Pharmacovigilance, clinical trials and essential medicines

Pharmacovigilance is the work of finding, judging, understanding and preventing the harms of medicines, together with any other problem a medicine causes. It begins with the recognition that all medicines carry risks alongside benefits, and that systematic monitoring is required throughout a product's lifecycle. Clinical trials provide the first structured safety data, but they are limited by sample size, duration, and selected populations. Post-marketing surveillance then ext ^[5]

Immune checkpoint inhibitors such as nivolumab, pembrolizumab, atezolizumab, avelumab and durvalumab have transformed oncology but introduced a distinct spectrum of immune-related adverse events. Proper clinical management requires understanding associated immune-related adverse events that can include organ specific adverse events, general adverse events related to immune activation, and adverse events consistent with musculoskeletal problems. Organ-specific events comprise colitis, hepatitis, pneumonitis, hypophysitis or hypopituitarism, and hypothyroidism. General events related to immune activation include fatigue, diarrhoea, and rash. Musculoskeletal problems encompass arthritis, arthralgia, back pain, musculoskeletal pain, and myalgia. Rates of these adverse events are unclear and publications from clinical trials may inconsistently report unexpected adverse events. This uncertainty stems partly from trial design: most phase II and III checkpoint inhibitor trials were open label with primary outcome of overall survival, and adverse event reporting was performed by the clinicians directly caring for patients, making reports unmasked. Because included studies were not designed primarily to assess adverse events, collection of adverse event information was poorly described, and all studies were deemed at high risk of bias with regard to incomplete outcome data and selective outcome reporting. ^[4]

Historical milestones

• The 1985 Nairobi Conference first proposed a comprehensive essential medicines policy framework^[7].
• Thirty years later the Lancet Commission revisited progress, challenges and future directions for these policies^[7].
• In 2004 WHO identified therapeutic gaps and coined “missing essential medicines”^[7].
• The 2007 World Health Assembly Resolution (WHA60.20) called for paediatric dose forms, prompting the first WHO Model List for Children^[7].

Five policy pillars (as identified by the Lancet Commission)

1. Financing a basket of essential medicines – mechanisms to raise funds for the whole set, distinct from pricing strategies^[7].
2. Affordability – ensuring prices are low enough for patients and health systems while remaining sustainable for manufacturers^[7].
3. Quality and safety – guaranteeing that all medicines, essential or not, meet regulatory standards^[7].
4. Quality use – promoting rational prescribing, dispensing and adherence^[7].
5. Development of missing essential medicines – stimulating research and public‑health‑oriented innovation^[7].

Key definitions for pharmacovigilance in the essential‑medicine context

• Adverse drug reaction (ADR): any noxious and unintended response to a medicine at normal doses.
• Signal: a hypothesis‑generating observation that a drug‑event association may exist, requiring further evaluation.
• Reporting rate: number of ADR reports per 1 000 patient‑exposures; a core metric for post‑marketing surveillance.
• Benefit‑risk assessment: systematic comparison of therapeutic benefits against safety risks, informing whether a medicine should remain on the essential list.

Regulatory and institutional framework

• WHO’s Model Lists (adult and paediatric) serve as global reference points for national essential‑medicine lists.
• The Lancet Commission’s recommendations (e.g., promoting affordability while preserving industry sustainability) guide policy‑makers^[7].
• Measurement of pharmaceutical expenditure often cannot separate essential from non‑essential medicines, but the Commission advises that all‑medicine data be collected where possible^[7].

Adverse event reporting in clinical trials follows standardised terminology to allow aggregation and comparison. The Common Terminology Criteria for Adverse Events (CTCAE) prompts investigators to note the presence or absence of a symptom or an abnormal laboratory value and grade it based upon its clinical significance. For data extracted from published reports, Common Terminology of Clinical Adverse Events (CTCAE) categorization to identify grades 3-5 as serious and CTCAE grades 1-2 as other. This grading underpins safety summaries and regula ^[4]

