Advertisement

Editors’ Choice: August 2022

      Despite some appearances to the contrary, the Coronavirus is still with us. Governments in many countries may have “moved on,” but those at the sharp end are still bearing witness to the daily toll of grief, exhausted resources (human and otherwise), and the continuing impact on the care available to non-COVID patients. In the August issue, ex Journal of Clinical Epidemiology (JCE) co-Editor André Knotterus et al. explore strategic frameworks that might inform long-term policy decision-making in the context of uncertainty around the future development of the pandemic [
      • Brom F.
      • de Hoog J.
      • Knotterus A.
      • Mampuys R.
      • van der Lippe T.
      COVID-19 scenarios for a long-term strategy under fundamental uncertainty.
      ]. They describe potential scenarios based on four crucial “driving forces.” These are immunity, vaccination, mutations, and human behaviour. They identify and describe five scenarios ranging from “return to normal” to “worst case.” Consideration of the relative likelihood of these scenarios can support the planning and communication that remain critical if the pandemic is finally to be brought to an end in the most efficient and humane manner possible. Readers may wish to compare the recent trajectory of the pandemic in their region with the various scenarios, whether for the outcome of overall caseload, hospitalization, or mortality, and consider what this implies for the future management of the pandemic.
      From its origin, evidence-based medicine challenged the presumed authority of eminence. This emphasis on scepticism and scrutiny may be relevant to current explorations of the challenges to medical science that have been seen across the world in response to the Coronavirus pandemic. Promoting a realistic appreciation of the limitations of health care is central to evidence-based medicine and is a key factor in delivering and maintaining trust between professionals and the population [
      • Fugelli P.
      Trust – in general practice James McKenzie lecture.
      ]. The editors of the JCE are passionate about reporting the methodologies that deliver such realism, and those that undermine it, and the August issue of JCE is no exception. In an original article, Ioannidis [
      • Ioannidis J.
      High-cited favorable studies for COVID-19 treatments ineffective in large trials.
      ] demonstrates the caution that needs to be applied to early, small observational studies and randomized controlled trials. In his analysis of articles relating to largely ineffective treatments for COVID-19 that have received more than 150 citations, he reports that many of these studies with favorable conclusions are uncritically cited and disseminated. This potentially fuels inappropriate optimism in readers, and if followed through into practice, disappointment.
      Clinical practice guidelines panels seek to provide professionals and the public with recommendations based on the best current evidence, patient and community preferences and values, and the resources available. In their original article published in this issue, a diverse team led by Holger Schunemann and Kevin Pottie reports on the multistakeholder process to prioritize and translate health recommendations for patients, caregivers, and the public, for the COVID-19 recommendation map (https://covid19.recmap.org/) [
      • Pottie K.
      • Smith M.
      • Matthews M.
      • Santesso N.
      • Magwood O.
      • Kredo T.
      • et al.
      A multistakeholder development process to prioritize and translate COVID-19 health recommendations for patients, caregivers and the public. A case Study COVID-19 recommendation map.
      ]. The authors note that the pandemic has created the demand for accessible evidence in plain language, which the map seeks to address. The team built on the GRADE Plain Language Review template and the multistakeholder process to create a digital Plain Language Recommendation tool that places recommendations into a broader context that includes advice such as “what does this mean for you?” including prompts of questions to explore with health professionals.
      The route to truth, and to achieving trust, is to avoid overpromotion of healthcare interventions, yet that is a frequent feature of scientific reports of their effects. For obvious reasons, over the past 2 years overdiagnosis has not been at the forefront of most people’s minds. However, as the pandemic begins to subside, and health systems desperately try to redress its impact on health systems and people with noncommunicable diseases, it should be a priority to be able to distinguish those in whom health care interventions are most likely to be effective. In their paper published in the August issue, Bell et al. [
      • Bell K.
      • Doust J.
      • Sanders S.
      • Buchbinder R.
      • Glasziou P.
      • Irwig L.
      • et al.
      A novel methodological framework was described for detecting and quantifying overdiagnosis.
      ] seek to provide a framework that can help to detect and quantify overdiagnosis beyond the area of cancer, where methods for this already exist. They describe two specific theoretical approaches to aid the consideration of overdiagnosis. First, the “prognostic approach,” which has become associated with cancer screening, and is based on whether the additional people identified through screening (or early diagnosis) benefit from being diagnosed and treated? Thus overdiagnosis represents those individuals who cannot be identified in advance, but do not benefit and therefore can only experience the harm associated with treatments. In the case of nonmalignant disease the second approach relates to utility. In this scenario, overdiagnosis may be considered as including identification of risk factors for disease in low risk people, or people with minor disease in which there may be small benefit from identifying and treating the disease, but the net benefit to harm ratio favours no treatment. The authors provide examples of each of these scenarios and focus particularly on situations where different diagnostic approaches might come to discordant results through changing thresholds of disease, or increased sensitivity of imaging for example. They use the information gained by these discrepant results and introduce the use of causal directed acyclic graphs and “fair umpires” to quantify the extent of overdiagnosis and inform decision-making. The authors recognize that there are some challenges in the application of this framework but hope that the study may act as a starting point for further research in this area.
      The August issue of JCE also highlights two important articles that relate to observational studies. Pufulete et al. [
      • Pufulete M.
      • Mahadevan K.
      • Johnson T.W.
      • Pithara C.
      • Redwood S.
      • et al.
      Confounders and co-interventions identified in non-randomized studies of interventions.
      ] explore approaches to identify confounders and cointerventions in nonrandomized studies of interventions. They conducted a systematic review of the evidence from randomized controlled trials and cohort studies of dual antiplatelet therapy, and also conducted interviews with cardiologists and cardiac surgeons, and administered two online surveys with professional organizations. All told they identified 10 cointerventions and 70 potential confounders, of which 34 of 70 were judged to influence dual antiplatelet therapy prescribing and risk of bleeding. The systematic review and surveys both identified a large proportion of confounders (31 of 34, 91%) and there was a fair degree of overlap between the different approaches. The authors note that the interviews identified “hard-to-measure factors such as adherence and resistance to certain antiplatelet agents.” They conclude that these methods could be applied more widely when designing nonrandomized studies of interventions but that researchers should evaluate the relative advantages of using such resource-intensive methods against other more efficient data-driven approaches.
      The use of inappropriate causal language in systematic reviews of observational studies is explored in a methodological study by Ah Han et al. [
      • Han M.A.
      • Leung G.
      • Storman D.
      • Xiao Y.
      • Srivastava A.
      • Talukdar J.R.
      • et al.
      Causal language use in systematic reviews of observational studies is often inconsistent with intent: a systematic Survey.
      ]. They note that irrespective of whether the intent of the review is to identify causation, the use of causal language is inconsistently applied and is also frequently inconsistent with intent. Across 199 reviews, just over half had clear causal intent. These were more likely to focus on therapeutic clinical interventions, and on mortality and functional outcomes rather than morbidity outcomes. Among the 86 reviews without causal intent about half of these used causal language somewhere in the manuscript, although none did so throughout. Those reviews that used GRADE were significantly more likely to use causal language in the title and abstract. The authors recommend that “authors should clearly present their study objectives and use language appropriate to those objectives” and that journal editors should take responsibility for ensuring this.

