The Clinical and Financial Transformation Of Asia''s Largest Private Hospital

PDF File: 

HCT2_wp_downing_schroeder.pdf

Global Care Solutions

With the “big bang” approach, Bumrungrad Hospital implemented an enterprise softwaresolution that provides doctors, nurses, radiologists, and lab technicians with immediate clinicalinformation and satisfies the billing and material management needs of back-office users.

Established in 1980, Bumrungrad Hospital is a recognized provider of premium private health care throughout Thailand and Southeast Asia. Bumrungrad is the first hospital in Asia to be accredited by the U.S.-based Joint Commission on International Accreditation. Located in central Bangkok, the hospital is a 554-bed tertiary care facility serving over 900,000 patients annually. This translates to a daily average of 2,400 outpatients per day with peak days touching 3,400. Patients come from over 130 countries and bring with them a large variety of languages and cultural diversity.

In January 1997, Bumrungrad Hospital opened its 554-bed replacement facility along with a newly commissioned computer system. A few months before the facility opened, the management team realized that the clinical and financial information systems they had purchased and customized would not be able to meet end-users’ needs and cope with the volume of patients. Unfortunately, due to long implementation cycles of hospital systems, they were left with no other choice but to open their doors with their new system. Their pre-opening worries were proved correct – the system’s functionality and performance were not enough for the new facility’s expanded capacity (at that point it was serving only 1,200 outpatients a day with 250 of 554 beds in use).

The Failed Computer System

Bumrungrad’s failed computer system was comprised of four integrated database servers/systems and a series of optical jukeboxes with disks containing 17 years of previously scanned patient medical records. In comparison, typical facilities of the time in the United States operated efficiently with more complex systems, sometimes involving dozens of integrated servers. Bumrungrad had accomplished a lot with just four systems, and was certainly moving in the right direction. Failure had been contained primarily to vendor selection and local vendor support.

A saving grace of the failed system was the decision to scan old medical records. Without these previous 17 years of scanned medical records, doctors at Bumrungrad would have been reluctant to use the new system. With any new system, it is not realistic to assume that all doctors will become touch typists overnight. With the ability for the doctors to enter their notes online, or to write on bar-coded sheets of paper, the system could straddle the old/new age of medical records without inconveniencing a large roster of current doctors and nurses with a shift to keyboard-based data entry.

Interfacing issues were enormous. The volume of mismatched data in the systems created a climate of mistrust of both the financial and clinical data. With a facility of over 700 practicing doctors and 2,000 other staff all calling management into question over the system, blame was unfairly passed to the computer department staff. In reality, those who poorly configured the brand name systems and the nonexistent vendor support were responsible. The problem was clear – no matter how much time and effort were spent on the interfaces and data access speeds, there would always be problems with the system.

The Search for a New System

Bumrungrad Hospital knew that its new facility’s IT system needed urgent replacing as it would be unable to cope with the projected rise in patient volumes. Sadly, the search for a new system started the day Bumrungrad opened the doors of its new facility.

Bumrungrad management drew up guidelines for a new system, and efforts to procure a new one were placed in the hands of senior management, including the CEO and managing director. CEO Curtis Schroeder commented, “We needed to create an environment that would allow our doctors to practice medicine in the best way and create the best relationships with their patients. We wanted clinical information at the fingertips of our doctors to enable them to make better decisions.” This translates into a need for a system that can satisfy the requirements of doctors, nurses, radiologists, lab techs, and pharmacists, and at the same time satisfy all back-office users with functionality including billing, materials management, and other ancillary departments’ needs. The system would also need to:

• Produce invoices in real time for inpatients and outpatients;
• Deliver a real-time electronic medical record (EMR);
• Process the huge number of patients;
• Deal effectively with both outpatient physician practice and inpatient information processing in a single database;
• Use a single database for patient data, images, financial information, and statistics;
• Keep all data online forever – no archiving;
• Have no interfaces;
• Have minimal downtime;
• Be easy and intuitive to learn (given the relatively limited experience of the local worker with information systems); and
• Not require the doctors to type.

