Diagnose and manage perinatal mental health disorders

In one line

Perinatal mental illness runs a spectrum from transient baby blues through perinatal depression and anxiety to puerperal psychosis, and the consultant task is to separate the self-limiting from the dangerous — recognising that suicide is a leading cause of maternal death and that puerperal psychosis is a psychiatric emergency demanding admission — while prescribing psychotropics whose benefit to a treated mother almost always outweighs the over-stated fetal risk.

Mechanism & pathophysiology

The puerperium is the steepest neuroendocrine cliff in human physiology. Within forty-eight hours of placental separation, oestradiol and progesterone fall from late-pregnancy levels — hundreds of times the non-pregnant concentration — to near-zero, and this abrupt withdrawal acts on a brain whose serotonergic, GABAergic and HPA-axis systems have spent nine months remodelling around a high-steroid environment. The baby blues — the tearful, labile, over-sensitive state peaking around days three to five and resolving by day ten in up to two-thirds of women — is best understood as the normal central nervous system response to that withdrawal, and its self-limiting course is the feature that distinguishes it from depression. The same steroid-withdrawal model underlies the recent interest in neuroactive steroids: allopregnanolone, a progesterone metabolite and positive allosteric modulator of the GABA-A receptor, falls precipitously after delivery, and the licensing of brexanolone and oral zuranolone (synthetic allopregnanolone analogues) as rapid-acting treatments for postpartum depression is the clinical vindication of the hypothesis that disordered GABAergic signalling, not simply "low mood", sits at the centre of the biology.

Depression and anxiety in the perinatal period are not a single disease with a hormonal trigger; they are the common final pathway of a genetic-temperamental vulnerability colliding with the steroid shift and a heavy load of psychosocial stress. The strongest single predictor of perinatal depression is a past history of depression or anxiety, and the second is the absence of social support; a previous perinatal episode, an unwanted or unsupported pregnancy, intimate-partner violence, and adverse life events stack the risk. In the South African setting these psychosocial determinants are not background noise but the dominant signal — poverty, food insecurity, HIV, the high prevalence of gender-based violence, and the particular vulnerability of the adolescent mother mean perinatal depression here is driven as much by the social world as by neuroendocrinology, which is why the SA evidence repeatedly shows prevalence figures far above the global average. The groundwork on screening, the social determinants and the first-line response is laid in the Intermediate chapter on GBV and mental health in pregnancy; the consultant-level task is the differential diagnosis, the psychotropic decisions, and the management of the emergency that the Intermediate chapter does not.

Puerperal (postpartum) psychosis is a biologically distinct entity and must not be filed under "severe depression". It is the most dramatic illness in obstetrics-adjacent medicine: an abrupt onset, usually within the first two weeks and often within the first few days, of a fluctuating, polymorphic psychotic state — grandiose or persecutory delusions, hallucinations, profound mood disturbance that swings between elation and despair, and a characteristic perplexity and confusion that gives it an almost delirium-like quality. It is, in the great majority of cases, a presentation of the bipolar spectrum: the single strongest risk factors are a personal history of bipolar disorder, a previous episode of puerperal psychosis, and a first-degree family history of either. A woman with bipolar disorder has roughly a one-in-five-to-one-in-two risk of a severe postpartum episode, and a woman with a previous puerperal psychosis carries a recurrence risk of the same order in a subsequent pregnancy — figures that turn the booking history into a screening question of consequence. The mechanism is incompletely understood but is thought to involve the interaction of the steroid-withdrawal and sleep-deprivation of the early puerperium with an underlying bipolar diathesis and, in some women, an immune or thyroid-autoimmune contribution. The clinical point that falls out of the biology is that this is an illness of rapid onset and rapid fluctuation, so a woman can look reassuring at one review and be floridly unwell hours later.

