We wish to reply to the editorial by Looi et al. We are glad that the authors agree that hyperbolic tapering is the most sensible approach to deprescribing antidepressants. However, they minimise evidence of overprescribing and misrepresent aspects of the Maudsley Deprescribing Guidelines. The editorial erroneously characterises the Maudsley Deprescribing Guidelines as advocating ‘population-level deprescribing of antidepressants’, – deprescribing even when against the wishes of the patient (as in cases where ‘patients maintain a preference for medication’) or where patients ‘are benefiting’. In fact, the book offers guidance on how to stop antidepressants in a safe manner when this is in the best interests of the patient and when they agree to the process through shared decision-making. It does not advocate deprescribing for those benefiting from medication or those who wish to continue it.

We are confused by the perspective of the authors on overprescribing. Early in their piece they state that ‘we agree that not all antidepressant prescribing is warranted’. Yet they then dispute that many people who take antidepressants do not have an evidence-based reason to continue them. They even seem to regard the issue of people continuing on antidepressants when they are no longer needed as trivial (‘the same can be said of almost all medications’). They seem also to shrug their shoulders at the issue of unnecessary prescribing when they say ‘it is difficult to determine to what extent this is true and thus far there is no substantive research to support this statement’. Finally, they imply that because ‘it is difficult to withdraw treatment’ it may be easier to continue antidepressants indefinitely.

We do not share this view of unnecessary prescribing. Australia is currently the fourth highest prescriber of antidepressants amongst OECD countries, with two antidepressants in the top 10 most prescribed drugs in the country. Most of this increase in prescribing is due to longer use  and it has been suggested that this longer use is explained in part by the difficulty patients have in stopping these drugs. Patient-centred, high-quality care requires that we use medications when they are, on balance, beneficial and cease them when they are no longer necessary. Sensible deprescribing aims to ensure high-quality use of medications by carefully balancing their harms and benefits; factors which can change over time.

The authors’ view stands in contrast to consensus in the UK (where antidepressant prescribing is similar to that in Australia) that antidepressants are overprescribed. In England the NHS has identified reducing inappropriate antidepressants as a goal for the health system and has called for the establishment of services to help people safely stop, with initiatives already emerging.

The authors argue that it is difficult to determine to what extent overprescribing is occurs but dismiss three studies demonstrating that 30% or more of patients currently on antidepressants do not meet criteria for continuing. The first study examined clearly showed that 31% of patients do not meet criteria for ongoing treatment.  In the second study Looi et al. have overlooked (in Table 1 in Ambresin et al (2015)) that 34.6% of patients prescribed antidepressants long term (>2 years) did not meet criteria for a DSM-IV diagnosis of depression in the past year and the relevant figure for short-term treatment (<2 years) was 30.4%. As most guidelines suggest patients could consider stopping after 6–12 months of remission these observations are consistent with 30% of patients not meeting criteria for ongoing treatment. The third study does not make the measure of inappropriate prescribing as clear as the main paper from this RCT (an oversight on our part) which reports that 37% of patients were judged to be on inappropriate antidepressants by their GP and so entered into a deprescribing trial.

Other studies also confirm these estimates. An Australian study found that in 32.6% of patients, antidepressants were unlikely to be indicated for ongoing use. Furthermore, the ANTLER study referred to by the authors demonstrated that 44% of patients on long-term antidepressants could discontinue and stay well. The tapering in this study was less than optimal, and so the true proportion might be higher. Looi et al. incorrectly report that 53% of patients who stopped medication in this trial returned to medication. In fact, 53% of patients who stopped medication and relapsed returned to their medication. So it was still the case that 44% of patients on long term medication can safely discontinue their antidepressant without relapsing. Overall, there seems to be clear convergence that more than 30% of patients on antidepressants have no evidence-based reason for remaining on their antidepressant and could consider deprescribing.

Although the authors state it is difficult to determine when ongoing medication is appropriate (a state of affairs that should invite concern rather than complacency) a recent study has operationalised somewhat conservative criteria for which patients no longer need antidepressants and could benefit from deprescribing, articulating 25 indicators for overprescribing through a consensus process. 