The vendor(s) also needed to provide complete data conversion of the existing systems; no easy task because:

• The data in the current system was of suspect quality (due to bad interfaces) and needed to be cleaned before it was exported to a new system;
• The current vendors’ support was lacking, leading to poor relations with the existing systems integrator (SI); and
• Seventeen years of scanned medical records had been placed into a proprietary FileNet system composed of six HP jukeboxes with more than 400 optical disks. This data needed to be converted during the final months of the old system’s operational life.

To make matters even more difficult, the logistics of not only converting a system but also making a new system operational in an existing high volume operation is not an easy task. The replacement system and SI would need to have a very good plan for bringing a new system live without interfering with patient volumes, which exceeded 2,000 outpatient ambulatory visits per day and a full house of inpatients.

The New System

Bumrungrad set the goals and objectives and chose Global Care Solutions’ (GCS) new Hospital 2000 product suite to run its facility. It was a new product, so Bumrungrad had a lot of operational input into the system to ensure it worked in a high volume Asian environment. The Hospital 2000 system is an enterprise software solution that integrates virtually every department in a hospital. This type of system is traditionally categorized as a hospital information system (HIS). What is unique is that it is an integrated solution that combines both the health care (frontoffice) and back-office operations in a single database, which works in virtually all languages.

The Hospital 2000 solution includes registration, clinic systems, ward, emergency department, laboratory, radiology, picture archiving and communication system (PACS), clinical image management, pharmacy, operating theater, enterprise appointments, billing system, purchasing, inventory, accounts receivable, accounts payable, general ledger, asset management, human resources, rostering/time and attendance, and payroll.

The new solution was to be a complete divergence from the existing multiple system architecture. The Hospital 2000 solution architecture was based on a single server and single database model containing the entire facility’s data, solely using Microsoft Windows 2000/2003 server and Microsoft SQL 7/2000 database platform technology. The then current and now failing system was based on a multiple server solution running Sun Solaris operating systems and databases from Oracle, Ingres, Sybase, and Informix. The new replacement system hardware architecture was simple; the database server was a Dell 4 processor Intel Xeon running at 500 MHz with 4GB of RAM. The database server was attached to fibre channel storage in the form of a Dell PV 660/650F.

With all of this data on one server and storage unit, many administrators would be concerned that it created one big single point of failure. But if this strategy is examined on a deeper level, the facility actually benefits. With a production system based on a single server/storage solution, it is now cost-effective to have a complete backup data center. This backup data center option would not be affordable in the old UNIX-based system with multiple databases, as the technology is fragile and budgets would have been unobtainable.

The Big-Bang Method of System Rollout

There are basically two ways of rolling out a new system in an existing high volume facility: either by parallel running systems and slowly turning off one of the existing systems, or by transitioning all in one night (i.e., big bang). There are literally no alternatives to these two options.

The parallel-run theory is about risk mitigation and reduction. This process theoretically allows one department at a time to switch over to a new system. But it also has some very large drawbacks, namely:

• Time. Parallel running takes years to roll out to a multidepartment system;
• While the parallel run process is going on, the computer team is generally working around the clock to keep interfaces going that now need to work in perfect synchronization between two disparate systems, often from competing and noncooperative vendors. All of this multi-year, around-the-clock work for the IS team makes them tired and prone to mistakes;
• When mistakes happen, the process is elongated, as the users blame the IS team; and
• Users will typically find every excuse they can to retain their existing systems – old systems are like security blankets. People are very reluctant to give up old systems no matter how bad they become.

The big-bang theory of bringing a system live is obviously more chaotic as many activities need to occur simultaneously, and therefore is often seen as a higher risk than a parallel run transition. But the big-bang method has some interesting features:

• It is fast and you can then take advantage of the new functionality that you purchased (and get a faster ROI);
• The disruption in the departments is intense, but it is also short-lived as the users will tend to adapt quickly to any changes and “get on with it,” as the old system is now turned off and there is no going back; and
• Many analysts would argue that a parallel run process is even more risky than a big bang due to the potential risks in parallel running systems (morale issues, data mismatches, vendor opposition and politics, long switchover times).

