Withdrawal reactions

Interruption of ongoing treatment with various psychotropic medicines sometimes is followed by clinically significant withdrawal reactions within hours or days, as well as later increases in morbidity, relapses or recurrences of the illnesses being treated. Early withdrawal reactions commonly follow discontinuation of benzodiazepines^{Reference Fluyau, Revadigar and Monobianco2} as well as some antidepressants, particularly selective serotonin reuptake inhibitors.^{Reference Fava, Gatti, Belaise, Guidi and Offidani3} They include rapidly emerging physiological syndromes as well as new neuropsychiatric symptoms including sensory changes, anxiety and agitation, all of which can persist for weeks.^{Reference Rosenbaum, Fava, Hoog, Ascroft and Krebs4,Reference Chouinard and Chouinard5} Adverse clinical effects of discontinuing psychotropic drugs also can arise in changing from active drug to placebo in clinical trials.^{Reference Baldessarini and Tondo6,Reference Récalt and Cohen7}

Pharmacodynamic mechanisms underlying discontinuation syndromes

Mechanisms underlying early reactions to withdrawal of various psychotropic agents, and particularly later relapses or recurrences, remain uncertain. It is likely that complex pharmacodynamic adaptations to long-term drug treatment are involved.^{Reference Baldessarini and Tondo6,Reference Blier and Trembley8,Reference Baldessarini9} Long-term exposure to antidepressant, antipsychotic, anxiolytic, mood stabilising and other psychotropic drugs leads to neuropharmacological adaptations that include changes in postsynaptic receptor and autoreceptor sensitivity, neurotransmitter synthesis and release, and various downstream molecular and genetic mechanisms in multiple brain systems.^{Reference Blier and Trembley8,Reference Baldessarini9} Agents with long elimination half-lives appear to carry lower risk of withdrawal effects, including fluoxetine among antidepressants and long-acting antipsychotics.^{Reference Baldessarini and Tondo6} Effects of adaptative changes during prolonged drug exposure evidently become manifest as clinically apparent neurobehavioural responses when treatment is removed. Restoring the withdrawn medicine, even temporarily and at lower doses, sometimes may reduce the clinical manifestations of withdrawal reactions.^{Reference Baldessarini and Tondo6}

Later adverse clinical effects of drug discontinuation

Of great clinical importance is the phenomenon of increased risk of earlier relapses (re-emergence of recent, and possibly not fully remitted, episodes) or recurrences (new episodes) of illnesses following discontinuation of their treatment, possibly related to a ‘rebound’ effect associated with removal of effects of a drug.^{Reference Franks, Macritchie, Mahmood and Young10} Such outcomes appear to represent more than a return of illness without treatment, and include an iatrogenic component arising from treatment discontinuation itself as a significant physiological and psychological stressor. That is, stopping treatment evidently is not equivalent to being untreated. That earlier relapses or recurrences are a consequence of treatment discontinuation itself is supported by differences in morbidity in patients after discontinuing a treatment versus their spontaneous illness course before starting treatment.^{Reference Suppes, Baldessarini, Faedda, Tondo and Tohen11} Such adverse clinical responses have been reported in association with discontinuing lithium or other treatments in bipolar disorder,^{Reference Öhlund, Ott, Berqvist, Oja, Lundqvist and Sandlund1,Reference Viguera, Whitfield, Baldessarini, Newport, Stowe and Cohen12–Reference Baldessarini, Tondo, Floris and Rudas14} antipsychotics in schizophrenia,^{Reference Viguera, Baldessarini, Hegarty, Van Kammen and Tohen15} and antidepressants in patients with major depression.^{Reference Baldessarini, Tondo, Ghiani and Lepri16} For selective serotonin reuptake inhibitors these effects may last for periods varying between a few weeks to a year or more.^{Reference Horowitz and Taylor17}

Pregnancy is commonly associated with interruption of ongoing medicinal treatments – often abruptly, typically driven by fear of teratogenic or toxic effects on the fetus and associated liability. Discontinuing lithium maintenance treatment in women with bipolar disorder has been associated with 60% risk of a new episode of illness within 9 months in age-matched women, whether pregnant or not, and the risk was higher following abrupt or rapid (<15 days) discontinuation.^{Reference Viguera, Nonacs, Cohen, Tondo, Murray and Baldessarini18} In addition, risk of new, mainly depressive or mixed, episodes of bipolar disorder was markedly increased after stopping various mood stabilisers with pregnancy, especially in the first trimester.^{Reference Viguera, Whitfield, Baldessarini, Newport, Stowe and Cohen12} Increased recurrences even among women with bipolar disorder who had remained stable without treatment in the year prior to pregnancy suggest that pregnancy itself is an additional stressor.

