An Introduction to Clinical Decision Making Dr Graham R. Nimmo Chair, Scottish Clinical Decision Making Special Interest Group Clinical Skills Managed Educational Network and Scottish Clinical Skills Network Hello and welcome to this interactive tutorial on clinical decision making. Once you have completed it you should have an enhanced idea of what is involved in CDM and hopefully will have surfaced and framed your own questions regarding cognition in the clinical context which will require further exploration both physically and into the literature. I would like to thank my colleagues in the Scottish CDM SIG for input to this work and particularly to thank Prof Pat Croskerry from Halifax, Nova Scotia for his advice, inspiration and slides!

Overview Intro to clinical decision making (CDM) The five second rule: a case based CDM challenge How do we think and make decisions? What affects our CDM? How can we improve CDM and enhance patient safety? Next? CDM is a massive subject and this short introduction aims to give you an insight to the practicalities of CDM and a window onto the theoretical basis, particularly current theories of how we make decisions. This might influence your clinical practice and teaching.

CDM and non-technical skills Team working Decision making Situation awareness Task management Diagnosis Prognosis Decision making is one of the so-called ‘softer’ skills in clinical practice. It sits within the wider group of skills which are embraced in the terms human factors or non-technical skills and which are listed here. If you are interested in learning more about these look at the links at the end of this tutorial. There are a number of viewpoints as to which of them is most important, with some authorities viewing situation awareness as the critical element. My own take on the NTSs is that they shouldn’t be parcelled up into separate boxes but that there is overlap between the so-called domains and that decision making and communication probably flow through all of them. Diagnosis is a separate issue. The construction of a clinical picture in terms of what the underlying disease process is lies at the heart of all medicine whether the therapy is surgical, psychological, pharmacological, palliative, curative. As GH Rodger stated eloquently in his 1906 book ‘an introduction to the study of medicine’ “Medicine is a science by its means of study: it is an art by its applications”. Two branches concerning the medical art are defined: diagnosis (from Greek to know, through) and prognosis (from the Greek from, before, to know). These involve some of the most complex and difficult decisions which need to be made and often involve elements of uncertainty and judgment. More about these later.

CDM exercise 1 Think about your choice of speciality, profession, job What influenced you in that decision making? Of course decisions are made continuously in life, some trivial (HP sauce or tomato ketchup?), some monumental like ‘should I accept this job in Alaska?’ Speaking of jobs have a think about why you chose your particular career.

CDM exercise 2 Look at the next slide and ask yourself: “Which beach would I rather be on?”

A or B ? How did you decide that? Did you spend time and effort deciding? Or was it virtually instantaneous? More about that later…………

CDM in the acutely ill 21 year old with known asthma In respiratory ward Phone call Decisions Recognition Let’s think about our assessment process for this patient and the decision making necessary. Actually seeing the patient can give a huge amount of information to the experienced clinician. Physically going to look at the patient is more valuable than hearing about them over the phone: the five second rule relates to that initial sight of the patient where their demeanour and context are added to the standard clinical and physiological assessment . This disciplined noticing can avoid cognitive biases of the telephone caller. (This is equivalent to the lesson in Dawn’s telephone scenario in the other on line introductory module : speak to/look at the patient rather than getting second hand information). Let’s go through the process. We have already made the critical decision to GO TO SEE THE PATIENT.

The five second rule So to the patient: in the first few seconds our decision making and diagnostics may work in a very fast, efficient but potentially hazardous way. This is named System 1 thinking and its characteristics are listed on the next slide…

System 1 thinking: (intuitive) CDM in the acutely ill In the ward Assessment and management Looks…. Obs…. Actions System 1 thinking: (intuitive) Cognitive style Heuristic Cognitive awareness Low Cost Low Automaticity High Rate Fast Reliability Low Errors Usually Effort Low Predictive power Low Emotional component High Scientific rigour Low The immediate impression is based in what we define as System 1 thinking: impulsive, immediate, urgent. In the hands of an expert this can be invaluable although still potentially risky. In the hands of a novice it is quite likely to be dangerous. It involves emotion and rapid almost ‘reflex’ decision making. If we are aware that this process is happening we can step back from our decisions and examine them, check for errors and calibrate our decision making.

Importance of CDM in managing sick patients Patients still die from ‘simple’ things either missed, delayed or done sub-optimally Decisions including diagnosis Approx 80% of clinical time spent in the cognitive domain Clinical decision making is one of the key activities which underpins patient safety and the quality of patient experience.

