spacer
home > ict > summer 2011 > tales from the city
PUBLICATIONS
International Clinical Trials

Tales from the City

Bruce D Schatzman of StudyManager talks with experience as he outlines the ways to overcome technology challenges when setting up a data centre for clinical trials in China

Making the move from the relatively tame sphere of conducting clinical trials in your native country to conducting them internationally is like standing on the shore of a murky swamp. From the shore, the water doesn’t look that deep, and there doesn’t appear to be anything dangerous under the surface. What could possibly go wrong?

After all, clinical trials follow a somewhat standard format and process worldwide, so if you know how to conduct them in one country, you’re likely to have success in another, or so the theory goes. Most of you reading this, however, are smart enough to know that this is a bad theory. In physics, laws that apply at one location in the universe apply everywhere else in the universe under the same conditions. However, clinical trials are definitely not physics.

When we decided to take the plunge into international clinical trials, we were under no illusion that it would be easy. We knew a thousand things could go wrong, and we took as many preventive measures as we could in an attempt to sail smoothly through rough waters. But no amount of planning ever prevents all problems – especially the technical ones.

As a software company, we pride ourselves in our deep technical knowledge of software development and enterprise IT infrastructure, so overcoming technical problems was not our biggest fear. We have 18 years of experience in clinical research software and data centre management. Despite this, we surprisingly found ourselves bogged down in the swamp and challenged by a range of unforeseen factors.

BABY STEPS

Among the first steps we took, of course, was to internationalise our EDC/CTMS application for language independence. Among many other things, this meant implementing the UNICODE international character standard within our database. UNICODE stores each character (that is, a letter or symbol) in a 16-bit representation, providing enough ‘width’ to store a very large number of characters. In languages such as English, French, German and Spanish, you can represent all the letters and symbols using only a single byte (8 bits) of storage. Eight bits allows you to store up to 128 characters. To encompass all languages, especially Asian forms like Chinese and Japanese which have thousands of characters, two bytes are required. Two bytes store a little over 65,000 characters – more than enough for any language.

For quality assurance and compliance reasons, making a fundamental change to an application’s data storage requires a top-to-bottom revalidation of the entire system. This process can take several months, but once it was completed, our application will be ready for international clinical data entry.

SETTING UP IN SHANGHAI

Based on a number of business factors specific to our situation, China made the most sense as our first international destination. To manage our China operation, we hired a bilingual senior manager of business development located in China who had a background in clinical data management, EDC software and technology.

We chose Shanghai to set up shop with our first co-located data centre. With a population of about 20 million, Shanghai is China’s largest city, with an educated talent pool and substantial activity in both commercial and governmental life sciences. Hosting databases within the US was not an option for our Chinese customers due to regulatory concerns, as well as internet latency issues that would have seriously challenged system performance or availability. Although latency between the US and EU is manageable, web traffic in and out of China is subject to substantial delay.

In the weeks following the initial setup of our data centre, we encountered a range of problems which slowed our progress. Among these challenges was that public-facing IP addresses are tightly controlled in China. IP addresses are the unique identifiers of internet-connected devices worldwide, including the servers that would power our application. The current format of these addresses – known as IPv4 – has been in existence for decades and was not designed to support the scope of technology growth we are now seeing. As a result, the world has essentially run out of network addresses. In fact, in April 2011, the Asia Pacific Network Information Centre (APNIC), which is the central registry for the Asia Pacific region, announced that it had no more blocks of internet addresses available. Although this problem is being resolved slowly with the new IPv6 standard, most data centres still use IPv4 and are gradually using up the initial blocks of addresses that APNIC granted them.

Due to the IPv4 problem, we found ourselves having to justify every server we wanted to use. Fortunately, we don’t require that many servers and were able to get around this issue, at least for now. It’s possible to solve this problem by connecting all your web servers to a load balancer that only requires a single public IP address. However, these devices are expensive and we didn’t have enough web servers to justify the cost. So, one message to takeaway is that if you’re headed to China with the desire to set up a data centre with a large number of servers, you may have a problem unless you are willing to spend a lot of money.


Once the servers were operational, our next challenge was domain name registration. We wanted a localised domain name in China and so selected www.studymanager.cn – the ‘cn’ being the main suffix for China. After spending time getting this set up, we were then notified by the Chinese government that we could not use this domain name, under Chinese law, only Chinese companies can use this suffix.

We had obtained a license from the Chinese government to do business in China, but as we were a US-based company and we had to reconfigure our servers to a different web domain.

