Diagnostic conversations: Clinical Decision Making in surgery – Part 2

Introduction

This paper focuses on the fourth stage of a clinical decision making model developed for surgery. As in Part 1 [1], the model is elaborated through an e-mail discussion between two surgeons: both Directors of Surgery in their own institutions. David is an experienced general surgeon specialising in colorectal surgery; Spencer, a senior paediatric surgeon is from a different specialty. At one level the conversation is about a particular patient, and the issues for the clinicians concerned. At another level the conversation addresses a broader collaboration in developing and promulgating a model of Clinical Decision Making (CDM). The third author (Wendy) is not a clinician: she has worked together with both surgeons in interpreting and commenting on the cognitive processes they and other surgeons employ to develop the CDM model.

The CDM model and the case so-far

The four major stages in the model incorporate the time from the patient’s first presentation, to review of performance at the completion of treatment:

1. The Working Diagnosis and Initial Management Plan (this can take minutes, hours, days or weeks depending on the nature of the case and its urgency).

2. Preparing for a procedure – this takes place before (and sometimes just before) a procedure is performed – but may continue even whilst the patient is being anaesthetised.

3. Monitoring progress of the management plan and procedure. This includes any need to modify the intended plan or to make a decision intra-operatively to vary the procedure being performed, in response to the operative findings or other factors such as patient physiology.

4. Reviewing what was done following the procedure in the light of the outcome for the patient, the team and the surgeon. At times, this stage can be carried out almost contemporaneously as a form of debrief, or later on, as part of a case review and outcome-based surgical audit.

In this paper we focus on the fourth stage. This stage, unlike the other three, may require a more conscious and retrospective analysis of decision making to ensure that any lessons are identified, learned and shared. A more detailed description of the first three stages, plus some of the theory underpinning the model, has been published elsewhere [2].

In Part 1, we discussed the initial decisions in relation to an 84-year-old cachectic patient who presented with a near obstructing lesion of the mid-transverse colon and who had a large 18 mm stone in a non-inflamed gall bladder. He had type 2 diabetes, had suffered marked weight loss and his serum albumin was 24 g/l. Investigations showed no evidence of metastatic disease. A colonoscopy failed to traverse the tumour but biopsies taken confirmed an adenocarcinoma. The patient consented for a laparotomy and large bowel resection, most likely an extended right hemicolectomy. The consent included the possibility of a stoma and that a decision about the gall bladder and its stone would be made during the operation.

The patient underwent an open extended right hemicolectomy with mobilisation of the splenic flexure, and a stapled side to side anastomosis. The gall bladder was removed without an operative cholangiogram: this was uncomplicated and took no more than 15 min. The whole procedure took 120 min.

The conversation continues

From: David

Dear Wendy and Spencer,

Postoperatively: The patient did well for 3 days but on the evening of the third day developed oliguria which responded to fluids. The urine output was then about 30 ml/h and there was extensive peripheral oedema. The serum albumin had fallen to 14 g/l. The patient was not in cardiac failure but medical advice was to supplement fluid resuscitation with 20 mg doses of Frusemide. The patient was apyrexial, fully conscious and communicative – though not perky. Vital signs were stable. Examination of the abdomen revealed a distended abdomen which was soft and non-tender. A paralytic ileus was diagnosed and a nasogastric tube inserted.

The surgical team returned to the ward after a weekend off to find the patient on day 4 postoperatively with a nasogastric tube draining 1400 ml, peripheral oedema, normal urea and electrolytes, a white cell count of 9×10⁹/l with bands of 3.8 and haemoglobin of 88 g/l.

On day 4 what decisions needed to be made for this elderly, frail, postoperative patient?