Data sources for adverse event reporting include ClinicalTrials.gov and peer-reviewed publications. For studies where full adverse event information was not on ClinicalTrials.gov, information from the publication was used and study authors or pharmaceutical sponsors were contacted directly for additional information. Adverse events were recorded as either serious or other on ClinicalTrials.gov. When data were available from both sources, ClinicalTrials.gov data were prioritised over publication data because it was more complete. If a study did not report a specific adverse event, it was assumed that the event did not occur. Data from different dosing arms within the same study were extracted and reported separately. ^[4]

Statistical synthesis of event rates across trials uses defined methods. Overall event rates are calculated by dividing the total number of patients across trials with a given adverse event by the total number at risk. For each included study, odds ratios and 95% confidence intervals for event rates in the intervention arm compared with control are calculated based on the reported number of events and sample size. Heterogeneity across trials is examined using the I² index and the Cochran Q statistic. If significant heterogeneity is not present (P > 0.1), a pooled odds ratio and 95% confidence interval are estimated with a fixed effects model using the inverse variance method. This approach allows quantitative comparison of safety profiles across different checkpoint inhibitors and dosing regimens. ^[4]

Despite high rates of adverse events consistent with musculoskeletal problems, reporting of these adverse events was inconsistent and incomplete across trials. Accessing adverse events data on ClinicalTrials.gov enabled inclusion of information that did not appear in publications, however, analysis remained constrained by the recording methods for adverse events in clinical trials. High inter-rater agreement for risk of bias assessments was observed (κ between 0.629 and 1.00 across domains), yet the underlying reporting limitations persist. Clinicians and students should recognise that published trial data may underestimate or misrepresent the true frequency and nature of adverse events, particularly for newer drug classes where safety signals are still emerging. ^[4]

Clinical investigation and post-marketing surveillance rely heavily on structured evaluation frameworks. When calculating production costs and generic pricing for 148 medicines on the WHO essential medicines list, calculation was possible for 148 out of 197 representing 75% of the medicines meeting inclusion criteria. Despite most medicines on the essential list being off-patent, 214 of 277 comparable prices in the UK, 142 of 212 comparable prices in South Africa and 118 of 298 comparable prices in India exceeded the price that would be expected based on cost of production and a 10% profit margin. ^[1]

Average active pharmaceutical ingredient price per kilogram was in the range of US1 to US10 per kilogram for 5 medicines, US10 to US100 for 43 medicines, US100 to US1000 for 58 medicines, US1000 to US10 000 for 19 medicines, and were greater than US10 000 per kilogram for 3 medicines including anastrozole, methotrexate, and entecavir. ^[1]

For injectable formulations on the World Health Organization essential list, medicines had prices above the estimated cost-based price in 77% of comparisons in England with a median ratio of 2.54, and 62% in South Africa with a median ratio of 1.48, while 85% of medicines in India had prices below estimated cost-based price with a median ratio of 0.30. Furthermore, 19% of injectable medicines in England, 9% in South Africa, and 5% in India had prices more than 10 times the estimated cost-based price. ^[3]

Clinical trial methodology often incorporates blood pressure difference trials and drug comparison trials to establish cardiovascular endpoints. Overall, 147 trial reports were included in specific analyses, where 108 were blood pressure difference trials and 46 were drug comparison trials. Estimates from trial meta-analyses demonstrated a 22% reduction in coronary heart disease events with a 95% confidence interval of 17% to 27% and a relative risk of 0.78, alongside a 41% reduction in stroke with a 95% confidence interval of 33% to 48% and a relative risk of 0.59. ^[2]

The WHO Essential Medicines List (EML) is a cornerstone of global public health, designed to identify the medicines that satisfy the priority health-care needs of a population. It is intended to be available at all times, in adequate amounts, in appropriate dosage forms, and at a price the community can afford. The list is updated periodically, with the 2017 edition adding several new patented medicines that had previously been excluded, including dolutegravir, raltegravir, velpatasvir, nilotinib and dasatinib. ^[1]

The concept of essential medicines is not merely about listing drugs; it is fundamentally about equity and access. High prices of essential medicines pose a significant barrier to their use in many different settings. This is particularly acute in low- and middle-income countries where out-of-pocket expenditure dominates health financing. For instance, over 75% of health expenditure in India is out-of-pocket, with the majority spent on medicines. In such contexts, even modest price differences can determine whether a patient can access treatment. ^[1]

The affordability challenge is compounded by patent protection. Depending on the country, as many as 15% of medicines on the current EML may be under patent protection. This creates a tension between innovation incentives and public health needs, particularly for newer, more effective treatments. ^[1]

For injectable formulations on the WHO EML, price-to-cost ratios vary dramatically across settings: 77% of medicines had prices above the estimated cost-based price in England, and 62% in South Africa, while 85% of medicines in India had prices below the estimated cost-based price. Therapeutic categories with notably higher ratios of lowest current price to estimated cost-based price include cardiovascular medicines, medicines for mental and behavioural disorders, hormones, other endocrine medicines and contraceptives, anticonvulsants/antiepileptics, antineoplastics and immunosuppressives, and oxytocics. ^[3]

The WHO commissioned a study to analyse production costs for medicines on the EML, identifying this as a gap in the literature. Notably, no studies had previously examined production costs for all medicines on the WHO EML or any similarly broad group of products. This highlights the ongoing need for transparency in pharmaceutical pricing to support equitable access to essential medicines worldwide. ^[1]

Generic drugs now dominate prescribing in many high-income countries, accounting for nearly nine out of every 10 prescriptions filled in the US and the UK. The fundamental principle is straightforward: a generic contains the identical active ingredient to its branded counterpart, yet can be substantially less expensive. This cost difference matters clinically because lower prices have the potential to improve medication adherence and persistence—patients are more likely to keep taking a medicine they can afford. ^[6]

For a generic to gain approval, manufacturers must demonstrate both pharmaceutical equivalence and bioequivalence to the reference product. Notably, no pre-marketing trials testing clinical equivalence are required. However, robust evidence from multiple post-marketing studies has demonstrated equivalent clinical efficacy for many generic products compared with their branded counterparts. This evidence spans diverse drug classes including β blockers, diuretics, calcium channel blockers, statins, and antiplatelet agents, showing equivalent benefits in both short-term endpoints (blood pressure changes, urine output, platelet aggregation) and long-term endpoints such as atherothrombotic events and healthcare utilisation. ^[6]

A unique category is the authorized generic: a product identical in composition and appearance to the branded drug, but marketed by the brand manufacturer (or its licensees) as a generic, usually around the date of market entry of other generic products. Unlike standard generics, authorized generics are marketed in the US under the branded product's New Drug Application licence without any bioequivalence testing requirements. ^[6]

Real-world data illustrate the clinical impact of switching. Among patients switched from branded to generic products, eight out of every 100 switched back to the branded product within a year. Switchback rates were 28% lower for patients who switched to an authorized generic compared with those who switched to a standard generic. The overall incidence rate difference was 1.5 per 100 person years, ranging from 0.7 per 100 person years for alendronate tablets to 5.4 per 100 person years for amlodipine-benazepril capsules. The similarity in appearance between authorized generics and branded products likely explains this difference. ^[6]

Drug regulation is the lifecycle governance of medicines—from preclinical development, through clinical trials, to marketing authorisation, post-marketing surveillance, and eventually market withdrawal. The goal is to balance patient access to effective therapies against the risks of harm from unsafe, ineffective, or poor-quality products. Regulatory agencies such as the US Food and Drug Administration (FDA) and the European Medicines Agency (EMA) set standards for efficacy, safety, and quality, and they decide whether a new drug may be sold.

A central feature of modern regulation is the distinction between branded (innovator) products and generic products. A generic contains the same active ingredient and is intended to be therapeutically equivalent, but it may differ in appearance and excipients (inactive ingredients). To encourage cost savings, many jurisdictions allow pharmacists to substitute a generic for a branded prescription. However, this practice is not always accepted by patients or clinicians.

The behaviour of patients after switching from a branded to a generic product is an important regulatory and economic concern. In a large US cohort, 94 909 patients switched from branded to authorized generic products and 116 017 switched from branded to generic products, contributing to a switchback analysis. Switchback—returning to the branded product—was common: the overall rate was 8.2 per 100 person years across all drug products, ranging from 3.8 per 100 person years for alendronate tablets to 17.8 per 100 person years for amlodipine-benazepril capsules. Importantly, patients who switched to authorized generics had lower switchback rates than those who switched to ordinary generics (pooled hazard ratio 0.72, 95% confidence interval 0.64 to 0.81). ^[6]

Since authorized generics are identical to the branded product in appearance and excipients, the lower switchback rate among their users suggests that negative perceptions of generics—not pharmacological differences—drive many switchbacks. The economic impact is substantial: switchbacks increase patient expenses by an estimated 1.2bn (£860m; €980m) and overall US healthcare system costs by an estimated 7.7bn (£5.5bn; €6.3bn) annually. One proposed intervention is for generic products to mirror the appearance of branded products, though courts would likely not allow brand manufacturers to claim intellectual property protection over appearance. ^[6]

Regulatory systems must therefore consider not only bioequivalence but also the real-world behaviour of patients and prescribers, since perceptions of product quality influence clinical outcomes and healthcare expenditure.

Pharmacovigilance and essential‑medicine decision‑making rely on simple quantitative tools that translate raw data into actionable metrics.

1. Reporting rate (RR)
   RR = (Number of ADR reports ÷ Number of patient‑exposures) × 1 000

2. Cost‑effectiveness ratio (CER)
   CER = (Cost of medicine per treatment course) ÷ (Quality‑adjusted life years gained)

3. Affordability index (AI)
   AI = (Annual cost of essential‑medicine basket ÷ Gross national income per capita) × 100 %

4. Signal detection (Proportional Reporting Ratio, PRR)
   PRR = (A / (A + B)) ÷ (C / (C + D))
   where A = reports of the event with the drug of interest, B = reports of other events with the same drug, C = reports of the event with all other drugs, D = reports of all other events with all other drugs.

Assume a national pharmacovigilance centre receives 45 ADR reports for a newly introduced paediatric antimalarial over a 12‑month period. The drug is dispensed to 15 000 children during the same interval.

RR = (45 ÷ 15 000) × 1 000 = 3 reports per 1 000 exposures.

Interpretation: a reporting rate of 3 / 1 000 is comparable to the median spontaneous reporting rates observed for well‑established essential medicines, suggesting adequate surveillance.

A generic amoxicillin regimen costs £2.50 per 7‑day course. Health‑technology assessment estimates that each course averts 0.015 quality‑adjusted life years (QALYs) in children with acute otitis media.

CER = £2.50 ÷ 0.015 QALY = £166.7 per QALY.

The WHO threshold for cost‑effectiveness in low‑income settings is often set at three times the gross domestic product (GDP) per capita (≈£1 500). The calculated CER is well below this threshold, supporting inclusion on the essential list.

A low‑income country’s essential‑medicine basket comprises 120 medicines. The total annual procurement cost is £12 million. The country’s gross national income (GNI) per capita is £1 200.

AI = (£12 000 000 ÷ (Population × £1 200)) × 100 %
If the population is 10 million:
AI = (£12 000 000 ÷ (£12 000 000 000)) × 100 % = 0.1 %.

An AI of 0.1 % indicates that the basket consumes a very small share of national income, satisfying the affordability criterion advocated by the Commission^[7].

During post‑marketing surveillance, 30 reports of severe neutropenia are linked to Drug X, while 270 other ADRs are reported for Drug X. For all other drugs, 150 reports of neutropenia and 4 500 other ADRs are recorded. ^[7]

A = 30, B = 270, C = 150, D = 4 500

PRR = (30 / (30 + 270)) ÷ (150 / (150 + 4 500))
PRR = (30 / 300) ÷ (150 / 4 650) = 0.10 ÷ 0.0323 ≈ 3.1

A PRR > 2 with a chi‑square > 4 is generally considered a signal, prompting deeper investigation of Drug X’s safety profile before it can remain on the essential list.

• Reporting rates inform the robustness of pharmacovigilance systems; low rates may trigger capacity‑building measures.
• CER and AI feed directly into the five‑pillar framework: affordability (pillar 2) and financing (pillar 1).
• PRR and other signal‑detection metrics support the quality‑and‑safety pillar (pillar 3) and guide decisions on whether a medicine should be retained, withdrawn, or moved to a “conditional” status on the essential list.

By applying these straightforward calculations, students can appreciate how raw clinical‑trial data, post‑marketing reports and economic figures converge to shape essential‑medicine policies that underpin UHC^[7].