The danger of the spectrum is not evenly distributed. Perinatal depression harms through chronicity — impaired mother-infant bonding, disrupted attachment, poorer infant cognitive and emotional development, and, at its severe end, suicide. Puerperal psychosis harms through acuity and through its two catastrophic outcomes: suicide and infanticide. Suicide in the perinatal period is characteristically violent and determined — hanging, jumping, drowning — rather than the impulsive overdose more typical outside pregnancy, which is why a perinatal suicide is so often completed, and why the soft reassurance that "she only has fleeting thoughts" is so dangerous. Maternal suicide is a leading cause of death in the year after birth in every well-conducted confidential enquiry, and the deaths cluster in the late postnatal period rather than in pregnancy itself.

Assessment

The first discipline is to ask. Perinatal mental illness is massively under-detected because women conceal symptoms out of shame and fear of having the baby removed, and because clinicians who are comfortable asking about bleeding and blood pressure go quiet around mood. Routine, structured screening at booking and across the perinatal contacts is what closes that gap.

• Screening instruments. The two-item Whooley questions ("During the past month, have you often been bothered by feeling down, depressed or hopeless? … by having little interest or pleasure in doing things?") are the recommended case-finding opener at first contact — quick, validated, and a positive answer to either triggers fuller assessment. The Edinburgh Postnatal Depression Scale (EPDS) is the workhorse self-report tool: ten items, completed in about five minutes, validated specifically in the perinatal population so that it deliberately omits the somatic items (fatigue, appetite, sleep) that are confounded by normal new-motherhood. A score around ≥13 flags probable depression, lower thresholds raise sensitivity, and — the item that is too often skipped — question 10 asks directly about thoughts of self-harm and any positive answer demands immediate risk assessment regardless of the total. The EPDS also picks up anxiety. Both tools are validated in the urban South African setting, but the SA validation work is explicit that a screening score is a flag, not a diagnosis: it must be followed by a clinical interview, because cultural idioms of distress, language, and the somatic presentation common in this population mean the numbers alone misclassify.
• History. The booking history is a risk-stratification exercise: past or current psychiatric illness and its treatment, and specifically any history of bipolar disorder, previous puerperal psychosis, or a first-degree relative with either — these are the questions that identify the woman at high risk of the emergency, and they change the antenatal plan. Then the present state: mood, anhedonia, sleep (distinguishing the can't-sleep of depression and the doesn't-need-sleep of incipient mania from normal infant-driven sleep loss), appetite, anxiety and intrusive thoughts, the quality of the bond with the baby, and the social scaffold — partner, support, housing, food security, immigration or documentation status, substance use, and a direct enquiry about intimate-partner violence.
• The risk assessment is the examination. Ask explicitly about thoughts of death, of self-harm, and — separately and without flinching — about thoughts of harming the baby. Distinguish the egodystonic intrusive thoughts of perinatal OCD (the mother is horrified by an unwanted thought of harm, recognises it as wrong, and takes steps to avoid the baby; these carry a low risk of acting) from the ego-syntonic, delusionally-driven risk of psychosis or severe depression (the harm is congruent with a delusional belief — that the baby is evil, or that mother and baby must die together to be spared some imagined horror; this is high-risk and an emergency). Altruistic or "extended" suicide, in which a severely depressed or psychotic mother kills the infant believing she is protecting it, is the mechanism behind maternal infanticide and is the scenario the assessment exists to catch.
• Examination and investigations are about exclusion. New-onset psychosis or confusion in the puerperium is puerperal psychosis until proven otherwise, but the differential includes organic causes that are dangerous to miss: sepsis and delirium, eclampsia, intracranial events, thyroid disease (postpartum thyroiditis), Wernicke's encephalopathy after hyperemesis, drug intoxication or withdrawal, and HIV-associated neurocognitive or opportunistic CNS disease in the SA setting. A first psychotic episode therefore warrants observations, bloods (FBC, U&E, calcium, glucose, thyroid function, CRP), and a low threshold for imaging and lumbar puncture if any feature points organic.

Interpretation matters as much as the score. A high EPDS with anhedonia, hopelessness and early-morning waking is depression; a high EPDS dominated by the anxiety items, with intrusive checking and reassurance-seeking, is perinatal anxiety or OCD; reliving a traumatic birth with flashbacks, avoidance and hypervigilance is post-traumatic stress after a frightening delivery, a genuinely under-recognised perinatal diagnosis. The presentation that must never be missed is the one where the woman is not low but elevated, over-talkative, not sleeping and not distressed by it — that is mania, and in a recently delivered woman it is the leading edge of puerperal psychosis.

