MedsScan, Issue 3, 2026

These reviews provide updates on the international literature on therapeutics. Expert pharmacy practitioners — via AdPha’s Specialty Practice Groups — scan major peer-reviewed journals in areas relevant to Australian pharmacy practice and present precis on major clinical trials, important pharmacoepidemiology studies and pharmacoeconomic research, and other updates relevant to practice. Interested readers are encouraged to explore the original publications in greater detail. 

Published online: 9 September 2026

COMPOUNDING SERVICES

MedsScan Editor for Compounding Services SPG: Rachel Berry

The effect of light exposure on bevacizumab infusions

Bevacizumab is a monoclonal antibody used in the treatment of a range of cancers including colorectal, gynaecological and lung cancer. Bevacizumab works to inhibit angiogenesis by binding to vascular endothelial growth factor A (VEGF-A). Originally available as Avastin, several biosimilar brands are now available.

Multiple opportunities exist during compounding, storage and administration for infusions to inadvertently be exposed to light. Manufacturer’s product information often doesn’t specify that infusions should be protected from light during preparation, storage or administration, only that vials should be protected during storage. Expiries from external licensed compounding facilities may also not include a recommendation to protect from light.

This study investigated whether realistic light conditions in the clinical setting affected stability, biological activity or immunogenicity of bevacizumab infusions. This study is of interest as it highlights factors that may adversely impact the effectiveness of treatment.

Two bevacizumab (Zirabev, biosimilar) infusion concentrations were studied; 1.4 mg/mL and 16.5 mg/mL diluted in sodium chloride 0.9%, matching the recommended concentration range in the product information. Light was simulated in a SunTest chamber, to replicate sunlight filtered through a window for approximately 9 hours on a sunny day. The temperature in the chamber was below 25oC during the irradiation. A light protected sample was also placed inside the chamber to control for any possible effect of temperature. Variables assessed in this study included chemical stability, aggregation formation, binding to VEGF-A, anti-angiogenic activity and immunogenic potential.

Of particular interest were the results showing that light-exposed bevacizumab demonstrated a reduced binding to VEGF-A, with reduced anti-angiogenic activity. Although light exposure in this study well exceeds the light an infusion of bevacizumab would usually be exposed to, given that the longest infusion time for a bevacizumab infusion is 90 minutes, it highlights that there can still be an adverse effect from inadvertent light exposure. This study is of relevance for infusions prepared in advance with an extended shelf life where there is possibly more opportunity for light exposure during storage. If bevacizumab infusions are prepared in advance and require storage, then storing them in the refrigerator will help to mitigate much of the potential light exposure. Care should be taken in infusion clinics where there is potential for excessive light exposure with consideration given to optimizing infusion production schedules and storage locations so that exposure can be minimised.

De Diana E, Sottoriva C, Zorzetto G, Crivellaro G, Rigamonti N, Coppola M, et al. Unveiling the effects of light exposure on bevacizumab stability in clinical settings. Int J Pharm 2026; 691: 126578.

Challenges in the comparison of closed system transfer devices

Closed system transfer devices (CSTDs) are components designed to reduce or eliminate vapour, aerosol or liquid escape, during the preparation and administration of hazardous medicinal products (HMPs).

There are several different CSTD products on the market both in Australia and internationally. Decisions on the extent of use of CSTDs are usually made on a local organisational level, with extent of use varying greatly both domestically and internationally.

International definitions of hazardous medicinal products also vary, and product risk assessments are generally not conducted at a national or international level. Various documents exist to aid in assessment of precautions that might be required for hazardous products (e.g. National Institute for Occupational Safety and Health [NIOSH], Victorian Therapeutics Advisory Group [VicTAG], eviQ Cancer Treatments Online [eviQ]), but there isn’t a definitive list of hazardous medicinal products and new products are constantly being introduced.  Clinical trials also require sponsor permission to use a CSTD with investigational medicinal products (IMPs), due to a lack of specific material testing compatibility.

Barriers to the implementation of CSTDs include the selection of an appropriate device, perceived benefit depending on where medicinal products are being prepared (pharmacy department versus nurse preparation), cost and required staff training.

This discussion paper provides a summary of international standards and regulations, inconsistencies, published evidence for CSTDs and a proposal for global harmonisation in the approach to handing hazardous medicinal products.

The discussion paper presents a summary of evidence published over the past six years, and advises that caution is required when interpreting CSTD product claims. There is no internationally recognised standard testing protocol for CSTDs that uses HMPs. The only globally accepted CSTD testing protocol was developed by NIOSH in 2015 using isopropyl alcohol (IPA) as a surrogate for a HMP. The obvious disadvantage of this is that no HMP is used in the CSTD testing procedure.  An updated protocol is in development.

An area of potential interest is whether CSTDs can maintain vial sterility allowing cost savings from storing partly used vials of costly medicinal products. Caution should be exercised when considering this, including considering the effect of temperature cycling, light exposure and the level of facility each vial has been manipulated in.

This review is of interest to institutions reviewing their current CSTDs, or any considering implementing the use of CSTDs.

Amari P, Campbell K, Henry M, Polovich M, Sessink P. Development of a unified approach to hazardous medicinal product management using closed system drug transfer devices. J Oncol Pharm Pract 2026; [online ahead of print]. DOI: 10.1177/10781552261434870.

Adverse drug events from paediatric extemporaneous compounded medicines

Extemporaneously compounded medicines are often required for paediatric patients due to a lack of commercially available products that can provide paediatric doses. This is primarily because children are not included in most clinical trials.

These compounded products may expose paediatric patients to an increased risk of adverse events due to extensive testing not being conducted on them and the fact they are often compounded in response to an individual prescription request.

This article is a scoping review of published literature relating to adverse drug events from community pharmacy compounded medicines in paediatric patients under the age of 18. A Medline search was used to identify studies describing adverse events from medicines compounded in community pharmacies. There were no restrictions based on country or publication date. Thirty-seven adverse drug events were reported in the 25 studies included in the review.

Most of the formulations compounded were oral suspensions, as expected. The most affected age group of children was the 0–5 year old age group. The most common errors detected were incorrect drug concentrations or the inclusion of an incorrect ingredient. The most severe outcome was the deaths of two children.

Recommended strategies to reduce the risk of harm from community pharmacy compounded medicines included using standardised concentrations to minimise the potential for calculation errors, special care with ingredient selection and calculations, the introduction of safeguards such as barcode scanning of individual ingredients, and maintaining vigilance regarding any unexpected variations at any stage of the preparation or administration.

Although it is very likely that adverse events are underreported, this review provides a reminder of the potential for serious harm from paediatric compounded medicines and strategies that might aid in minimising risks in the compounding process, to provide safe and effective formulations.

Li J, Buckley NA, Lee J, Cairns R. Adverse drug events from paediatric use of extemporaneously compounded medicines in community pharmacy settings: a scoping review. BMJ Paediatr Open 2026; 10: e004327.

Back to top

DISPENSING AND DISTRIBUTION

MedsScan Editors for Dispensing and distribution SPG:  Anna Wood and Ashley Crawford

The impact of numerical counselling strategies on medication compliance

Poor medication adherence is a recognised contributor to worsening chronic disease and avoidable hospitalisation. For conditions such as hypertension and hypercholesterolaemia, the benefits of treatment are often not immediately apparent because these conditions are frequently asymptomatic. In contrast, adverse effects may be experienced directly, leading some patients to perceive the risks of treatment outweigh its benefits.

