Sign Up
Login
Try for Free
Login
The EPDS (Edinburgh Postnatal Depression Scale) is a 10-item perinatal depression screen published by Cox, Holden, and Sagovsky in 1987 and scored 0 to 30 over the past seven days. Obstetric, midwifery, pediatric, and primary care teams use it through pregnancy and the postpartum year. It screens; it never diagnoses. This page covers how to document EPDS results defensibly, with a fictional sample note.
Patient self-report (may be administered verbally); scored and actioned by the clinician; partners screen as separate respondents
OB and midwifery practices, pediatric well-child teams, primary care, perinatal mental health programs, quality reviewers
A result line plus interpretation and follow-up in the note · patient completion under 5 minutes
Perinatal self-report screen (10 items scored 0 to 3, total 0 to 30, seven-day window)
Antenatal and postpartum depression screening, well-child-visit caregiver screening, serial rescreening, partner screening
RCPsych copyright, not public domain: own-use photocopying only; distribution or republication needs written permission
The Edinburgh Postnatal Depression Scale (its exact name; it is not the "Edinburgh Postpartum Depression Scale") is a 10-item self-report screen for perinatal depressive symptoms, published by Cox, Holden, and Sagovsky in the British Journal of Psychiatry in 1987. Each item is scored 0 to 3 for a total of 0 to 30, over the preceding seven days, a shorter window than the two weeks many pages misremember. It was built for pregnancy and the postpartum period and deliberately omits somatic items, because the developers found it too hard for a self-completed scale to separate the physical changes of childbearing from mood disorder. The result is a screen, never a diagnosis: a positive total warrants clinical assessment, and the scale does not detect bipolar disorder or postpartum psychosis at all.
Two facts frame the documentation. First, thresholds are tradeoffs, not truths. In the largest individual-participant meta-analysis (58 studies, 15,557 participants), a threshold of 10 or higher ran sensitivity 0.85 with specificity 0.84, 11 or higher gave the best combined balance at 0.81 and 0.88, and 13 or higher reached specificity 0.95 while missing roughly one-third of major-depression cases at sensitivity 0.66, with accuracy broadly similar in pregnancy and postpartum. The chart therefore names the threshold applied and its purpose. Second, copyright: the scale belongs to the Royal College of Psychiatrists and is not public domain. The College's terms allow an individual clinician or researcher to photocopy a complete, unmodified, attributed copy for their own use, and require written permission for distribution to others or republication in any medium, which is why this page describes the instrument without reproducing it and links out rather than embedding a form. The free PHQ-9 is the usual general-purpose comparison; the licensed BDI-II is the instrument whose somatic items make it a poor perinatal fit.
Obstetric and midwifery practices screen through pregnancy and postpartum: US convention (ACOG's 2023 guideline) calls for a validated instrument at the initial prenatal visit, later in pregnancy, and at postpartum visits, inside a system that can assess, treat, and follow up. Pediatric teams are the second major user, and the setting with the most documentation traps: AAP recommends caregiver screening at the 1-, 2-, 4-, and 6-month well-child visits, records it in the infant's chart because the screen is performed for the infant's benefit, and advises against opening a maternal chart solely for it. Australia is the jurisdiction that names the EPDS itself: the NHMRC-approved 2023 national guideline recommends universal EPDS screening early and later in pregnancy and again postpartum, and Medicare's obstetric items require a mental-health assessment to be offered and the result, or the decision not to participate, recorded. Canada runs the instructive contrast: the national task force recommends against universal instrument-based screening while the obstetric society and several provinces operationalize risk- and concern-based use, so a Canadian chart names whose guidance it follows. Positive item-10 responses route to a suicide risk assessment, brief triage findings sit in a risk screening note, and serial totals follow the conventions in the outcome measure note.
No US federal law prescribes an EPDS note format, and no jurisdiction makes a score diagnostic. What survives review is a record with the version, the timing anchor, the threshold and its purpose, a separate item-10 line, a modest interpretation, and a closed follow-up loop. Each element below carries the pitfall that most often undermines it.
