The article reviewed here is ‘The Relationship Between Antipsychotic Medication Adherence and Patient Outcomes Among Individuals Diagnosed with Bipolar Disorder: A Retrospective Study’ by Maureen Lage and Mariam Hassan and freely available here. The abstract reads
‘Conclusion: Patients with lower antipsychotic adherence were at greater risk of hospitalizations and ER visits. Thus, any efforts to increase adherence, even in small increments, can be helpful in decreasing these risks‘
So the first question I would ask here is ‘do the results from this study justify these conclusions?’. There were a number of points about this study that I didn’t understand completely and may have misinterpreted but i’ll just work through them.
Firstly one of the primary outcome measures was something the authors referred to as the ‘Medication Possession Ratio’ (MPR). From the description
‘was utilized as a measure of adherence, and was calculated as the number of unique days an antipsychotic medication was prescribed in the postindex period divided by the number of days in the same period‘
it looked as though the MPR was a marker of the prescriber’s prescribing behaviour rather than the patient’s adherence to medication. It does not necessarily follow that if the medication is being repeatedly prescribed that this is because it is being taken. Indeed this may be a marker of the system in which the prescriber operates (e.g. use of automated reminders may predict prescribing behaviour). Thus I would argue that the MPR does not necessarily measure adherence.
The next question I have is why were Emergency ER visits (for all causes) and hospital admissions (for all causes) used as the primary outcome measures? Surely for a study of this type, the primary outcome measures that would make most sense would be Emergency ER visits for mental health related conditions or admissions for acute exacerbations of Bipolar Disorder. Nevertheless these primary outcome measures produce some interesting findings – namely that the higher the MPR (which have been interpreted as adherence), the lower is the risk of admission or ER visits. Indeed an analysis of the suggested primary outcome measures didn’t produce such interesting results as for the actual primary outcome measures.
So the next point is about the Stepwise Regression Analysis. There is some more information on this over at Wikipedia as well as the video explanation on YouTube below (quite a nice explanation if you ignore the ?Black and Decker drill in the background)
At the time of writing, the Wikipedia article contains some criticisms of Stepwise Regression Analysis including what they describe as a ‘false discovery rate’ (and also the difficulties associated with multiple comparisons). Watching the video makes it clear that this is an iterative process and the point is that it’s difficult to interpret the p-values since they are dependent on previous steps in this iterative process. Also it is clear from the video that there are several approaches – backwards, forwards and a combination of the two. I didn’t notice an elaboration of the approach used within the methods section and given that a large number of variables were being taken into consideration I wondered if this was an exploratory approach.
The authors have also state that
‘the risk of hospitalization or an ER visit with an accompanying diagnosis of bipolar disorder was also examined‘
However the data, analysis and results are not included although at the time of writing the submission policy for the journal states that it does not ‘restrict the length and quantity of data in a paper’. It would have been interesting therefore to see this data in more detail within the paper particularly as the conclusion in the abstract does not explicitly mention that hospitalisation is for all causes.
Within the conflicts of interests section of the paper, it is reported that one of the authors is an employee of Astra Zeneca and that there has been financial assistance from the company also. The journal is open access and at the time of writing operates using a model where the authors pay a fee for submitting their articles (with some exceptions) although the journal has strict stipulations on conflicts of interests and for authors who are employed within the pharmaceutical industry. I find it difficult to interpret conflict of interests. On the one hand, good faith should be assumed. On the other hand, it is reasonable to ask what effect the conflict of interests would have on the final paper. It would be useful to have a systematic process for examining this although I suspect that the publishing process would need to be modified in order for readers to gain a better understanding of what is happening ‘behind the scenes’. In this study for instance, a video of the authors undertaking the data analysis would be interesting to follow.
In conclusion, on the basis of the above, I wasn’t sure what to make of the study. The most interesting question would be the relationship of adherence to admission for relapse of Bipolar Disorder but the primary outcomes were MPR and hospital admissions for all causes and ER visits for all causes. Even if we could say that MPR correlated exactly with adherence, it is still difficult to interpret. Why would adherence to antipsychotics reduce admission for all causes which presumably would involve medical and surgical causes? Maybe there is an interesting finding here but I would be more persuaded if there is a closer examination of the relationship of MPR to adherence in another study.
You can listen to this post on Odiogo by clicking on this link (there may be a small delay between publishing of the blog article and the availability of the podcast).
If you have any comments, you can leave them below or alternatively e-mail firstname.lastname@example.org
The comments made here represent the opinions of the author and do not represent the profession or any body/organisation. The comments made here are not meant as a source of medical advice and those seeking medical advice are advised to consult with their own doctor. The author is not responsible for the contents of any external sites that are linked to in this blog.