Business intelligence rejuvenates cancer research

When child cancer charities Cancer and Leukaemia in Childhood and Sargent Cancer Care merged in 2005, the new organisation, called CLIC...

When child cancer charities Cancer and Leukaemia in Childhood and Sargent Cancer Care merged in 2005, the new organisation, called CLIC Sargent, seized the opportunity to boost its research and services by implementing business intelligence.

"Previously, both charities had been pretty much restricted to using Excel spreadsheets as a means of storing and interrogating research data," explains CLIC Sargent's Dr Pilar Gonzalez-Doupe. Given the basic nature of the intelligence collating and data mining that had been undertaken up to then, no major data merger and integration exercise was required.

Even the methods of both charities were compatible. Both worked with family groups, although in different areas, one concentrating on Bristol and the South West, the other focusing on London and Scotland. As a result, the legacy databases held similar types of data structure and were oriented towards family support.

"The issue for us was to show that we could respond to needs according to what the families were telling us. We needed faster tools to do this and more flexible means of collecting information," says Gonzalez-Doupe.

To do so, CLIC Sargent had to transform the legacy databases so they could query the needs of families and take their feedback rather than just issue them with grants.

The biggest change required was to source and implement a tool that would enable impacts to be measured.

"Excel didn't enable us to acquire this critical intelligence, because you need two reference points - a start point and a finish point along an axis - to measure an impact against time," explains Gonzalez-Doupe. "We also needed a tool that would allow us to collect a large volume of data over time to do time series analysis."

CLIC Sargent wanted a tool that could compare impacts within one family's feedback form and also between families.

Choosing the SPSS business analytics tool was the most straightforward element of the implementation. Gonzalez-Doupe had personal experience of SPSS from her academic research background, and SPSS was well known to the universities and hospitals with which CLIC Sargent works closely.

The charity's research team worked with the IT department to locate suitable hardware to run SPSS on. Because the merged organisation had already made the business decision to host as much data as possible centrally at its head office in London, a couple of servers in the Bristol office were surplus to requirements one of these, an HP Proliant DL360 G3 with a single Intel Xeon 2GHz processor and 4Gbyte of RAM, became the server for the SPSS installation. All the client machines were less than three years old.

The IT team's database manager raised the issues of possible duplication with the charity's main database and how to export or import data between SPSS and the database. Given the very basic level of data contained in the legacy Excel files, a greenfield approach was adopted.

Because the charity uses unstructured (free text) as well as structured data, it considered using text analysis tools but has decided against for the time being. "We export the free text from the forms we receive back and code them into themes manually, which we can then search using SPSS. This method seemed to be as good as the automated alternative," explains Gonzalez-Doupe.

However, it did opt for the MrInterview tool, which helps researchers create online surveys and structure questions, including the if/then type. In particular, the feature has been useful for generating the quick-fire questions of online surveys, which are useful for campaigns.

"We've increased our survey production by 50%. We've a database of 1,300 families we can poll for this quick-hit data on topics such as the quality of hospital food," says Gonzalez-Doupe.

Two days were taken to prepare the server, then another two days for the SPSS and MrInterview installation.

"Apart from applying occasional patches, SPSS is a fairly low-maintenance application," says Darren Hurst, the IT team systems manager at CLIC Sargent. "The impact on the network has been fairly low as users are given the opportunity to complete surveys in their own time, so this gives a staggered usage of network resources over a period of days.

"IT staff needed to be trained in Windows to MCSE level and SQL to MCDBA level but apart from this the installation was carried out with an onsite specialist from SPSS, so no additional training was necessary in the IT department."

The research department needed specialist training to use SPSS. Despite the intuitive interface, the biggest learning curve has been to double the amount of time taken to write an interview script because testing takes as long as the script writing.


Business intelligence rises up IT agenda >>

Read more on Business intelligence software