Instrument, version, and mode. Chart the full 10-item RCPsych EPDS, or name the variant explicitly (an authorized translation with its language, the EPDS-US adaptation, a verbal administration, interpreter assistance, and whether completion was private). Pitfall: unverified forms. A 2003 journal letter documented circulating copies with altered item wording and missing preambles, so confirm the local form matches the original; and a modified or translated version charted as plain "EPDS" hides exactly what a reviewer needs to know.
The timing anchor. Record gestational age or weeks postpartum, and at a well-child visit, the infant's age and the respondent's relationship to the infant. Pitfall: a bare score with no perinatal timepoint, which makes serial comparison and guideline mapping impossible; and the seven-day recall window misdescribed as two weeks.
The total, with its threshold and purpose. Chart "x/30" plus the threshold applied and why: 10 or higher where sensitivity is the priority, 11 or higher as the pooled best balance, 13 or higher where specificity is intended, per the local program or guideline. Pitfall: "EPDS positive" with no number or threshold, or 13 treated as a diagnosis; at that threshold the pooled data miss about one-third of major-depression cases.
Item 10, separately, every time. Record the self-harm item as zero or nonzero on its own line, apart from the total. Any nonzero response gets a same-day, direct risk assessment regardless of the total, with the disposition matched to assessed acuity rather than an automatic transfer, documented in a suicide risk assessment and cross-referenced. Pitfall: arithmetic in either direction (a low total used to skip follow-up, or every nonzero response treated as an emergency), "denies SI" charted against a nonzero item without reconciling the two, and method or means detail written into a screening note.
Interpretation kept modest. A positive screen means elevated symptoms warranting assessment; among screen-positive women in the largest screening cohort, 22.6% of those assessed had bipolar disorders, and the scale cannot rule out postpartum psychosis, which is an emergency on clinical signs alone. The items 3 to 5 anxiety cluster is a conversation aid, not an anxiety diagnosis. Pitfall: a negative EPDS charted as ruling out depression, bipolar disorder, psychosis, or risk.
Action and the closed loop. Document the same-visit assessment or referral, the warm handoff and its receipt, treatment connection verified at follow-up, the rescreen date, and who owns each open item; US quality measures count follow-up within 30 days of a positive screen. Pitfall: a positive screen that ends at a passive referral order, which quality reviewers and plaintiffs both read the same way.
Whose chart, the billing line, and the declined screen. A caregiver screen at a well-child visit is recorded in the infant's chart with the minimum necessary (instrument, result, discussion, action), no maternal chart created for it, proxy visibility explained beforehand, and the caregiver pathway billed under the appropriate code (96161 for a caregiver instrument administered for the infant's benefit; 96127 is for the billed patient's own assessment). A declined screen is charted as offered and declined with any volunteered reason and a re-offer plan. Pitfall: a detailed maternal mental-health narrative in the infant's record, the two codes swapped, or a declined screen entered as a zero or negative result.
EPDS DOCUMENTATION BLOCK Date: [ ] Setting: [ ] Clinician: [ ] Instrument: full 10-item EPDS Version/language: [ ] Mode: [self-completed privately / verbal / interpreter] Timing: [gestational age / weeks postpartum / infant age + respondent relationship, for well-child screens] Total: [ ]/30 Threshold applied + purpose: [>=10 / >=11 / >=13 + local program or guideline] Item 10: [zero / nonzero -> same-day assessment documented at: ] Interpretation: [screen result only; not a diagnosis; bipolar and psychosis not excluded] Action: [assessment / referral + warm handoff / crisis response / rescreen date] Closed loop: [handoff received; treatment connection verified; follow-up owner + date] Chart + billing: [infant chart, minimum necessary (well-child) / own record; bare code noted] Declined: [offered + declined + reason if volunteered + re-offer plan; never entered as a score] Clinician signature / credentials: Date:
Free to use and share, no signup. The PDF includes a one-page cheat sheet with element-by-element pitfalls and a pre-sign checklist; the DOCX is the blank documentation block, ready to adapt. Neither reproduces the scale itself.
Scenario: a caregiver screen at a two-month well-child visit, positive at the program threshold with a zero item 10, documented in the infant's chart with the minimum necessary and a closed referral loop. All details are fictional.
