Baseline characteristics

The average age of our study population (37.4 years) is generally higher than findings from other countries in Africa and the UK. ^{Reference Mishra, Patil, Sathe and Reshamvala24,Reference Gureje, Appiah-Poku, Bello, Kola, Araya and Chisholm25} This higher age of presentation probably reflects the prolonged duration of untreated psychosis in low- and middle-income countries, and the multiple pathways to care in Ghana. ^{Reference Appiah-Poku, Laugharne, Mensah, Osei and Burns26} Additionally, because Accra Psychiatric Hospital is a referral centre, there may be delays between the onset of psychosis and hospital presentation.

The mean BMI of 22.2 kg/m2, with 25.9% of patients being either overweight or obese, is comparable to the general Ghanaian population mean BMI of 24.7 kg/m2. ^{Reference Titi, Id, Morrison and Annan27}

Disengagement from service

Psychotic disorders generally follow a chronic course, usually requiring prolonged treatment. It is recommended that a first-episode psychosis be treated until symptoms have been in remission for a minimum of 1 year, to prevent exacerbations. ^{Reference Üçok28} Our finding that over half (53.1%) of patients with non-affective psychosis disengaged within 1 month, and 75.5% by 6 months, is alarming. This rate is significantly higher compared with other regions, ^{Reference O’Brien, Fahmy and Singh29} highlighting a particular issue in Ghana.

Several factors may have contributed to this high disengagement rate. Patients may transfer to community mental health services due to proximity, although this is unlikely to account for most cases because standard practice requires that patients be clinically stable before transfer to community care. Another factor is resorting to religious and faith healers, especially prayer camps, which is common and widely practised in Ghana. ^{Reference Gyimah, Ofori-Atta, Asafo and Curry30} During the period under this review, patients and their caregivers had to bear the cost of care for mental health services, and this is another possible explanation for the high rates of disengagement. Additionally, some patients may have achieved full recovery and returned to baseline functioning, reducing their perceived need to continue to engage with mental health services, with some evidence suggesting that patients with early psychosis show good treatment response. ^{Reference Kang, Piao, Li, Kim, Kim and Lee31} Furthermore, our review included all patients diagnosed with a non-affective disorder, and this includes patients with acute and transient psychotic disorders. These disorders are known to resolve quickly, usually within 1 month, ² and these patients are unlikely to feel the need to return for review, especially considering the stigma associated with a psychiatric diagnosis. It is now a priority to understand the reasons for patient disengagement, to help to develop strategies aimed at improving retention and treatment outcomes.

Cardiometabolic considerations

Despite growing evidence concerning the metabolic consequences of olanzapine, ^{Reference Pillinger, McCutcheon, Vano, Mizuno, Arumuham and Hindley32} 63% of patients in our review received this medicatiion during their first visit. Clinicians probably prefer olanzapine due to its perceived superior efficacy in resolving psychotic symptoms, ^{Reference Davis and Chen33} prioritising immediate clinical response over the risk of long-term metabolic dysregulation. Other countries recommend that olanzapine not be used as a first-line treatment for psychosis because of its metabolic profile. ^{Reference Crlenjak, Curtis, Fraser, Fraser, Hughes and Mcgorry34} The low levels of use of long-acting injectable (LAI) antipsychotics (depot) in our study population could represent a cost issue. While second-generation antipsychotics are the preferred first-line treatment, LAIs are expensive and many patients cannot afford them. Clinicians might opt against first-generation LAIs due to their extrapyramidal side-effects.

Proactive prevention of cardiovascular-related mortality requires regular monitoring of cardiometabolic risk factors. The Royal College of Psychiatrists recommends cardiometabolic tests to monitor physical health in the early stages of antipsychotic treatment, including weight, BMI, blood pressure, plasma glucose and lipid profiles. ^{Reference Perry, Holt, Chew-Graham, Tiffin, French and Pratt35} Our study’s finding of low levels of cardiometabolic monitoring, even among those receiving olanzapine, means that these standards are not being met, exposing patients to risk. Under half of patients had their blood sugar levels recorded before initiating antipsychotics, and this proportion did not improve significantly over time. Similarly, lipid profile monitoring was insufficient. The absence of established local protocols for cardiometabolic monitoring probably contributes to this issue, as well as the cost of laboratory investigations, which are borne by the patient or their caregiver. However, this cannot explain the failure to record patients’ weight during initial consultation.

