Giving patients the whole truth

I am guilty of being the eternal optimist in my practice. I can’t recall ever telling a patient that the outlook isn’t great for them, even though on many occasions, I had that sinking feeling in my gut. Many patients in pain are distressed enough – they don’t need me to give them more to worry about, do they?

Maybe I’m not the only clinician who feels a bit allergic to the idea of discussing poor prognosis. Prognosis is the likely course of a health condition and, unsurprisingly, it’s what patients want to know about when they see a health professional. It’s also the number one expectation of the patient to go unaddressed by a clinician.

When we omit information about our patient’s prognosis, either because we don’t know or don’t have time to discuss it, or because of a paternalistic desire to protect our patients from bad news, we deprive patients of the right to make an informed decision about their healthcare.

I can hear you saying:

“Really?”

“Isn’t it part of our job to be enthusiastic and positive with our patients, especially with the ones who are distressed?”

Not always. ‘Unknowns’ in health make patients worry. Unfortunately, in low back pain, there are many unknowns. A diagnosis, for example, is often unclear or unavailable. The mechanisms of many treatments are also unclear. And prognosis? We know the average prognosis is positive – recall the ubiquitous “70-80% recover” mantra. What we don’t know is whether the patient in front of us will be one of the lucky 70-80%.

We intended to remove some of the uncertainty surrounding individual prognosis by developing a new prognostic model for patients with recent onset low back pain (pain for less than 6 weeks). The model would be developed using the highest methodological standards. We would publish a protocol in advance. It would be a model that can give an accurate risk estimate to a patient using an outcome that is high on the list of patient concerns: persistent pain.

Despite there being over 30 prognostic models in low back pain, none have been developed with this purpose in mind: to provide risk estimates to patients. Even the most popular prognostic models, namely the OMPSQ and STarT Back Tool, were not designed to predict the onset of persistent pain. Perhaps not unexpectedly, when these models are tested in samples with recent onset pain, their predictions aren’t perfect and may not be much better than an educated guess.

Which brings me to the aim of our most recent study, which was to a build a prognostic model that would: 1) predict persistent pain in patients with a recent onset of low back pain with acceptable accuracy, and 2) have potential to inform shared decision making. Informed decisions lead to more appropriate healthcare. An accurate prognostic model would also give researchers and clinicians a better chance at targeted prevention.

We developed the model using patient data from a cohort study on 1230 patients in Sydney, Australia. All patients had seen their GP or physio about a recent episode of low back pain. We used statistical methods to identify the 5 most important questions that predict the onset of persistent pain. You can see these 5 questions and the risk calculator here. We called the model PICKUP, which stands for “Preventing the Inception of Chronic Pain.” After we built the model, we tested how well PICKUP could predict outcome in a separate sample of 1528 patients with recent onset low back pain.

We found that in this separate group of patients, the validation sample, the area under the curve statistic, a measure of how well a model can discriminate between those who did and those who did not develop persistent pain, was only slightly better than existing models (sigh). Fortunately, the area under the curve statistic is only one measure of accuracy. No need to throw the baby out just yet. When we looked at the ‘calibration’ of the model, that is, how well risk estimates matched actual outcomes, things got a little more interesting. For the patients with low-risk estimates, for example, below 30% risk of persistent pain, the predictions were quite accurate indeed. That means if you were to tell a patient that they had a 5% risk of persistent pain, around 5% of patients with the same risk score would actually develop persistent pain. Until now, this is not something that these prognostic models have been able to do very well.

When we looked at measures of clinical usefulness, that is, whether or not the model would lead to better decisions in practice, the results became even more interesting. We estimated that if clinicians screened patients with PICKUP, they could reduce the number of unnecessary interventions they recommended to low-risk patients by 40%. Considering the scale of low back pain, we thought this finding was a very big deal.

Where to from here? Well, we are planning a trial to test whether PICKUP does actually lead to more appropriate healthcare for patients with recent onset low back pain. For example, if clinicians use PICKUP, do they recommend intensive interventions to high-risk patients and minimal intervention to low-risk patients at a higher rate than if they were left to their own devices? Do patients feel more informed about their decision to pursue further intervention for their back pain?

Also, how do we sensitively communicate these risk scores to patients? For example, is it helpful to tell a patient: “You have a 40% risk of persistent pain?” What about: “You have a 60% chance of recovery?” or “You have some risk factors for a slower than average recovery. Lets talk about why that is, and what your treatment options are…?”

We are researching these questions as I write this blog. In the meantime, rather than struggling blindly through the uncertain quagmire that is non-specific acute low back pain, perhaps it’s time to look to the data from prognostic models such as PICKUP for help. The next step is to get the conversation started.

Adrian Traeger

Adrian Traeger Body In Mind

Adrian is right in the thick of his PhD research at NeuRA looking at the prediction and prevention of chronic back pain. He is particularly interested in the effects of the clinical consultation. Like most physios, Adrian likes talking to patients and, thanks to Carl Rogers, is getting better at listening.

Adrian’s musical taste has changed recently – from moody indie rock to “Giggle and Hoot’s Giggleicious Favourites”.

Reference

Traeger AC, Henschke N, Hübscher M, Williams CM, Kamper SJ, et al. (2016) Estimating the Risk of Chronic Pain: Development and Validation of a Prognostic Model (PICKUP) for Patients with Acute Low Back Pain. PLoS Med 13(5): e1002019. doi: 10.1371/journal.pmed.1002019