This study investigated whether counselling that provided numerical information about the likelihood of adverse effects could improve willingness to take medication. Participants were randomly assigned to one of four counselling scenarios. Adverse effects were described either qualitatively (for example, ’rarely‘) or numerically (for example, ’less than 5%’), with each approach delivered either with or without a ’nudge‘ intervention that emphasised social norms to encourage adherence. Outcomes were analysed according to participants’ baseline numeracy skills.

Participants with higher numeracy skills were significantly more willing to take medication regardless of the counselling approach (59.6% versus 48.0%). In this group, willingness was highest when numerical risk information was provided without a nudge intervention. Among participants with lower numeracy skills, willingness was highest when numerical information was combined with the nudge intervention. Use of the nudge technique was significantly associated with willingness to take treatment in the less numerate group.

The findings highlight the importance of tailoring communication when counselling patients. Numerical presentation of risk information may improve understanding for some patients, while behavioural approaches such as nudging may be beneficial for those with lower numeracy skills. The study however has several limitations which could impact the applicability of the results. Firstly, the study only assessed hypothetical willingness to commence treatment rather than actual adherence, and it was conducted in a non-patient population. In addition, the effectiveness of the nudge intervention may not be directly generalisable to Australia given the different cultural setting in which the study was undertaken. Despite these limitations, the research reinforces the value of presenting medicines information in different ways to support consumers’ informed decision-making. Importantly, non-adherence can itself result in harm, highlighting the need for effective risk communication.

Yoshida A, Horii N, Oshima S, Oshima S, Kobayashi D. Effects of pharmacists’ communication strategies and patients’ numeracy skills on willingness to take medication. Int J Pharm Pract 2026; 34: 35–40.

Pharmacy automation in outpatients: patient perceptions

The evaluation of pharmacy automation has typically focused on safety and efficiency along with clinician acceptance of the technology. While the driver for pharmacy automation is generally the reduced risk of errors and harm, little is known about whether patients feel this technology has improved their experience of attending an outpatient hospital pharmacy.

This cross-sectional survey sought to assess the patient acceptance of, and satisfaction with, automated pharmacy systems across four hospitals in Kuwait. The term ’automated pharmacy‘ was defined as the use of electronic prescriptions and automated dispensing software such as in pharmacy robotics. More than 400 patients completed a 40-item survey which was based on other validated pharmacy survey tools, providing valuable insights into the patient experiences.

The patients reported their satisfaction with other aspects of the pharmacy service, including medication supply, pharmacist counselling and prescription wait times. A high proportion of patients reported satisfaction with these aspects of the pharmacy service. When looking specifically at the results of the survey associated with pharmacy automation, patients perceived positive impacts from the automation including time savings, reduced likelihood of medication errors and greater access to prescription information. Overall, patients reported very high satisfaction with the automation services. The areas that patients showed lower levels of satisfaction with were around the completeness of counselling provided by pharmacists and whether pharmacists spent sufficient time with the patients explaining their medications.

The use of automated technology in outpatient pharmacy services has the potential to positively impact clinicians, patient safety and potentially, based on the results of this survey, patients’ experiences and their perception of benefits. However, whilst pharmacy automation is often promoted as a way to free up pharmacists’ time for more complex clinical tasks, this example suggests this benefit may not have resulted in improved patient counselling or communication. Despite the limitations of the study, ultimately it provides a reminder that no matter the automation that is available, pharmacist to patient communication remains the cornerstone of high-quality clinical care.

Buabbas A, Alsaleh F, Alsanafi M, Safar M, Al-Sultan A, Zaman T, Patient experiences with outpatient pharmacy services in hospitals using automated pharmacy systems: cross-sectional study. JMIR Med Inform 2026; 14: e80963

Does artificial intelligence have a role to play in minimising the risk of medication dispensing errors?

There has been increased utilisation of artificial intelligence (AI) in healthcare over recent years to support efficiency and patient safety. One promising application within pharmacy practice is reducing the risk of medication dispensing errors. Such errors have the potential to cause significant patient harm and there are several technological innovations such as barcode scanning and robotics already integrated into standard pharmacy workflows to reduce risk. The aim of this study was to apply a systems thinking approach to analyse the impacts of the utilisation of AI as an additional risk mitigation strategy.

Researchers performed a literature search to identify studies representative of diverse AI approaches to dispensing error management for further analysis. Seven different studies were considered relevant for further exploration. AI approaches utilised included: Natural Language Processing, Logistic Regression, Deep Learning and Ant Colony Optimisation. Each different AI approach was assessed using a systems thinking framework focused on four pillars: people, systems, design and risk. This allowed for holistic consideration of how the technology could be incorporated through the assessment of potential benefits, barriers to adoption and unintended consequences.

AI was found to demonstrate promise for supporting improved dispensing accuracy by a variety of mechanisms including: identifying name and terminology errors in medication labels, application of imaging classification to prevent medication selection errors, and optimisation of pharmacy inventory control. Authors emphasised the importance of Human-AI integration in any approach implemented, to ensure AI is viewed as a tool to support the dispensing process and that it does not replace the need for clinical judgement and accountability. Beyond clinical safety, authors highlighted that consideration of technical factors such as security of personal data and systems integration is also critical.

The information in this article is limited to a small subset of research in a broad and rapidly evolving field. Authors acknowledged that the studies analysed focused on individual AI interventions applied in single institutions or simulation-based environments, and that application of AI into a real-world pharmacy environment requires further exploration and analysis. Applying a systems thinking approach to the introduction of AI proved useful in identifying gaps and potential solutions to implementation and adoption. Based on the information gathered in this article, it is apparent that a multimodal AI approach would be beneficial to address the various sources of medication dispensing errors that exist within pharmacy workflows. Further research into real-world, multimodal AI implementation in a pharmacy environment will be a valuable contribution to this field.

Ouda E, Chaabi I, Abualola H, Ramadan MA, Patro PK, Kaya GK, Simsekler MC. The role of artificial intelligence in reducing dispensing errors for patient safety and quality: a systems approach. Risk Manag Healthc Policy 2026: 573762.

Back to top

EDUCATION AND EDUCATIONAL VISITING 

MedsScan Editors for Education and educational visiting SPG: Hilai Ahmadzai and Diana Sandulache

Structures and processes supporting interprofessional education in Scottish pharmacy placements

This study explored stakeholder perspectives on the structures and processes supporting interprofessional education (IPE) during experiential learning (EL) placements for student pharmacists in Scotland. Using a qualitative design underpinned by systems theory and the Biggs 3P (presage–process–product) model, the authors conducted semi-structured group interviews with academics, practice educators and preceptors (n = 12). Thematic analysis identified three overarching themes.

Most IPE during placements was described as informal or opportunistic, typically occurring through ward rounds, multidisciplinary team meetings, or ad hoc professional interactions. Few examples of formally planned, structured IPE involving multiple student groups were identified. Participants reported limited visibility of explicit IPE learning outcomes within placement documentation, reinforcing a predominantly single professional focus.

Preceptor characteristics (enthusiasm, confidence, integration within MDTs), student factors (motivation, confidence, perceived competence), and logistical constraints (short placement duration, timetabling, workforce pressures) were significant determinants of IPE exposure. Regulatory and governance requirements such as fixed working hours and emphasis on ’active doing‘ rather than observation, were perceived as unintentionally restricting participation in collaborative activities. Sector-specific differences emerged, with hospital and primary care settings often perceived as more conducive to IPE than community pharmacy, although this was not universally agreed.