Patient: Infant A.R., 8 weeks (caregiver screen: mother, 29) · Setting: Pediatric clinic, 2-month well-child visit · Clinician: S. Okafor, MD · Note date: 08/12/2026
Screening (infant chart entry): Caregiver postpartum depression screening offered per the 2-month well-child protocol and accepted after explanation that the result is recorded in the infant's chart. Full 10-item English EPDS self-completed privately by the infant's mother, 8 weeks postpartum. Total 12/30; clinic follow-up threshold is 10 or greater: screen positive. Item 10: zero. Result reviewed privately with the mother.
Discussion and action: Brief same-visit conversation: she describes low mood, poor sleep independent of the infant's schedule, and reduced enjoyment for about three weeks; no thoughts of self-harm on direct inquiry, consistent with the zero item 10. Screen interpreted as elevated symptoms warranting assessment, not a diagnosis. With her consent, warm handoff completed by phone during the visit to the perinatal behavioral health intake line; intake appointment scheduled 08/19/2026, and she agreed to notification of her own obstetric clinician, sent today.
Infant-focused plan: Infant growth, feeding, and development reviewed and normal today; caregiver support and infant-care resources provided. Rescreen at the 4-month visit per protocol, earlier if concerns emerge. Follow-up call by clinic nursing within one week to confirm the intake appointment was kept.
Administrative: Caregiver screening documented in the infant's record with the minimum necessary detail; the mother's own assessment and treatment will be documented by her receiving clinicians in her record. Caregiver-instrument administration reported under 96161 for this encounter per payer rules.
This sample is fictional and for educational purposes. It does not describe a real patient or record; the scores, threshold, dates, and details are invented to show documentation structure and are not clinical guidance.
Writing these after every session? BastionGPT drafts complete notes from bullets, dictation, or a transcript.
Generate a note from bulletsUnited States: screening is convention plus a patchwork of real law. USPSTF's 2023 Grade B recommendation covers depression screening in adults including pregnant and postpartum persons (CONVENTION, and a coverage driver), ACOG's 2023 guideline sets the initial-prenatal, later-pregnancy, and postpartum cadence with systems for assessment and follow-up (CONVENTION), and AAP guidance puts caregiver screening at the 1-, 2-, 4-, and 6-month well-child visits, in the infant's chart, billed under the caregiver-instrument pathway (CONVENTION and PAYER POLICY). Several states legislate directly (LAW): New Jersey requires screening around delivery and early postnatal visits; Illinois's Perinatal Mental Health Disorders Prevention and Treatment Act requires clinicians to invite and review a validated questionnaire, with pediatric invitations through the first year and consent before results go to the mother's clinician absent acute danger; California requires prenatal and postpartum clinicians to screen or offer screening; Arkansas requires facilitation of screening in the first six postpartum weeks and, distinctively, requires the record to state when a patient refused, with a liability consequence attached to that documentation; Louisiana conditions pediatric caregiver screening on consent and provides that screening alone creates no caregiver-clinician relationship. On the payer side, 12-month postpartum Medicaid coverage had been adopted by 49 states and DC as of July 2026 (Arkansas the holdout), a 2016 CMS bulletin lets state Medicaid programs cover the caregiver screen as a service to the child under EPSDT, and NCQA's prenatal and postpartum measures count a standardized screen and, for positives, follow-up within 30 days, with the EPDS threshold set at 10 or greater in measure logic, a quality data element, not a diagnosis (PAYER POLICY).
Canada and Australia pull in opposite directions, which is itself the documentation point. Canada's national task force conditionally recommended against universal instrument-based perinatal screening in 2022, while the SOGC's 2024 guideline operationalizes validated tools when symptoms, risk factors, or concern are present, and provincial programs (British Columbia, Alberta, Manitoba) run their own cadences (CONVENTION, divergent; name whose guidance the chart follows). Australia's NHMRC-approved 2023 guideline recommends universal EPDS screening early and later in pregnancy and at six to 12 weeks postpartum plus once more in the first year, treats 13 or greater as the further-assessment threshold with 10 to 12 generally repeated within two to four weeks, and requires immediate further assessment on any positive item 10 (CONVENTION); Medicare's obstetric and postnatal items (16590, 16591, 16407, with telehealth equivalents) require a mental-health assessment to be offered and the result, or the decision not to participate, recorded (PAYER POLICY), and New South Wales publishes authorized translations with cautions that lower thresholds may fit some migrant and refugee populations. If you or a client needs immediate support: call or text 988 (US), 9-8-8 (Canada), or Lifeline 13 11 14 (Australia).
