Home / Resources / Psychological Measures / EPDS
Edinburgh Postnatal Depression Scale:
Scoring and Interpretation
The Edinburgh Postnatal Depression Scale (EPDS) is a ten-item self-report questionnaire that screens for depressive symptoms after childbirth, scored from 0 to 30 over a seven-day recall period (Cox et al., 1987). This article sets out EPDS scoring for raw response positions and for pre-keyed data, the trade-off between the thresholds of 10, 11 and 13, the separate safety step for the self-harm item, and SPSS and R code that reverses only when it should.
What the EPDS Measures
The EPDS was developed to screen for depressive symptoms after childbirth. It assesses recent affective and cognitive symptoms over a seven-day recall period and avoids some common somatic postpartum complaints (Cox et al., 1987).
It is extensively studied in US and international maternal-health research, in antenatal and postpartum samples alike. Researchers use the sum as a symptom severity indicator or classify positive screens with prespecified thresholds, and an individual participant data meta-analysis showed how sensitivity and specificity move with the selected cut-off (Levis et al., 2020). Two avoidable mistakes follow from this: reversing items that were already keyed, and reading a threshold as a diagnosis.
Structure of the EPDS
- Items: ten respondent-rated questions, each converted with an item-specific 0 to 3 severity key into a 0 to 30 total.
- Reverse keying: on forms where response position is coded 0, 1, 2, 3 from top to bottom, questions 3 and 5 to 10 require reverse ordering of these positional codes, while questions 1, 2 and 4 do not.
- Pre-keyed data: some electronic datasets already contain severity-keyed 0 to 3 scores. In that case no further reversal may be applied (HEALthy Brain and Child Development Study, n.d.).
- Subscales: none in the original standard score, which is a single depression-screening total. Other subscale models are research adaptations.
- Safety item: the tenth item concerns possible self-harm and requires an independent safety response process, whatever the total score.
The instrument itself is available from the original British Journal of Psychiatry publication and instrument documentation used in perinatal research.
How to Score the EPDS
- Identify whether epds1 to epds10 are raw response positions or already keyed severity values. Do not run a second reversal on pre-keyed scores.
- If raw positions are coded 0 to 3 in printed top-to-bottom response order, reverse items 3 and 5 to 10 (new value = 3 minus raw), leave items 1, 2 and 4 unchanged, and sum all ten severity-scored items (0 to 30).
- The code below assumes raw positional coding. If your inputs already hold severity scores, use the original variables in the sum without reversing.
- Standard administration requests a complete response, so the example requires all ten scored values. Document any missing-item approach.
* ONLY for questionnaire response POSITION coded 0,1,2,3 from top to bottom.
* If data are already keyed by severity, do NOT reverse them again.
DO REPEAT src=epds1 epds2 epds3 epds4 epds5 epds6 epds7 epds8 epds9 epds10
/dst=epds1_sc epds2_sc epds3_sc epds4_sc epds5_sc epds6_sc epds7_sc epds8_sc epds9_sc epds10_sc.
COMPUTE dst=src.
END REPEAT.
COMPUTE epds3_sc=3-epds3.
COMPUTE epds5_sc=3-epds5.
COMPUTE epds6_sc=3-epds6.
COMPUTE epds7_sc=3-epds7.
COMPUTE epds8_sc=3-epds8.
COMPUTE epds9_sc=3-epds9.
COMPUTE epds10_sc=3-epds10.
COMPUTE epds_total=SUM.10(epds1_sc, epds2_sc, epds3_sc, epds4_sc, epds5_sc, epds6_sc, epds7_sc, epds8_sc, epds9_sc, epds10_sc).
EXECUTE.# ONLY if original response positions are coded 0-3 in printed order.
# If already keyed by severity, skip reverse conversion.
cols <- paste0("epds",1:10)
x <- data[cols]
rev <- c(3,5,6,7,8,9,10)
x[rev] <- lapply(x[rev],function(z) 3-z)
data$epds_total <- rowSums(x,na.rm=FALSE)Interpreting EPDS Scores
| Score band | Interpretation |
|---|---|
| 0 to 9 | Below commonly used screening thresholds; does not exclude depression |
| 10 to 12 | Intermediate zone relative to the common cut-offs of 10 and 13 |
| 13 to 30 | At or above a commonly used high-specificity screening threshold |
Source: Cox et al. (1987), who proposed 13 or more for probable depression and 10 or more for possible depression; Levis et al. (2020). These descriptive rows are not validated diagnostic severity bands.
Higher EPDS scores indicate more depressive symptoms, but screening thresholds are context-sensitive. Against semistructured interviews, the individual participant data meta-analysis of Levis et al. (2020) reported sensitivity and specificity of 0.81 and 0.88 at 11 or more, and 0.66 and 0.95 at 13 or more. A cut-off of 11 therefore gave the better combined balance in that synthesis, while 13 was more specific but less sensitive. Neither threshold confirms a diagnosis.