Stakeholders advocated for a continuum of interprofessional learning embedded throughout the MPharm curriculum and aligned with practice placements. Greater coordination between universities, placement providers and MDT members was identified as essential. Suggested strategies included clearer IPE learning outcomes, enhanced facilitator training (including interprofessional case discussions), improved communication between higher education institutions and sites, and shared planning across professions.

This study identified that IPE in placements remains largely dependent on individual initiative rather than structured curricular design. Strategic alignment, faculty development and system-level coordination are required to ensure consistent development of collaborative competencies essential for contemporary pharmacist practice.

Depasquale C, Arnold A, Cunningham S, Jacob SA, Boyter A, Portlock J, et al. Exploring structures and processes supporting interprofessional education during experiential learning placements for student pharmacists. Am J Pharm Educ 2024; 88: 101267.

Appraising the quality of a US pharmacy technician training program

In this study, the authors evaluated the effectiveness of an employer-based Pharmacy Technician Training Program (UHPTTP) implemented within a large US health system to address workforce shortages. The primary objective was to assess program success using employment retention metrics.

This quality improvement study included all trainees enrolled between November 2020–September 2023 (n = 95; nine cohorts). The 16-week paid program was accredited by the American Society of Health-System Pharmacists (ASHP) and the Accredited Council for Pharmacy Education (ACPE), and combined didactic instruction with experiential rotations within hospital and community settings totalling 500 hours. Completion of the program qualified graduates to pursue registration with the Pharmacy Technician Certification Board.

Of the 95 participants, the overall program completion rate was 77.8% (n = 74). Of program graduates, 93.2% were retained in pharmacy technician roles at graduation. One-year retention (assessed for cohorts with available follow-up) was 68.3%, compared with externally hired technicians at 61.3%. Although graduates demonstrated a 7% higher retention rate, the difference was not statistically significant (p = 0.688). Retention at graduation was strong and comparable or superior to several published institutional programs. However, one-year retention declined, reflecting broader workforce dynamics rather than program-specific shortcomings. It was postulated that the absence of financial retention incentives, such as tuition forgiveness, may partially explain lower long-term retention compared with programs using contractual commitments.

Under this initiative, UHPTTP successfully created an internal pipeline of qualified candidates during a period of national technician shortages, and filled vacant positions that may otherwise have remained unfilled. While the program did not significantly outperform external hiring in one-year retention, it demonstrated comparable workforce stability and addressed recruitment gaps. Limitations included a small sample size, cohort variability and potential confounding from COVID-19 related workforce pressures. Overall, internal accredited technician training programs represent a viable workforce stabilisation strategy for health systems.

Wat SKS, Roberts P, Rogers, J. Evaluating the success of a pharmacy technician training program at a large health system. J Am Pharm Assoc 2026; 66: 103002.

Podcasts in pharmacy education: a systematic review of learning outcomes

This systematic review investigated how podcasts affect knowledge, learner engagement and empathy outcomes in pharmacy education. A structured search across five databases identified nine experimental studies that met the eligibility criteria for inclusion.

Five studies evaluated knowledge acquisition or retention. Four demonstrated statistically significant pre-post improvements following podcast exposure, including improved retention of commonly prescribed medicines and enhanced topic-specific knowledge such as urine drug testing and ECG fundamentals. However, the only head-to-head comparison with traditional didactic lectures found that students receiving podcast-based instruction performed modestly worse on assessment, particularly among lower-performing learners. These findings suggest podcasts may support knowledge acquisition but may not match the effectiveness of structured, face-to-face teaching when used as a primary delivery modality.

Five studies assessed learner engagement. A strong preference for podcast formats was reported, with over 80% of students favouring podcasts over traditional lectures in one study. Learners valued concise episodes, conversational tone, authentic and relatable guest speakers, question-and-answer segments, and personal storytelling. Student-created podcasts were associated with professional growth, interprofessional collaboration and reflective learning, suggesting an additional benefit when podcasts are used as an assessment modality.
Two studies explored empathy. Although validated empathy instruments were not employed, learners reported increased perspective-taking, reflection on biases and deeper consideration of professional identity after engaging with podcast content.

This review suggests that podcasts appear most appropriately positioned as supplementary educational tools within pharmacy curricula. They may enhance engagement and reflective capacity, but current evidence does not support replacing traditional didactic teaching with podcast-based instruction alone. To clarify the educational value of podcasts, more rigorous comparative studies and validated outcome measures are required.

Vu A, Diec S, Wang E, Rosario N, Davis S, Guerrero T, Wollen J. The impact of podcasts on knowledge, engagement, and empathy outcomes in pharmacy education. Pharm Educ 2025; 25: 442–449.

Back to top

GENERAL MEDICINE

MedsScan Editors for General medicine SPG: Salwan Al-Saour and Gail Edwards

Duloxetine use for pain and neuropathic-like symptoms in end-stage osteoarthritis

Special contributor: Wei Mann Chew

Background: Osteoarthritis is one of the most prevalent forms of arthritis in Australia and has been associated with high health care expenditure and lost workforce productivity.1,2  Paracetamol and non-steroidal anti-inflammatory agents (NSAIDs) are commonly used to manage osteoarthritic pain.3-5 A proportion of patients, however, do not gain adequate relief from these first-line agents.3-5 It has been suggested that duloxetine may play a role in knee osteoarthritis where knee pain is persistent, but evidence is lacking in other types of osteoarthritis.5

Duloxetine, a serotonin–noradrenaline reuptake inhibitor with a central mechanism of action, has shown analgesic efficacy in placebo-controlled knee osteoarthritis trials and in other pain states involving central sensitisation.6 Limited previous randomised trials in knee osteoarthritis in patients with end-stage disease awaiting joint replacement have compared duloxetine with placebo rather than with usual care.

Aim: The study aimed to assess the effectiveness of duloxetine compared to usual care in improving pain outcomes, functional status, quality of life and reducing symptoms of knee and hip osteoarthritis.

Method: Subjects with knee or hip osteoarthritis experiencing pain with neuropathic features and waiting for total joint arthroplasty were recruited to this open-label study. In total, 111 patients were randomised. Knee osteoarthritis patients accounted for 55% of the study group. Subjects were randomised to receive either duloxetine or standard care with medications already prescribed by their physician for 10 weeks. For subjects receiving duloxetine, dosing commenced at 30 mg daily for the first week and was then uptitrated to the target 60 mg dose.

Towards the end of the study, the dose was weaned to 30 mg daily for two weeks prior to discontinuation, specifically to reduce the risk of withdrawal symptoms.

Results: At the end of the treatment phase, the subjects who received duloxetine reported greater perceived improvement in pain (mean difference 11.3 points, p < 0.001), other functional symptoms (mean difference 9.2 points, p = 0.004), ability to perform activities of daily living (mean difference 10.5 points, p = 0.001), neuropathic-like symptoms (mean difference 3.6 points, p = 0.001) and perceived change in overall well-being (p < 0.001) compared to the control group. After the tapering phase, subjects receiving duloxetine were still demonstrating statistically significant perceived improvement in pain (mean difference 6.7 points, p = 0.017), neuropathic-like symptoms (mean difference 2.1 points, p = 0.04), pain at rest (mean difference 17.7 points, p < 0.001) and during movement (mean difference 10.4 points, p = 0.004), and perceived change in overall wellbeing (p < 0.001). The overall benefit was driven by the knee subgroup, with subgroup analyses demonstrating that there were no statistically significant differences in outcomes measured between the duloxetine intervention and the care-as-usual groups at any time point in the hip osteoarthritis subgroup.

