Waiting for Lorenzo

Jon Hoeksma

EHI editor Jon Hoeksma provides at a detailed look a confidential report into iSoft’s development of Lorenzo. The extent of the problems catalogued raises questions about the NHS IT programme and whether Lorenzo will ever be delivered to the NHS’s requirements.

A detailed review of iSoft’s development of Lorenzo, carried out by Accenture and Computer Sciences Corporation this year concluded that there is a “significant risk” the software will not meet NHS requirements as defined by NHS Connecting for Health.

EHI has obtained a copy of the confidential report, which indicates the development of the Lorenzo system bought for the NHS IT programme remains fraught and is still at an alarmingly early stage. By February no module had yet been completed or tested and development plans for more complex later releases were sketchy at best.

Overall the report paints a bleak picture of iSoft’s approach to project management and rigorous software development. It also reveals the company’s limited readiness to share development plans with its prime contractors Accenture and CSC.

Urgent

The iSoft review warns that urgent steps must be taken “if we are to avoid the delivery of Lorenzo in a timeframe that will inevitably be far too late for CfH”. It further suggests the NHS may wind up with a solution “whose scope does not match that required by CfH, as it has not been defined from the top down with LSP in respect to the CfH requirements”.

Lorenzo is the core clinical software at the heart of the NHS IT modernisation programme, and is meant to be delivered to 60% of the English NHS. The first versions of Lorenzo are now running two years late, having due to be delivered from 2004.

The February review says 2008 is now the earliest possible date for Lorenzo to be delivered, though even this looks “highly optimistic” and likely to be subject to “schedule slippage”. Once the software is provided to Local Service Providers (LSPs), up to nine more months will be needed before it becomes available to NHS trusts.

It concludes that due to incomplete scope definition, design, build test estimates are tentative at best. As a result the release dates provided to the LSP “must be viewed as indicative” at best.

E-Health Insider has been told that the review was ordered by NHS Connecting for Health boss Richard Granger to determine whether iSoft’s Lorenzo product “was likely to be fit for purpose as the solution for the NHS and in what timescale”. It also sought to identify “whether Accenture/CSC and iSoft can form the type of working relationship that will foster successful delivery of this solution”.

Red flags

The Lorenzo review, which involved a team visiting iSoft’s Chennai development facility in India, assessed 39 matters relating to Lorenzo. Nineteen were flagged up as "red" – meaning they required immediate work. Of particular concern were questions over iSoft’s ability to plan, produce credible roadmaps for products, and estimate how long the development process would take.

Damningly, the Lorenzo review found "no evidence for the development, nor testing of, technical procedures that would be required for operation and maintenance of the live system … this is the main risk to the successful delivery of a fit-for-purpose solution."

One of the red flags was the absence of robust change control mechanisms. “There does not appear to be a fully functioning change control process that impacts all of the Lorenzo teams before approval is given. It appears entirely feasible that one of the product managers can add features without consulting other product managers for impact.”

Another red flag was methodology and work product breakdown. “The overall approach of producing a generic approach and only then considering market (CfH) requirements results in significant rework and additional time being identified when these are considered.”

Yet another red was planning and estimating: “iSoft have high confidence in their ability to recover the time lost to date in the next few builds, though there is no evidence base to assume these will run counter to historic experience.”

The design methodology is also flagged as red. The report notes that it was “heavily weighted on the IP harvesting of a limited number of products and the reliance on domain knowledge, predominantly within iSoft, with restricted influence of clinical users.”

Development models

Alarmingly for NHS trusts awaiting the Lorenzo clinical software the report says the software is being developed from the bottom up, instead of being built from the top down with features tailored to meet the requirements set by the NHS National Programme for IT (NPfIT).

The Lorenzo solution is based on four layers and is designed as a generic product to be tailored to particular markets, including the NHS. But one of the central conclusions of the review is that there is no clear road map for the development of later releases of Lorenzo: “There is no mapping of features to release, nor detailed plans. In other words, there is no well-defined scope and therefore no believable plan for releases beyond Lorenzo GP.”

The review adds the strategy iSoft has pursued has been to attempt to develop a generic Lorenzo solution and only then adapt this to the requirements set out in its LSP Connecting for Health contracts. As a result, it says, the likelihood of further delays is increased. “There is a significant risk that an evaluation of the ‘gap’ between the needs of CfH and the capability of the generic solution will require significant re-work of the product and the platform layers. This will likely lead to schedule slippage.”

Despite the LSPs being responsible for actively managing the delivery of their contractors the review says that the scope of the Lorenzo solution is not defined by the LSPs, nor based on the requirements set out by the NHS National Programme for IT – against which contracts were awarded.

“NPfIT requirements and priorities were not directly reflected into the solution or its phasing”, and as a result the current Lorenzo scope and release strategy “does not therefore reflect imperatives, requirements or priorities of CfH to an appropriate level”.

“The LSPs have had little or no influence on the scope and release strategy of Lorenzo, and this has been dictated by iSoft. It was apparent from the meetings with the outbound product managers that the LSPs are not really seen as customers.”

Changes needed

This must change, says the Lorenzo report. “The primacy of the CfH release needs to be established. For delivery to CfH in acceptable timescale, the fundamental approach must change to one where the scope and content of each release is driven by CfH requirements from the outset.”

The report states that some improvement had been made since Accenture undertook reviews in 2005, but says it is unclear whether issue and risk management tools have been universally implemented across the Lorenzo programme.

The authors of the report state that the only software demonstrated by iSoft was “elements” of the Lorenzo generic Clinicals solution, which was claimed to be about 40% complete.

Although this is described as good on configurability and flexibility, usability issues of “a significant number and magnitude” are also noted. It states there is a serious problem with the strict organisation hierarchy “which does not appear to allow for logical groupings of teams which cross organisational boundaries”. The report says this will be a problem.

CfH and iSoft say that steps have since been taken to address the problems identified, including local service provider CSC seconding staff into iSoft. CSC said “a joint action plan was prepared and had been implemented” and reaffirmed its continuing commitment to working with iSoft. Accenture, meanwhile, simply said steps had been taken “to enhance the development process”.