In discussing the management options the following questions should have been considered by the treating unit:

1. Could there have been an anastomotic leak?

2. As the WBC bands are 3.8 – was this of any significance?

3. Did this patient only need to be actively observed?

4. Did he need imaging, for example, a CT-scan?

5. Did he need a relaparotomy?

From: Spencer

Dear David and Wendy,

You could even go back a step, to identify and priortise the actual issues for this patient at the end of day 3, and how they might be inter-related. For example, as a clinician, how might you interpret the high NG losses in conjunction with the distended soft and non-tender abdomen; the “OK” but not perky general condition; the hypo-albuminaemia, peripheral oedema, and other laboratory results? How uneasy should the team have felt?

Re: Q1, anastomotic leak: in considering this, there could be some earlier consideration of the operative assessment of the ease and quality of the anastomosis – no tension, good blood supply, technically satisfactory vs. technically difficult/tension/marginal blood supply. The reason is that most surgeons incorporate their view of the quality of the anastomosis to influence their assessment of the likelihood of a leak. Discussion could also include how to distinguish paralytic ileus in the absence of a leak, from the likely clinical features that would be evident if a leak is present. To what extent could an ileus be a manifestation of leak, but if so, why is there no leukocytosis, fever, pain and tenderness? Would any imaging be indicated at this point? And what would its limitations be, and why? To me, this patient’s postoperative course seems to have reached a critical clinical moment.

Another thought: should the choice of antibiotics and their duration of use be mentioned?

Re: Q3 about the management plan: The answer would be yes – but why? Alternatively, would it be better worded: “How might the underlying age and frailty influence the clinical manifestations of any complications that need to be excluded?” We could tease out how his general condition might obscure or “underplay” features of a leak, and that his ability to heal is already significantly compromised: age, diabetes, low albumin, etc.

Q5 – more surgery: on the evidence provided most would answer “probably not” at this stage. An option would be to ask: what are the features that might encourage you to do a re-laparotomy? This would force them to consider exactly what clinical or investigative parameters would influence them to re-look, and be helpful to them in their further surveillance of the patient, and where might the traps be that make the interpretation of these features more difficult? It fits in with Q3 very well.

From: David

Dear Wendy and Spencer,

The team was confident of the blood supply, lack of tension and colour of the anastomosis.

The unit decides to actively observe. An anastomotic leak is not thought to be likely due to absence of leukocytosis, no fever, no abdominal tenderness, only an expected distension due to ileus. They do not feel uneasy enough.

For 4 further days there was no real progress or deterioration in the above findings. Anastomotic leak was obviously the worst-case underlying pathology, which admittedly might not have been definitively diagnosed by CT-scan. Yet they did not think this likely and the patient was presumed to have an ileus with consequent fluid imbalance and further complicated by his hypoalbuminaemia.

Outcome: Eventually on day 7–8 and over a 14-h period he deteriorated, a Medical Emergency Team (MET) call was made but resuscitation was unsuccessful and within a few hours the patient arrested and died. A coroner’s post mortem found an anastomotic leak. The serum albumin then was 11 g/l, the white count never rose above 10 and he never developed a fever or obvious abdominal signs of peritonitis.

Review – We reach stage 4 of the model: case and mortality review:

What can be learned about the decision making in this case?

From: Spencer

Dear David and Wendy.

Sad, but familiar. It illustrates well the problems even experienced clinicians face. I suppose it highlights exactly some of the aspects that we should get the course attendees to consider in the earlier part of the discussion. The practical problem here is that even with hindsight management may have been similar. Would it have been different if the first few days had not been over a weekend? Certainly to re-operate without reasonable evidence of a surgically-correctable problem itself would carry some risk and have raised fears of the same ultimate outcome.

David, this case illustrates well how there may be no “right” answer, and that decision-making is very dependent on the information and resources available. It highlights the management of uncertainty and conflicting information that challenges clinical interpretation.

Commentary

Given the ultimate outcome, the questions in Figure 1 need to be honestly addressed by all of the people responsible for this patient’s postoperative care as well as the surgeon and surgical team. Every step and every decision needs to be scrutinised. All clinicians involved need to understand how they made the decisions they did, and the degree to which they were justified.