The psychometrics and rights reward precision. The original validation involved 84 Edinburgh mothers at a 12/13 division (sensitivity 86%, specificity 78%); the pooled modern evidence supports threshold choice as a tradeoff and no universal pregnancy-versus-postpartum split; father and partner cutoffs in the literature run lower and vary with the target construct, so partners screen as separate respondents with their own thresholds, consent, and records. The scale is translated into more than 60 languages, but translation availability is not local validation, reviews have found uneven translation quality, and evidence in Indigenous populations is notably limited, so the chart names the exact version and language. The EPDS-US adaptation and the proposed EPDS-R revision are named versions, not silent substitutes. A nonzero item 10 is a strong relative signal with low absolute predictive value, which is why the rule is assess every nonzero response, neither ignore nor automatically hospitalize. On rights: the Edinburgh Postnatal Depression Scale is copyright of the Royal College of Psychiatrists; current terms permit individual own-use photocopying of complete, attributed copies and require written permission for distribution or republication in any medium. BastionGPT is not affiliated with the Royal College of Psychiatrists or the instrument authors. This page describes the instrument in original prose and reproduces no items, response options, or scoring materials.
The numbers behind these errors are precise. In the pooled individual-participant data, a threshold of 13 or higher runs sensitivity 0.66 against specificity 0.95, missing about one-third of major-depression cases, while 11 or higher gives the best combined balance; in a screening cohort of 10,000 postpartum women, 22.6% of assessed screen-positive women had bipolar disorders; a 2026 Danish cohort of 170,218 births found a nonzero item 10 carried an adjusted odds ratio of 10.91 for later recorded self-harm while the absolute rate stayed at 0.27%; and measured US screening rates remain in the single digits to low teens across health plans. The BastionGPT Clinical Advisory Board sees the same errors most often in EPDS documentation reviews:
BastionGPT is specifically trained, tuned, and clinically tested on behavioral health progress notes and screening documentation.
See how clinicians use it day to day on the AI therapy notes page.
Many BastionGPT users report saving more than 90 minutes per day on documentation.
HIPAA-compliant with a signed BAA on every plan. Your data is never used to train models. BastionGPT drafts, you review and sign.
Ten items scored 0 to 3 sum to a total of 0 to 30 over the past seven days. Thresholds are sensitivity-specificity tradeoffs, not diagnostic lines: in the pooled individual-participant data, 10 or higher runs 0.85 sensitivity with 0.84 specificity, 11 or higher gives the best combined balance, and 13 or higher reaches 0.95 specificity while missing roughly one-third of major-depression cases. Accuracy is broadly similar in pregnancy and postpartum. A positive screen means elevated symptoms warranting clinical assessment; it is never itself a diagnosis, and the defensible chart line names the total, the threshold applied, and its purpose.
Any nonzero response to the self-harm item requires a same-day, direct risk assessment regardless of the total score, a rule that traces to the instrument's developers and is operationalized by both US and Australian guidance. The disposition matches the assessed acuity: not every nonzero response is an emergency transfer, and a large 2026 cohort shows why both extremes fail (the relative association with later recorded self-harm was strong, an adjusted odds ratio near 11, while the absolute rate stayed near 0.27%). Record item 10 on its own line even when zero, document the assessment separately without method or means detail in the screening note, and reconcile any apparent contradiction rather than charting "denies SI" against a nonzero item.
