Evidence about reassurance for back pain is not reassuring

Giving reassurance may be one of the most commonly used strategies for non specific low back pain (LBP) in primary care. We set out to review the evidence on reassurance in low back pain. We wanted to look at this neglected area because the majority of guidelines mention reassuring patients, especially at early stages of back pain. We wanted to find out how best to reassure patients in consultations in primary care. It turned out to be a tough job, nevertheless we systematically reviewed the literature.

Very early on, we realised that there were hardly any publications dedicated to reassurance in LBP. To find the evidence we had to look across broader groups that resembled LBP in primary care, and we had to develop some operational definitions for ‘reassurance’.

For the purposes of the study we defined reassurance as something that the therapist does, with the aim of reducing anxiety and changing beliefs and behaviours. We used a broad patient-centred approach to divide the consultations into data gathering followed by information giving, and focused on this latter stage as the major opportunity for reassurance. Finally, we were able to identify a model, developed from evidence of effective persuasion, which classified reassurance into affective and cognitive components. This model was a major breakthrough for us, because it proposed a counter-intuitive and highly controversial hypothesis: Affective reassurance was defined as a swift emotional response to a message from the therapist that suggested that the patient was understood, heard, cared for and supported. This message resulted in immediate reduction in anxiety. In contrast, cognitive reassurance is slow, iterative and demanding. It involves a neutral transmission of information from therapist to patient. According to the Coia and Morley model, only cognitive reassurance can result in changed beliefs and behaviours. The motivation for engaging in change is reduced with affective reassurance, as the urgency for change is reduced alongside the reduction in anxiety. So, once the patient is away from the consultation in the real world and the pain strikes again, they have no new tools to deal with it. The model suggests that patients will engage with affective or cognitive reassurance, but that they will not engage with both simultaneously. In a radical statement, the authors called for clinicians to abstain from their ubiquitous attempts to deliver affective reassurance, as they may cause more harm than good.

We considered that the model was testable, by looking at evidence from prospective cohorts that measure the interactions during the consultations, and looked at patients’ outcomes.

We considered that a key aspect of reassurance was the uncertainty that patients and therapists had to deal with, about unclear aetiology, variable prognosis, and weak evidence on effective interventions. We therefore restricted our inclusion criteria to studies of groups that didn’t have a clear diagnosis and management plan. We also excluded groups where management included regular tests and referrals. The majority of guidelines recommend avoiding these for LBP, and we considered that it might be an important aspect of reassurance (or failure to reassure).

Hence the systematic review, and the surprising findings – generally supporting the model. Cognitive reassurance (giving clear explanations about aetiology, prognosis, discussing treatment plan etc.) improves outcomes not only in the short term, but also in the long run. Affective reassurance at best improves patients’ satisfaction at consultation exit, but there was also evidence (from three studies with decent methodology) for an association with poorer outcomes.

We have to emphasise that no causal path can be inferred, and the likeliest explanation is that therapists identified the extra needs in patients who were more likely tohave worse outcomes, and tried to provide emotional and pastoral support.

We would like to hear views from readers of this blog in response to the following questions: :

  1. How do we provide clear and acceptable cognitive reassurance in the context of uncertainty?
  2. Presumably patients’ needs and ability to digest cognitive reassurance differ between sub-groups. Can research identify sub-groups and their needs without huge studies and expense? (We recognise here that the knee jerk response of supplying anxious patients with banal advice to avoid worrying appears to increase worry and health care seeking.)
  3. What is happening  with affective reassurance in the final stage of the consultation? We think that empathic, sensitive communication is essential in early stages of the consultation, to foster rapport and trust and elicit patients’ concerns. Could Coia and Morley be correct about switching to a neutral style in later stages? Or is some level of empathy needed to deliver cognitive reassurance effectively?
  4. Assuming the findings are replicated and confirmed, what next? Trials to change clinicians’ behaviour seem to show modest if any effects!

About Steven Vogel

Steven VogelSteven Vogel is Vice Principal (Research) at the British School of Osteopathy in London. Although most of his working time is spent on research, he has held a clinical post as an osteopath in an NHS primary care setting since 1993. His main research interests focus on back pain, clinicians’ beliefs and attitudes and more recently safety and manual therapy. Steven was a member of the development group for the NICE clinical guidelines for persistent non-specific low back pain and is an editor for the International Journal of Osteopathic Medicine. Spare time is a thing of the past, as his other full time job involves taking his three boys to various sports related activities 🙂