Any nonzero self-harm response requires a study or clinical safety protocol, irrespective of total or cut-off, and local escalation criteria should be established by qualified clinical teams. A negative screen is not proof that depression is absent.
Psychometric Properties
| Property | Published finding | Source |
|---|---|---|
| Original instrument | Designed as a 10-item community screening scale for postpartum depressive symptoms; at 13 or more, sensitivity 86% and specificity 78% against RDC major depression (n = 84); split-half reliability .88, standardized α .87 | Cox et al. (1987) |
| Diagnostic accuracy, 11 or more | Sensitivity .81 and specificity .88 versus semistructured interviews in the IPD meta-analysis | Levis et al. (2020) |
| Diagnostic accuracy, 13 or more | Sensitivity .66 and specificity .95 in the same reference-standard subset | Levis et al. (2020) |
| Meta-analysis size | 58 contributing studies; 15,557 participants overall | Levis et al. (2020) |
| Factor structure | Different factor solutions proposed; standard clinical use remains one overall score | Cox et al. (1987); subsequent factor-structure research |
These are source-specific results and cannot be assumed to hold as reliability or accuracy estimates in a new sample.
Using the EPDS in a Study
In cohorts and intervention trials, use the continuous 0 to 30 EPDS score where appropriate, and add screening prevalences only when the chosen threshold has a clear rationale. In diagnostic-accuracy research, compare the EPDS with a structured interview and present uncertainty around sensitivity and specificity, as covered in Sensitivity & Specificity (Diagnostic Accuracy). Perinatal timing, language and setting should be prespecified. Because accuracy estimates differ by study and reference standard, pooled evidence of the kind in Individual Participant Data (IPD) Meta-Analysis is best read alongside the cut-off you chose.
Estimate reliability in the current sample with Reliability Analysis (Cronbach’s Alpha), and describe how the self-harm item and missing responses were handled. For maternal mental-health interventions, a baseline-adjusted comparison such as ANCOVA, or Generalized Estimating Equations (GEE) for repeated measures, can analyze change without assuming that all screen categories have equally spaced clinical severity.
Related field: Nursing
Reporting the EPDS
Depressive symptoms were measured using the ten-item Edinburgh Postnatal Depression Scale (EPDS; Cox et al., 1987), which assesses experiences over the prior seven days. Responses were converted using the version-specific 0 to 3 severity key and summed to yield a total from 0 to 30. Screening was defined using the prespecified threshold of [XX], and any endorsement of the self-harm item was managed according to [protocol].
References
Cox, J. L., Holden, J. M., & Sagovsky, R. (1987). Detection of postnatal depression: Development of the 10-item Edinburgh Postnatal Depression Scale. British Journal of Psychiatry, 150, 782-786. https://doi.org/10.1192/bjp.150.6.782
HEALthy Brain and Child Development Study (HBCD). (n.d.). Postnatal Depression: EPDS scoring documentation. https://docs.hbcdstudy.org/latest/instruments/pregexp/mh/epds/
Levis, B., Negeri, Z., Sun, Y., Benedetti, A., & Thombs, B. D., for the DEPRESSD EPDS Group. (2020). Accuracy of the Edinburgh Postnatal Depression Scale (EPDS) for screening to detect major depression among pregnant and postpartum women: Systematic review and meta-analysis of individual participant data. BMJ, 371, m4022. https://doi.org/10.1136/bmj.m4022
Frequently Asked Questions
Which EPDS items are reverse-scored?
When the dataset records printed response positions 0 to 3 from top to bottom, items 3 and 5 to 10 require reversal. If the items already contain severity scores, do not reverse them again, and check the codebook before you run any syntax.
Is 13 or more the only valid EPDS cut-off?
No. A large individual participant data meta-analysis found that 11 or more balanced sensitivity and specificity more closely, while 13 or more emphasized specificity. Choose and justify the threshold in advance for your population.
What about a nonzero response to the self-harm item?
It warrants a prespecified safety assessment or escalation process regardless of the total score, set by qualified clinical teams.
Can EPDS scores be analyzed as continuous?
Yes, subject to an appropriate research design and statistical assumptions. Screening categories should be used only with a stated rationale.
NEXT STEP
Need EPDS Keying
and Totals Verified?
OnlineSPSS is a statistical consultancy for researchers. Share your design and dataset, and a statistically trained Research Coordinator reviews it. The scope and one fixed project fee are confirmed before work begins.
No Commitment Required
Strictly Confidential
Fixed Project Fee
In-House PhD-Level Statisticians
Same-Day Response
Serving Researchers Since 2014