Effects of discontinuation rate

We have reported on adverse clinical effects of discontinuing long-term treatment with antidepressants and antipsychotics as well as lithium.^{Reference Baldessarini and Tondo6,Reference Baldessarini9,Reference Viguera, Whitfield, Baldessarini, Newport, Stowe and Cohen12–Reference Baldessarini, Tondo, Floris and Rudas14,Reference Viguera, Nonacs, Cohen, Tondo, Murray and Baldessarini18}  Such effects may also occur in patients with bipolar disorder following discontinuation of maintenance treatments other than with lithum.^{Reference Horowitz and Taylor17} Of particular interest, abrupt or rapid (1–14 days) versus slower discontinuation of lithium was consistently associated with earlier recurrences of illness and increased risk or suicide attempts or fatalities in patients with bipolar disorder.^{Reference Faedda, Tondo, Baldessarini, Suppes and Tohen13,Reference Baldessarini, Tondo, Floris and Rudas14} Similar risks also followed rapid discontinuation of antipsychotic drugs in patients with psychotic disorders,^{Reference Baldessarini, Tondo, Ghiani and Lepri16} and antidepressants in those with major depression or other disorders.^{Reference Baldessarini, Tondo, Ghiani and Lepri16,Reference Khan, Musgnung, Ramey, Messig, Buckley and Ninan19} The impact of abrupt or rapid treatment discontinuation in these studies and that of Öhlund et al ^{Reference Öhlund, Ott, Berqvist, Oja, Lundqvist and Sandlund1} was most apparent in initial months, with more similar risk × time rate functions thereafter that paralleled risk × time before treatment.^{Reference Öhlund, Ott, Berqvist, Oja, Lundqvist and Sandlund1} Hospital admission increased non-significantly after stopping lithium abruptly or rapidly in the new study by Öhlund et al,^{Reference Öhlund, Ott, Berqvist, Oja, Lundqvist and Sandlund1} although contrasts were limited by including conditions other than BD-I and late spontaneous recurrences probably unrelated to treatment discontinuation. The preponderance of available evidence supports the view that gradual discontinuation of most psychotropic drugs over at least several weeks is likely to reduce early risks of relapses^{Reference Baldessarini and Tondo6,Reference Baldessarini9,Reference Suppes, Baldessarini, Faedda, Tondo and Tohen11–Reference Baldessarini, Tondo, Ghiani and Lepri16,Reference Horowitz and Taylor17} as well as risk or severity of early withdrawal reactions^{Reference Baldessarini and Tondo6,Reference Baldessarini9,Reference Horowitz and Taylor17,Reference Khan, Musgnung, Ramey, Messig, Buckley and Ninan19,Reference Fava, Bernardi, Tomba and Rafanelli20} and should be considered a preferred clinical practice.

Implications of treatment discontinuation for therapeutics research

In addition to its clinical importance, treatment discontinuation has important implications for the design, conduct and interpretation of clinical treatment trials. Instead of relying on long-term, randomised controlled trials, which are challenging to carry out, it has become common to study effects of treatment discontinuation following initial clinical improvement in randomised trials for acute psychiatric illnesses. Typically, such designs involve removing an apparently effective treatment, often relatively rapidly, to a placebo-control condition, often when patients may not be fully recovered from the index episode of illness.^{Reference Baldessarini, Lau, Sim, Sum and Sim21} Such circumstances make scientific interpretation of findings uncertain and also may raise ethical issues.

More generally, complications often arise in modern therapeutic trials in relation to previous treatments, to which most contemporary research participants have been exposed up to trial entry. Terminating preceding treatments can lead to both discontinuation and carry-over effects. That is, benefits of prior treatments can continue into the early days after randomisation to a new treatment. In addition, drug discontinuation effects can contaminate the outcome of placebo-control conditions, exaggerate comparisons with responses to a new active treatment, and complicate assessment of adverse effects.^{Reference Baldessarini9,Reference Gallagher, Strzinek, Cheng, Ausmanas, Asti and Seljan22} Trial designs involving treatment discontinuation currently are prevalent, and can, potentially misleadingly, be considered to offer evidence of long-term treatment efficacy. In such trials, additional bias is introduced by ‘enrichment,’ or allowing only responders to a particular experimental treatment to continue into a more prolonged treatment-discontinuation trial phase.