|
Abstract
|
Richard Armitage reports a 3·9% increase in antidepressant prescribing in England during April to September, 2020, compared with the same period a year earlier. Armitage attributes this change to the COVID-19 pandemic: "These data suggest the predictions that mental health would be severely affected by COVID-19 were correct. They also indicate that this increased burden of mental ill-health is being disproportionately managed by pharmacological treatments in primary care…".
We have serious concerns about this analysis for a number of reasons.
First, it ignores all pre-existing trends in antidepressant prescribing. As previously reported in The Lancet Psychiatry, our team provides the OpenPrescribing.net service, which provides an openly accessible explorer for the same underlying National Health Service (NHS) data cited by Armitage. Data from OpenPrescribing show the ongoing upward trend in antidepressant prescribing over the preceding 5 years (appendix). There is no obvious change in this trend during the period of the pandemic. Indeed, the 2019 to 2020 increase of 3·9% is actually smaller than previous years: the same comparison between 2018 and 2019 yields a 6·1% increase in prescribing. It is therefore highly misleading to attribute this increase to the pandemic.
Second, there are no methods or code available for Armitage's analysis. It is a tremendous amount of work to convert the raw data files cited by Armitage into a finished analysis. Our team has invested resources over many years to create online tools and open source analytical code that transparently describe the complete pipeline from raw data to completed analyses, in line with best practice for all research. The figures from Armitage seem broadly correct but there is no analytical code, nor even a brief description of the methods used. Is the analysis restricted to the individual drug classes such as SSRIs? Are tricyclic antidepressant drugs, which are more commonly used for pain, included or were all antidepressant drugs included? There is no way to evaluate these data. It would therefore be helpful if Armitage could share the code for his data management and analysis.
We strongly agree with Holmes and colleagues about the immediate need to establish and implement research priorities and funding to address the effect of the pandemic on mental health services. Opaque analyses that apparently misinterpret longstanding trends in the data risk distracting patients, the media, and commissioners of health services. The COVID-19 pandemic
Language: en |