The authors hypothesised that duloxetine treatment could result in postponement of joint replacement surgery in responsive end-stage knee osteoarthritis patients.

Adverse effects occurred in 54 out of 57 subjects in the duloxetine group, but the absence of corresponding adverse-effect data for the control group limits comparative assessment.

Key limitations: There was a substantive number of eligible subjects that declined to participate in the study, leading to selection bias. Because this was an open label study a potential contribution to the placebo effect of duloxetine in improving osteoarthritis symptoms must be taken into account.

Participants in the duloxetine arm had a higher mean body mass index than those in the control arm and had suffered osteoarthritic pain for a longer duration but had better functional status and quality of life at baseline and lower baseline pain symptoms.

Impact on practice: This study highlights the efficacy of duloxetine in improving osteoarthritis symptoms and overall wellbeing in patients with knee osteoarthritis when receiving an adequate treatment dose. Further studies are required to confirm these findings.

Duloxetine is not Therapeutic Goods Administration-approved for osteoarthritis pain; the approved pain indication is diabetic peripheral neuropathic pain.

People who smoke require a higher dose of duloxetine to achieve therapeutic effect due to cytochrome P450 1A2 (CYP1A2) induction and increased drug metabolism.7 Therefore, the treatment effects of duloxetine in this study may have been underestimated as over a quarter of the subjects in the duloxetine arm were smokers.

Follow up will include the collection of patients’ postoperative results to evaluate whether duloxetine could contribute to a reduction in residual postoperative pain in knee osteoarthritis patients, given it is known that a lower level of preoperative central sensitisation is linked with patients requiring less pain relief post joint replacement.8

References

  1. Australian Institute of Health and Welfare (AIHW). Chronic musculoskeletal conditions: osteoarthritis [updated 17 June 2024]. Canberra: AIHW; 2014. Available from: https://www.aihw.gov.au/reports/musculoskeletal-conditions/osteoarthritis. Accessed 11 August 2026.
  2. Arthritis Australia. Time to move: osteoarthritis. A national strategy to reduce a costly burden. Broadway, NSW: Arthritis Australia; 2014. Available from: https://arthritisaustralia.org.au/wp-content/uploads/2026/06/Final_Time_to_Move_Osteoarthritis.pdf. Accessed 11 August 2026.
  3. The Royal Australian College of General Practitioners (RACGP). Guideline for the management of knee and hip osteoarthritis. 2nd edition. East Melbourne: RACGP; 2018. Available from: https://www.racgp.org.au/getattachment/71ab5b77-afdf-4b01-90c3-04f61a910be6/Guideline-for-the-management-of-knee-and-hip-osteoarthritis.aspx. Accessed 11 August 2026.
  4. Kolasinski SL, Neogi T, Hochberg MC, Oatis C, Guyatt G, Block J, et al. 2019 American College of Rheumatology/Arthritis Foundation guideline for the management of osteoarthritis of the hand, hip, and knee.  Arthritis Rheumatol 2020; 72: 220—233.
  5. Queensland Government. Chronic Conditions Manual: Section 4 – management of diagnosed conditions. Osteoarthritis. Brisbane: State of Queensland; 2024. Available from: https://www.ccm.health.qld.gov.au/management-of-diagnosed-conditions/osteoarthritis. Accessed 11 August 2026.
  6. Chappell AS, Ossanna MJ, Liu-Seifert H, Iyengar S, Skljarevski V, Li LC, et al. Duloxetine, a centrally acting analgesic, in the treatment of patients with osteoarthritis knee pain: a 13-week, randomized, placebo-controlled trial. Pain 2009; 146: 253-260.
  7. Fric M, Pfuhlmann B, Laux G, Riederer P, Distler G, Artmann S, et al. The influence of smoking on the serum level of duloxetine. Pharmacopsychiatry 2008; 41: 151-155.
  8. Thakur M, Dickenson AH, Baron R. Osteoarthritis pain: nociceptive or neuropathic? Nat Rev Rheumatol 2014; 10; 374-380.

Blikman T, Rienstra W, van Raaij TM, ten Hagen AK, Dijkstra B, Zijlstra WP, et al. Duloxetine in osteoarthritis (DOA) study: effects of duloxetine on pain and function in end-stage hip and knee OA – a pragmatic enriched randomized controlled trial. BMC Musculoskelet Disord 2022; 23: 115.

Back to top

GERIATRIC MEDICINE

MedsScan Editors for Geriatric medicine SPG: Anna Jennings and Jess Coddo

Therapy for atrial fibrillation in patients with drug‑eluting stents

Guidelines recommend non-vitamin K antagonist oral anticoagulant (NOAC) monotherapy for patients with atrial fibrillation (AF) who are more than 12 months post drug‑eluting stent (DES) implantation. Despite this, evidence remains limited and many patients remain on combination therapy (NOAC plus antiplatelet). For older adults, the risk of bleeding complications from antithrombotics is high, and unnecessary use of multiple agents should be avoided.

This multicentre, open label noninferiority trial conducted in South Korea randomly assigned patients with AF and stable coronary artery disease (CAD) at least one year post DES in a 1:1 ratio to receive either NOAC monotherapy or NOAC plus clopidogrel. The primary endpoint was net adverse clinical events (NACE), which included a composite of death from any cause, myocardial infarction, stent thrombosis, stroke, systemic embolism, major bleeding or clinically relevant nonmajor bleeding.

Of the 960 patients included, older patients were well represented with a mean age of 71 years. At 12 months, 46 patients from the monotherapy group had suffered from a primary endpoint event compared to 82 patients from the combination therapy group (hazard ratio [HR] 0.54, 95.2% confidence interval [CI] 0.37-0.77, p < 0.001 for noninferiority, p < 0.001 for superiority).

These findings align with previous trials and suggest that discontinuing antiplatelet therapy at least 12 months post DES in patients with stable CAD concurrently taking a NOAC for AF, is not associated with an increased risk of ischaemic events and will reduce the risk of bleeding. Geriatric medicine pharmacists should actively assess the ongoing need for combination therapy in their patients, especially given the increased risk of bleeding in this population. Decision making should always be individualised, with other indications for antiplatelets (such as unstable disease) considered prior to deprescribing.

Lee SJ, Yu HT, Lee YJ, Lee SH, Heo JH, Ahn SG, et al. Therapy for atrial fibrillation in patients with drug‑eluting stents. N Engl J Med 2026; 392: 1234‑1245.

To sleep or not to sleep: is melatonin the answer?

Up to 60% of older adults with mild cognitive impairment (MCI) or dementia have sleep-wake disturbances, which can lead to accelerated cognitive decline, impaired function and increased carer burnout. There have been numerous studies evaluating the use of melatonin in healthy adults, however the benefits of melatonin use in older adults with cognitive impairment are still widely unknown.

This systematic review and meta-analysis were conducted according to the Preferred reporting items for systematic reviews and meta-analyses (PRISMA) guidelines. Inclusion criteria comprised randomised controlled trials (RCTs) evaluating the effects of melatonin in adults older than 65 years of age with cognitive impairment. Primary outcomes examined were total sleep time in minutes, sleep efficiency (%) and global cognitive performance (such as mini-mental state examination [MMSE], Montreal cognitive assessment [MoCA]).

