On May 18th 2017 the Dutch Ministry of Health, Welfare and Sports in cooperation with the Academic Medical Centre Amsterdam organized a meet-up event called “Big Data in Healthcare: a combination of an un-conference and a hackathon”.
Approximately 50 participating experience experts were asked: “What should policy regarding big data in healthcare look like?” The organizers were looking for inspiring ideas and initiatives in which big data is used in healthcare in a safe and transparent way. Notwithstanding the benefits of big data in healthcare being clear to many, but the obstacles or barriers less and, perhaps for this reason, more feared. Ten break-out sessions were held wherein main obstacles to working with big data in healthcare were discussed. The audience was divided in groups each handling one of the obstacles. Groups were challenged to provide three concrete steps that could possibly diminish the obstacle.
To sum up, the output of all 10 break-out sessions shows four main themes:
- A central control and a leadership role of the government is desirable;
- It is crucial to emphasize the ‘Why?’ of data usage and its context. Involved parties such as patients and healthcare providers are willing to share data when the ‘Why’ (context) is understood and the players are aware of the added value of big data;
- Both financial and non-financial incentives are suggested to stimulate cooperation, standardized registration and data sharing;
- Control of the relevant data must lie with the patient/ health consumer; the government needs to hold the responsibility of facilitating and informing the relevant stakeholders about big data usage.
Below, three of the obstacles are discussed in more detail.
Lack of cooperation
In general, big data offers great opportunities for healthcare, however, the lack of cooperation between stakeholders represents one of the biggest challenges to work with big data. Even though the different stakeholders are convinced of the need to cooperate, they are not willing to share their data that easily. One of the reasons for this is self-protection, as they do not want to give away their own data without receiving something in return. Distrust also plays a role as they do not know what the other party might do with their data. Moreover, where healthcare organizations focus on obtaining business insights from large datasets, science uses large datasets to publicly publish findings and algorithms. To overcome the hurdle ‘lack of cooperation’, the main solution suggested was to create a common goal to stimulate the cooperation between the different stakeholders. Examples of these are the Biden cancer moonshot (USA) or the war on diabetes (Singapore). To realise this, a coalition can be created where all stakeholders play an important role by contributing to such a cooperative platform. A top-down leadership role by the government is desirable whereby financial incentives can be granted for cooperation between players.
Quality of data
Ensuring that data is of high quality sounds very self-evident, but is not always realised even though it is a necessary condition for any form of analysis or operation to be performed on the data. To obtain data of high quality, it is crucial to focus on two aspects:
- The 4V characteristics of data: volume, velocity, variety, and value;
- Data integrity: the maintenance of, and the assurance of the accuracy and consistency of data over its entire life cycle.
Quality of data can be promoted by telling registrars and data users why it is important to share and register relevant data and by working towards a shared vision and goal. Users of data need to know in which context data are collected. Moreover, uniformity of data processing is an aspect that is a possible solution to enhance the quality of data. All in all, there is an urgent need to establish quality criteria of big data.
If you think of big data in healthcare, one often discusses privacy issues, as it concerns personal medical information. Questions such as “Who own the data? Is it the patient, the healthcare organization, the insurer, the payer or the company that performs the analyses?” play an increasingly important role. To overcome the ‘privacy’ hurdle, technological solutions on a national level need to be offered and more focus on developing and integrating solutions from other domains is needed (e.g. privacy by design, encryption, central organized data broker). Ownership of data needs to be in hands of the patient. Anonymizing data, whereby analysis software ensures that users can only see anonymous data, is widely used as an easy solution to overcome privacy invasion.
From this intriguing meeting, one gets the impression that the Ministry of Health, Welfare and Sports finds itself being a two-headed animal: she needs to lead the flock of stakeholders, whereas she also needs to protect the weakest ones who are potentially in danger. From the animal kingdom we know however that two headed animals are a developmental accident. I am not sure yet which one to fear more.
Anyway, I look forward to the follow-up meeting later this year to find out more about the policy making process regarding big data in Dutch healthcare.
Consultant Health Economics and Outcomes Research at Panaxea b.v.