THE FOUR KEY DOMAINS OF EMERGENCY CARE 1. Acute assessment + stabilisation with immediate investigations and support. Targeted secondary exam 2. Monitors: reassess Surface Invasive Real time or Delayed Illness severity 3. CDM Team work Task Mx Situation Awareness Critical Thinking 4. Differential diagnosis/ definitive diagnosis Immediate, medium term and long term treatment We can think of a number of particular activities which need to happen simultaneously when we are dealing with a sick patient such as our young man with acute asthma: the familiar ABCDE approach is employed to help identify life threatening abnormalities which require immediate attention: usually respiratory failure (hypoxaemia with or without hypercapnia, shock, impaired conscious level or a combination of these as in severe sepsis or major trauma. Even within this standardised, structured approach there is a plethora of decisions to be made as you can see in the next few slides. In addition an assessment of illness severity must be made in order to guide the speed and level of intervention. In the next section relating to ABCDE approach and ISA the key clinical decisions are highlighted in yellow font.

1. Advanced First Aid ASSESSMENT Hello, how are you ? Response Airway: patent ? Breathing ACTION Look: obstruction Listen: ? noise Clear or secure: headtilt/chin lift or jaw thrust? Airway: oral or nasal? High concn oxygen: mask type? Flow?

1. Advanced First Aid ASSESSMENT Sounds ? Common 3 are ? Causes are ? ACTION Clear and keep open Get help 2222? Advanced airway management required? Tracheal tube? Size? Cut to what length? Are drugs needed for anaesthesia and intubation? If yes, which?

For each ask ‘what is the diagnosis?’ 1. Advanced First Aid ASSESSMENT Sounds Nil: complete obstruction or not breathing Snoring/gurgling: reduced GCS, foreign material Stridor: anaphylaxis, burns/thermal; tumour; abscess/infection For each ask ‘what is the diagnosis?’ ACTION Clear and keep open Get help 2222 Positioning Advanced airway management required

1. Advanced First Aid plus 2. OBSERVE Rate Volume Symmetry Character Work of breathing Compromise Ix & MONITOR CXR, PEFR, ABGs Repeat observations Pulse oximetry TREAT Oxygen Nebulisers

1. Advanced First Aid ASSESS Pulse: which pulse? Skin: cap refill time, temperature BP: where? Which method? PHYSIOLOGY MAP = CO x SVR CO = HR x SV Low BP = decompensation

1. Advanced First Aid iv access iv access Site Upper limb Femoral Size Blood sampling iv access Upper limb Femoral High flow: short and thick Fluids Drugs

Wide bore peripheral cannulae The bottom one is a straw!

1. Advanced First Aid Disability ? Conscious level, focal neurology DEFG Difficult bit

IMMEDIATE INVESTIGATIONS Arterial blood gases: O2, CO2, acid-base Potassium Glucose can all be done on a Haemoglobin blood gas sample 12 lead ECG CXR Targeted investigations What should we do having analysed this information? Having a system and a routine can help reduce the cognitive load. Having a blood gas machine which can rapidly perform the most critical investigations which result in instability, and which are also rapidly improvable (usually), takes away the need to decide on them individually. The decision to obtain an arterial blood sample for analysis needs to be made and then a decision as to which artery to sample from is required. Once these and other investigations are available multiple decisions on supportive treatments (eg oxygen, ventilation, potassium, glucose, iv fluids, blood transfusion, analgesia), diagnosis and definitive treatments (eg antimicrobials, heparin, anticonvulsants and so on), prognosis, illness severity, where should they be looked after?

1. Advanced First Aid Evidence Environment: context Targeted secondary Examination Explanation Everything else… For completeness here is the E of the acute approach. Where the patient is and what is happening around them may affect CDM. See the reading list to access a paper on the impact of context on decision making.

Advanced First Aid=Phase 1:abcde abcde, treating as you go Repeated assessment and continuous monitoring: patient better or worse ? Do we need enhanced abcde ? Targeted secondary examination

THE FOUR KEY DOMAINS OF EMERGENCY CARE 1. Acute assessment + stabilisation & immediate investigations and support. Targeted secondary exam 2. Monitors: reassess Surface Invasive Real time or Delayed Illness severity 3. CDM Team work Task Mx Situation Awareness Critical Thinking 4. Differential diagnosis/ definitive diagnosis Immediate, medium term and long term treatment At the same time as all of this is going on with our asthma patient we need to establish monitoring: this will be determined by what is available? (context, ergonomics) and your experience, competence and capability. In severe or LT asthma a large bore iv cannula is mandatory and (ideally) an arterial line although time pressure may negate the latter.