Management

The structure is immediate → ongoing → long-term, and the single organising principle is that the management is stratified by severity and by where the woman sits on the spectrum — reassurance for the blues, a stepped psychological-then-pharmacological approach for mild-to-moderate depression and anxiety, and emergency admission for psychosis or active suicidality.

Immediate. The emergencies are puerperal psychosis and the acutely suicidal mother, and both are managed by getting her to safety. Puerperal psychosis is a psychiatric emergency that requires urgent admission — it is not a community diagnosis to be managed with an outpatient prescription and a follow-up appointment. The risk of rapid deterioration, suicide and infanticide means she needs assessment and admission the same day, on the strength of clinical suspicion, before the picture has fully declared itself. In the United Kingdom the standard is admission to a mother-and-baby unit (MBU) so that mother and infant are kept together while she is treated and the bond protected. In South Africa, dedicated MBUs are scarce-to-absent outside one or two tertiary centres, so the realistic plan is acute psychiatric admission, with separation from the infant handled as compassionately and safely as the facility allows and a clear safeguarding plan for the baby's care — a difference in resource, not in urgency. Acute pharmacological treatment is with an antipsychotic (in SA practice typically haloperidol or an available atypical such as risperidone or olanzapine), with a benzodiazepine for acute agitation and to restore sleep, lithium as a mood stabiliser once organic causes are excluded and renal/thyroid function checked, and electroconvulsive therapy held in reserve for severe, refractory, catatonic or high-suicide-risk cases where it is rapidly and reliably effective. The acutely suicidal woman without psychosis is likewise admitted under psychiatric care; a positive answer on EPDS question 10, or a disclosure of active plans, is escalated, not deferred.

Ongoing — depression and anxiety, stratified. For mild-to-moderate perinatal depression or anxiety, the first-line treatment is psychological, not pharmacological: structured psycho-education, guided self-help, cognitive behavioural therapy or interpersonal psychotherapy, and, where this is what the service can deliver, lay-counsellor or community-health-worker support integrated into routine maternal care. Antidepressants are reserved for moderate-to-severe illness, for failure of psychological therapy, or where the severity makes waiting unsafe — and here the decision a consultant must own is the prescription of a psychotropic in pregnancy or lactation.

The governing principle is that untreated maternal mental illness is itself a risk to the pregnancy and the infant, and that the relevant comparison is never "drug versus nothing" but "treated illness versus untreated illness". Untreated antenatal depression is associated with poor attendance, substance use, preterm birth, low birthweight, impaired bonding and, at its extreme, maternal death; the fetal risks of the modern antidepressants are, by contrast, small and often statistically attenuated once maternal depression is properly controlled for. The counselling task is to put both sides of that ledger honestly to the woman and reach a shared decision.

The drug a consultant should be able to name and defend is sertraline as the first-line SSRI in both pregnancy and breastfeeding — effective, long human safety record, and the lowest breast-milk transfer of the class. The drug a consultant must be able to refuse is valproate in any woman of reproductive potential.

Ongoing — the high-risk woman and prophylaxis. The booking identification of a woman with bipolar disorder or a previous puerperal psychosis converts management into prevention. The evidence supports two distinct strategies depending on her diagnosis. A woman with bipolar disorder needs continuous mood-stabiliser prophylaxis through pregnancy and the puerperium, because stopping medication for pregnancy carries a high antenatal and postpartum relapse rate; the prophylaxis is chosen to minimise fetal risk (favouring an antipsychotic or carefully-managed lithium over valproate). A woman whose only illness has been an isolated puerperal psychosis — well between episodes, well during pregnancy — can reasonably remain medication-free through pregnancy and start prophylaxis (lithium or an antipsychotic) immediately after delivery, which avoids fetal drug exposure while covering the highest-risk window. Either way she needs a written perinatal plan, intensive monitoring in the early puerperium, protected sleep, and a named specialist contact.