From the 10 studies analysed, there were 516 patients (57% of those were female, with a mean age of 68.8-86.0 years) included in the overall results. Doses of melatonin ranged from 0.5 mg to 10 mg, and these were given over 2-24 weeks. Total sleep time was reported in eight studies, where the pooled main difference was +12.4 min (95% CI 8.1-16.7, p < 0.001) when compared with placebo. Sleep efficiency was reported in six studies. Sleep efficiency with melatonin was improved by 5.8 percentage points (95% CI 3.2–8.4, p < 0.001). Global cognitive performance was evaluated in five studies, with a pooled mean difference of +1.8 point (95% CI 0.9–2.7, p = 0.002) on the MMSE in patients given melatonin. Melatonin was well tolerated, with mild headache and dizziness occurring in 5% of patients. 

Although the results show statistical significance in favour of melatonin, its clinical significance is yet to be seen in older adults with cognitive impairment; suggesting larger RCTs are required to provide some definitive answers to the age-old question: is melatonin the answer?

Mdluli NT, Banda KJ, Chang Y-C. Melatonin for sleep and cognitive outcomes in older adults with cognitive impairment: a meta-analysis of randomised controlled trials. Age Ageing 2025; 54: afaf333. 

The OZ-ABCD: consensus developed mnemonic for high-risk medicines in residential aged care   

Special contributor: Hannah Bakker

The Australian residential aged care population continues to grow each year. Among this growing population there is a median of three medication-related adverse events per patient over a 12-month period. Many high-risk medication lists are already available: this study aimed to generate a consensus list of high-risk medications specifically targeted to this Australian patient group.

The study used a scoping review to select literature that included a list or examples of high-risk medications in a residential care setting, and an invitational sampling strategy sent via email to healthcare practitioners who were experienced in residential aged care in Australia. In all, 42 participants from various health professions completed round 1 and 35 (83%) completed all 3 rounds of this modified Delphi consensus method study. The consensus method led to 15 groups of medications/medication classes which were subsequently ranked by participants in the final stage.

This final stage developed the mnemonic ‘OZ-ABCD’ to identify the top 15 high-risk medications/medication classes for residential aged care patients in Australia.  While the study identified the usual suspects found on high-risk medication lists around the world (opioids, chemotherapeutic agents, insulin, anticoagulants), anti-infective agents and potassium were not prioritised, whilst hypnotics were identified as a much higher risk due to overall commonality of use in this particular population group.  Given that commonality of use was a key difference between other more generalised high-risk medication lists, the OZ-ABCD tool may allow pharmacists to have a greater impact in minimising medication-related harms specifically for the residential aged care population in Australia.

Readers are encouraged to view the OZ-ABCD high-risk medication list for Australian residential aged care (Box 1) in the study, available via https://onlinelibrary.wiley.com/doi/10.1111/ajag.70141?af=R.

Cross AJ, Chaudhry M, Goordeen D, Breen JL, Clark M, Daly S, et al. Development of a high-risk medication list for Australian residential aged care: A modified Delphi study. Australas J Ageing 2026; 45: e70141.

Back to top

MENTAL HEALTH 

MedsScan Editor for Mental health SPG: Amy Sieff

Concurrent lithium and electroconvulsive therapy: evidence of adverse effects on cognition

Standard clinical guidelines typically recommend withholding a lithium dose prior to electroconvulsive therapy (ECT) due to potential adverse cognitive effects if combined. 1 This systematic review and meta-analysis sought to examine available evidence on neurocognitive changes and confusional states in the context of ECT and lithium use.

Eight databases were searched from inception to late 2025. Studies met inclusion criteria if they compared adults undergoing lithium therapy and ECT with adults having only ECT, and if they examined either cognitive changes as assessed by standard neurocognitive tests or occurrences of confusional states. Reference lists were checked for further relevant articles and case reports were included in the analysis if notable. This yielded 16 studies and 50 case reports for review.

The authors did not find a statistically significant increase in adverse effects on cognition with ECT and lithium compared to ECT alone. However, they were hesitant to recommend changes to current usual practice and highlighted the importance of individualised care. The authors noted the limitations on their work such as difficulty directly comparing the small number of studies included due to their methodologies, and patient-specific factors and variables that were challenging to account for. The significant proportion of case reports included may detract from the strength of conclusions drawn, given they rank lower in standard evidence hierarchies.

References

  1. NSW Health. Electroconvulsive therapy handbook: implementing the practice requirements for ECT. Version 1.1 [24 March 2026]. St Leonards, NSW: State of New South Wales NSW Ministry of Health; 2026.

Van den Eynde L, Takamiya A, Vansteelandt K, Obbels J, Denayer N, Verspecht S, et al. Lithium, electroconvulsive therapy and cognition: A systematic review and meta-analysis. Acta Psychiatr Scand 2026; 154: 116-130.

Effectiveness of hormone therapy for mood symptoms in perimenopause and menopause

The significant impact that conditions such as, but not limited to, endometriosis, perimenopause and menopause can have on women’s mental health and wellbeing has increasingly been recognised in recent times. Specialised women’s health clinics continue to emerge in Australia and around the world.

This retrospective cohort study occurred at one large UK specialist clinic over a seven-month period. It examined the effect of body-identical transdermal oestradiol and testosterone, and synthetic progestin delivered via intrauterine device (IUD), on depressive and mood symptoms associated with menopause. Nine hundred and twenty peri-menopausal and post-menopausal women already using, or about to commence using, hormone therapy were included. The menopause depression rating scale (Meno-D) questionnaire was used for evaluation at baseline and follow up.

The authors concluded that all studied forms of menopausal hormone therapy were linked with notable improvements in mood for participants. However, this study had several drawbacks due to both its design and execution. For example, while it was a real-world, retrospective study, this did not preclude the use of a control group and authors chose not to have one. Given the finding that participants who were hormone naïve at baseline were more likely to experience a bigger improvement, a placebo or standard treatment arm may have therefore been particularly valuable. There was also insufficient evaluation of the effect of concurrent antidepressant use, despite use being documented at baseline reviews; and nearly one-third of participants reporting a past history of, or current depression. The authors appropriately acknowledged the need for randomized clinical trials of longer duration to comprehensively assess the effects of this therapy.

Glynne S, Kamal A, McColl L, Newson L, Reisel D, Mu E. Transdermal oestradiol and testosterone therapy for menopausal depression and mood symptoms: retrospective cohort study. Br J Psychiatry 2026; 5: 409-418.

Use of cannabinoids in mental illness or substance use disorders

Special contributor: Judy Longworth

Given the increase all over the world in access to cannabinoids, and their use in a wide range of mental health and substance use disorders (SUDs), there is a need for high-quality research to substantiate the claims that are made. This paper revealed a need for more high-quality research and identifies a substantial level of bias in the available studies to date. Bias was measured using Cochrane Risk of Bias 2.0, and the quality of the studies was assessed using the GRADE [Grades of Recommendation, Assessment, Development and Evaluation] framework.

The products studied included registered cannabinoids (high quality) that could increase the risk of developing cannabis use disorder as a long-term harm, which is not addressed at all in the short-term studies included in this systematic review and meta-analysis. Efficacy of cannabinoids in comparison to long standing therapies such as cognitive behavioural therapy (CBT), which have well established efficacy, has also not been addressed.