About Tamar Pincus

Tamar Pincus

Professor Tamar Pincus holds a PhD (University College London), as well as Masters degrees in experimental research methods in psychology (UCL), and epidemiology (Cambridge University). Her research has embraced a variety of methodologies, including experimental, epidemiological and qualitative. The research has included investigation of attention and recall in pain patients; the psychological predictors for poor outcome in low back pain, and the study of clinicians’ beliefs and attitudes in low back pain. Recently the focus of her research has moved to investigating the effectiveness of interventions through randomised controlled trials, and throughout she has collaborated closely with researchers from many disciplines, including doctors, physiotherapists, osteopaths, chiropractors and clinical psychologists, from a multitude of institutions, including Warwick, Keele, QMW, Leeds, Manchester, The British School of Osteopathy and many others. She also convened the international consensus group to establish what factors and measures should be included in prospective cohorts investigation the transition from early to persistent back pain. Most recently her research has focused on delivering effective reassurance to patients in primary care.


Pincus T, Holt N, Vogel S, Underwood M, Savage R, Walsh DA, & Taylor SJ (2013). Cognitive and affective reassurance and patient outcomes in primary care: A systematic review. Pain, 154 (11), 2407-16 PMID: 23872104


  1. Diane Mulholland says:

    Hi Mike,

    Loved your post, and just wanted to say I also feel strongly about getting patients/clients to take control of their recovery. I see it as my job to foster their independence and ownership of their own health, rather than the ‘you need to come back every week forever’ model I see from so many instructors and clinicians.

  2. For me, theory has receded with the years. Too often I see ideas from ancient times (before everything was digitised) proposed as new.
    What remains is that we are all on a voyage and, as travellers, we can respect each others’ search.
    The research on old people shows that helping them makes them worse, but encouraging them to do for themselves reduces depression.
    So the question that we ask might be “what do you do to cope with that?”

  3. I’m not surprised at the results but still think reassurance is important but with a few caveats;
    Paul Salkovskis (big in the CBT world) talks about assurance being OK but when you get to reassurance it’s a problem. He’s referring to the health anxiety model in which reassurance is a safety seeking behaviour. As such repeated reassurance will just maintain the strongly held catastrophic belief as it doesn’t effectively challenge it. Patient feels better after reassurance but it’s short lived before they are back to their old beliefs. Challenging strongly held beliefs takes time through education +/- questioning techniques +/- behavioural experiments. Since it is difficult to establish in the first couple of sessions how strongly someone holds an unhelpful belief, and there is a time pressure, I go for reassurance which is by far the quickest way to change a not-so-strongly-held unhelpful belief. If those simple messages haven’t had the effect I require, i.e. I haven’t seen behaviour change which would indicate belief shift then I know I need to use the big guns to challenge the belief and I need to spend adequate time doing so. A couple of things I do to make the reassurance more effective; I always make sure it’s founded in an understanding of exactly what the patient thinks is wrong, no point delivering generic reassuring messages (usually), more effective to get to the nitty gritty of what they think is wrong and then make my reassurance more specific. Where possible I often end my reassurance with a question so that the patient has to engage cognitively and will be more likely to remember what we’ve been talking about and can therefore self reassure next time the unhelpful thought crops up. I like to keep in mind Louis Giffords concept of pink flags, the opposite of risk factors, so all the things that the patient has got going for them such as fitness, coping strategies they’ve used successfully before, and of course the fact that they’ve got a fabulous physio (followed by a wink!).

  4. Tamar Pincus says:

    Thanks Nick! We went in to this study very much from this point of view, but there reasons to at least take a pause for thought. The three studies that found empathy / listening/ creating rapport to be associated with worse outcomes did adjust for baseline factors, including mood. Mood is an insufficient marker for resilience, granted, but I think it would useful to focus on how we increase cognitive reassurance- information, education, to empower people to a better sense of control over their life / pain. This isn’t easy (as mentioned above) in the context of uncertainty.
    To my knowledge there is no evidence on whether people ‘learn’ better from a neutral delivery or an empathic delivery. Coia and Morley argue for the neutral, based on a CBT approach. My understanding is that this approach increases the cognitive work the patient has to do, and the ownership over the information gained. Could be a nice little experiment here to test out the contradictory theories.

  5. An interesting topic for sure! I wonder if this is partly tied up with resilience? If we take it that patients care seek because they are not coping, either with the presenting symptoms or the level of psychosocial distress in their lives at the time, or both, then reassurance is about regaining a sense of coherence, meaning and manageability even if psychosocial chaos abounds. Reducing the level of threat one way or another.
    Porges’s theory of Neuroception and the need of social interaction as a pathway to decreasing SNS activity might suggest empathy remains important throughout all clinical encounters? If the mind indeed is focussed on what is wrong, lacking or threatening then the ‘internal narrative’ may hear a statement of positive affective reassurance but if it considers the threat to significant it is unlikely to believe it or act on it. As Rick Hanson would say ‘the mind is like Teflon for positive experience’
    I don’t think we necessarily need to try and identify sub groups given our success to date with non specific musculoskeletal pain, extending compassion and empathy to all comers would be a great start.
    There has been some work recently here in New Zealand on the decreasing capacity of primary care practitioners to feel compassion towards their patients, the use of Mindfulness seemingly promotes empathy and compassion in those who practice it. Perhaps MBI’s may also help the patient as well as the practitioner to decrease the influence of the cognitive component of the presentation? Decreased reactivity in the autonomic nervous system and an increase in parasympathetic tone is another recognised aspect of Mindfulness based interventions.

  6. I like the comments above – as usual I have a spare 5c in my pocket. I do my best NOT to give reassurance of the kind that says “I think you’ll get better” – though some people seem to demand it as a precondition to trusting whatever is being done. After having tested everything out in an initial treatment/consultation, I do often say thinks like – “quite often I’ve seen this kind of problem get better in maybe X sessions”, but the main reason for even doing this is a statement that “I think I can help, so therefore I think you would benefit from giving this a go” in a strong enough manner so that they do give it a go. I would never out of choice use reassurance that was flabby – i.e.has no congruency behind it, so I really have to believe what I’m saying. Though as I said earlier occasionally reassurance is demanded so strongly that it’s difficult not to let some mis-spoken woffle out of the mouth. This latter is not reassuring, because on a very fundamental level the patient will know it is just placatory and contains no real expectation of success. so from my pov, although there are psychological factors in the patient, the congruency and real non-faked and non-inflated professional confidence that backs up any statement is the real measure of any reassurance. And here is the rub, because given that level of professional confidence, was I giving a truly accurate assessment, or did my (misplaced) confidence invoke a placebo effect? Not sure how that would be unpicked.

  7. Just a quick post to say how much I have enjoyed this discussion – you all bring such wisdom and insight. This is EXACTLY the sort of chatter we hope to facilitate and we are grateful to you for making the time and sharing with the world!

  8. Mike Caruso says:

    I find this discussion very interesting.
    I see about 30 – 40 injured workers (IW) with chronic pain each year. The reassurances they received from their primary care physician did not help much. Neither did the reassurance from their PT, surgeon or pain specialist (opioids), or the reassurance they received from the programmer of their spinal cord stimulator.
    I view these folks (IWs) as people with movement problems and pacing problems. I aim to shift their thinking from that of an anatomical problem, which they can not do nothing about, to problem of tissue mobility and motor control, with a whole host of things that can contribute.
    I use Tom Meyers (Anatomy Trains) understanding… ‘anger at your boss changes your posture, your movement patterns, and overtime tissue mobility”. I show them Anatomy Train posters on the wall and point out how the shoulder is attached to the opposite knee. I then use Gray Cook’s functional movement patterns, and others, as target patterns to aim for using many varied exercises.
    Importantly for this discussion, they work in a group with individual programs so they can reassure each other along the way. They are also given home work.
    I am a manual therapist but now ‘never’ perform a manual therapy until they they have made the shift to thinking about their problem as something they can change.
    Empathy is important all along the way. But what I withdraw slowly is what I contribute to their self-care. When a flair-up occurs I point out how they managed in the past and ask ‘what do you think?’ Empathy is not doing more to the patient…. perhaps that is more likely sympathy (which is about ourselves, not the patient).
    There is more on empathy from Marshal Rosenberg (www.cnvc.org)
    Thank you raising this discussion.
    All the best

    Matt Reply:

    Interesting post Mike, thanks for sharing on this topic, which clearly I also find “very interesting”…
    I too draw from non-violent communication skills in my approach – which also dovetails off the mind-body facilitation models I’m trained in – present moment, what’s happening here and now… greatly assists the patient/provider relationship dynamics (most of the time!)