Long-term. Recovery is the rule with treatment, but the arc extends well beyond the acute episode: continuation of an antidepressant for the months that prevent relapse, attention to the mother-infant relationship and the child's development (the lasting harm of perinatal depression is mediated through disrupted attachment), and pre-pregnancy counselling for the next pregnancy — recurrence risk, a prophylaxis plan agreed in advance, and the chance to review medication while not pregnant. Safeguarding runs throughout: where the mother's illness, substance use or risk compromises infant safety, the social-work and child-protection pathway is engaged early and proportionately, with the explicit aim of supporting the mother to parent safely rather than removing the child by default.

Guidelines compared

The major bodies agree on the architecture — screen, treat illness rather than fear the drug, treat psychosis as an emergency — and diverge on screening intensity and on the resources they assume.

The substantive contrast is between the comprehensive specialist model NICE assumes and the task-shared, primary-care-integrated model the SA NDoH and WHO mhGAP build for a system with a handful of psychiatrists per million people. The point a consultant should make is that the principles are universal — screen, treat the illness, escalate the emergency — but the delivery in South Africa is necessarily through trained non-specialists embedded in maternal services, with specialist psychiatry reserved for the severe and the psychotic.

The evidence & the controversy

The defensible modern position rests on a few firm points and several genuinely open ones. The firm ground is that treating perinatal depression works and that the modern antidepressants are, on balance, safe — the fetal signals (PPHN, cardiac defects) are real but small, frequently attenuate once maternal illness is controlled for, and are outweighed in moderate-to-severe illness by the harms of leaving the mother untreated. The firm ground also includes the hard rules at the dangerous end: puerperal psychosis is an emergency, and valproate has no place in a woman of reproductive potential.

The most genuinely contested area in the SA context is whether the task-sharing model that the system depends on actually delivers. The argument for it is irresistible on resource grounds — there is no other way to reach the burden — and process evidence supports its feasibility. But the most rigorous local trial, a double-blind randomised study of community-health-worker-delivered counselling for antenatal depression in Khayelitsha, found that the intervention did not outperform enhanced usual care on depression outcomes at three months postpartum. That negative result is not a reason to abandon task-sharing, but it is a sharp corrective to the assumption that a short lay-counselling package is sufficient on its own; it suggests that in a population whose depression is driven by ongoing poverty, violence and HIV, a few counselling sessions cannot fix what the social world keeps breaking, and that screening without a treatment pathway that actually changes outcomes risks being an empty gesture. The honest consultant position is that perinatal mental health in South Africa is a structural problem as much as a clinical one.

The newer controversy is the neuroactive-steroid drugs — brexanolone and oral zuranolone — which produce rapid antidepressant effects in postpartum depression and validate the GABAergic-withdrawal mechanism, but at a cost (brexanolone a 60-hour infusion; both expensive) that places them far outside routine SA practice. They are worth knowing as proof of mechanism and as the direction of travel, while recognising they change nothing in a district hospital today. A live ethical thread worth being able to argue is the screening paradox: it is unethical to screen for a condition you cannot then treat, so the SA debate is rightly about building the treatment pathway before scaling the screening, not screening for its own sake. And the safeguarding tension — protecting an at-risk infant without reflexively separating a treatable mother from her baby — is a values question a consultant should approach with a framework (proportionality, the least-restrictive option, keeping mother and baby together wherever safe, multidisciplinary decision-making) rather than a fixed verdict.

Landmark trials & key evidence

Reading the prophylaxis evidence together gives the high-risk plan its numbers. Wesseloo's 23% relapse on prophylaxis versus 66% without it, and Bergink's zero relapses among 20 women given postpartum lithium for isolated puerperal psychosis versus 44% of those who declined, are the figures that justify both the booking question and the antenatal plan — they convert "history of bipolar disorder" from a line in the notes into an active, evidence-based intervention.

