• Chinese (Simplified, PRC)

In 2009 whilst at the European Medicines Agency, I had the great fortune, with Xavier Kurz and Jim Slattery, to start a ground-breaking program that aimed to strengthen the monitoring of the benefit-risk of medicines in Europe and to test new tools for collecting data directly from consumers of medicines.  This was PROTECT: the acronym of Pharmacoepidemiological Research on Outcomes of Therapeutics by a European Consortium — a public private partnership program of 34 partners created as part of the Innovative Medicines Initiative. As part of PROTECT, one of the research areas was a pilot study that could change the way the industry collects real world outcomes data — and  we demonstrated this by gathering information from two thousand pregnant women across Europe. Outcome Sciences, which became part of Quintiles, was involved in this study as one of the partners right from the start. It was only natural that when I joined Quintiles, I continued working on this with my colleague, Nancy Dreyer, who had been the original lead at Outcome.

The PROTECT Pregnancy Study was an observational study of lifestyle factors, health and use of medicines by pregnant women throughout their pregnancies. This is particularly important to study because women still need to take medicines when they are pregnant and there is limited information about the safety of medicines and their effects on the unborn baby. Traditional electronic health records (EHR) capture limited data about medication use by pregnant women. These systems only record what medications are prescribed, but not what patients actually take. They do not capture data about use of over-the-counter medications, herbals or homeopathic medicines or medication acquired by other means — e.g., “borrowed” from friends. As a result it is difficult, using existing data sets, to get a holistic view of all the medications pregnant women actually take.

We also wondered whether what is missing from the electronic record actually matters.  The only way to collect such information is to ask the concerned person directly — at least that was our theory. The Pregnancy Study was designed to see whether it was possible to collect data of good enough quality to use in medication research directly from pregnant women using data collection tools developed by PROTECT.

At this year’s European Conference on Rare Diseases & Orphan Products (ECRD 2016) I presented some of the data that emerged from this initiative, and the valuable lessons we learned.

Answer questions today for the pregnancy of tomorrow!

It took quite a while to get permission from all the different ethics and data protection bodies involved and to build the systems. In 2012, we received final approval for the study and began recruiting. Our original goal was to recruit 5600 pregnant women in the UK, Poland, Denmark and the Netherlands who would anonymously answer questions about their lifestyle, health and medication use during and immediately before pregnancy. A key component of the project was to recruit and work with women outside of the healthcare environment, and to avoid recruiting them the “traditional” way via their doctors and midwives. We recruited through leaflets in pharmacies and advertised through relevant websites, Facebook and other social media outlets. We also adhered to natural language in our communications with participants to make the experience more engaging and user-friendly. We felt that such an approach would make the opportunity to participate in the research more appealing for women, and that it would create an environment where they felt comfortable reporting honestly about their experiences.

Participants who took part in the online format could choose to complete surveys either every two or four weeks during their pregnancy, and also complete two short questionnaires after the end. Women who opted to take part via telephone, using an electronic voice automated system, only completed two questionnaires: one at the start of the study and another after the date of delivery.

Not as easy as you’d think

In the end, the study was a great success, generating valuable data both about lifestyle factors and the choices made by pregnant women regarding medication as well as the benefits of these data collection techniques in enhancing our knowledge about adverse drug reactions and real-world behavior of different patient populations. At the same time, this project provided us with valuable lessons about the challenges of recruiting and retaining participants into such real-world studies.

  • It is possible to recruit women directly – but it isn’t easy. When we began this project, we imagined that women would see our ads on Facebook or in a leaflet in their OB’s waiting room and jump at the chance to participate. That wasn’t the case. It was hard work and took extensive recruiting via an array of online, print, radio and TV ads to generate enough interest and even then we didn’t reach our original goals. Ultimately, recruiting took much longer, and was much more difficult than we anticipated. 

  • Women are willing to provide data, to a point. These women are busy, and they aren’t easily persuaded to give up lots of their free time for altruistic initiatives. Unfortunately, we learned that lesson the hard way. Our introductory questionnaire was rather long and we experienced a high rate of drop-outs as a result. In retrospect, if we had kept those first questionnaires shorter, we may have had better early rates of retention.  

  • They need a reason to come back. Keeping participants engaged with the study over several months proved challenging, and we saw high drop-out rates throughout the study as a result. While 2521 women originally signed up to participate, we only collected useful data from 2065 of them; of those whose pregnancy ended during the study, only just over a quarter were still providing data. If we had had a larger budget to build incentives tools, such as online games, user feedback, emoticons for participation and personal dashboards, we believe that it might have helped improve retention. We also considered token gifts as a way to maintain participation, though we recognize that it is important to always maintain boundaries and avoid any incentive that might be considered inappropriate to the integrity of the study.  

  • They will be honest in this kind of setting. The resulting data from the study shows that most pregnant women take at least one medication during their pregnancies.  Over 40% of women took medicines that were available over the counter.  Women also admitted to drinking alcohol, smoking and a few, recreational drug use –which are unlikely to be recorded accurately (if at all) in their EHR. Some of this use was habitual, while for others it was occasional. These results reinforce our belief that these kinds of studies are necessary to gain insight into how pregnant women use medication. It also validates that patients are willing to share this information in an anonymous environment.  

  • They don’t need a phone option. Going into this study, we assumed that at least 10 percent of the participants would choose to use the phone-based system to answer questionnaires, but we were wrong. Among all of the participants who signed up, only 14 chose the phone system, and only one completed the study. The phone option is much more expensive than the web-based platform and much more cumbersome to use, so this was good news.  

  • These data sets are best combined with other sources. This kind of study would be most valuable if we could link it directly to the participants’ EHR and other health data, especially when it comes to tracking the outcomes of their pregnancies. We attempted to make such connections, but as it was a pilot study, we were only able to do this successfully in Denmark where there are national databases.

In an ideal world we would know whether every medication was safe or not safe for use during pregnancy, but due to the limitations of research into medication and pregnancy, such a vision isn’t possible – particularly as some unwelcome effects may be very subtle and only become apparent as the child grows up. There are some very talented and dedicated researchers looking at the safety of medicines during pregnancy, but there are so many different factors that can affect the outcome of pregnancy that obtaining complete data is very difficult.  But, we could do better. The Pregnancy Study demonstrates the impact that collecting data directly from women and combining it with other observational data could have on helping us understand the safety of medicines used during pregnancy.  This is terribly important so that we can improve the advice given to pregnant women who need to take medicines. That kind of reassurance can be of great value to these women and to the healthcare community at large.