What are the specific challenges of the big project?
Well, first up, you are generally dealing with a much wider group of stakeholders, many of whom you are unlikely to personally know. So the relationships aren’t there in the same way as when a project is being rolled out in just one hospital, or just one small area.
This creates both design issues and ownership issues.
From a design point of view, you just simply have more people involved, and more opinions as a result. Too many opinions inevitably lead to some level of over-engineering of your system.
From an ownership perspective, it becomes a lot easier for staff to say afterwards that they either had no input into the design of the system, or that they tried to have input and were not listened to. Note the last comment may apply whether they actually were listened to or not – it’s about perception, not about reality.
I’ve certainly heard a combination of these two factors expressed by frustrated clinicians.
And we have the problem of distance from the workforce. If you are rolling out a system across a whole state or a large region, then communication, both of the project’s detail and the broader strategy, becomes a lot more difficult.
In the oil industry there is a saying, “Anyone who is more than five steps from the pump doesn’t understand anything.” I am sure many clinicians across our health system feel the same way about some of the people in the Ministry, the Department, or the eHealth branch.
So you may have the best strategy around, but it’s a lot more difficult to communicate effectively and ensure that it gets out to all the people that need to hear it.
This is compounded by a lack of existing relationships. That is, you do not know all of these people and they do not know you. So you do not necessarily have any credibility attached to your message by virtue of a past relationship. And there’s no easy way to fix this. No executive can know everybody in the whole organisation.
The next problem is the higher level of scrutiny from stakeholders both within the organisation and most probably externally as well. This is a key difference visible in very large projects such as the recent Queensland health payroll project, which had all sorts of people in other government departments having some input as well.
By contrast, an implementation of a payroll system in just one hospital is probably not going to have nearly so much scrutiny, and in many ways that makes the management of the project far simpler.
The issue may be compounded by political pressure to deliver in order to “cut the ribbon”. If the minister has announced that a new system is going to go live on a particular date, the pressure to stick to that date will be at a different level of intensity entirely from the pressure to stick to a date on an ordinary project. Very rational reasons for delay may not stand up against political pressure to be seen to deliver.
Is also true that scale results in some level of unhappy customers. This is simply a mathematical equation. If you are rolling out a system to 100 users, a 1% group of people who are desperately unhappy with the system chosen, or with the way it was implemented, is relatively easy to manage. But if you have 10,000 users, this can translate to 100 people who are desperately unhappy. This dissatisfied group of customers are probably emailing each other and combining their voices to make a lot more noise than if they were operating within the walls of a single hospital or district.
The last issue I want to address with big projects is the way in which any infrastructure issues are magnified and are harder to identify.
I saw this at first hand with the very first EMR rollout at St George Hospital in New South Wales in 2008. We had a couple of instances where the system went down, and it was very difficult to determine the cause of the problem – was it the software, the locally run hardware, the Citrix layer, or the network hardware run by the state? Over time we got better at working out the answer to these sorts of issues, but they never entirely disappear with very large systems.