Although anxiety, depression, post-traumatic stress disorder (PTSD) and SUDs are among the leading reasons for Australians’ use of medical cannabis, included studies showed that there were no significant effects on any primary outcomes, except for cannabis use disorder. The authors found no randomised controlled trials (RCTs) assessing cannabinoids in the treatment of depression.

Looking at some mental health disorders from insomnia through to Tourette’s and autism spectrum disorder where cannabinoids were assessed as primary trial treatment, there were no benefits shown, except for a reduction in autistic traits, which disappeared when subgroup analysis was performed based on cannabinoid (mixed cannabidiol and delta-9-tetrahydrocannabinol [THC] versus cannabidiol alone). There was also a high risk of bias, as there was in 24 (44%) of the 54 studies included. Four studies looked at the efficacy of cannabinoids in patients with insomnia. There was some reduction in insomnia symptoms and slight improvement in sleep time which diminished when the studies at high risk of bias were excluded. Additionally, there were high rates of adverse events in the cannabinoid groups in these insomnia studies.

With little evidence of efficacy in this large systematic review and meta-analysis, except for promising evidence in treatment of cannabis use disorder, it is clear further long-term studies are needed to justify the use of cannabinoids in mental illnesses. This is a significant counselling point when pharmacists are discussing their potential use with patients and other health professionals.

Wilson J, Dobson O, Langcake A, Mishra P, Bryant Z, Leung J, et al. The efficacy and safety of cannabinoids for the treatment of mental disorders and substance use disorders: a systematic review and meta-analysis. Lancet Psychiatry 2026; 13: 304-315.

Back to top

PAIN MANAGEMENT

MedsScan Editor for Pain management SPG: Jeremy Szmerling

National standards aim to improve pain management education

The Faculty of Pain Medicine (FPM), Australian and New Zealand College of Anaesthetists (ANZCA), has released the Australian standards for health practitioner pain management education, the first national standards developed to guide pain management education and training across Australian health professions. Developed with funding from the Australian Government Department of Health, Disability and Ageing, the standards form part of the National strategy for health practitioner pain management education1 and are intended to support consistent, evidence-based pain management education across all healthcare disciplines and levels of training.

The standards apply to health practitioners working across health, aged care and disability services and encompass both acute and chronic pain. Six overarching standards are outlined: person-centred care, best practice education, evidence-based content, reflective practice, communication, and a collaborative approach to care. Together, these standards are intended to guide the planning, development, implementation and evaluation of pain management education programs. A strong emphasis is placed on recognising the diversity of people who experience pain and ensuring that education supports culturally safe, psychologically safe and person-centred care.

Although implementing the standards is voluntary, they provide a framework that may be used by universities, professional colleges, healthcare organisations, educators and regulatory bodies when developing or reviewing pain management curricula. For pharmacists, the standards reinforce the importance of contemporary pain science, interdisciplinary collaboration, effective communication and reflective practice in pain management. The document may serve as a useful benchmark for evaluating existing pain education programs and identifying opportunities to strengthen workforce capability in pain management.

References

  1. Faculty of Pain Medicine, Australian and New Zealand College of Anaesthetists (ANZCA). National strategy for health practitioner pain management education. Melbourne: ANZCA; 2026. Available from: https://www.health.gov.au/sites/default/files/2023-10/national-strategy-for-health-practitioner-pain-management-education.pdf.

Faculty of Pain Medicine, Australian and New Zealand College of Anaesthetists (ANZCA). Australian standards for health practitioner pain management education. Melbourne: ANZCA; 2026. Available from: https://www.anzca.edu.au/australian-standards-for-health-practitioner-pain-management-education.

Buprenorphine and its place in post-operative pain management

Special contributor: Tina Xiao

In Australia, full agonist opioids remain key components in the management of post-operative pain. Alternatives like buprenorphine, a partial μ-opioid agonist, may offer comparable analgesia with potential safety advantages.

This systematic review and meta-analysis evaluated buprenorphine versus other opioids for adult post-operative pain management, and included comparisons with full agonist opioids. Reviewers independently screened 2421 records against inclusion criteria and included 58 studies reporting validated quantitative post-operative pain assessments in adults who received either buprenorphine or a full agonist opioid. These studies included a variety of opioids in multiple formulations, with morphine being the most studied comparator opioid, in 33 studies.

Analgesic efficacy was evaluated across 24 randomised controlled trials, of which 23 studies found that buprenorphine provided analgesia equivalent to full agonist opioids, with a mean duration of analgesia of 8.5 h (standard deviation [SD] ±1.84). Mean pain intensity was lower in the buprenorphine group than in patients receiving other opioids (3.86 ± 3.05 versus 4.66 ± 3.18, 41 comparisons, n = 2587). In 19 studies (n = 1535), buprenorphine also demonstrated a reduced need for additional rescue analgesia (odds ratio [OR] = 0.40, 95% confidence interval [CI] 0.26–0.63). Adverse effects were similar between groups, with no significant differences noted.

This review highlights that buprenorphine has comparable efficacy to full agonist opioids in the management of post-operative pain, while offering lower pain scores and reduced rescue analgesia requirements. For Australian pharmacists, these findings suggest that buprenorphine may have a broader role in pain management than its traditional use in post-operative patients who are nil by mouth. However, the results do not support buprenorphine as a universal replacement for full agonist opioids. Rather, they support its consideration as an alternative opioid analgesic in selected patients, alongside assessment of the setting, suitability of the route of administration, expected analgesic needs and local prescribing experience.

Hickey TR, Costa GP, Oliveira D, Podosek A, Abelleira A, Avila-Quintero VJ, et al.Buprenorphine versus full agonist opioids for acute postoperative pain management: a systematic review and meta-analysis of randomized controlled trials. Reg Anesth Pain Med 2026; 51: 1-16.

Why rural opioid prescribing persists

Opioid prescribing for chronic non-cancer pain remains higher in rural Australia than in metropolitan areas, despite growing awareness of opioid-related harms and limited evidence supporting long term efficacy. To better understand this disparity, the authors conducted qualitative interviews with 17 rural Australian general practitioners (GPs) who had prescribed opioids for chronic pain within the previous 12 months.

Participants represented a range of rural settings and experience levels. Interviews explored prescribing decisions, barriers to alternative pain management strategies, and potential solutions. Two dominant themes emerged. Firstly, systematic constraints such as short consultation times and Medicare funding arrangements encouraged continuation of existing opioid therapy rather than time-intensive discussions about deprescribing. Secondly, limited access to multidisciplinary pain management services, including physiotherapy, pain specialists and other allied health providers, restricted the availability of evidence-based alternatives.

Importantly, participants demonstrated awareness of contemporary chronic pain guidelines and the limitations of long-term opioid therapy. However, many described feeling trapped between patient expectations, limited resources, and a lack of practical alternatives. Several reported that justifying opioid deprescribing was often more difficult than continuing treatment.

This study shifts the focus away from clinician knowledge deficits as the primary driver of rural opioid prescribing. Instead, it highlights the influence of structural barriers within the healthcare system. For pharmacists, the findings reinforce the importance of advocating for multidisciplinary pain management services and recognising the contextual challenges faced by rural prescribers. Interventions aimed solely at reducing opioid prescribing may have limited impact unless accompanied by improved access to non-pharmacological treatments, specialist support, and models of care that facilitate complex pain management discussions.