Illness Severity Assessment Speed of action needed Level & type of expertise: resuscitation; diagnostic; therapeutic Where should the patient be ? Nursing intensity, monitoring, medical input? Definitive treatment: speed ? Rapid illness severity assessment is needed to answer these questions…However further information may help with this as the assessment, investigations and management proceed.

SEVERITY SCORING 1:CLINICAL ABNORMAL PHYSIOLOGY Airway compromised Resp rate Pulse rate SBP GCS OBSERVATIONS Bad <10 or >30 <45 or >120 <100 (110) or >200 Fall of 2 points, <15 Initially the ‘vital signs’ may be abnormal and indicate severity.

SEVERITY SCORING 2: INVESTIGATIONS ABNORMAL INVESTIGATIONS Hypoxaemia Hypercarbia Potassium Glucose H+ Base excess Lactate RESULTS <3 or >6 (ECG) <3 or >20 >50 or <30 < -5 or > +10 Diagnosis ? As investigation results become available these add further to the assessment of illness severity.

Relationship between base excess and mortality in ICU Some specific results such as base deficit, lactate indicate tissue hypoxia and risk of death. Base Excess and Mortality Red = Dead Green = survivor ICM 2001;27:74-83

SEVERITY SCORING 3: organ failures Clinical: cardiovascular (shock) CNS reduced conscious level Urea and creatinine: renal ABGs: respiratory (oxygenation+/or CO2 clearance) Clotting: coagulation WBC: bone marrow Gut/liver: glucose; lactate; clinical Thirdly, as measures of specific organ systems function are reported a picture of how many organ systems are failing is built up. We know that the more systems involved the worse the patient’s outlook so this helps with prognostic decision making although

4. Differential Diagnosis, ultimate diagnosis and definitive treatment Get more history Trachea Chest JVP and heart Abdomen Skin, CNS GP, family, SAS Deviation Lateralising signs, wheeze, crackles HS III or IV, murmurs Swelling, pulsation Rashes, neck stiffness, lateralising signs the actual diagnosis (or diagnoses) and the response to both supportive and definitive treatment refines the prognostic decisions.

ASSESSMENT A clinical + B investigations C organ failures D diagnosis So you can see that a number of factors (which become available over time) are required to make decisions about prognosis. An understanding of this helps in decisions to refer patients for intensive care and in discussions with relatives and friends.

Decision making Diagnosis (and treatment) Is the diagnosis correct (complete) ? Prognosis Admit ? Discharge ? Stop ? Distributed So what other kinds of decisions are we making: some of these are listed above but I want to spend some time considering diagnostic decision making.

Diagnostic Error Ranked 2nd cause of adverse events (Harvard study, 1991) Diagnostic failure highest in EM, GP, Gen Med Passing on to specialists in wards, ICU 2/3 of claims against UK GPs are for diagnostic failure This area is one of uncertainty with major patient safety implications.

Type 2 thinking (analytical) Cognitive style Systematic Cognitive awareness High Cost High Automaticity Low Rate Low Reliability High Errors Few Effort High Predictive power High Emotional component Low Scientific rigour High Diagnosis History: full Examination: complete Investigations Differential Dx Treatment Refine diagnosis So in order to make the diagnosis accurate we have developed and learned a painstaking approach which involves detailed history taking, systematic clinical examination, the performance of appropriate targeted investigations to allow us to create a differential diagnosis. Treatment is then started and the response of the patient helps to inform the subsequent refining of the diagnosis. This process can be neatly aligned to the profile of type 2 thinking. It requires high cognitive awareness, is costly etc

Categorising Clinical Decision Making Cognitive theory: traditional Technical Professional Distributed Traditionally theory related to clinical practice has been psychologically based on a number of cognitive paradigms. However when you look at your clinical practice you will see that there are a number of domains into which most decisions can be placed: t, p, d.

Traditional cognitive taxonomy or “how you think it” Problem solving Pattern recognition Decision analysis theory Hypothetico-deductive reasoning There is a vast literature on this area but how useful these individual theories are on informing, supporting, helping our day to day CDM is not always clear.

CDM: a universal model of diagnostic reasoning Intuitive Experiential-inductive Bounded rationality Heuristic Pattern recognition Hard wired response Thin slicing Unconscious thinking theory Analytical Hypothetico-deductive Unbounded rationality Normative reasoning Robust CDM Acquired, critical, logical thought Multiple branching/arborisation Deliberate, purposeful thinking However if you fit these theories into a universal model which is built around system 1 and system 2 thinking and we start to examine our day to day decision making you can start to work out which kind of thinking you are engaged in for specific decisions. And if you are making intuitive decisions you should then ask “what influences my rapid fire decision making? Remember the beach question?

A or B ? How did you decide that? Using system 1 processes. But what about context? If you were an international art thief and you wanted to be inconspicuous which beach would you prefer to hide on?

What affects clinical decision making ? CDM What affects clinical decision making ? Knowledge and skills Behaviours: attitude (multiple selves), emotions (affect: self, family, patients, relatives, colleagues), values. If you use system 1 processes as an expert you are often correct (remember the skin rashes? most experienced clinicians get the right diagnosis in a blink). However if you use these hardwired processes as a novice it could be way off the mark.

What affects clinical decision making ? Context Values Affect Knowledge Critical thinking Interruptions Clinical reasoning Words Non-technical Skills Physical factors Stress and Fatigue Ergonomics Experience What we hear What we think Cognitive biases Heuristics Epiphanies Geography Numerous factors affect decision making and this far from exhaustive list (in no particular order) gives an idea of the complexities underlying what is such an integral component of everyday clinical activity.

30 Cognitive Errors after Croskerry Aggregate bias Gender bias Psych-Out Errors Anchoring Hindsight bias Representativeness Ascertainment bias Multip.Alternatives Search satisficing Availability Omission bias Sutton’s Slip Base rate neglect Order effects Triage-Cueing Commission bias Outcome bias Unpacking principle Confirmation bias Overconfidence Vertical line failure Diagnostic creep Playing the odds Visceral bias Attribution error Posterior prob. Ying-Yang Out Gambler’s Fallacy Premature closure Zebra retreat One particular area which has been surfaced over the last few years is that of cognitive dispositions to respond. Pat Croskerry has catalogued and defined these innate human ‘failures of cognition’. Some are negative biases, others can be useful or detrimental. A detailed analysis of them is beyond the scope of this tutorial but to give a couple of examples: confirmation bias is where the clinician shoe-horns the patient into their preferred diagnosis; Ying-Yang Out is where boredom sets in: we have investigated this patient up and down the Ying Yang river so stop; Availability is that situation where you saw a patient with a certain condition last week so it is at the forefront of possibilities when you see a patient with similar symptoms and so on. The importance of knowing about these is that we can recognise when we are straying down one of these paths and by applying corrective processes known as cognitive forcing strategies we can avoid the biases and improve our CDM and enhance patient safety.

1 2 TYPE Calibration Diagnosis TYPE Intellectual ability Hard wiring Heuristics and Biases Ambient conditions/Context Task characteristics Age and Experience Affective state Gender Personality TYPE 1 processes RECOGNIZED Pattern Recognition Patient Presentation Pattern Processor Rational override Dysrationalia override Calibration Diagnosis Repetition TYPE 2 processes Putting it all together we work using both S 1 and S 2 processes all the time toggling back and forth between them with influences from all of the factors included in this schema. If we understand where we are situated cognitively we should be able to improve CDM and patient management. NOT RECOGNIZED Intellectual ability Education Training Critical thinking Logical competence Rationality Feedback

THE FOUR KEY DOMAINS OF EMERGENCY CARE 1. Acute assessment + stabilisation & immediate investigations and support. Targetted secondary exam 2. Mons: reassess Surface Invasive Real time or Delayed Tissue hypoxia 3. CDM Team work Task Mx Situation Awareness Critical Thinking 4. Differential diagnosis/ definitive diagnosis Immediate, medium term and long term treatment So this is where we are working clinically

Clinician factors: Evidence Based judgment, affect, Medicine Patient experience Patient Factors these are the elements which need to be included in our approach

Solutions Training in critical thinking Training in major cognitive and affective biases Training in logical thought Awareness of self and metacognition Timely feedback Training in cognitive forcing strategies And these are some of the potential solutions and that will be where we go in the next tutorial ….

You can find further information including related citations at