The Trouble With Looking Backwards
The Lindsay Clancy case raises an interesting question about how we judge clinical decisions after we know the outcome. It is also relevant to how parents make sense of their children's behaviour. How?
In both situations, we are trying to make decisions with incomplete information. We form an explanation for what we are seeing, act on that explanation, and then receive more information. The difficulty is that once we know what happened, it becomes much easier to believe that we could have predicted it.
That is where hindsight bias becomes important. And I do a version of this work in forensic psychiatry. When reviewing a case after something has happened, part of the task is to establish what was known at each point in time. What information was available to the clinician? What had the patient reported? What did other people know? What could reasonably have been concluded from that information?
You then have to separate all of that from what became known later. THis can be surprisingly difficult. Once you know the outcome, earlier information acquires a different significance. A symptom that seemed nonspecific can look like a warning sign. A medication change can look more significant than it did at the time. A clinical decision that appeared reasonable can look very different when viewed with knowledge of what happened afterwards.
The information hasn't changed. Our interpretation of it has. This is why retrospective judgement requires some discipline. The question isn't simply whether something could have been recognised. The question is whether it could reasonably have been recognised at the time. Those are very different questions.
This distinction is relevant in medicine because clinical decisions are made without complete information. A clinician sees a patient, takes a history, makes observations and considers whatever other information is available. They then make a judgement about what is most likely to be happening and what should be done. They don't know what will happen next. That is the environment in which the decision should be assessed.
Outcome bias creates another problem. We tend to judge a decision according to what happened afterwards. If the outcome is good, the decision can look sensible. If the outcome is terrible, the decision can look foolish or negligent. But the outcome doesn't establish whether the decision was reasonable. A reasonable decision can be followed by a poor outcome. A poor decision can be followed by a good outcome. To assess the decision, you have to return to the information that was available when it was made. That is less satisfying than having a clear explanation for what happened.
When something terrible occurs, there is a strong desire to identify a cause. Perhaps it was the medication. Perhaps the diagnosis was wrong. Perhaps a clinician missed something. Perhaps somebody should have intervened earlier. Those possibilities can then become the framework through which all of the available information is interpreted. aThat is where confirmation bias can enter.
Once we have settled on an explanation, we tend to notice information that supports it. Information that doesn't fit receives less attention. Over time, the explanation can become increasingly convincing because we are no longer evaluating all of the information equally.
There is another problem that comes up frequently in discussions about psychiatric diagnosis and medication: the difference between possibility and probability.
For instance, a medication can cause a particular adverse effect. That doesn't establish that it caused a particular symptom in a particular person. A symptom can occur in a particular psychiatric disorder. That doesn't establish that the disorder is the most likely explanation for that symptom.
Clinical reasoning requires consideration of the alternatives and the base rates. A headache can occur with a brain tumour, but that doesn't make a brain tumour the most likely explanation for a headache. The possibility is worth knowing about. Its probability is a separate question. The same principle applies to psychiatric symptoms. Knowing that an explanation is possible doesn't tell you how much weight it should have been given at the time.
The difficulty becomes greater after a catastrophic outcome because we already know which explanation turned out to be relevant. It can then become difficult to imagine a time when that explanation was only one possibility among several.
This brings us to the question of learning from history.
We should learn from bad outcomes. Reviewing what happened allows us to identify errors, reconsider clinical practice and change systems where necessary. But learning from an outcome doesn't mean assuming that we would have recognised it in advance. If we had been making the decision at the time, we would have had the information available at the time. We wouldn't have known the ending. We wouldn't have known which apparently minor details would later become significant. We wouldn't have known which decision would eventually be criticised. We would have been dealing with the same uncertainty. There is an important distinction between saying, "Knowing what we know now, we should do something differently in future," and saying, "Knowing what we know now, I would have recognised this at the time." The first is a reasonable use of hindsight. The second requires us to imagine that we would somehow have possessed information that nobody possessed. We can't know that.
There is also a tendency to locate responsibility in one person. The psychiatrist, the GP, the therapist, the patient or the family can become the focus of an explanation. Sometimes an individual has made an error. That should be examined properly. But clinical situations can involve multiple assessments, different clinicians, changing symptoms and information that becomes available at different points. Each person may have only part of the picture. When we review the case afterwards, we have the advantage of seeing all of those pieces together. The person making the decision didn't. That distinction is relevant to parenting too.
Parents are constantly trying to interpret behaviour. A child refuses school and the parent decides that the child is anxious. From that point, subsequent behaviour can start to be interpreted through that explanation. The child doesn't want to attend a birthday party, so that becomes anxiety. Their stomach hurts before school, so that becomes anxiety. They have a meltdown getting into the car, so that becomes anxiety. The child may be anxious. That doesn't mean anxiety explains everything. The same thing can happen with defiance. A child refuses to get dressed and the behaviour is interpreted as defiance. They argue about dinner and that is also defiance. They won't turn the television off and that confirms the same explanation. They come home from school exhausted and fall apart, and the behaviour is still interpreted as defiance. The behaviour hasn't necessarily changed. The model being used to interpret it has. This is one reason mental models can be useful for parents.
A mental model gives you a way of making sense of information, but it doesn't make the explanation certain. You can think that anxiety is contributing to a child's refusal to attend school and respond on that basis. You can then pay attention to what happens. If the response is helpful, that gives you information. If it isn't, that gives you information as well. You can reconsider the model.
The same applies to explanations such as tiredness, hunger, sensory overload, avoidance or defiance. They can all be useful ways of understanding behaviour in some circumstances. None of them needs to become a permanent explanation for the child. This is where I think parents can benefit from thinking in terms of hypotheses rather than conclusions. You have some information. You make the best assessment you can. You act on it. You observe what happens and update your thinking. That is essentially decision-making under uncertainty. It is also what clinicians do. The difference is that parents are often doing it while their child is refusing to put their shoes on and they are already late. The point isn't to find a perfect explanation for every behaviour. Children's behaviour is often ambiguous, and several things can be happening at once. A child can refuse something because they are anxious, tired, frustrated or simply don't want to do it. One explanation doesn't always exclude the others.
In my view, what is important is whether the model you are using continues to be useful. If it helps you understand the behaviour and make a better decision, keep using it. If the behaviour repeatedly fails to fit the explanation, reconsider it. That is a more reliable approach than deciding what the behaviour means and then treating everything that follows as evidence that you were right.
The Lindsay Clancy case provides an extreme example because the outcome is so serious and because we now have the advantage of knowing how the story ended. That makes it particularly easy to look backwards and decide that the important information should have been obvious. Perhaps some of it should have been. That is something that has to be established from the evidence rather than assumed from the outcome. We should examine decisions after bad outcomes. We should learn from them. We should change practice when the evidence supports doing so. But we should also recognise the limits of hindsight. Knowing what happened gives us information that the people making decisions at the time did not have. It can help us improve our mental models. It doesn't tell us that we would have made a better decision ourselves. That is useful to remember when thinking about clinical decisions. It is also useful to remember when thinking about our children.
Listen to the podcast
I explore this in more detail in my podcast episode, The Trouble With Looking Backwards. I look at hindsight bias, outcome bias and confirmation bias, and why knowing the outcome can change the way we interpret the decisions that came before it.
[Listen to The Trouble With Looking Backwards ]
Want to think differently about your child's behaviour?
This is one of the reasons I wrote The Parent's Mindset.
The book looks at the mental models we use when making sense of children's behaviour, particularly when we don't have enough information to know exactly what's going on.
It isn't a book of rules for getting your child to behave differently. It is about having better ways to think when the explanation isn't obvious.
[Read The Parent's Mindset ]