Thomas JA, Benson J, Davidson P, Ward PR. Opioids and the challenges of managing chronic non-cancer pain in rural Australia: a qualitative study. Med J Aust 2025; 223: 467–471.

Back to top

PHARMACY INFORMATICS AND TECHNOLOGY

MedsScan Editor for Pharmacy informatics and technology SPG: James Grant

Hospital pharmacist-conducted medication order verification shows improved safety and potential financial benefits

Special contributor: Wen Kin Huang

Medication order verification involves reviewing medication orders to ensure appropriate drug selection, dose, frequency, route of administration, allergies, drug interactions, contraindications, therapeutic duplication, patient-specific factors and medication compatibility before medicines are supplied or administered.

This systematic review evaluated the effects of pharmacist-conducted medication order verification in hospital settings. Thirty-five studies published between 2000–2023 were included, with studies conducted across Europe, North and South America, Asia, the Middle East and Africa. Most studies were cross-sectional (n = 28), with seven pre-post studies included in the analysis. The overall quality of evidence was low, with only three studies rated as moderate quality, using the Grading of Recommendations, Assessment, Development and Evaluation (GRADE) framework. Studies evaluating antimicrobial or anticoagulant stewardship programs were excluded, meaning the findings relate to routine medication order verification rather than specialised stewardship interventions.

The review found that pharmacist-led medication order verification consistently identified clinically significant medication-related problems, including incorrect dosing, inappropriate medication selection, prescribing outside clinical guidelines, contraindications, administration errors, incomplete medication orders and medicines without a clear indication. Physician acceptance of pharmacists' recommendations ranged from 26% to 100%, while interventions were required for between 0.8% and 45.2% of verified medication orders.

Economic outcomes were favourable in the seven studies that evaluated them. Reported benefits included reduced medicines expenditure through avoidance of unnecessary high-cost therapies, reductions in prescribing errors and preventable adverse drug events, and reported net economic benefits related to pharmacy staff time investment. 

No studies evaluated humanistic outcomes such as patient satisfaction or quality of life. The authors concluded that pharmacist-conducted medication order verification contributes positively to medication safety and may provide economic benefits, although the current evidence base is limited by the predominance of low-quality observational studies. Further high-quality research is needed to evaluate clinical outcomes, patient-centred outcomes, economic benefits and the impact of medication order verification in outpatient settings.

This review supports the important role of pharmacists in medication order verification as a strategy to identify prescribing errors and prevent medication-related harm. However, the evidence remains predominantly observational and of low certainty. Review of the included studies suggests that most were conducted in electronic prescribing environments although digital solutions were not consistently reported. Future studies in this space would benefit from the global trend of an increasingly digitised medications management environment.

Kallio MM, Kumpu-Huhtala A, Kvarnström K, Niittynen I, Lapatto-Reiniluoto O, Airaksinen M, et al. Effects of pharmacist-conducted medication order verification in a hospital setting: a systematic review. Eur J Hosp Pharm 2026 [online ahead of print]. DOI: 10.1136/ejhpharm-2025-004825.

Controlled substance dashboard: could combining pharmacy inventory systems and electronic medical records completely replace our registers?

Special contributor: Stewart Cockram

This study evaluated whether a new model for controlled substance accountability could provide an equivalent or better system to paper registers: integrating pharmacy distribution and electronic medical record (EMR) administration data into a dashboard. The study was undertaken in the largest tertiary teaching hospital in the Netherlands using a pre-, during- and post-implementation design between 2018–2024. The intervention involved development of a dashboard, built using Power BI (Microsoft Corp., Redmond, WA, USA), integrating pharmacy dispensing transactions with EMR medication administration records. A ’delta‘ value was calculated as the difference between quantities dispensed and administered to approximate ward-level balances, with a discrepancy threshold set as a delta of 30 units (approximately one pack) over a three-month period.

The dashboard was piloted in a single ward over six months, running concurrently with paper registers, before optimisation and hospital-wide rollout. The dashboard underwent extensive validation, with no discrepancies identified between source electronic health records data and dashboard outputs, supporting technical reliability.

The analysis included eight generic medicines (in several forms), across 21 clinical wards, 44 820 dispensing/return transactions and 432 133 medication administrations. Paediatrics, intensive care units and operating theatres were excluded, as were a small number of anomalous transactions (e.g. clear data-entry errors). Only oral solid dosage forms were included, with doses standardised into comparable units.

The study demonstrated the feasibility of the dashboard, with the integrated data able to approximate controlled substance balances at a systems level. Measured accountability improved from 88% pre-intervention to 91% post-implementation, indicating better alignment of dispensing and administration records compared to the prior paper-based system. Paper-based registers were discontinued following implementation. Nursing staff preferred the new workflow. Pharmacists also preferred the new system, though noted that discrepancy investigations remained time-consuming.

However, application of this study to an Australian context has some challenges due to differing regulatory expectations. While every Australian state and territory has slightly different regulations, all require extensive transaction-level recordings of controlled substance balances. The dashboard concept is far less granular, and the delta threshold to trigger an investigation is unlikely to meet expectations in any Australian jurisdiction. In addition, while the authors reported improved efficiency and reduced administrative workload, these outcomes were not quantitatively reported.

Nevertheless, this study highlights an important distinction between legislative compliance and meaningful monitoring for misappropriation. In current Australian practice, pharmacy and nursing staff spend considerable time and effort resolving minor documentation discrepancies that are unlikely to represent true misappropriation. The dashboard model considers accountability as system-level surveillance, potentially allowing identification of significant discrepancies, rather than chasing every small documentation error. While ’non-compliant‘ with current Australian legislation, it arguably aligns more closely with the actual intention: identifying genuine misappropriation in significant quantities. 

This study provides an intriguing proof-of-concept for controlled substance accountability through integration of EMR and dispensing data. Implementation in Australian systems would depend not only on technical aspects, but on whether legislative frameworks can be adapted to support more risk-based approaches to controlled substance governance.

Botma FE, Wilkes S, Abdulla A, et al. Digitising controlled substance accountability: implementation and use of a dashboard-based monitoring system in a university hospital. Eur J Hosp Pharm 2026 [online ahead of print]. DOI: 10.1136/ejhpharm-2025-004735.

Specialisation not always more expert: broader artificial intelligence tools beating the narrower

Special contributor: Michael Bakker

This study tackled a critical dilemma for modern health systems: balancing the perceived safety and institutional legitimacy of specialised clinical artificial intelligence (AI) against general-purpose frontier large language models (LLMs). 

The researchers evaluated the performance of leading generalist AI tools against sophisticated clinical AI tools, which typically utilise Retrieval-Augmented Generation (RAG), across an array of benchmark tests and real-world clinical scenarios. One might ask how a RAG-enabled clinical AI differs from a frontier model like ChatGPT? Simply put, a RAG architecture retrieves relevant external information (such as medical literature or clinical guidelines) to augment the prompt it receives, though it still relies on its underlying memory to reason and generate the final output. While this approach promises domain-specific accuracy, the authors note that RAG can negatively impact performance if irrelevant material is retrieved or poorly integrated by the base model. 

