Answers are grounded in this paper and cite their sources.
Over the past few years the development of registries for patients with psoriasis has flourished. So far, the majority of published results from biological registries in psoriasis have focused on safety, effectiveness and adherence to treatment (‘drug survival’) – the probability of discontinuing the drug. Being confronted with these publications, dermatologists need to discuss how to present and interpret the data and where to put it in the hierarchy of evidence‐based medicine. Although there is no consistent definition in current use,1 registries may be considered longitudinal observational prospective studies of multiple drug cohorts, with the purpose to evaluate long‐term outcomes (safety and effectiveness) in clinical practice. While registries provide valuable information on real‐life use of biologics and patient characteristics in a particular disease area, we feel that the frequently performed between‐group comparisons and presentation of drug survival curves within the same graph questionable and should be done only with extreme caution. Generating reliable data on efficacy, effectiveness and safety in any study setting has its limitations. Randomized controlled trials often lack external validity. A study by Garcia‐Doval et al. found that only about 30% of the patients receiving a systemic therapy in 13 dermatology departments in Spain (1042 patients) would have been eligible for the respective randomized clinical trial.2 In contrast, registries are noninterventional studies, allowing the treating physician to take a plethora of different aspects into account when choosing the best treatment from today's large armamentarium of psoriasis treatments. Now, where does this lead us with regard to the comparability of the results of different treatment options in registries? A patient included in a registry who has very severe disease is not very likely to be put on a treatment with only moderate efficacy. In the BIOBADERM registry, as an example, the mean baseline Psoriasis Area and Severity Index (PASI) scores varied largely among the different cohorts of patients treated with each therapy. The mean baseline PASI for patients being treated with acitretin was as low as 9 ± 6, and for etanercept it was 13 ± 9, whereas patients treated with infliximab had a mean baseline PASI of 18 ± 12. These differences reflect the standard of care, but in our opinion they also pose a strong barrier to the frequently performed direct comparisons of different treatment options and representation of the results within the same graph, which can insinuate direct comparability. Several other factors should raise caution for any attempt of direct comparisons of registry drug cohorts, and may serve as an explanation of apparent disaccord in drug survival between different cohorts (even within the same registry).3 4 Some clinicians may have autonomy to choose a specific therapy, and certain countries have strict guidelines dictating in which patients the biological agents can be used, as well as unspecified cut‐offs for determining when a therapy has failed. Treatments can be stopped for a number of reasons, including remission or pregnancy, and may not be due only to adverse events or lack of efficacy. In addition, when a new therapy is approved and available, patients and clinicians may want to initiate such therapy (the ‘novelty factor’) even though the patient has not clearly failed on the previous therapy. Similarly, lack of compliance and missing data (e.g. skipped consultations) may skew the results of drug survival analyses. Moreover, some treatments may be chosen, for example in high‐risk patients, due to a particular safety profile or length of the drug half‐life, which may result in incomparable patient populations between therapies. Along these lines, the beneficial effects of biologics on cardiovascular outcomes, which are reported from registry data,5 could be biased, as younger and more healthy individuals may be chosen as candidates for biological treatment over high‐risk patients. Similarly, the choice of therapy may vary between biologic‐naive patients and those who have previously failed on other biologics, which could lead to differential dropout and may result in a ‘survivor cohort’ that is much more likely to do well with the therapy. Furthermore, pricing and reimbursement of drugs may also play an important role. For example, dose escalation or shortening of the dosing intervals may occur more frequently with some drugs than with others, due to price agreements such as flat pricing or price per patient, rather than the price per unit of drug.3 As seen in countries such as Norway, the introduction of biosimilar agents has provided clinicians with a therapeutic alternative, with a 72% discount compared with the originator.6 Indeed, the prospect of being able to reduce cost, or treating more patients without increasing expenses, may drive clinicians to switch not only from the bio‐originator to a biosimilar, but also from other biological agents, again affecting drug survival analysis. Authors of systematic reviews and meta‐analyses give great attention to the question of which study groups can be pooled. Heterogeneity of study results is analysed and conclusive summaries are written with extreme caution. In our view, the different cohorts of which registries are composed are currently presented very eagerly within the same graphs, insinuating direct comparability. It is the aim of this article is to stimulate a discussion on the ‘critical perspective’ of how registry data should be presented and published. The treatment with the longest drug survival, even if the reasons for drug discontinuation are broken down into different graphs such as safety, lack of treatment response or other, is not necessarily the ‘champion’ of psoriasis treatment and should not be presented as such.
No takes yet. Share an insight, caveat, or question.
Egeberg et al. (2017) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: