National Institute for Health & Clinical Excellence

Bipolar Affective Disorder

Stakeholder Comments

Please use this form for submitting your comments to the Institute.

  1. Please put each new comment in a new row.
  2. Please do not paste other tables into this table, as your comments could get lost – type directly into this table.
  3. Please fill in the document you are commenting on in the first column, for example, the Full version, the NICE version or the Information for the Public. Please insert the section number in the 2nd column (see examples). If your comment relates to the document as a whole, please put ‘general’ in this column. Please refer to section numbers and not page numbers.


D B Double & Joanna Moncrieff




Indicate if you are referring to the

Full version, the NICE version, or the Information for the Public.

Section number

Indicate section number or ‘general’ if your comment relates to the whole document



Please insert each new comment in a new row.




We have found the deadline very tight for this second consultation. We do not think we have had enough time to give your guideline full consideration.



The history of the term bipolar disorder is potentially misleading. It suggests that it became recognised that psychosis was not required for the diagnosis of bipolar disorder. No reference is made to this theory. We think the modern concept of bipolar disorder could be said to have originated independently in the 1960s in three monographs by Carl Perris, Jules Angst and George Winokur et al. Of course the term was used previously by Kleist and Leonhard, from which the modern concept moved on by for example not mentioning cycloid psychosis. It also included unipolar mania in Leonhard's terms.

The qualitative difference between bipolar and unipolar disorders arises out of evidence for different morbidity risks among first degree relatives. Kleist and Leonhard had assumed that the genetic loading in bipolar disorders was greater than in monopolar disorders. The modern studies showed a significant bipolar morbid risk was present in the families of bipolar probands and not in the families of their unipolar depressive counterparts. However, the three monographs vary widely in their conclusions as to the overall morbidity in first degree relatives.

No mention is made of the advantages of the term manic-depressive illness over the term bipolar disorder. For example, it is generally easier for patients to appreciate its meaning, because it is much more specific about the phases of the illness. Moreover, although you say you broadly agree with our concerns about the overinclusive use of the term bipolar disorder, you have done nothing to correct this bias and imbalance. We are not (on this point) arguing about whether your recommendations are evidenced-based. The point is about the unreliable use of the diagnosis. You could at least make reference to the need for reliability studies on this issue. You may even know some. It is surprising to us that research was so dominated by issues about the reliability of psychiatric diagnosis in the 1970s, for example, but since the introduction of DSM-IV in particular, perhaps, this research has dried up. Perhaps this is why you seem to think that the term bipolar disorder is "cut and dried".

Although you have included further discussion of the term bipolar spectrum you are not specific enough in making clear that prescription (and other treatments) may well be beyond the evidence if a specific diagnosis of bipolar I (and possibly bipolar II) is not made.

Also, although you make reference in the section on bipolar spectrum to the distinction between bipolar I and II, you do not make clear at this point that ICD10 does not recognise bipolar II as a separate category. Is NICE driven more by American or international classification of diseases? Surely it should be the later. In which case



We thought the whole point of NICE guidelines was to assess the evidence. There is no attempt to do so on the point of whether there is evidence that biological factors are more primary than psychosocial factors in aetiology. You could make this point more clearly. It may well be the current view that biological factors are primary. But what is the NICE view?

Even though discussion of aetiology includes psychosocial factors as well as biological factors, there is at least a potential bias in the guideline that needs to be corrected. One potential way of correcting tis would be to discus psychosocial factor first, followed by biological factors. You could even make explicit that you are doing this to correct an imbalance.



Why are American criteria mentioned before ICD? As mentioned previously, surely NICE should be driven more by ICD than DSM.

We have also looked at the second paragraph on ICD10. This does not explicitly state that ICD10 does not recognise bipolar II as a separate category, and it should. It is not explicit enough to state that ICD10 differs from DSMIV because a single episode of mania does not merit a diagnosis until another mood episode is experienced. We think most people reading this will not appreciate its significance.



We would be interested to know how common the use of the dagger for marking drugs that do not have a marketing authorisation is in other guidleines. Is it used more commonly in this bipolar guideline than other mental health or indeed other physical disorder guidelines? It should be possible to measure this easily and maybe we should do a joint publication in the journals, if this is an original idea on our part. If more common in this guideline, the reason should be fully explained in the guideline.



You say that you have included text on the potential bias in trials of acute mania because of restricted selection and high drop out rates. Where? We have even searched the document using the search terms "restricted selection" and "drop out" and found no entries. (Please note our general comment at the beginning of this document, but we would like clarification on this issue.)



See our comments on 7.3 above. Surely if there is some acknowledgement of the methodological limitations of studies in acute mania we would find it in this section, and there does not seem to be any qualification in the claims because of methodological implications here. Quite simply, the claims made for treatment in acute mania are overstated.


