Real World Studies – A powerful research tool for COVID-19June 10, 2020
Real-World Studies (RWS) are conducted using real-world data (RWD) on patient health status routinely collected in clinics and hospitals. Observational, retrospective, descriptive studies using RWD have provided valuable information on the natural history of COVID-19 – clinical picture, complications, mortality and prognostic factors. RWD from a retrospective cohort of 1,590 hospitalised patients with COVID-19 from 575 Chinese hospitals was analysed to develop validated predictors of risk for the occurrence of critical illness (JAMA 12 May 20). The predictive factors identified were: 1) chest X-ray abnormality 2) age 3) haemoptysis 4) dyspnoea 5) unconsciousness 6) number of comorbidities 7) cancer history 8) neutrophil-to-lymphocyte ratio 9) lactate dehydrogenase and 10) direct bilirubin.
During a pandemic, conducting randomised clinical trials is scientifically and ethically challenging, costly and time-consuming. RWS provide a good option to assess the effectiveness and the safety of a therapy. Data from the registry of lupus patients showed that patients on long term hydroxychloroquine (HCQ) developed COVID-19 (Annals of Rheumatic Disease 7 May 20). Human COVID-19 convalescent plasma was reported to be safe in 5,000 hospitalised COVID-19 patients (medRxiv 14 May 20). However, the utility of such descriptive, open, non-randomised RWS is limited by confounding factors, e.g., baseline patient characteristics that can influence study outcomes. A propensity score, which considers the impact of the relationship between treatment assignment and baseline characteristics, is effective in reducing bias. Factors that are different between two treatment groups and are associated with treatment preferences are weighted to estimate the probability of any subject in the cohort being assigned a specific treatment. This estimate-propensity score is used to match the subjects across two treatment groups.
An observational comparative study, using routine care RWD, compared HCQ versus standard care without HCQ (BMJ 5 May 20). The study used propensity score considering all variables that can influence treatment choice, e.g., age, sex, comorbidities, chronic kidney disease, liver cirrhosis, cardiovascular disease, diabetes mellitus, treatment with immunosuppressive drugs, anticancer chemotherapy; uncontrolled HIV infection, body mass index, pregnancy, the use of angiotensin converting enzyme inhibitors or angiotensin receptor blockers, time since symptom onset and the severity of the condition at admission. HCQ treatment added to standard care did not reduce admissions to intensive care units or deaths 21 days after hospital admission compared with standard care alone.
A retrospective study compared in-hospital mortality between COVID-19 patients with well controlled blood glucose levels < 10 mmol/L (180 mg%) to those with poorly controlled blood glucose > 10 mmol/L (Cell Metabolism 1 May 20). Propensity score matching analysis included age, sex, hypertension, cardiovascular disease, cerebrovascular disease, chronic liver disease and chronic kidney injury. The study showed that the death rate of 1.1% in the well-controlled group was significantly lower compared to 11.0% in the poorly controlled group.
In a pandemic setting, RWD are of immense value for understanding the disease and influencing clinical practice. The collection of critical high-quality medical data using Electronic Medical Records and the collaboration of medical institutions are vital for generating robust medical evidence from real-world studies.
Writer is a consultant on clinical research & development from Mumbai. firstname.lastname@example.org