Exam traps & red flags

• Treating puerperal psychosis as an outpatient depression. It is a psychiatric emergency — same-day assessment and admission (mother-and-baby unit where one exists, acute psychiatric admission in most of SA), on clinical suspicion, before the picture is "confirmed". Delay risks suicide and infanticide.
• Calling florid baby blues a depression, or missing the depression behind quiet blues. The blues peak around day 3–5 and resolve by day 10; anything persisting, deepening or carrying suicidal thoughts after that window is depression and needs assessment.
• Missing mania. A recently delivered woman who is elevated, over-talkative and not sleeping (and not distressed by it) is not "coping well" — that is the leading edge of puerperal psychosis.
• Ignoring EPDS question 10. The total can be sub-threshold while the self-harm item is positive; any positive answer triggers immediate risk assessment regardless of the score.
• Conflating egodystonic and ego-syntonic harm thoughts. The horrified, intrusive, resisted thought of OCD is low-risk; the delusionally-congruent intent of psychosis or severe depression (the baby is evil; we must die together) is high-risk and an emergency. Ask about thoughts of harming the baby explicitly.
• Valproate in a woman of reproductive potential. Contraindicated for a mental-health indication — major malformations and dose-dependent loss of childhood IQ. Knowing this hard rule is non-negotiable.
• Reflexively stopping an effective antidepressant for pregnancy or breastfeeding. Relapse risk is high and the harm of untreated illness usually exceeds the small fetal/infant drug risk; sertraline is first-line in both pregnancy and lactation.
• Forgetting the organic differential. New-onset puerperal psychosis or confusion is psychosis until proven otherwise, but exclude sepsis, eclampsia, thyroid disease, intracranial events, Wernicke's, and (in SA) HIV-related CNS disease before settling on the psychiatric label.
• Underestimating perinatal suicide. It is a leading cause of maternal death in the late postnatal period, the method is often violent and lethal, and the woman may not look obviously ill — which is exactly why the structured risk question exists.
• Screening with no pathway. Screening for perinatal depression without a treatment route that actually changes outcomes is an empty and arguably unethical gesture in a resource-limited service — build the pathway alongside the screen.

Evidence anchors

• Cox JL, Holden JM, Sagovsky R — Detection of postnatal depression: development of the 10-item Edinburgh Postnatal Depression Scale, Br J Psychiatry 1987
• Patorno E et al. — Lithium Use in Pregnancy and the Risk of Cardiac Malformations, N Engl J Med 2017
• Huybrechts KF et al. — Antidepressant use late in pregnancy and risk of persistent pulmonary hypertension of the newborn, JAMA 2015
• Meador KJ et al. — Fetal antiepileptic drug exposure and cognitive outcomes at age 6 years (NEAD), Lancet Neurol 2013
• Meador KJ et al. — Cognitive function at 3 years of age after fetal exposure to antiepileptic drugs (NEAD), N Engl J Med 2009
• Bergink V et al. — Prevention of postpartum psychosis and mania in women at high risk, Am J Psychiatry 2012
• Wesseloo R et al. — Risk of Postpartum Relapse in Bipolar Disorder and Postpartum Psychosis: a systematic review and meta-analysis, Am J Psychiatry 2016
• Dennis CL, Dowswell T — Psychosocial and psychological interventions for preventing postpartum depression, Cochrane Database Syst Rev 2013
• Lund C et al. — Task-sharing of psychological treatment for antenatal depression in Khayelitsha, South Africa, Behav Res Ther 2019
• Marsay C, Manderson L, Subramaney U — Validation of the Whooley questions for antenatal depression and anxiety among low-income women in urban South Africa, S Afr J Psychiatry 2017
• NICE CG192 — Antenatal and postnatal mental health: clinical management and service guidance, 2014 (updated 2020)
• South Africa NDoH National Integrated Maternal and Perinatal Care Guideline and the Saving Mothers reports (NCCEMD), and the WHO mhGAP Intervention Guide — perinatal mental-health screening integrated into routine maternal care and non-specialist (task-shared) delivery in a resource-limited setting.
• South African Medicines Formulary (SAMF) / NDoH Essential Medicines List — perinatal psychotropic availability and dosing (sertraline/fluoxetine, amitriptyline, haloperidol, lithium).
• MBRRACE-UK Saving Lives, Improving Mothers' Care — maternal suicide a leading cause of late maternal death; "red-flag" symptoms mandating urgent review.