Generally, treatment is individualised according to various factors such as history of the illness course, treatment response and tolerability, adherence profile, comorbidities and availability.^1,9,10,11 Medications effective across different bipolar phases are preferred.^1,9,10,11 The South African National Department of Health Standard Treatment Guidelines (STG)⁹ recommend lithium monotherapy as first-line treatment in acute mania.⁹ The second-line options are valproate, olanzapine and risperidone.⁹ In contrast, the National Institute for Health and Care Excellence (NICE)¹ guidelines recommend antipsychotics as first-line treatment.¹ National Institute for Health and Care Excellence endorses olanzapine, quetiapine and risperidone but regards lithium and valproate as second-line options.¹

The greatest consensus exists between the South African Society of Psychiatrists (SASOP),¹¹ the Canadian Network for Mood and Anxiety Treatments (CANMAT) and International Society for Bipolar Disorders (ISBD),¹⁰ both MS (lithium and valproate) and antipsychotics are recommended as first-line interventions.^10,11 The following atypical antipsychotics are effective in a manic episode: risperidone,^1,10,11 quetiapine,^1,10,11 aripiprazole,^10,11 paliperidone,^10,11 asenapine¹⁰ cariprazine¹⁰ and ziprasidone.¹¹ However, the use of the latter three agents is scarce in South Africa because of no availability in the public sector.⁹ Although haloperidol is effective in acute mania, the risk for extrapyramidal side effects and poor efficacy in relapse prevention limits its use.^1,9,11 Consensus remains elusive on the use of carbamazepine in manic disorders. It is recommended by the CANMAT guidelines as a second-line agent while the local STG reports no efficacy in any mood episode.^9,10

Monotherapy treatment with first-line agents is effective in roughly half of the patients.¹⁰ However, combination therapy of lithium or valproate and an atypical antipsychotic is generally preferred because of the evidence of higher efficacy and a more rapid response, despite its cumulative side effect risk.^8,10 This is an important factor in resource-limited settings where rapid recovery is desirable.¹³

Bipolar I disorder (BD I) has at least one manic episode lasting ≥ 7 days (or requiring hospitalisation), often with psychosis and major impairment, with or without depressive episodes. Bipolar II disorder (BD II) is defined by at least one hypomanic episode—lasting ≥ 4 days and characterised by an elevated or irritable mood without marked functional impairment or psychosis—and at least one major depressive episode. BD I and BD II are distinct conditions with unique treatment requirements.^9,10,11 The STG recommends lithium, valproate and olanzapine as first line for acute BD I depression depending on tolerance and response to them.⁹ Additionally, quetiapine, lamotrigine and lurasidone monotherapy are endorsed by CANMAT and ISBD and SASOP.^10,11 Lurasidone and lamotrigine are effective as add-on therapy in cases of lithium-poor response.¹⁰ Quetiapine and lamotrigine may be added to lithium or valproate in cases of relapse while on optimal doses.^9,10,11

There is no consensus regarding the treatment of acute BD II depression.^9,10 The Canadian guidelines recommend quetiapine as first-line treatment because it is effective in both BD I and BD II depression.¹⁰ In contrast, lithium therapy is the first-line treatment in the local STG.⁹ Lamotrigine and quetiapine are the alternatives in cases of lithium non-response.⁹ Lamotrigine requires slow titration because of the risk of a severe drug reaction, thereby causing delays in achieving clinical response.^9,10,11

The role of antidepressants in BD depression is not yet confirmed.^1,8,9,10,11 Antidepressant monotherapy is not recommended in BD I depression because of the risk of a manic switch.^1,8,11 National Institute for Health and Care Excellence guidelines recommend a combination of olanzapine and fluoxetine as first-line treatment therapy because of its greater efficacy over lamotrigine monotherapy.¹ This combination is recommended as a second line by others^10,11 and demonstrates no efficacy in BD I depression prophylaxis.^9,11

The CANMAT and ISBD guidelines recommend bupropion, sertraline and venlafaxine monotherapy as second-line options for acute BD II depression, while venlafaxine and fluoxetine are recommended for prophylaxis.¹⁰ Notably, these recommendations are based on studies in pure depression,¹⁰ whereas bipolar depression typically presents with atypical or mixed features,³ in which the role of antidepressant monotherapy is ambiguous. In contrast, NICE, STG and SASOP advise against using antidepressant monotherapy for BD I or II depression because of associated uncertain benefits and mood switch risk; if used, these antidepressants should be combined with a MS, and mood is monitored closely.^1,9,11

Generally, agents effective in the acute phase are continued in the maintenance phase.^1,9,10

Lithium presents the best evidence for mania and depression prophylaxis, with a unique anti-suicidal effect.^1,8,9,10 Other agents, such as valproate, quetiapine, asenapine and aripiprazole (oral or injectable), are recommended for the prevention of mania and lamotrigine for the prophylaxis of depression.^1,10 The combination of quetiapine and lithium or valproate has also proven effective.^1,10,11