      References

        • Brom F.
        • de Hoog J.
        • Knotterus A.
        • Mampuys R.
        • van der Lippe T.
        COVID-19 scenarios for a long-term strategy under fundamental uncertainty.
        J Clin Epidemiol. 2022;
        • Fugelli P.
        Trust – in general practice James McKenzie lecture.
        Br J Gen Pract. 2001; 51: 575-579
        • Ioannidis J.
        High-cited favorable studies for COVID-19 treatments ineffective in large trials.
        J Clin Epidemiol. 2022; 148: 1-9
        • Pottie K.
        • Smith M.
        • Matthews M.
        • Santesso N.
        • Magwood O.
        • Kredo T.
        • et al.
        A multistakeholder development process to prioritize and translate COVID-19 health recommendations for patients, caregivers and the public. A case Study COVID-19 recommendation map.
        J Clin Epidemiol. 2022; 148: 104-114
        • Bell K.
        • Doust J.
        • Sanders S.
        • Buchbinder R.
        • Glasziou P.
        • Irwig L.
        • et al.
        A novel methodological framework was described for detecting and quantifying overdiagnosis.
        J Clin Epidemiol. 2022; 148: 146-159
        • Pufulete M.
        • Mahadevan K.
        • Johnson T.W.
        • Pithara C.
        • Redwood S.
        • et al.
        Confounders and co-interventions identified in non-randomized studies of interventions.
        J Clin Epidemiol. 2022; 148: 115-123
        • Han M.A.
        • Leung G.
        • Storman D.
        • Xiao Y.
        • Srivastava A.
        • Talukdar J.R.
        • et al.
        Causal language use in systematic reviews of observational studies is often inconsistent with intent: a systematic Survey.
        J Clin Epidemiol. 2022; 148: 65-73