Pharmacovigilance, clinical trials and essential medicines converge on a single clinical reality: patients need safe, effective and affordable treatments, and each pillar of this topic exists to protect that need. The diseases most directly correlated with this topic are those where treatment access, drug safety monitoring and evidence generation intersect—most prominently cardiovascular disease, HIV and hepatitis C.

Cardiovascular disease illustrates the clinical trial dimension. Despite widespread use of blood pressure lowering drugs and many randomised trials, uncertainty remains about which drugs to use and who to treat. A key clinical question is whether β blockers have a special effect beyond lowering blood pressure in preventing coronary heart disease (CHD) events in people with a history of CHD. The evidence now suggests no reason to conclude that β blockers in general, or atenolol in particular, are one trial which implausibly suggests no effect at all (relative risk 1.01, 0.78 to 1.31 ^[2]

HIV and hepatitis C exemplify the access dimension. The TRIPS agreement does not prevent individual patients importing medicines for personal use, regardless of patent status, although national legislation must also permit this. This route is being used by patients to access affordable HIV pre-exposure prophylaxis and hepatitis C drugs, with evidence of good clinical outcomes. However, this approach requires patients to be organised, assertive, and have high-level access to information and financial resources—it cannot be considered a feasible large-scale solution. ^[1]

Pharmacovigilance protects patients taking these medicines. The FDA Adverse Event Reporting System (FAERS), started in 1969 and last majorly revised in 1997, is the largest repository of spontaneously reported adverse events in the world, with more than 4 million reports. Public access allows external researchers to conduct pharmacoepidemiological studies and pharmacovigilance analyses. ^[5]

Finally, essential medicines pricing affects disease outcomes. Of 414 unique medicines on the 2015 EML, 145 fit the inclusion criteria of one analysis, with active pharmaceutical ingredient data available for 96 (66%). Most injectable essential medicines can be manufactured at very low cost; in England, about one in five injectable essential medicines was priced at more than 10 times the estimated cost-based price, in South Africa about one in ten, and in India about one in twenty. High generic prices may reflect anticompetitive practices or single-supplier price hikes. ^[3]

Pharmacovigilance relies on systematic data collection and analysis to detect, assess, understand and prevent adverse effects or any other drug-related problems. The science operates across two complementary domains: pharmacoepidemiology, which studies drug use and effects in large populations, and pharmacovigilance proper, which focuses on signal detection and safety monitoring.

The FDA Adverse Event Reporting System (FAERS) database is a cornerstone resource for analysing associations with adverse events. However, it has two major deficiencies when used for pharmacovigilance, where early and timely detection is critical: the lag time from collection to public release of the data, and the form in which the data are released (ASCII or SMGL files), which is not readily amenable to query or analysis. Spontaneous reports also suffer from missing data, and more than 2 reports can be submitted for one patient, creating duplicated reports that distort signal detection. ^[5]

Data mining algorithms have been proposed to enhance signal detection in large spontaneous reporting databases. In 2003, the Pharmaceutical Research and Manufacturers of America–FDA Collaborative Working Group on Safety Evaluation Tools reviewed best practices for these methods. Their conclusion was measured: there is evidence that data mining may be useful, but the evidence is not sufficient to fully judge its value in pharmacovigilance, and time and experience will reveal its utility. Understanding both strengths and weaknesses of these algorithms is essential to minimise their misapplication and misuse. ^[5]

Regional pharmacovigilance networks complement national databases. In France, activities are carried out by 31 regional centers based in clinical pharmacology departments of university hospitals, and the French Pharmacovigilance Database has been used to analyse safety profiles of various drugs. Two European databases are also frequently used for analytical studies: the General Practitioners Research Database (GPRD) in the UK and the PHARMO Record Linkage System in the Netherlands. Emerging information technology may soon enable real-time analysis at a doctor's office or by a patient. ^[5]

Pharmacological treatment operates across a continuum—from the individual patient's prescription to the population-level impact of drug availability. Understanding how drugs work here requires appreciating three interlocking domains: the evidence base for efficacy, the safety profile in real-world use, and the economic forces that determine access.