The infant's chart, because the screen is performed for the infant's benefit: pediatric guidance says to record the instrument, result, discussion, and follow-up plan there, and not to open a maternal chart solely for the screening encounter. Keep it to the minimum necessary, since proxies with access to the child's record may see it, and explain that before administering. The caregiver-instrument pathway is billed under the infant with 96161; 96127 is for a brief assessment of the billed patient themselves, and some payers direct one pathway or the other, so the code follows the payer. Detailed maternal assessment and treatment belong in her own record with her own clinicians; consent rules for sharing results (explicit in Illinois and Louisiana law) are documented where they apply.
Not without permission. The scale is copyright of the Royal College of Psychiatrists and is not public domain: the current terms allow an individual clinician or researcher to photocopy a complete, unmodified copy for their own use with the full citation, and require written permission for copying or distribution to others or republication in any medium, which covers websites, downloadable libraries, and EHR embedding shared across users. The older attribution-only footer still circulating on legacy PDFs was superseded. Two practical checks: verify the form in local use matches the original (a published letter documented circulating copies with altered item wording), and route republication requests through the College's permissions process rather than relying on what other sites appear to get away with.
Choose by purpose and program, then document the choice. A threshold of 10 or higher favors sensitivity and is what US quality-measure logic counts as positive; 11 or higher maximized the combined balance in the pooled data; 13 or higher favors specificity and is Australia's further-assessment threshold, with scores of 10 to 12 generally repeated within two to four weeks there. The pooled evidence found broadly similar accuracy in pregnancy and postpartum, so there is no universal timing split, and population matters more: authorized translations may carry locally validated thresholds, and lower thresholds may fit some migrant and refugee populations. What fails review is not the choice of cutoff; it is a chart that never says which one was applied.
Offering partner screening is supported (Australia's guideline recommends it), but the maternal threshold does not transfer. Father cutoffs in the literature run lower and vary with the target: one study found 5/6 useful for broader distress, another found above 10 best for major depression, a systematic review found candidate cutoffs from 7 through 10, and measurement-invariance work shows the scale does not function identically across parents. Screen each partner as a separate respondent with their own consent, their own documented threshold and rationale, and their own referral pathway, and do not place one partner's detailed mental-health information in the other's record or in the infant's chart beyond the minimum necessary.
A positive screen is not a diagnosis, and it is not even specific to unipolar depression: in the largest screening cohort, 22.6% of assessed screen-positive women had bipolar disorders, which changes treatment materially. A negative screen rules nothing out, and the scale does not detect postpartum psychosis at all; acute confusion, mania, delusions, hallucinations, or rapidly changing behavior is an emergency on clinical grounds regardless of any total. The items 3 to 5 anxiety cluster can flag anxiety content for conversation, but the evidence does not support it as a stand-alone anxiety screen; where anxiety is the question, use a validated anxiety instrument or a diagnostic assessment.
A declined screen is charted as offered and declined, with any reason the patient volunteered, resources provided, and the re-offer plan, and it is never entered as a zero or a negative result. Arkansas law makes this concrete, requiring the record to state that the patient was not screened because she refused; Australian Medicare items likewise require the result or the decision not to participate to be recorded. Cadence follows the named authority: ACOG's initial-prenatal, later-pregnancy, and postpartum pattern; AAP's 1-, 2-, 4-, and 6-month well-child visits; Australia's early-and-later pregnancy plus six-to-12-weeks and once more in the first year; and in Canada, name whose guidance applies, since the national task force and the obstetric society genuinely diverge.
Yes. Give it the facts (version and language, mode, perinatal timing, total, threshold and purpose, item-10 status, actions, referrals, and the chart placement) and it drafts the full entry: the screening line, modest interpretation, separate item-10 documentation, the closed referral loop with owners and dates, and the well-child conventions, ready for your review. It can also check a finished note for thresholds charted as diagnoses, item 10 buried in the total, maternal detail in an infant chart, declined screens entered as scores, and positives with no follow-up. BastionGPT is HIPAA-compliant with a signed BAA on every plan, and your data is never used to train models.
The instrument facts and compliance claims on this page trace to these sources, last verified August 2026:
Educational content, not legal or billing advice. Sample notes are fictional. Follow your organization's policies and your board, payer, and jurisdiction requirements.