitting into existing misconceptions some have about the USA. Which isn't to say that some supporters are bigots and that Trump seems happy to feed their needs. 

But Trump supporters, like all Americans, are much more diverse than that, as shown by this Pew Research Center graphic (July7, 2016):
7-7-2016 2-30-10 PM
People are not going to unravel Trump's success without fully examining the spectrum of why people vote for him. Settling on a simple narrative (fallacy) won't do it.

The same applies to the 'whys' of anything people perceive as an 'adverse event'. We need to understand before we can learn from mistakes made and identify contributing factors. Of course, in the political examples above, some may see the events as good, not bad. 

In contrast, the transfusion medicine examples from SHOT 2015 below are clearly adverse events. But I hope you will see that the narrative fallacies are analogous in that the stories used to explain them stop further investigation and work against learning from errors made.

SHOT 2015 introduces its use of narrative fallacy as follows:
'Humans have a tendency to construct stories around facts, which serves a purpose in making sense of the world that might otherwise be seen as too complicated. The natural instinct is to make patterns in order that the world is seen as a simple place, so a narrative is often constructed to explain the facts. Humans are hard-wired to try and turn chaos into order, so they can feel in control of their world. However, this can be termed ‘narrative fallacy’ (Taleb, 2007) because these rationalisations come after the effect and are not based on empirical data. 
...Narrative fallacy means that against all logic, individuals often do not learn from adverse events. Instead of seeing the error as a learning opportunity, the event is rationalised in a more comforting way....'
'Error Reports: Human factors' (pp. 24-5) 
Case 6: Three narrative fallacies add to confusion when grouping a patient after an allogeneic haemopoietic stem cell transplant (HSCT). 

CASE SUMMARY: In brief, the patient was a known group O and, unknown to the laboratory, had received a group A HSCT. A series of ABO grouping anomalies were incorrectly explained by 3 separate narrative fallacies outlined below. 

Besides the role of narrative fallacy, the case serves to stress the importance of communication between institutions, and also between colleagues, patients, manufacturers, everyone on the health care team. I love this clip from Cool Hand Luke:

Non-laboratorians may also glean how complex pretransfusion testing can be and the need for well educated, experienced medical laboratory technologists/biomedical scientists. 

Readers not well versed in the technical details of blood grouping or protocols for transfusing patients with HSCT may wish to scan the nitty-gritty of fallacies and focus on [my musings in brackets] after each fallacy. 

It's not the technical details that are key, it's the way our minds work, the stories we tell ourselves, and how often we misinterpret the evidence, typically in ways that are convenient to our needs.

Narrative fallacy 1
A known group O patient's pretransfusion blood sample gave mixed field (MF) with anti-A several times on an analyser suggesting the presence of transfused group A donor rbc. 

A second analyser gave the same MF result. Because fibrin was noticed on the top of the reaction well, fibrin was removed from the sample and it was re-centrifuged, leading to a negative result with anti-A. 

Staff concluded (narrative fallacy 1) that fibrin was responsible for the initial MF because the new result now agreed with the patient’s historical group.

The patient was correctly transfused group O rbc but also group O platelets, incorrect for a group O patient receiving a group A HSCT, which the lab staff were unaware of. 

[Because fibrin is a cause of blood testing anomalies in laboratory tests, the second result which agreed with the patient's historical blood group, seemed logical. Perhaps as importantly, lab workers always prefer inconvenient, unexpected test results to disappear as it makes our life simpler. This trait surely applies to MDs and RNs too.]

Narrative fallacy 2
When it was later known that the patient was post-transplant, the analyser manufacturer was asked to explain the discrepancy of a MF group A in instrument 1, but an eventual 'normal' group O using instrument 2.

The manufacturer wrongly proposed that repeat centrifugation of the sample might have concentrated donor rbc lower in the tube, seemingly logical, because transfused rbc would usually be older and heavier than patient rbcs and be at the bottom of blood samples. 

That could cause the O grouping result if instrument 2's sampling tip was testing rbc lower in the patient's blood sample than instrument 1's, a common explanation for missing post-transfusion MF groups on analysers.

But the manufacturer's narrative does not fit the facts in any way. When disparate grouping results occurred the ‘donor’ rbc would have been from the engrafting group A HSCT and not transfused group O donor rbc, because the group O cells were the patient’s.

[My guess, a guess shaped by biases, is that manufacturers often get asked about their instruments missing MF in blood grouping tests. In this case the manufacturer may have been on auto-pilot because they gave a standard answer even knowing the patient had a stem cell transplant. 

In a way it reminds me of how physicians (indeed all health professionals) often decide and diagnose based on intuition and pattern recognition ('I've seen this before'), not on all the presenting evidence and possibilities before them, certainly not on analysis of recent research. Narrative fallacies save time for busy practitioners but occasionally may have tragic results for patients.]