Lastly, depot antipsychotics play a role in cases of poor adherence.^1,8,11 Risperidone long-acting injectable (LAI) bi-weekly monotherapy or as an adjunct and aripiprazole LAI are effective against mania but not depression.^10,11

Monotherapy prophylaxis is recommended at the lowest effective dose when feasible.¹⁰ Treatment should be individualised while prioritising safety and efficacy.^1,10,11 Antipsychotics are considered to be the safest option in pregnancy and breastfeeding, especially quetiapine.⁹ The local STG recommends olanzapine because of efficacy in both mania and depression.⁹

The risk of weight gain and diabetes in pregnancy is maximum with olanzapine intake.⁹ Therefore, the Canadian guidelines advised olanzapine use for second-line treatment.¹⁰ One study reported an increased congenital risk associated with risperidone, thus not for first-line use in pregnancy.¹⁴ Additionally, risperidone may increase prolactin levels in non-pregnant WRA, which may impair fertility.¹⁰

Regarding MS, carbamazepine is associated with a risk of neural tube abnormalities and therefore not recommended in pregnancy.¹ Lithium and valproate are teratogens associated with congenital malformations.^15,16 Lithium is associated with a 2.4 cases per 100 live births risk of cardiac defects if intra-uterine exposure occurs in the first 3 months; however, this risk is lower than previously suggested.^9,15 Furthermore, lithium requires regular blood level monitoring, more so during pregnancy because of its narrow therapeutic index.^8,9 It is excreted in breastmilk and has a risk of toxicity in the infant, thereby making it unsuitable during breastfeeding.^8,9

Valproate is contraindicated in WRA because of its high risk for congenital malformations and poor neurodevelopmental outcomes in exposed infants.^1,9,10,16,17 If valproate use is unavoidable, a risk acknowledgement declaration should be signed and pregnancy prevention methods implemented.^1,8,9,10,18 Both the South African Health Products Regulatory Authority (SAHPRA)¹⁷ and Sanofi¹⁸ recommend regular counselling on teratogenic risk and contraceptive use at every visit.^17,18 Additionally, folic acid supplements of 5 mg/day are recommended.^19,20

The literature demonstrates a considerable risk associated with BD in WRA.^7,8 Hence, there is a need to follow evidence-based clinical guidelines to improve the treatment outcome in this population.^1,7,9,10 International and local studies have highlighted a discrepancy between treatment guidelines and clinical practice in managing women with BD.^4,21,22 Therefore, this study aims to describe the range of psychotropic medications prescribed in WRA with BD at a tertiary psychiatric unit. The documentation of discussions on effective pregnancy prevention, current contraceptive practices, prescription of folic acid and informed consent for valproate use was evaluated to determine adherence to prescribed guidelines.

The research objectives of this study were:

A quantitative, retrospective record review was conducted at Helen Joseph Hospital (HJH) psychiatric unit in Johannesburg, South Africa, from 01 January 2016 to 31 December 2023. Helen Joseph Hospital (HJH) is a public general hospital under the Gauteng Department of Health, affiliated with the University of the Witwatersrand. As a tertiary institution, HJH provides a variety of specialised services, including psychiatric care, to a diverse population from different parts of the greater Johannesburg metropolitan region.

Patients’ files were screened for WRA, defined as those aged 18–49 years^5,6 with a diagnosis of bipolar and related disorder as per Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition, Text Revision (DSM-5-TR) criteria,³ that is, BD I (mania, depressive or mixed episode), BD II (hypomanic or major depressive episodes), substance/medication-induced bipolar and related disorder³ and bipolar and related disorder due to another medical condition.³

Females treated at the HJH psychiatric unit between January 2016 and December 2023 only were included in the study sample. Those individuals with unclear or provisional diagnosis and having records with a missing prescription chart were excluded from this study. We conveniently sampled the individuals to include in the study. We considered everyone who presented from 01 January 2016 to 31 December 2023 at the institution. Overall, 339 files were screened and only 141 met the inclusion criteria, and this is the study sample size.

Data were extracted from the patients’ files manually by the primary investigator, using the data capture sheet (Online Appendix 1). The following variables were extracted from the file and the latest clinical notes:

A 10% quality check was done to the files retrieved to check for errors and consistency in data capturing.

The data for this study were captured in Microsoft Excel™ and statistically analysed in R software (version 4.00; www.R-project.org). The Shapiro–Wilk test was used to assess the normality of continuous variables. Continuous data are presented as the mean and standard deviation (s.d.), and categorical data are presented as counts and percentages. Bar charts were also used to display the distribution of the psychotropic agents.

Ethical clearance was obtained from the Human Research Ethics Committee of the University of the Witwatersrand (Ref. no. M240347 M240430-A-0001). The requisite permission was granted by the HJH research committee. Patient confidentiality and anonymity were maintained by not recording identifying data.