To test these architectures, investigators developed a comprehensive evaluation comparing frontier models (OpenAI GPT-5.2, Google Gemini 3.1 Pro, and Anthropic Claude Opus 4.6) against commercial clinical LLMs (OpenEvidence and UpToDate Expert AI). The methodology spanned three stages: 

  • MedQA: testing baseline medical knowledge. 
  • HealthBench: measuring alignment with expert clinicians via a cross-model meta-evaluation, where a panel of LLMs acted as judges to aggregate consensus. 
  • Real Clinical Queries (RCQ): a real-world test utilising 100 actual, deidentified queries from physicians in a live clinical environment. 

For the RCQ benchmark, the researchers utilised 12 blinded American clinicians who produced 1800 manual model-question annotations to evaluate the outputs. A real-world control, the auto-enabled Google Search AI Overview, was also included in this stage as it is frequently encountered by physicians. 

The findings were decisive: general-purpose frontier LLMs outperformed the specialised clinical AI tools on both the MedQA and HealthBench benchmarks. Furthermore, in the RCQ evaluation, all 12 evaluators consistently ranked the frontier models above the clinical AI tools. 

The specialised clinical AI tools lagged in every evaluation and performed only comparably to the auto-enabled Google Search AI Overview. Qualitative feedback revealed that clinical AI tools sometimes struggled with communication, with OpenEvidence scoring lowest on ’clarity for clinicians’, and occasionally provided disorganised answers, incomplete clinical content, or omitted safety-critical information.

The superior performance of frontier models may stem from faster iteration cycles, larger training corpora, and better inherent cross-domain reasoning capabilities. Importantly, the authors are keenly aware of the hype surrounding these tools, cautioning that these results represent a snapshot of a rapidly evolving landscape rather than a permanent hierarchy. They also raise valid concerns regarding benchmarking itself; there is an ongoing risk that industry-created benchmarks may systematically favour the systems developed by their creators, reinforcing the urgent need for independent, health-system-grounded evaluations. 

Overall, this study raises a vital strategic question for pharmacy practice and health systems investing in AI infrastructure. Currently, there appears to be no distinct performance benefit to utilising these likely more expensive, specialised clinical applications. If general frontier models outcompete specialised applications on clinical knowledge and clarity, the institutional legitimacy presumed to accompany clinical AI may be insufficient to justify their adoption. 

Is the primary barrier to utilising general LLMs in a clinical context no longer one of competency, more one of enterprise data protection and regulatory oversight? As the authors suggest, the most viable path forward may involve hospital-specific LLMs that leverage local institutional data for sensitive operational tasks, paired with the careful use of frontier models for broader, less-sensitive clinical reasoning. While generative AI will play a massive role in clinical care, the question of where best to place your dollar when planning for enterprise AI utilisation is becoming increasingly complex. 

Vishwanath K, Alyakin A, Ghosh M, et al. General-purpose large language models outperform specialized clinical AI tools on medical benchmarks. Nat Med 2026; 32: 2405–2409.

Back to top

RESEARCH

MedsScan Editors for Research SPG: Nazanin Ghahreman-Falconer and Elizabeth McCourt

The importance of considering the limits of statistical analysis when establishing causality, and the need for robust study design

Special contributor: Benita Suckling

This feature article, authored by Feeney and Zivich, explains that medical research relies heavily on causal inference for clinical decision making. We are constantly making decisions about how different factors, or variables as they are referred to statistically, are likely to influence outcomes of interest. This process of causal inference is central to health research but remains challenging due to complex relationships between variables.

Establishing credible cause-and-effect relationships requires more than simply identifying associations between variables. It needs careful consideration of how those variables interact within a broader system. Feeney and Zivich argue understanding cause and effect in health is essential to answering clinical queries and the article describes how it is imperative to understand the concept of ’estimands’. These are proposed by a novel framework known as the potential outcomes (counterfactual) framework, which provides a formal way to define and reason about causal effects.

As described in this commentary ‘estimands’ can be defined as “[a] precise description of the treatment effect reflecting the clinical question posed by the trial objective. It summarises, at a population level, what the outcomes would be in the same patients under different treatment conditions being compared.”2(p6) Or put more simply it defines what effect you want to estimate in a clinical trial. It links:

  1. the clinical question (what we want to know), and
  2. the treatment effect (what we measure and compare).

The paper highlights how crucial it is to do this correctly as simply applying statistical methods to a dataset without understanding the underlying interplay of variables (i.e., the timeline of events, other confounding factors and potential sources of bias) can lead to incorrect, and potentially dangerous, conclusions.

For example, the authors highlighted a study that initially found that among low‑birthweight babies, those whose mothers smoked seemed to have lower mortality risk.3 To most clinicians and researchers such a finding is counterintuitive. The mistake came from only analysing low‑birthweight babies where low birthweight may be due to a variety of conditions.  Birthweight is not just a background variable, but it is caused by smoking and related to infant death.  This makes birthweight a mediator (or a variable on the causal pathway).

Causal pathways can be explored by mapping relationships using tools such as directed acyclic graphs (DAGs). Using such tools, researchers can understand potential confounding pathways and make more robust causal inferences. This is vital in rigorous scientific research.

References

  1. International Council for Harmonisation (ICH) of Technical Requirements for Pharmaceuticals for Human Use. ICH E9 (R1) addendum on estimands and sensitivity analysis in clinical trials to the guideline on statistical principles for clinical trials. Step 5. Amsterdam: European Medicines Agency; 2020. Available from https://www.ema.europa.eu/en/documents/scientific-guideline/ich-e9-r1-addendum-estimands-and-sensitivity-analysis-clinical-trials-guideline-statistical-principles-clinical-trials-step-5_en.pdf. Accessed 11 August 2026.
  2. Hernández-Díaz S, Schisterman EF, Hernán MA. The birth weight “paradox” uncovered? Am J Epidemiol 2006; 164: 1115-1120.

Feenay, T, Zivich PN. Causal inference is hard, and advanced statistical analysis is not enough. BMJ 2025; 391: r2618.

A different perspective on research worth publishing

The JAMA Health Forum journal most commonly receives study submissions with the most common being retrospective cohort or cross-sectional designs: approximately 50% descriptive and 25% quasi-experimental. However, this editorial refutes the assumption that any single design is superior for policy evaluation, and highlights room for nuance. According to the authors, “[t]he best method is the most appropriate method to answer the study question with the data available and where there is transparency in the method and its assumptions.”

To answer the complex challenges we face in healthcare, some of the less-commonly received study designs can yield valuable insights:

  • qualitative (e.g. semi-structured interviews, focus groups or consensus methods)
  • modelling or simulations
  • reviews (systematic, scoping, meta-analyses) relevant to ongoing policy debates.

Clinical pharmacists are well-positioned with practical experience, interdisciplinary colleagues and access to consumers, to be able to contribute to robust qualitative research. The value of these contributions should not be underestimated, as “[q]ualitative studies not only generate hypotheses for future quantitative studies but also can generate actionable findings that can be translated to other policy, clinical, or public health problems”.

Alternatively, scoping or systematic reviews represent an accessible entry point to research with tasks that can easily be shared across a team. This builds interdisciplinary networks and foundations for future collaboration, particularly if leveraging organisational resources like health librarians and Covidence, where available.

So, if you are a researcher at any stage of your journey focusing on policy or services, here is the invitation to design your health service research in a novel, pragmatic way. Just be sure to save some word count to robustly discuss the assumptions, limitations and confounders in your methods and results!

Kilaru AS, Donohue JM, Galea S. Study designs and methods for health policy and health services research: a look back, a look ahead. JAMA Health Forum 2025; 6: e255191.

Back to top