Blood pressure lowering drugs illustrate the principle that class effects often matter more than drug selection. The five main classes—thiazides, β blockers, angiotensin converting enzyme (ACE) inhibitors, angiotensin receptor blockers (ARBs), and calcium channel blockers—are equally effective in lowering blood pressure and equally effective in preventing coronary heart disease (CHD) events and stroke. Claims that one class is superior or inferior are not supported by broader meta-analyses; β blockers, for instance, have a similar protective effect on CHD to other drugs and a greater protective effect after myocardial infarction. The practical implication is that the choice between classes matters less than achieving blood pressure reduction itself—blood pressure reduction alone explains the preventive effect of these drugs. ^[2]

The magnitude of benefit depends on the number of drugs, dose, pre-treatment blood pressure, and age. A two-stage analysis combining trial data with cohort studies allows a doctor to estimate how much a given treatment will reduce an individual's risk of stroke and CHD. The proportional effect of blood pressure reduction diminishes with increasing age. Notably, the case for individualising therapy disappears with low dose combination therapy based on three drugs; the greater efficacy compared with selective monotherapy or dual therapy avoids the need to choose between drugs. ^[2]

Cancer immunotherapy with anti-PD-1 drugs such as pembrolizumab has revolutionised care. Pembrolizumab is approved for renal cell carcinoma, head and neck cancer, Merkel cell carcinoma, and tumours of any organ with high microsatellite instability, with further indications under investigation in at least 135 clinical trials. However, these drugs carry a distinct adverse event profile: they are more likely than standard treatments to cause pneumonitis, colitis, rash, and hypothyroidism, but not fatigue or diarrhoea. Musculoskeletal problems are likely under-reported, and true rates of adverse events are probably underestimated due to short follow-up and incomplete capture. ^[4]

Access and essential medicines are shaped by market dynamics. Generic competition achieved massive price reductions for antiretroviral drugs in the early 2000s. Non-communicable diseases (NCDs) now represent a larger disease burden than communicable disease, and similar price reductions in medicines for cancer, type 2 diabetes, and anticoagulation may be valuable in enabling wider treatment of NCDs in low- and middle-income countries. ^[1]

Pharmacovigilance and access to essential medicines look very different depending on where in the world a patient is treated. The WHO Essential Medicines List (EML) contains medicines that can, in most cases, be profitably manufactured at very low cost^[1]. Yet the prices patients and health systems actually pay vary enormously between countries. A comparison of injectable essential medicines found that prices were higher in England than in South Africa in 60% of comparisons, with a median ratio of 1.44; higher in England than in India in 91% of comparisons, with a median ratio of 7.08; and higher in South Africa than in India in 84% of comparisons, with a median ratio of 3.69^[1]. These striking disparities mean that a medicine that is affordable in one setting may be entirely out of reach in another, directly affecting which treatments are available to groups with particular needs: young patients, older adults, people who are pregnant and people on low incomes. ^[3]

Why do such gaps persist? Data on production costs are not publicly available, and health systems have limited negotiating power when a medicine is sold in a monopoly situation^[1]. However, enabling generic competition can produce dramatic falls in price—for example, reductions of 99% in first-line HIV/AIDS medicines^[1]. Generic price estimation, based on manufacturing costs in India, can empower government price negotiations and support cost-effectiveness calculations^[1].

Certain medicines repeatedly appear in top-20 lists for high price-to-cost ratios across multiple countries, including oncology drugs such as irinotecan, ifosfamide, daunorubicin, filgrastim and mesna, as well as ampicillin, acyclovir, quinine, valproic acid and ciclosporin^[1]. For clinicians, this means that even essential, off-patent drugs may be priced far above their production cost, and that prescribing choices should consider local affordability. ^[3]

For special populations, the evidence base is often thinner. Trials frequently exclude pregnant women, children and the frail elderly, so safety data must be extrapolated cautiously. In cardiovascular trials, for example, participants with a history of coronary heart disease were generally not selected by blood pressure and had no blood pressure target; treated patients were allocated a specified drug in fixed dose, varied only to avoid adverse effects. This means that real-world patients—who may be older, frailer or have more comorbidity—are not always well represented in the trials that guide their treatment. A global perspective demands that we ask not only "does this drug work?" but "is it available, affordable and safe for the patient in front of me?" ^[2]