  9. Hi Maxi, thanks for posting,
    As for social engagement, Porges work is very good and highly recommended (for those interested the entire body of published work has been edited and compiled into a textbook/hardcover format) – and informs some of what I mean by social engagement –
    The core (and solid, I might add) science that underlies his “theory” (I personally do not consider it to just be theory but understand the nature of the scientific world…) represents an organizing principle for the a portion of the social engagement brain, particularly an organizing principle for affective/emotional expression, as well as our orienting and neuroceptive systems (environment/afferent/sensory). However, his published work is largely focused on phylogenetic and developmental aspects of the bulbar and associated cranial nerve relationships and relation to evolutionary autonomic processes, and is not particularly heavy on the cortical aspects… and also to some degree leaves out more thorough discussion on non-pharyngeal/branchial arch derivatives to all that is involved in neurophysiological correlates of social, inter-personal engagement. I do believe the theory also provides a organizing principle for aspects of the neurophysiological correlates of empathy and compassion. Like all things in the scientific world, the theory is not without criticism, but my observation is that is very minor. I would also include additional mechanisms and related cortical connectivity between visual processing, (additional) auditory mechanisms, vestibular, sensory and motor/movement, and neurovisceral integration (see Thayer and Lane, 2000, Journal of Affective Disorders, for latter) in a more expansive view of the neurophysiology of “social engagement”.
    BTW, many on this forum have read, but would recommend Dr. Quinter’s articles on empathy, and pain medicine models, and am reminded of this statement from the 2008 Pain Medicine article summary: “We suggest a project in which pain may be apprehended in the clinical encounter, through the engagement of two autonomous self-referential beings in the inter-subjective or so-called ‘third space’, from which new therapeutic possibilities can arise.”
    I’m sure John is reading – might want to post a link to your articles if so, I don’t have the other references handy, but recall one was on the extinction of empathy in pain treatment…

  10. Interesting, post, but yikes, that Coia/Morley model: “In a radical statement, the authors called for clinicians to abstain from their ubiquitous attempts to deliver affective reassurance, as they may cause more harm than good.”
    The final questions posed are good food for thought with 3 and 4 being most important to my perspective. Question 1, for me personally in clinical practice, is not so difficult – but I think it would take me too long to present my awareness around that question in this forum except for the following: I would say that there is never anything wrong in telling the truth (in truly vague/unclear etiologies) and saying “I don’t know for sure,” (,”but I will do everything I can to help you change the physiology of this”…); but would also say there is a point at which we MUST attempt to navigate activating the biology of belief through clever yet truthful, well chosen semantics as this is often a missing piece of the puzzle (and in my professional opinion, an oft missing area of cortical activation)… We must assist in activating the biology of empowerment.

    I certainly believe cognitive “reassurance” (good education) is critical – but believe that a clinic visit without genuine empathy, compassion, is a huge concern – and published research around physicians affect reinforces that (e.g.’s: Del Canale 2012, Hojat 2011, and also recommend overview from Dr Riess: http://jama.jamanetwork.com/article.aspx?articleid=186692). But it seems you either have it, or don’t (empathy)? – can it be taught? And much of it is through the social engagement system, and operating unconsciously through facial expressivity, posturing and body language, not just in a validating statement like “oh, wow, that sounds very difficult, I can’t imagine”…
    And why would it matter if the empathic presence (again, not just verbal content) is present at the start or the end of the session – I disagree that holding such a space would deter the ability to integrate a cognitive model – in fact would suggest the opposite – when the social engagement system is ‘engaged’ it facilitates higher cognitive processing – if the clinician’s affect is rushed, cold/overly “clinical”, etc. it will support a more sympathetic state in the patient and reduce the ability to integrate such content. In this, we are assisting not just in activating the biology of empowerment, but balancing it the activation of the biology of nurturance, and effective combination for healing (in my opinion!). So, to question 4, can this be taught in a way that changes the combination of a clinician’s presence/affect and expertise in clinical delivery? I would say yes, but as always, “needs more research” 🙂 (see Riess et al 2012 Empathy Training for Resident Physicians: A Randomized Controlled Trial of a Neuroscience-Informed Curriculum)
    Love this forum, thanks for providing it –

    Maxi Reply:

    I agree with Matt on a couple of points re: question #3. Here are my thoughts:
    1) I do not know that you can apply a ‘reassurance model’ based on the stage of the consultation (e.g. beginning, middle, end). This is because the stage of consultation is a linear description of the clinical encounter whereas patient responses during an encounter may not be linear – there can be ups and downs at various spots for various reasons. For example, I may provide some affective reassurance, which helps settle a patient, then move onto providing cognitive reassurance, which may activate (i.e. increase anxiety) the patient depending on the nature of the information I am providing and the meaning it has for the patient. I then may need to engage more affectively in response to the patient.
    2) I agree with Matt’s comments re: social engagement system (Matt, I think you are speaking about Stephen Porges’ polyvagal theory, correct? Porges, 2003, 2009) This theory assumes that higher cognitive process are challenged when an individual has higher anxiety or fear. As humans, we have an ancient autonomic system that regulates this sympathetic activation via social engagement (e.g. eye contact, tone of voice, proxemics). So, in order for an individual to digest information via cognitive reassurance, they have to be in a less anxious/more regulated state. Therefore, affective reassurance would be a necessary component depending on the patient and their state. That would mean that a ‘certain degree of empathy’ may be necessary at all points of the interaction. At least, there must be an intent to be empathetic as needed, both verbally and non-verbally.
    3) I do understand the sense of ‘being neutral’. I think that we have to be mindful of how we are conveying information – therapists are the grounding. If we are gushing over the patient without adequately challenging their beliefs or are perpetuating them re-living their pain story by being affectively reassuring, we may not be effective. I believe our job is to listen and make every effort to understand and support the patient with the intent of providing resource to help them jump out of the pain story groundhog day. This would likely involve a balancing act between affective and cognitive reassurance.