Narrative fallacy 3
Three days after the first blood groupings a fresh sample was received, but laboratory staff were still unaware of the patient’s HSCT. MF again occurred with anti-A, but this time the person doing the grouping explained (narrative fallacy) that the MF result was due to group O donor rbc transfused over the weekend. 

The test result was modified to a 3+ positive, giving a group A result. However, authorisation failed, because the patient was historically group O.

A repeat sample also grouped as A with MF. Lab staff eventually discovered that the patient had received an ABO-incompatible HSCT at another Trust, which had not been communicated to them. The transplant was reason for the MF: as the transplant engrafted, transplanted donor-origin group A rbc were present along with the patient’s own group O rbc. 

This narrative fallacy could have led to the patient being mis-grouped as an A patient who was transfused with O donor rbc, instead of being a post-transplant group O patient in the process of becoming group A.

[To me, in the absence of more case details, the lab technologist concluding the patient was group A, 'modifying' (falsifying?) a blood group test result to fit the narrative that transfused group O cells caused the MF, is mind boggling. I suspect that significant contradicting evidence would need to be ignored to create this narrative fallacy. More information is needed. If you are a UK TM colleague and know more details, I'd love to hear them. 

Again, a guess about how this narrative fallacy could happen relates to something else that SHOT 2015 discusses: Laboratory errors have increased and many laboratories are under pressure with vacancies and increased workloads.]

Techie stuff - if you are so inclined: 
Unusual that a MF test result would be 'modified' to 3+ positive, as MF is nothing like a strong 3+. Wonder what 'modified' means because it suggests something totally unacceptable - recording a false result to fit a preferred narrative. Just guessing but was that because the lab's SOPs or information system (LIS) require anything less than a 3+ result in a blood group test to be investigated further? 
Regardless, in this case, the historical record (probably in the LIS) produced a failed authorisation. In general, mistyping a patient's ABO blood group, especially a group O patient, can have catastrophic consequences if they are then transfused after an electronic (non-serological) crossmatch with ABO-incompatible rbc. Of course, an ABO discrepancy, including MF in a patient's ABO group, would prevent use of the electronic crossmatch.  
Not mentioned in the SHOT report is how the patient's reverse group tested. Group O recipients would have anti-A in their serum/plasma for months, making concluding the patient is group A difficult. The patient's anti-A would be continually 'mopped up' by adsorbing to donor-produced A antigen-bearing cells, possibly creating delayed hemolysis. But immediately post-transplant the anti-A should be relatively strong, precluding transfusion with group A rbc until engraftment occurs and a serological crossmatch is compatible. 
Think back for a moment to the HIV/AIDS 'tainted blood' scandal of the 1980s and 1990s and the narrative fallacies it spawned.
At first the narrative fallacy was the disease was gay-related immune deficiency (or GRID), comforting to those who were not gay because, 'Hey, it can't affect us.' 
After it was reported in hemophiliacs and Haitians, a narrative fallacy proposed that the disease originated in Haiti. 
The Canadian Red Cross BTS and its medical experts believed that probably all hemophiliacs were already infected by FVIII concentrates made from tens of 1000s of blood donors and conveniently opted to use up its large stockpile of untreated (unsafe) FVIII concentrates.  
With these and other narrative fallacies, the transfusion medicine community took an exceedingly long time to accept the disease could be transmitted via blood transfusion.
I encourage you to think about the comforting and convenient stories we in the TM community tell ourselves, whether in the lab, on the wards, in blood donor clinics and clinician's offices, and wonder if our stories could be narrative fallacies.

Remember that grappling with complex information and mysteries strains our brains and brains like to save energy. Stories allow us to process and retain information more easily by showing a pattern that solves a mystery. Often our stories comfort us by confirming pre-existing beliefs or just making life simpler.
For techies who've stuck with the blog to the end, an inside joke: Think of all the trouble we go to in the lab to get rid of troublesome (unexplained) test results in antibody screens or antibody identification panels that don't show a pattern. The go-to narrative fallacy is ? Clinically insignificant cold agglutinins, of course. 
The blog's theme reminded me of a CCR 1970 ditty, sometimes associated with opposition to the Vietnam War. SHOT 2015 discusses narrative fallacies - stories we tell ourselves - in the context of phenomena that prevent us from learning from our mistakes. 
Long as I remember the rain been comin' down.  
Clouds of mystery pourin' confusion on the ground.  
Good men through the ages tryin' to find the sun. 
And I wonder, still I wonder, who'll stop the rain

As always, comments are most welcome. 

UK SHOT website 
SHOT: 2015 Annual Report 
Miller S. Myth based medicine. Br J Gen Pract. 2015 Jun; 65(635): 313. E-pub 2015 May 26.  
Taleb NN. The Black Swan (2007) 
Related TM blog (Aug. 2015): The early days (Musings on educating young TM professionals)