Weight and BMI changes over time

Patients demonstrated progressive weight and BMI increases over the course of this review, consistent with findings from other studies in patients with first-episode psychosis. ^{Reference Tek, Kucukgoncu, Guloksuz, Woods, Srihari and Annamalai36} People with higher baseline weight continued to record further increases.

Our analysis revealed that weight change associated with the other medications was not different from that of olanzapine. This finding of insignificant difference in the contribution of medication type to weight changes conflicts with the existing literature, which typically shows olanzapine to be associated with the greatest weight gain. ^{Reference Tek, Kucukgoncu, Guloksuz, Woods, Srihari and Annamalai36,Reference Bernardo, Rico-Villademoros, García-Rizo, Rojo and Gómez-Huelgas37} The lack of randomisation in our study may have introduced selection bias based on clinician preferences, influencing the observed effects on weight changes. Olanzapine was preferentially prescribed for patients with lower baseline weight and, because lower baseline weight is associated with less weight gain in our population, people on other medications are likely to have differentially gained more weight, potentially masking the effects of olanzapine on weight. This would be plausible given that almost all antipsychotics are known to cause weight gain. ^{Reference Campforts, Drukker, Crins, van Amelsvoort and Bak17,Reference Wu, Siafis, Hamza, Schneider-Thoma, Davis and Salanti18,Reference Bak, Fransen, Janssen, van Os and Drukker38} However, this could also be a type II error, with a lack of power to detect differences between medications. Our finding that 40% of patients experienced CSWG at 3 months and 58% at 6 months is higher than those of a recent meta-analysis on the association between antipsychotics and CSWG. ^{Reference Campforts, Drukker, Crins, van Amelsvoort and Bak17} At baseline, one in four patients was either overweight or obese and, by 6 months, this proportion had increased to one in two. Research has pointed to racial differences in the experience of antipsychotic-induced weight gain with people of black ancestry recording higher weight gains than their caucasian counterparts. ^{Reference Jerome, Pulley, Sathe, Krishnaswami, Dickerson and Worley39} Genetic variability in the expression of antipsychotic target genes and antipsychotic metabolism has been proposed as a reason for these observed racial disparities, ^{Reference Zhang, Lencz, Zhang, Nitta, Maayan and John40,Reference Pardiñas, Kappel, Roberts, Tipple, Shitomi-Jones and King41} and this could be the reason why our study population, which is entirely African, recorded higher rates of weight increase. This calls for early, proactive efforts when initiating antipsychotics to prevent weight gain, identify those gaining weight and initiate measures to mitigate the impact on their physical health. Because there are no established guidelines for dealing with antipsychotic-induced weight gain, implementation of mitigation measures is at the discretion of the clinician. Having guidelines allows the hospital to plan and make resources for lifestyle or pharmacological interventions available for all patients.

Future research directions

This study highlights significant patient disengagement and lapses in both cardiometabolic risk monitoring and weight management in patients with psychosis at Accra Psychiatric Hospital. Even though feedback has been provided to the hospital, addressing these challenges requires robust monitoring protocols to ensure that patients’ cardiovascular risks are routinely evaluated, especially in the first few months of initiating antipsychotic treatment. Evidence-based interventions should be recommended to mitigate the cardiovascular risks associated with exposure to antipsychotics, and these should be timely, adopting both preventive and therapeutic approaches. There is also a need for ongoing research to understand the factors contributing to the high disengagement rate, and to develop interventions to address this problem. While these findings are restricted to the study site, the contributory factors are common to most low- and middle-income countries and these findings are likely to be replicated in similar settings.

Study limitations

This study utilised real-world clinical data, with significant missing data, which is a limitation to its interpretation. Additionally, the absence of randomisation potentially introduces bias in the allocation of medications, thus affecting the interpretation of the relationship between antipsychotic type and weight/BMI changes. Furthermore, there are probably unmeasured confounding variables such as genetic, lifestyle and dietary factors that could potentially account for a significant proportion of the variation noted in weight and BMI. Finally, because our study was a retrospective review, we were unable to obtain comprehensive follow-up data to ascertain the cause of some findings, such as the high disengagement rate.