Objectives The Psychopharmacology Algorithm Project at the Harvard South Shore Psychiatry Department has been preparing evidence-derived algorithms for choosing medication for psychiatric disorders for 25 years. They provide clinicians with syntheses of the best evidence for efficacy, effectiveness, safety, and cost of pharmaceuticals in a format that is more prescriptive than guidelines. The aim is to present very specific and very context sensitive best practice. The algorithms start with recommendations for medication-naïve patients. After that, suggestions for patients who had one failed trial are offered, and then we address deeper levels of treatment-resistance. Important comorbidity is considered. They are usable on cell phones. Practitioners can obtain an actionable recommendation in 1-2 minutes. We will demonstrate the major depression algorithm, most recently published in the February 2019 issue of Harvard Review of Psychiatry.

Method Small groups of clinicians create new algorithms or update older ones and then submit papers describing them for peer review in psychiatric journals. The review process involves reaching consensus with the expert reviewers who may not agree with all the interpretations of the evidence leading to the recommendations in the initial draft. This review process enhances the validity of the final version. After acceptance, the authors construct the on-line versions of the algorithms. Users see the entire flowchart and click on the node representing their patient’s current clinical status -- ranging from first use of medication to very treatment resistant, while considering the impact of various illness subtypes and comorbidities. The website (www.psychopharm.mobi) currently has 9 algorithms. All algorithms emphasize that they advise on choice of pharmacotherapy only, but non-medication approaches are valid, important, and may be preferred.

Results For most depressed outpatients, sertraline, escitalopram or bupropion are reasonable first choices. If there is no response, the prescriber has many choices for the second trial in this algorithm because there is no clear preference based on evidence, and there are many individual patient considerations and variations in patient preference to take into account. The prescriber and patient may decide to either switch (to one of the above options not previously tried, or to venlafaxine, or to a nutraceutical antidepressant such as St. John’s Wort or S-adenosylmethionine, or to transcranial magnetic stimulation), or to augment (with nutrients including l-methylfolate, or second-generation antipsychotics, or mirtazapine, or lithium or triiodothyronine). If there is no response to the second medication trial, the patient is considered to have a relatively medication-resistant depression. More recommendations follow. Comorbidities such as chronic pain, obsessive-compulsive disorder, attention-deficit hyperactivity disorder and posttraumatic stress disorder are considered.

Conclusions Utilization of this consultative tool for picking medication treatment for depression could help minimize unproductive variation in clinical care and improve clinical outcomes and produce remissions in shorter times and with fewer medication changes than with treatment as usual. Also, the algorithm encourages more cost-effective practice when generic options are recommended over expensive, brand-name products (when there is no apparent disadvantage in outcome or safety). Clinicians often overlook the large role their beliefs play in medication selection, ignoring placebo effects that contribute to their experience-based biases. These online tools provide evidence-supported ways of thinking that are available rapidly at the point of care in time to influence decision-making.