Recent opinion pieces have reflected on the neglect of the principles of evidence-based medicine in the context of the covid-19 pandemic,¹ particularly focussing on trial conduct, with opaque mid-trial changes to protocols and early terminations. Recent highly publicised retractions of hospital database studies investigating different therapies for covid-19 draw scrutiny also towards observational studies.¹ The assumption seems to be that since the SARS-CoV-2 pandemic was unanticipated and progressed rapidly, these changes and errors can be explained by a need to design and implement studies rapidly.

Our national cohort study of covid-19 in pregnancy, published in The BMJ,² was conceived and designed in 2011. In the wake of the 2009 influenza A/H1N1 pandemic, when similar lessons were learned in terms of the need for timely, quality research, the National Institute for Health Research (NIHR) put out an open call for applications for studies to be conducted in the event of a future pandemic. Eight studies, including ours, were funded, set up and then mothballed in preparation for activation in the context of a future pandemic.³ On several occasions over the next eight years we indicated when asked that the study could be rapidly adapted for other emerging threats (including SARS at the time). We were asked in mid-February 2020 to make the adaptations to enable the study to be activated for covid-19, although it was not formally activated until 20th March. 

The study uses an existing platform, the UK Obstetric Surveillance System, which we set up in 2005 to conduct robust national observational studies of severe pregnancy complications. With collaborating clinicians in all hospitals with obstetric units in the UK, it, and similar systems internationally,⁴ represents an ideal means to obtain rapid and comprehensive epidemiological information about the severe impacts of emerging infections in pregnancy. The preparation undertaken to set up and hibernate the study in 2011-12 allowed us to begin to collect specific, nuanced data across all obstetric units in the UK within three days of activation. With the relatively high infection rate and high burden on reporting clinicians who were also maintaining active clinical services, the UK’s existing infrastructure of dedicated research teams within the four UK nations’ Clinical Research Networks were crucial to the rapid collection of high quality information. 

If nothing else, the covid-19 pandemic has crystallised the need for reliable basic epidemiological information, and that this is equally important in pregnancy must be recognised. From the earliest days of the pandemic there was a call for pregnancy-specific information, with a first publication of a case series of nine affected women in February⁵ exploding by the end of May into more than 250 overlapping case reports, case series, reports from pregnancy registries or small groups of hospitals. In none of these series can we be certain of the underlying population denominators nor the degree to which they are affected by the biases, such as case ascertainment or reporting biases, inherent in hospital series and registries. The results are thus almost impossible to interpret. To me, our study clearly illustrates the benefits of thinking ahead, designing and setting up study infrastructure when threats are only hypothetical. This is not only because it allowed for rapid collection and analysis of information for pregnant women and their clinicians, but because it allowed for the collection of high quality data and robust study design—thus minimising the risks of publication of, at best, misleading, or, at worst, erroneous, information.

Nevertheless, there are some aspects of research we can never plan or prepare for. Two days after our study was activated, and the day before we began data collection, my father was admitted to hospital with covid-19. In common with many other families, we never saw or spoke to him again before he died two weeks later. Adding this personal aspect to the professional only made me even more determined to complete the research and deliver robust information about covid-19 in pregnancy for women and clinicians, but certainly required all my reserves of resilience. 

Beyond covid-19, setting up and hibernating high quality research projects ready for the next emerging public health emergency must be the way forward—even if, in a best case scenario, there is never any need for them to be activated. The confusing mire of information on covid-19 in pregnancy emphasises that a study, or programme of studies, to address the basic epidemiological questions around pregnancy must be one of them.

Marian Knight is Professor of Maternal and Child Population Health, National Perinatal Epidemiology Unit, Nuffield Department of Population Health, University of Oxford, UK.

Competing interests: see declaration on research paper. 

1. Aronson J. When I Use a Word . . . Retraction. BMJ; 2020.
2. Knight M, Bunch K, Vousden N, et al. Characteristics and outcomes of pregnant women admitted to hospital with confirmed SARS-CoV-2 infection in UK: national population based cohort study. BMJ 2020; 369:m2107.
3. Simpson CR, Beever D, Challen K, et al. The UK’s pandemic influenza research portfolio: a model for future research on emerging infections. The Lancet Infectious diseases 2019;19(8):e295-e300. doi: 10.1016/S1473-3099(18)30786-2
4. Knight M, INOSS. The International Network of Obstetric Survey Systems (INOSS): benefits of multi-country studies of severe and uncommon maternal morbidities. Acta Obstet Gynecol Scand 2014;93(2):127-31. doi: 10.1111/aogs.12316
5. Chen H, Guo J, Wang C, et al. Clinical characteristics and intrauterine vertical transmission potential of COVID-19 infection in nine pregnant women: a retrospective review of medical records. Lancet 2020;395(10226):809-15. doi: 10.1016/S0140-6736(20)30360-3