    Thanks for opportunity to post on a great topic.

    Porges SW. Social engagement and attachment: A phylogenetic perspective (2003). Ann NY Acad Sci 1008: 31-47.

    Porges SW. The polyvagal theory: New insights into adaptive reactions of the autonomic nervous system (2009). Cleve Clin J Med Apr; 76 Suppl 2:S86-90.

    Tamar Pincus Reply:

    Many great points raised in the previous two postings. The area is incredibly under-researched, and we are hampered by poor measurement of the complex interactions that take place during the consultation. There is only one study, to my knowledge, that looked at practitioners’ disclosing their uncertainty to patients, and that study suggests that patients are less satisfied when this happens. I also agree that the ‘early-middle-late’ model of consultation stages is probably lacking in validity. I’ve also been trying to get my head around the evidence for anxiety interfering with cognitive function in relation to Coia and Morley’s model. Perhaps there is a U-shaped curve, in which high anxiety interferes with learning, but a certain level of anxiety? Worry? is needed in order to engage with learning.

    Matt Reply:

    Hi Tamar,
    I would clarify – expressing uncertainty about diagnosis versus uncertainty on clinical course/outcome, expectations –
    How often do we truly “know” the cause, even with a coded diagnosis –
    And I believe slapping labels on often reinforces disability via mind-body processes –
    I actually think its more helpful to have “non specific low back pain”, or just “low back pain” come in as diagnosis than to have 5 different MRI generated codes….
    When I have patients come in all focused on “herniated disc” or “pinched nerve” or “osteoarthritis”, and consistent with so much of the work done by the Body in Mind group, I immediately remind people to not attach to that and relay the studies of “normal” MRI findings in pain-free populations.
    It’s a balancing act – because sometimes diagnosis does drive treatment, but more often, it does not, and actually reinforces disability.
    Anyway, what I was referring to when saying “I’m not sure, but I am here to help” (however I worded that) wasn’t specifically referring to diagnostics, rather was referring to that we don’t really always know “why” this is happening, how quickly you might respond to therapy, and that we can move towards “what can I learn from this experience and what am I going to do to move into healing…” (and I use a mind-body-spirit paradigm when I refer to “healing”)…

    Also, as for “social engagement”, I would add that perception of safety is a definite (with its biological correlates) precursor to engaging in relationship, including the provider-patient interaction, and empathy (affective reassurance in this posts semantics) is a definite requirement for that (in my opinion), and should be present alongside very well-crafted education based on mind-body science (cognitive reassurance).

    Lloyd Reply:

    Wot You said Matt
    almost exactly but I reckon you said it better, in fact im a little jealous.

  11. Doug Scown says:

    As you’d be aware there is always an issue if a dignosis is absent. People do not have a condition called NSCLBP for example. They are merely grouped with other individuals who experience ‘back pain’ which lasts beyond an expected tissue healing time. Therefore how can a clinican even approach reassurance? What are you reassuring them of? How can you set reasonable expectations if you assume ‘they’re all basically suffering the same ‘thing’? Are they predisposed to chronic pain. Is there a family history of such disorders (migraine, fibromyalgia or other centralised processing disorders? Were they in fact accurately diagnosed or is it just assumed that the refering specialst got it right (they often do not) There is too much emphasis on this classification system of chronic back pain and too little on educating clinicians regarding accurate diagnoses. My 5c worth. Thankyou for your efforts in promoting evidence informed practice.

    Tamar Pincus Reply:

    Thanks, Doug! In fact, we have come to the same conclusion, and have now started a strand of research into the impact that uncertainty about diagnosis may have on patients’ outcomes. Of interest, it seems that patients’ perceptions are more important than practitioners’ actions. One of the studies we reviewed audio-taped consultations for independent coding, but also measured patients’ perceptions of the consultation. Only the latter were linked to outcomes (Stewart et al., 2000). With this in mind, we are now asking patients not only whether they received a diagnosis / explanation, but also whether they agree with it. We will keep you posted on whether these factors predict outcomes!