Hospital Anxiety and Depression Scale:
Scoring and Interpretation

The HADS was developed to screen for anxious and depressive symptoms in patients attending general medical outpatient clinics (Zigmond & Snaith, 1983). It gives little weight to somatic symptoms, which makes it useful when physical illness could otherwise inflate a psychological score.

Its two seven-item scales, HADS-A and HADS-D, are read separately. In an updated literature review, Bjelland et al. (2002) found that most factor analyses supported an anxiety factor and a depression factor, although accuracy estimates varied between studies. The scale is therefore a symptom screen for medical and community samples, and its scores are not diagnoses.

• Items: fourteen, alternating between the two domains in the standard order.
• HADS-A (anxiety): items 1, 3, 5, 7, 9, 11 and 13.
• HADS-D (depression): items 2, 4, 6, 8, 10, 12 and 14.
• Response format: each item has four item-specific response options, which are converted to keyed severity values from 0 to 3. Some items run in the reverse direction, so the printed order of the options is not the score.
• Scores: two subscale totals, each from 0 to 21. A combined 0 to 42 total appears in some research, but the standard interpretation applies to each subscale on its own.
• Recall period: the past week.
• Reverse-keyed positions: these depend on how the options were encoded in your dataset, so use the scoring key supplied with the form your participants completed.

1. Convert every recorded answer to its keyed severity value from 0 to 3, using the scoring key of the exact form administered. Do not assume that the numeric position of an answer already equals its score.
2. Name the seven anxiety items hads1, hads3 and so on to hads13, and the seven depression items hads2, hads4 and so on to hads14, holding the keyed values.
3. Add each set of seven to obtain hads_a_total and hads_d_total, both ranging from 0 to 21.
4. Choose a missing-item rule that the instrument’s documentation supports. The cited HADS sources set no universal prorating policy, so the code below requires all seven keyed values on each subscale. State any alternative rule in the methods section.

A higher HADS-A score means more anxiety symptoms and a higher HADS-D score means more depressive symptoms. Across studies a subscale threshold of 8 or more often balances sensitivity and specificity, at about .80 in the review by Bjelland et al. (2002), yet it remains a screening convention and not a diagnosis. Performance shifts with the case definition and the medical population, so a positive screen calls for clinical assessment.

In most studies HADS-A and HADS-D work best as separate continuous variables, as outcomes or as predictors. The two subscales correlate, so report that association with a Correlation Analysis (Pearson & Spearman), check collinearity before placing both in one regression model, and ask whether the research question needs both. A binary screen at 8 or more can be added as a prespecified, supplementary descriptive outcome.

Report reliability for each subscale in the medical or community sample you actually studied. When a study evaluates the threshold against a diagnostic interview, a ROC Curve Analysis shows how sensitivity and specificity trade off across the whole range of scores. Be explicit about data entry and recoding, since that is where most HADS errors arise. Power depends on the primary analysis and design, so no sample-size minimum can be inferred from the scale alone.