approvals , because of the significant potential effect on timelines and development costs . For example , the FDA approved palbociclib , originally licensed for use in women with advanced breast cancer , to include the treatment of male patients , based on RWE from three commercial databases that provided information on the established safety profile of the drug in men . 6 Nevertheless , while RWE is potentially of enormous value , there are limitations which in general relate to a lack of randomisation ( hence studies are at higher risk from both confounders and treatment allocation bias ), the problem of missing data and the simple fact that RWE was not primarily designed for research purposes . These limitations have been used as barriers and disincentives for using RWE in the process of clinical decision making and disease management .
Poor clinical research is poor clinical research , whether it is randomised or non-randomised evidence . Unfortunately , there are many examples of both , even in recent years . For example , as a response to the COVID-19 pandemic , hundreds of RCTs evaluating COVID-19 treatment have been put forward , many of which have sampling errors and sample bias ( lack of proper randomisation methodologies ; the population included is not the population at a higher risk of a bad outcome ) and measurement bias ( small underpowered studies ; short follow-up ; absence of hard outcome measurements ). 7 Similarly , hundreds of nonrandomised studies have also been carried out , some using wrong methodologies and / or databases and / or low-quality data , giving rise to unreliable RWE . One example of the most intense media coverage is hydroxychloroquine , in particular the study conducted in Marseille , France , which studied the use of hydroxychloroquine combined with azithromycin to treat a cohort of 1061 COVID-19 patients . 8 It concluded that the low mortality rate ( 0.5 %) was proof of the treatment ’ s efficacy . However , the mean age of the participants was only 43.6 , meaning that the reported low mortality rate was probably due to the efficient immune system response of younger people and not due to hydroxychloroquine / azithromycin efficacy . 8
In an effort to enhance credibility , the RWE transparency initiative was formed and represents a collaboration between the International Society for Pharmacoeconomics and Outcomes Research ( ISPOR ), the International Society for Pharmacoepidemiology ( ISPE ), Duke-Margolis Center for Health Policy and the National Pharmaceutical Council . The RWE transparency initiative has made recommendations on good procedural practices to enhance to decision makers ’ confidence in RWE . 9 In an effort to limit some of the sources of bias in real-world studies , a prevalent new-user study design can be employed and used to evaluate the effect of a treatment . For example , using this design in a cohort study allows for a comparison between a new drug and a comparator including participants who have been switched to the new drug . This improves the generalisability of RWE 9 and conceptually mitigates biases including confounding by indication and other unmeasured patient characteristics . The use of robust analytical techniques such as propensity scores , which estimates the effect of a treatment by accounting for covariates , represent a powerful tool to control measured confounders and allows for an estimation of different overall treatment effects in different populations . 10 The inconsistent use of terminology to capture information in registries , through the use of medical synonyms can lead to data fragmentation and this problem can be easily remedied by standardisation of data entry through the use of common data elements and allows for seamless aggregation of patient information . 11 Finally , it is crucial to ensure clinician engagement in the collection of RWD to enable accuracy in data entry . 11
The use of RWE can generate new and informative data that can impact on clinical decision-making . For example , in an analysis of registry data collected twice yearly from 5000 – 6000 patients with rheumatoid arthritis over a 19-year period , Yamanaka et al 12 were able to establish the incidence of osteoporosis among patients with rheumatoid arthritis , the impact of treatment changes on disease activity as well as quantifying clinician and patient experiences and preferences in the routine management of the condition . RWE can also allow researchers to examine disease trajectories over time , which is more difficult in RCTs due to time constraints and budget restrictions . This was illustrated in a comprehensive review of the RWE in rheumatic disease by Misra and Agarwal , 13 who showed that the actual behaviour of rheumatic diseases in practice differed from the evidence gathered in RCTs and called for guidelines to take account of this RWE when developing healthcare management decisions .
Although the evidence from a real-world study
hospitalpharmacyeurope . com | 2021 | 9