Understanding how regulatory authorities and international advisory panels evaluate healthcare interventions is essential for clinical practice and therapeutics exams. Comparative clinical and cost-effectiveness is assessed both by national regulators such as the UK’s National Institute for Health and Care Excellence and the WHO Expert Committee reviewing proposed additions to the EML. Cost-effectiveness calculation depends on the efficacy and available price for the drug and the comparator. Where a novel drug is being compared with a drug that is near patent expiry, a cost estimation exercise can help anticipate the generic price of the latter. Recent examples where this analysis could have been applied include novel oral anticoagulants compared with warfarin, dasatinib as a second-line after imatinib for chronic myeloid leukaemia and dolutegravir compared with efavirenz for the treatment of HIV. ^[1]

When evaluating medicine costs across different healthcare settings, researchers utilize specific reference materials and government procurement repositories. For England, prices were collected from the British National Formulary (BNF) and the electronic market information tool (eMit). The BNF lists ‘indicative prices’ and is a reference used by clinicians and pharmacists, while eMit provides actual government-purchase prices. The lowest price available across the two sources was used. For South Africa, prices were collected from a database of prices in the public healthcare system as well as a database of prices in the private market, both published by the Department of Health, and the lowest price available across the two sources was used. For India, prices reported in public tenders of the state of Tamil Nadu were used, and where prices were not available in this source, the price reported in an online database of Indian Maximum Retail Prices was used. ^[3]

Pharmacovigilance database mining relies on disproportionality analyses using a two-by-two contingency table as the framework for analysis. A frequent pitfall in postgraduate examinations is confusing the mathematical definitions and cells of this contingency table. The number of co-occurrences of interest is defined as n11. The number of co-occurrences with a drug of interest, but without an adverse of interest, is defined as n10, and the number n01 is assigned to those without a drug of interest, but with an adverse event of interest. The number of co-occurrences without either is defined as n00. ^[5]

Different statistical tests have rigid scoring thresholds that must be memorised for clinical trials safety modules. Using the Proportional Reporting Ratio (PRR), a signal is detected if the number of co-occurrences is 3 or more and the PRR is 2 or more with an associated chi-squared value of 4 or more. For the Reporting Odds Ratio (ROR), a signal is detected if the lower limit of the 95% two-sided confidence interval exceeds 1. Signal detection using the Information Component (IC) is done using the IC025 metric, a lower limit of the 95% two-sided confidence interval of the IC, and a signal is detected if the IC025 value exceeds 0. Finally, for the Empirical Bayes Geometric Mean (EBGM), the EB05 metric, a lower one-sided 95% confidence limit of the EBGM, is used and a signal is detected when the EB05 is greater than or equal to the threshold value 2.0. ^[5]

In routine clinical practice and post-marketing surveillance, adverse event reporting data are subject to temporal and behavioural biases that can distort signal detection. The number of reports and signals, and the signal scores are influenced by various factors. The number of reports increases over the first 2 years after launching, and then starts going down, which is known as the Weber effect, although it is not always observed. Furthermore, the number of signals and signal scores also possibly fluctuate during several years after launching, and the number of unlabeled supported signals depends on the time window after launching. ^[5]

Reporting rates are frequently altered by media attention and public safety announcements. Generally speaking, reporting can be accelerated after a drug-associated adverse event is highlighted, and this is known as the notoriety effect. Additionally, the notoriety for a drug can accelerate the reporting of other drugs in the same class, known as the ripple effect. In contrast, signal scores can be suppressed by a large number of reports in which the same adverse event is connected with other drugs, which is the masking or cloaking effect. Close attention should be paid to the results of signal detection, especially for drugs launched only recently, and it is important to investigate a temporal axis when planning a pharmacovigilance analysis. Whether an adverse event is detected as a signal or not depends on the algorithm used for signal detection. For instance, haematemesis ranked 559th, 375th, and 581st, respectively, according to the ROR score, but based on the ROR and IC, the analysis indicated the association of these drugs with haematemesis; however, it was not associated with warfarin when the PRR was used, and the EBGM failed to detect the associations for warfarin and clopidogrel. ^[5]