The Hamilton scale at 66: how a Leeds rating sheet became depression's yardstick

When a trial reports that a new antidepressant beat placebo by "two points," the clinician reading it usually knows, without being told, which ruler was used. For more than half a century that ruler has been the Hamilton Depression Rating Scale, a one-page instrument published in 1960 by a Leeds psychiatrist who wanted something more consistent than clinical impression. Its history explains much of how depression came to be measured.

A scale for a new kind of drug

Max Hamilton was born in Germany in 1912 and brought to England as a child. He trained in medicine in London and in psychiatry at the Maudsley Hospital, and joined the University of Leeds in the 1950s. He had an unusual (for a psychiatrist) interest in statistics, and later served as President of the British Psychological Society.

The timing mattered. Iproniazid and imipramine had just shown that depression might respond to medication, and investigators needed a way to say how much a patient had improved. Hamilton's paper, "A rating scale for depression," appeared in the Journal of Neurology, Neurosurgery and Psychiatry in 1960. It set out 21 items rated by a clinician after an interview. Seventeen of them were summed into a severity score. The other four (such as diurnal variation and paranoid or obsessional symptoms) were meant to describe the type of depression, not its severity.

Hamilton was explicit that this was not a diagnostic test. It was meant for patients already judged to have a depressive illness, to quantify how ill they were and to track their course. He revised the scale in 1967, and the 17-item version became the standard form.

From one man's sheet to the trial standard

No regulator decreed the HAM-D the yardstick. It spread because psychiatrists running drug trials in the 1960s and 1970s adopted it. The 1960 paper is among the most cited in psychiatry. Over time, the 17-item total became the usual primary endpoint in placebo-controlled antidepressant studies, with entry often requiring a score of around 20 or more, "response" defined as a 50 percent drop, and "remission" as a score of 7 or less.

In a 2014 editorial marking the scale's legacy, David Nutt asked whether it had sped up or slowed down antidepressant discovery. A scale that defines the target also shapes which drugs look effective.

The case against

The fullest critique came in 2004, when R. Michael Bagby and colleagues reviewed decades of psychometric studies in the American Journal of Psychiatry under the title "has the gold standard become a lead weight?" Their main points:

• It is not one thing. Factor analyses and item-response studies find several dimensions, not a single depression construct, so the total score adds unlike items together.
• Uneven weighting. Items carry different maximum scores. Insomnia alone has three items (early, middle and late), worth up to 6 points. Anxiety, gastrointestinal, general somatic and genital symptoms, hypochondriasis and weight loss add more. A sedating drug can lower the total without touching mood, and medically ill patients can score high for physical reasons.
• Weak item reliability and poor fit with modern criteria. Some DSM features, such as hypersomnia and weight gain, are missing, while items like loss of insight are not diagnostic criteria at all.

Bagby's group concluded the instrument could not be fixed by revision and called for a new gold standard. Defenders replied that Hamilton built a clinical index of illness burden, not a pure measure of one trait.

The successors

The best-known challenger came early. In 1979 Stuart Montgomery and Marie Åsberg published "A new depression scale designed to be sensitive to change" in the British Journal of Psychiatry. They rated 17 symptoms in 64 patients taking four different antidepressants and kept the 10 that changed most with treatment. The resulting MADRS separated responders from non-responders better than the Hamilton in their sample, and it leaves out most somatic items. It is now a common primary endpoint, particularly in trials of newer and rapid-acting agents.

The PHQ-9, published by Kurt Kroenke, Robert Spitzer and Janet Williams in 2001, solved a different problem. It is self-reported, maps onto the nine DSM criteria, and takes a couple of minutes. In about 6,000 primary care and obstetrics-gynecology patients, a score of 10 or more had 88 percent sensitivity and 88 percent specificity for major depression against a structured interview. That made routine measurement possible in busy clinics.

Hamilton died in 1988. His scale has outlived many of the drugs it first tested, and knowing how it was built, and for whom, is still part of reading the antidepressant literature well.