Eventually, we were able to configure and test our data centre in Shanghai. Customers were already waiting to use our software, so we went into production. Things went very well at first, but over time we began to experience performance issues. We were aware of internet bandwidth variability on a worldwide basis and assumed this would be a problem in China as well. As an EDC vendor, this is a fact of life you learn to deal with in various ways. However, we never anticipated how variable the internet connection would be in a major industrial centre, as you usually expect this type of issue in rural areas.

The problem turned out to be subtle and hard to diagnose. Averaged out over a 24-hour period, bandwidth appeared sufficient, but it would inexplicably and dramatically fall at random times in the middle of the day. After a full investigation, we determined that it was not a data centre issue, but a problem relating to Shanghai’s infrastructure. As such, no amount of resources could be thrown at the problem to resolve it.

We lost time by making a number of assumptions. We thought that something might be wrong with our application, or with the internet connections of the customers trying to access it. We also thought there might have been issues with the customers’ browsers. We thought there may have been a problem with the system software on the Chinese-installed web and database servers. Perhaps it had something to do with the engineering work we did to support the UNICODE standard: this is part of the problem with global technology deployments – there are many more hardware and software variables that are potentially different in each country. As a result, when a technical problem arises, you’re forced to investigate far more root causes than you normally do in your native country. Problems that normally took a few hours to resolve now took days as we had to pursue many more variables.

MOVING TO BEIJING

The bandwidth problems in Shanghai eventually compelled us to move our primary data centre to Beijing, where we found internet connectivity to be much more consistent and reliable. In Beijing however, we encountered other problems. Despite having chosen one of the most reputable collocation vendors in the capital, one of the structured query language (SQL) servers that had been installed in our rack was generating errors. Once again, we found ourselves investigating potential variables that would not have been a factor in our native country. The errant server was a very reputable brand, but manufactured and sold in China. We had used Hewlett Packard servers for many years in the US and were now faced with systems we had never used before. Theoretically, the components used by the Chinese servers and the HP servers were nearly identical, but even the slightest differences could be the source of the errors. Trying to uncover these differences was like looking for a needle in a haystack.

Should we have simply used HP servers in China? Perhaps, but Chinese data centres tend to use Chinese servers, and building our own data centre in China with US equipment would have been expensive, time consuming and risky. When doing business in China, you can’t avoid having to shave costs. EDC is still not very well established, and a large educated population at a comparatively low pay scale still makes paper trials economically viable. Because of this, there is significantly less overall business volume in China for EDC/CTMS software compared with the rest of the world. This will slowly change in the coming years, but represents a barrier for investment in the meantime.

After days of fruitless analysis on what was wrong with the server, we exchanged it with a new one that worked – the new one was the same brand as the old one, but had a different hardware configuration.

LESSONS LEARNED

It’s difficult to include all the challenges we faced, or get a sense of the timeframes involved. The events described above actually took place over a period of nearly two years. Even with the usual ‘time padding’ that a good project manager includes in an undertaking of this scope, to get to where we are today took us far longer, and the total investment was far higher, than we thought it would be.

Because of all the challenges, we learned many lessons along the way, and jumping through this first ‘ring of fire’ has changed our approach to international business.

The road to profitability for international business in the clinical trial marketplace is long and arduous. If you’re not in it for the long haul (many years), you’ll end up wasting a lot of time and money. Although the international EDC/CTMS market is in some respects still in a nascent stage, we’re glad we made the decision to take the leap.


Read full article from PDF >>

Rate this article You must be a member of the site to make a vote.  
Average rating:
0
     

There are no comments in regards to this article.

spacer
Bruce D Schatzman is President and Founder of StudyManager. Bruce founded StudyManager in 1993 with the launch of StudyManager version 1, the first commercially available clinical trial management software designed to help staff do their jobs faster and more efficiently. He spent 20 years in technical and business positions within companies including Microsoft Corporation, Xerox and Tektronix. Bruce received his degree in Applied Mathematics from the University of California at Berkeley. Email: bschatzman@studymanager.com
spacer
Bruce D Schatzman
spacer
spacer
Print this page
Send to a friend
Privacy statement
News and Press Releases

Signant Partners with Loftware for Adaptive Investigational Product Labeling

Collaboration reinforces Signant's end-to-end clinical supplies management solutions and supports the complex manufacturing and distribution models inherent in modern clinical research
More info >>

White Papers

Igb Tamper Evident Folding Box

IGB

The easiest, safest and less expensive solution in compliance with what requires Dir. 62/2011/EU, FDA and TGA.
More info >>

 

 

 

©2000-2011 Samedan Ltd.
Add to favourites

Print this page

Send to a friend
Privacy statement