The mean (s.d.) age was 33.58 (s.d. = 7.69) years. Most women were single (n = 101, 71.6%) and were staying with their families (n = 116, 82.3%). The majority had a high school education (n = 85, 60.3%), only 31.9% (n = 45) were employed and 9.2% (n = 13) received a disability grant (Table 1).

Analysis of clinical characteristics revealed that over a third of women (n = 59, 34.7%) had a substance use disorder, 20 (11.8%) had human immunodeficiency virus (HIV), while 10 (5.9%) had epilepsy and 6 (3.5%) had anxiety disorders (Table 2). Only one woman was pregnant, and 13 (9.2%) women were on contraception using injectables (6, 4.3%), intra-uterine device (n = 3, 2.1%) and oral contraceptive pill (n = 4, 2.8%) (Table 2). The majority of the women showed an undocumented pregnancy status (55%) and contraceptive use (79%).

Most women (n = 123, 87.2%) had received MS. Valproate was predominantly prescribed (n = 88, 50.6%) and only nine (6.4%) women received folic acid co-prescription. This was followed by lithium (n = 51, 29.3%), lamotrigine (n = 14, 8.0%) and carbamazepine (n = 3, 1.7%) (Figure 1). Furthermore, 133 (94.3%) women received oral antipsychotics. Some received more than one antipsychotic; therefore, the total number of antipsychotics exceeds the number of participants. Risperidone was most commonly prescribed (n = 57, 39.0%), followed by olanzapine (n = 36, 24.7%), quetiapine (n = 30, 20.5%), amisulpride (n = 5, 3.4%), haloperidol (n = 4, 2.7%) and aripiprazole (n = 3, 2.1%). Besides, 14 (9.9%) women received depot antipsychotics, predominantly flupentixol decanoate (n = 11, 7.8%). Only two (1.4%) women were given paliperidone and one (0.7%) zuclopenthixol decanoate. Fifteen (10.6%) women received antidepressants. Six women (4.3%) received citalopram, five (3.5%) fluoxetine, three (2.1%) duloxetine and one venlafaxine (0.7%).

Only 21 (14.9%) women received monotherapy, predominantly with atypical antipsychotics (n = 12, 8.5%). This includes the one pregnant woman who received olanzapine monotherapy. Mood stabilisers monotherapy was employed in five women (3.5%) and typical antipsychotics in three (2.1%).

The majority (n = 120, 85.1%) of women received two or more agents, predominantly a MS and an antipsychotic (n = 115, 81.6%). The most common combination was valproate and an antipsychotic (n = 58, 41.1%), followed by valproate, lithium and an antipsychotic (n = 26, 18.4%) and lastly lithium and an antipsychotic (n = 20, 14.2%). Antidepressants were prescribed with antipsychotics and/or MS (n = 15, 10.6%).

The results demonstrate that none of the women receiving valproate had documented risk acknowledgement, that is, consent (n = 88, 100%). Only seven (5%) women had documented counselling on contraception use at the latest consultation. These seven women, who were documented to have received counselling, later received valproate. There was no statistically significant difference in age between participants who had documented contraception counselling (median = 34.0 years, interquartile range [IQR]: 14.0) and those who did not (median = 25.0 years, IQR: 16.0), as indicated by a Mann-Whitney U test (U = 64.0, p = 0.164). Given the highly skewed distribution of the documented counselling, (7 vs 133 not documented), bivariate analysis across socio-demographic variables did not yield any statistically meaningful inferences. The profile of those seven women who had documented counselling is provided as a supplement (Online Appendix 2).

This retrospective study design had several strengths. It was cost-effective and time efficient. Reviewing existing clinical records provided insights into the routine clinical practice. The results obtained may form a foundation for future research and contribute towards improving clinical practice. However, the study was largely dependent on the accuracy and completeness of documented information in the files, and inconsistent documentation may have affected the validity of the findings. As the current study was conducted at an academic institution, where clinicians undergo rigorous training and continuous education, it was assumed that the clinicians at HJH are well versed in the DSM-5-TR and have accurately applied it while making the diagnoses and recording notes in the file.

The current study focused on the general prescribing approach in WRA, a specific subgroup of patients. However, it did not consider the BD subtype, the mood episode being treated or the factors influencing medication choices. Medication data were obtained from outpatient prescription charts; therefore, it was assumed that most patients were in the maintenance phase of the illness. In addition, only the most recent clinical notes and prescription charts were reviewed for documentation of counselling, contraception use, and pregnancy status. Consequently, information recorded during previous consultations was not analysed, which may have introduced potential bias into the study findings. Furthermore, inferential statistical analysis was not meaningful because of the highly skewed comparison groups, thereby limiting the ability to conduct robust statistical comparisons. Lastly, as a single-site, cross-sectional study, the results cannot be generalised to other institutions.

It is recommended that clinicians consistently document reproductive health discussions such as pregnancy status, family planning and teratogenic risk associated with specific psychotropic medications. Informed consent should also be documented, and treatment guidelines closely adhered to when valproate is prescribed.