Mood Disorder Questionnaire (MDQ): Questions, Scoring, Accuracy

The Mood Disorder Questionnaire (MDQ) is a one-page self-report screen for bipolar disorder that asks whether a person has ever had a period of manic or hypomanic symptoms. It was published in the American Journal of Psychiatry in 2000 by Robert Hirschfeld and colleagues, who drew its 13 yes/no items from DSM-IV criteria and clinical experience^[2].

It is one of several short questionnaires clinics use to flag possible mental illness for a fuller assessment. The form states that it takes less than 5 minutes and is "designed for screening purposes only and is not to be used as a diagnostic tool"^[1].

The MDQ targets bipolar spectrum disorders and screens for a lifetime history of a manic or hypomanic syndrome, not for current mood^[2]. A person who is depressed today can still screen positive if earlier episodes fit the pattern.

Two of its questions still use the older term "manic-depressive illness"^[1]. NIMH notes that this was the former name for bipolar disorder^[7].

Question 1 opens with "Has there ever been a period of time when you were not your usual self and..." and lists 13 experiences, each answered yes or no^[1]. The items cover elevated or irritable mood, self-confidence, sleep, speech, racing thoughts, distractibility, energy, activity, sociability, sex drive, risk-taking and spending^[1].

Four shorter questions follow^[1]:

• Question 2 asks whether several of the checked experiences happened during the same period of time.
• Question 3 asks how much of a problem they caused, such as trouble working, family, money or legal troubles, or fights, rated as no, minor, moderate or serious problem.
• Question 4 asks whether any blood relative has had manic-depressive illness or bipolar disorder.
• Question 5 asks whether a health professional has ever said the person has manic-depressive illness or bipolar disorder.

Question 2 checks whether symptoms co-occurred and question 3 rates functional impairment, and the scoring rule uses both^[1]. That is what separates the MDQ from a plain checklist of mania symptoms.

Under the standard rule, a screen is positive only when all three conditions are met^[1]:

1. Yes to 7 or more of the 13 items in question 1^[1].
2. Yes to question 2, meaning several symptoms happened in the same period.
3. "Moderate problem" or "serious problem" on question 3.

Questions 4 and 5, on family history and past diagnosis, are not part of the scoring rule^[1]. The form directs a physician, nurse or medical staff assistant to score it after the patient completes it^[1].

The cutoff of 7 comes from the original validation^[2]. Of 198 patients at five outpatient clinics that mainly treat mood disorders, 109 (55%) received a bipolar spectrum diagnosis in a blind telephone SCID interview^[2].

At 7 or more items, the MDQ had a sensitivity of 0.73 and a specificity of 0.90^[2]. Higher cutoffs lost sensitivity without a clear gain in specificity, and lower cutoffs cost considerable specificity^[2].

Put another way, about 7 in 10 people with a bipolar spectrum disorder in those clinics screened positive, and 9 in 10 without one screened negative^[2].

Several studies have tested what happens when the impairment requirement in question 3 is relaxed^[5][8]. In a Rhode Island study of 480 psychiatric outpatients, 52 (10.4%) of whom had bipolar disorder, standard scoring caught 63.5% of bipolar cases^[5].

When moderate or serious impairment was not required, sensitivity rose to 75.0% and specificity fell from 84.8% to 78.5%^[5]. The authors concluded that standard scoring gave inadequate sensitivity in that setting^[5].

A French validation found the same pattern for bipolar II disorder. Standard scoring identified 52.4% of patients with bipolar II, and lowering the level of impairment required raised that to 76.2%^[8].

A 2011 review of nearly two dozen MDQ reports found that lowering the threshold markedly improved sensitivity with only a modest loss of specificity^[4]. Studies searching for the single best symptom cutoff, however, produced inconsistent results^[4].

A 2015 meta-analysis suggested that the cutoff be adjusted to raise sensitivity when the MDQ is used with depressed patients who have no previous bipolar diagnosis^[3]. The adolescent version uses a lower threshold of 5 items on the parent-report form^[6].

Pooling 21 studies, a 2015 meta-analysis in Depression and Anxiety found a summary sensitivity of 0.62 and specificity of 0.85 at a cutoff of 7^[3]. In the 11 studies that included both patients with bipolar disorder and patients with unipolar depression, sensitivity was 0.76 and specificity 0.81^[3].

The weak spot is the group screening is meant to help. In 6 studies that excluded patients already known to have bipolar disorder, sensitivity fell to 0.37, with specificity of 0.88^[3].

Subtype matters as much as setting. Across studies in the 2011 review, the MDQ detected 66.3% of bipolar I cases but 38.6% of bipolar II cases^[4]. The French validation showed the same gap, 90.3% for bipolar I against 52.4% for bipolar II^[8].

In a U.S. general population sample of 695 adults interviewed by telephone, the MDQ flagged 28.1% of those with bipolar I or II disorder and correctly cleared 97.2% of those without it^[10]. The authors argued that the diagnostic interview's limited reliability in community samples meant measured sensitivity could not exceed about 0.4^[10].

For a side-by-side look at other bipolar screens and online quizzes, see the bipolar test overview.

The original English MDQ had a Cronbach's alpha of 0.90, a measure of how consistently its items track the same construct^[2]. The French translation reached 0.89^[8].

When French patients repeated the questionnaire a month later, agreement was a kappa of 0.79, with similar stability for bipolar I (0.75) and bipolar II (0.77)^[8].

The 2015 meta-analysis found no significant difference in accuracy between studies from Eastern and Western countries after adjusting for clinical differences between samples^[3].

The 32-item Hypomania Checklist (HCL-32) is another self-report screen for bipolar disorder^[11]. A 2019 meta-analysis pooled 9 studies in which the same patients completed both questionnaires^[11].

Sensitivity was similar, 82% for the HCL-32 and 80% for the MDQ, but the MDQ had higher specificity, 70% against 57%^[11]. In those studies, the HCL-32 caught a similar share of cases while flagging more people who did not have bipolar disorder^[11].

The authors concluded that both tools have acceptable psychometric properties for identifying bipolar disorder^[11].

The adolescent version (MDQ-A)

The MDQ-A adapts the questionnaire for adolescents and has self-report and parent-report forms. In a 2006 study of 104 adolescent psychiatric outpatients and their parents, the parent form at 5 or more items had a sensitivity of 0.72 and specificity of 0.81, better than the adolescent self-report^[6].

A positive MDQ means bipolar disorder is worth assessing, not that it is present. In the Rhode Island outpatient study, 33.7% of positive screens had bipolar disorder under standard scoring, and 29.8% when impairment was not required^[5].

A negative screen was more informative in that sample: negative predictive value was 95.0% under standard scoring and 96.3% without the impairment requirement^[5]. Predictive values depend on how common the disorder is in the group screened, and 10.4% of that sample had bipolar disorder^[5].

The case for screening rests on missed diagnoses. In a University of Texas family medicine clinic, 21.3% of adults taking an antidepressant for depression screened positive, and nearly two thirds of them had never received a bipolar diagnosis^[9].

Evidence on whether screening improves outcomes is thin and dated. A 2011 review concluded that routine clinical use of the MDQ could not be recommended, because no studies had weighed the benefits of better detection against the costs of overdiagnosis^[4].

The diagnosis itself comes from a clinical evaluation; the form states that an accurate diagnosis can only be made through a personal evaluation by a doctor^[1]. The steps are covered in how bipolar disorder is diagnosed.