We did not receive a satisfactory answer to our question about why recommendations about benzodiazepines have crept in to a guideline about bipolar disorder. We can at least now see an explicit reference that there is no RCT data for benzodiazepines (whereas we do not remember seeing this in the first draft - see general comment at beginning of this document). However, there still does not seem to be any answer to our question of why benzodiazepines seem to have a place in the guideline in the management of behavioural disturbance and insomnia with atypical neuroletics, but not traditional neuroleptics. Surely, following the logic of this should raise the profile of traditional neuroleptics in the guideline, because they avoid the use of "combination" treatment. And anyway, where is the evidence?



Reference to Ciprani et al (2005) on the suicide risk of lithium may not give adequate emphasis to evidence that lithium can increase suicide rates.



We could make more of the problems of discontinuation effects in the interpretation of clinical trials but we think we will be ignored. (Also see general comment at beginning of document). We believe it is a general problem with the assessment of evidence in NICE guidelines that methodological problems are not given sufficient weight. For example, just because the problem of unblinding/unmasking is very difficult to deal with in the interpretation of the data does not mean that the potential bias should not be fully acknowledged. Small effect sizes could easily be due to an exaggerated placebo effect due to unblinding/unmasking. (See our comments on the Depression guideline and the subsequent reference in an article on antidepressants by Moncrieff and Kirsch (2005)).



When some of us trained, the rule of thumb for starting prophylactic treatment was 2-3 episodes in 5 years. We think this "rule" may be as good as yours. You are not explicit enough that this recommendation in the guideline is without evidence, more so than most recommendations you make at the level of consensus and expert opinion in NICE guidelines.


Atypical antipsychotics are recommended as first line treatments. We are curious as to how this recommendation was reached. The only RCT evidence relates to olanzapine. This does not demonstrate that olanzapine is superior to lithium or valproate, or indeed to conventional antipsychotics, which have not been trialled in bipolar disorder prophylaxis. The placebo controlled RCT demonstrates a clear discontinuation effect with relapses concentrated in the first 50 days, and a high drop out rate meaning that conclusions are based on assumptions about how to statistically deal with dropouts. We are concerned that this emphasis on atypicals is premature, and that the guideline may appear to support Eli Lilly's marketing efforts rather than reflecting the evidence.


Why should not a shorter period such as 2 years be sufficient to allow evaluation of the long term benefits? We think patients see this as sufficiently long-term and that their view should be sought on this issue. (See our comments on 8.2.4 and the general comment at the beginning of the document - we are prepared to give more time to this issue if you are interested.) There is little doubt that there is evidence of a vulnerability to relapse created through being on medication. We think this is due to psychosocial factors. There is therefore an indication that psychosocial management may be more likely to lead to recovery than maintenance on medication. (Again, we need more time to elaborate this point if you are really interested - and probably a few more publications in the literature!).



Who is it that was incorrect in the first draft - you or us? We object to being told we made an error on the evidence of the issue of effective lithium levels without making it clear where we have made an error. As we have mentioned several times, we have not had sufficient time to consider all of the guideline in sufficient detail. Could we at least know where we are supposed to have made an error when we suggested that there is evidence that lower levels of lithium than 0.6 (or 0.8 for those that have relapsed) may be effective? We are not the sort of psychiatrists that get too hung up about this sort of thing, but we are interested in the evidence, for what it is worth!


It's all very well you saying your own views about lithium monitoring, but is this supposed to be a consensus or expert view, because as far as we can see there are several experts in the field? One of them should be Schou, who is responsible for having got the drug on to the market. Do you know what his views about lithium monitoring are?



Whatever does lithium presents unique challenges mean? Overzealous recommendations about lithium discontinuation are clearly not determined by the evidence. Are there no dicsontinuation problems with other psychotropic medication. We think you should reflect on the history of the poor recognition of discontintinuation problems by doctors. Lithium is not the only drug that we have found that patients have problems discontinuing.

Also we do not appear to have received a reply to whether the figure of 2 years of prophylaxis conflicts with the 5 recommended elsewhere in the guideline.



The guideline should say that discontinuation problems are most commonly reported with paroxetine, not that they are more common. Reliance on yellow card reporting is notoriously difficult.

We should be interested to know how much of an overlap you think there is between tricylic and SSRI discontinuation symptoms. Again, we think it better to record that these symptoms have been reported, rather than imply they occur inevitably. Some patients do not experience discontinuation problems - perhaps particularly those that do not get psychologically dependent on their medication. We welcome the recent recommendation elsewhere that doctors should warn patients of the risk of discontinuation problems when starting medication. This recommendation should be repeated here.



We remain concerned about drug company stakeholder involvement. Certainly they have more resources as stakeholders in the guideline than other stakeholders. We remain convinced that a more sceptical guideline would be produced without their involvement and that NICE should give serious consideration to this issue