There are other indicators that not all prescribing is appropriate. As the authors state most guidelines advocate prescribing antidepressants for the more severe end of the spectrum but the vast majority of people who are prescribed antidepressants have mild to moderate conditions. The authors also neglect to note that rising rates of prescribing have been found to be caused by longer duration of treatment for each patient, rather than increased frequency of initiation.  As they state, there is no evidence for treatment beyond 3 years and now the average duration of antidepressant use is 4 years. The authors also state that patient preference is revealed by rising prescription rates but this is untrue. There is no need to speculate on motivation when meta-analysis finds that 75% of patients prefer psychological to pharmacological treatment when asked. 

Looi et al. downplay the potential harms of antidepressants. They state that side effects ‘usually….are transient’ but many side effects of antidepressants persist in the long term. For example, in one cohort on antidepressants for more than 12 months, many patients reported several side effects. Many symptoms are more common in longer-term use than in short-term regulatory trials. For example, 40–80% of patients demonstrate treatment-emergent sexual dysfunction on meta-analysis when treated with common antidepressants. Indeed, the TGA has recently warned clinicians that some adverse effects of antidepressants, such as profound sexual dysfunction, may persist for ‘weeks to years’ even after cessation of the drug – called post-SSRI sexual dysfunction (PSSD). 

They minimise the physical health consequences in older people where studies find an excess all-cause mortality for every year of antidepressant use of 3.6% for SSRIs, and 4.4% for ‘other’ antidepressants. SSRIs have been found to increase risk of bleeding by at least 36% in meta-analysis, especially when used in combination with warfarin (where SSRIs cause an extra serious bleed in every 55 patients exposed), DOACs and NSAIDs.

The authors also downplay withdrawal risks. Withdrawal effects from antidepressants are now recognised to be common and can often be severe, with a recent flawed meta-analysis, despite focusing on short-term exposure and relying on spontaneous reporting, still finding that withdrawal effects are commonly severe. Looi et al. state that withdrawal effects are ‘short term’ but most now recognise that withdrawal effects can be long-lasting and they can be debilitating, leading to job loss, relationship breakdown and serious financial and personal consequences in some patients. As risk of withdrawal effects increases with long term use, this is one reason to be cautious of unnecessary long term exposure.

The authors also over-state the evidence for the long-term use of antidepressants for relapse prevention, when referring to the ANTLER trial and other similar discontinuation trials. In the ANTLER trial, patients well on antidepressants were randomised to continue or stop (tapered over 4–8 weeks) and relapse measured using depression scores. In this study relapse was not differentiated from withdrawal. Withdrawal effects (e.g. worsened mood, increased anxiety, insomnia, etc.) overlap with all items on the depression rating scale used in the study. These withdrawal effects register on depression scales and therefore artificially inflate rates of detected relapse in the discontinuation group, tending to inflate the apparent relapse prevention properties of antidepressants. Indeed, the ANTLER study detected a spike in withdrawal effects in the discontinued group which persisted for 9 months after discontinuation but it did not make efforts to exclude these effects from detection of relapse. Therefore this study, like other discontinuation studies, does not present robust evidence that antidepressants prevent relapse because of confounding by withdrawal effects.

Overall, Looi et al. overlooked clear evidence that a substantial proportion of patients no longer meet criteria for ongoing use of antidepressants. They minimised the potential harms of antidepressants (including withdrawal effects and serious physical health consequences) especially in older people, over-emphasised their benefits in the long term and seem complacent when regarding patients being exposed unnecessarily to the harms of no longer needed antidepressants. We would rather encourage a patient-centred approach guided by shared decision-making to carefully evaluate the benefits and harms for each patient. If this balance is found unfavourable – as is likely in at least a third of patients – and the patient is willing to stop we should help them to safely decrease their medication using the latest evidence, such as is summarised in the Maudsley Deprescribing Guidelines.