Figure 1

A model of clinical decision making – Stage 4.

For example, at what level of suspicion of an anastomotic leak should further investigation (which might include imaging and/or re-operation) have been done? Did the team(s) give excessive weight to some of the available information and give insufficient attention to other information. To what extent were the clinical features modified by co-existing disease? Or to what degree could this have been factored in? And what were the risks of repeat surgery versus continued observation, given the dangers of delayed intervention? What was the worst case scenario they were facing? Did they address this adequately? Were other options for treatment available, and at what point should they have been considered?

Cognitive errors contribute to a higher proportion of diagnostic errors than lack of knowledge or skills [3]. In this case a number of cognitive biases may have contributed to the outcome. ‘Anchoring’ and ‘under adjustment’ [4] on their earlier assessment and ignoring the developing signs of lack of progress (in this case not considering the pointers toward a possible anastomotic leak). The team also showed indications of ‘search satisficing’ in that they called off any further consideration of an anastomotic leak once they had identified fluid imbalance which required treatment. It seems that they were unable to resist what is termed ‘diagnostic momentum’: this means that they were biased by prior decisions and the label placed on the patient as being hypoalbuminaemic, fluid imbalanced, and in need of fluid challenges to maintain urine output. Whilst the correct diagnosis that there was an anastomotic leak was all the more difficult because the classic presentation (fever, leukocytosis, peritonitis) were masked by the patients age, hypoalbuminaemia and comorbidities, they did not change their assessment of what was wrong and what needed to be done until it was too late, and the patient was unsalvageable. Therefore, whilst it is possible that they were responding to the patient using a false perception of the situation, arrived at most probably through intuitive, pattern recognition thinking, they were too comfortable to accept gaps and could have worked harder analytically to work out why there was so little progress. This is challenging for all clinicians as some patients are just “slow” even when there is nothing else going on.

The result of this case review and subsequent discussion did show that this was a difficult anastomotic leak to diagnose but that there were clues present: if these cues had been recognised they may have led to an abdominal CT-scan or a second opinion, and perhaps an earlier decision as to whether to risk a relaparotomy. For example, the lack of understanding of the signficance of the bands resulted in this information being ignored, albeit acknowledging that there is not a great deal of evidence in the literature on the relationship between WBC bands and anastomotic leak. They also needed to recognise that the patient was not progressing, and needed to take a step back and reconsider all their decision making, probably around day 5, rather than carrying over the same presumptions from day to day.

Dear Spencer and Wendy,

In reviewing this case with the team concerned, we had to conclude that the team missed their opportunity to image this elderly man around day 4 or 5. They should have felt more uneasy and been suspicious. Experience should have told them that oliguria can also be a sign of sepsis, not only of fluid imbalance; that a patient with hypoalbuminaemia can fail to heal even when there is a perfectly constructed, well-vascularised anastomosis. The presence of an ileus should have worried them, unsettled them, driven them to image and to seriously consider getting another independent senior surgical opinion. The images obtained might have been hard to interpret, but that does not mean an abdominal CT was not indicated. It might have identified the leak. There were also the bands – these were ignored because the members of the team were not sure how to interpret them in the absence of a high white count. The bands meant there was significant sepsis for which the three likely possibilities were intra-abdominal sepsis (related to leak or abscess), pneumonia or wound infection. The white count does not need to be raised, bands are associated with infection, and hypoalbuminaemic patients would be expected to not always have a high wbc.

Therefore, in retrospect, how did they do? Their postoperative decision making was impaired because it was limited to a presumption of fluid imbalance and ileus whilst they ignored other pointers to the fact that the anastomosis could have been leaking. My experience tells me – no one wants a relaparotomy but, if necessary, the earlier the better if the patient is to survive a leak. I acknowledge they might have been thinking the patient was not in great shape to survive any further surgery. However, patients are less likely to die from a negative relaparotomy, than from a late necessary one.