Strength in numbers: can crowdsourcing speed up health care improvement?

From counting galaxies to spotting cancer cells, crowdsourcing is already being used in all sorts of sectors to speed up processes, solve problems and engage the public in research. We spoke to Rebecca Simmons, Deputy Director of THIS Institute, about their plans to invite NHS staff, patients and the general public to get involved in health care improvement research.    

What is crowdsourcing and how can it help? 

Crowdsourcing simply means engaging a group to gather inputs to reach a common goal. It can be used in citizen science research projects, where members of the public provide valuable contributions despite not being formally trained experts. 

At THIS Institute we’ve been looking at using crowdsourcing methods to engage NHS staff, patients and members of the public in our health care research and improvement work. We think it has the potential to offer big benefits, speeding up research processes and giving us access to knowledge and ideas from a much wider pool of expertise, while also helping with awareness raising and public engagement. 

How has crowdsourcing been successfully used in other industries?

It’s being used everywhere, in some really interesting ways. Zooniverse, at University of Oxford, hosts over 50 citizen science projects from around the world. People can help classify galaxies, transcribe manuscripts and count penguins. 

Cancer Research UK have also done some amazing things. They recruited nearly 100,000 volunteers to help identify and analyse cancer cells, doing jobs that can’t yet be fully automated and can only be done with human input. They’ve seen high levels of accuracy, and massive chunks of work done in a short space of time. 

Crowdsourcing is also used commercially, with employers like Starbucks using it to ask employees for ideas about how to improve the company – a process that also builds consensus and engagement. 

Can you explain how you’re using crowdsourcing at THIS Institute and what benefits you’ve seen so far?

We’re starting by testing out using crowdsourcing to help speed up systematic reviews. 

Systematic reviews provide a detailed summary of all published primary research in response to a particular research question. They are really important, but very expensive and time consuming. Researchers have to go through thousands of abstracts to check if they should be included, but new papers are published every day, so often by the time you’ve finished your review is already out of date. 

A couple of months ago we ran our first pilot review using crowdsourcing in partnership with Cochrane Crowd. We asked people to sort through abstracts for us. This approach has been well tested by Cochrane Crowd showing that a crowd can, with a high degree of accuracy, decide which abstracts should be in and out.

We ran the same review in the traditional gold standard way so that we could compare results, and the reduction in time was just astonishing. The crowd managed to go through all the abstracts in 24 hours, whereas it took us 17 days to do the same work in-house. 

By inviting the public to be more involved in research, we’re developing scientific literacy while also democratising research. Within health care crowdsourcing has the potential to allow contributions from the people at the sharp end of care. If thousands of NHS staff around the UK have worked on a research study with us and helped come up with an improvement solution, they’re already aware and engaged when it comes to wider spread and implementation of the ideas that work. 

What are some of the challenges people have encountered when using crowdsourcing? 

One of the main challenges is ensuring data quality and scientific rigour. How do we know that the crowd are picking the right abstracts for our review? Providing training in how to complete a task is a good start. And you can ensure that the same abstract is reviewed by several different people and check they are in agreement. 

The whole question of how to encourage and maintain engagement is another challenge. It’s important to understand people’s motivations for taking part, make the tasks as straightforward as possible, and make sure you give feedback on how the research is progressing. We know that dropout can be pretty high. Around 90% of people will ‘lurk’ and do not contribute, 9% contribute a little, and the final 1% are the super users, who give vast amounts of their spare time. 

Why do people want to be involved?

Some just want to help out and be involved in science. Others are motivated because of the topic, particularly if it’s linked to their own experience. 

I’d always assumed it would be younger IT-savvy people taking part, but it attracts a broad range of people. We’re finding online crowdsourcing is good for engaging with hard to reach people – including those with health challenges who might find it difficult to attend a focus group in the traditional way you might engage the public. 

How else could this approach help to build our knowledge about improving health and care? 

Crowdsourcing isn’t just about analysing data, but also providing it. For example, a recent systematic review identified over 1,200 different quality indicators for perioperative care. Half are not supported by any evidence base, and few are patient-reported. Rationalising these indicators would have benefits for care, research, and health service monitoring. 

Using citizen science, we plan to invite NHS staff and patients to take part in a consensus-building process. We’re exploring a tournament-style method of multiple rounds to determine what matters in perioperative care quality and safety. We think this is a situation where a crowd of experts knows better than a few people working in-house.

Citizen science has been used lots in other areas, but no one has systematically applied it in health care until now. We’re still in the pilot phase with many of our projects, but we’ve already published several learning reports, including one on crowdsourcing for research. We plan to share all our learning about the use of citizen science in health care, so you can expect to hear lots more from us.