The biggest risk in the big-bang scenario is that the system the facility is turning on is a very poor system (in respect to features and performance.) But, if the facility has done its homework correctly and mitigated this risk, then big bang certainly has its advantages.

Going Live

Bumrungrad Hospital went live with the GCS Hospital 2000 system using the big-bang approach. Curtis Schroeder commented, “We turned on the new system instantly after we switched off the old one ... there was no parallel running. It was a challenge preparing the hospital staff – approximately 2,000 of them – to manage an instant switchover to the new system.”

Leading up to the go-live date, there were obvious challenges, including:

• Employee training;
• Performing the data conversion; and
• Rolling out the new PCs and not disturbing the current system.

Data conversion was conducted the old-fashioned way – dump existing system data to text files, clean up the data, and then load the data into the new system. This is, of course, easier said than done, as the existing system had a lot of bad data inherited from the failing interfaces and subsequent data mismatches. The existing system also had data in English and Thai languages, presenting an additional challenge to the mainly European developers who had only basic Thai language skills. In addition to the data issues, the Buddhist calendar was used inconsistently in many parts of the system; almost the entire data conversion process was affected by date consistency issues.

The system was to go live on Dec. 17, 1999; it was predicted that this would be a slow time at the hospital with the Christmas holidays approaching. As luck would have it, it happened to be the busiest day they ever had with a full inpatient census and 2,400 outpatients! At 6:30 a.m., the new system was turned on, and the existing systems were all turned off. Everything worked and the hospital never looked back.

PACS at Bumrungrad

GCS also supplied the PACS at Bumrungrad. The GCS Amalga PACS operates on the same server and storage hardware as the hospital information system. No new servers or interfaces were needed, only some additional storage for the large radiology data generated by the 150,000 studies per year. The PACS system was implemented in January 2002.

One of the most challenging issues of a PACS installed in a Southeast Asian facility is the local language support for non- Western character sets. Imagine seeing a patient’s name in the Thai language on the operator console of an MRI made in the United States – this generally will never occur as most Western modality vendors tend to ignore the fact that a few billion people do not use English as their native language.

The PACS has an “everything online” approach to image storage. This is a very different approach than the traditional PACS that has just 30 to 60 days of recent studies online. Traditional PACS products archive older studies on a hierarchical storage media such as an optical jukebox. With the optical jukebox approach to image storage, there are several issues that are typically faced, including:

• Image retrieval time is typically very slow with systems based on hierarchical optical/tape storage (at least 30 seconds per study, if there is only one user on the system);
• Doctors tend to avoid looking at historical studies if they take too long to retrieve from the system, so patient care and the doctors’ decision-making processes are affected;
• Cost of hard disk storage has been dropping dramatically for the past few years, and the trend is predicted to continue at a 50 percent per year price drop per megabyte; and
• When radiology studies are stored on online media (hard disks), all studies are retrieved at the same speed.

Bumrungrad Hospital conducts 150,000 radiology studies per year. This consumes approximately 2 terabytes (TB) of storage in the production data center with a duplicate data set also located in the backup data center. With online storage media, the cost in 2004 for 4 TB of storage is approximately $100,000. This figure compares well to the previous annual film budget of $500,000 per year. Not all of the savings can be put back on the bottom line of Bumrungrad, as other PACSrelated equipment such as expensive diagnostic workstations and other hardware requirements consume some of these savings. With the cost of storage and diagnostic workstations dropping at considerable rates (50 percent per annum), more money will be put back into the Bumrungrad facility to improve patient care and customer service. Actual savings are only part of the story in justifying a PACS solution – patient care, medical staff career enhancement, and customer service is where to look for the real ROI of a PACS.

TheDays to close the monthly books reduced from 30 to four; Success Story

Bumrungrad Hospital has experienced double-digit growth for the past six years to become one of the largest private clinics in the world. In Thailand, the most successful private hospitals are those that have been able to bring good quality health care to the customers in a timely, efficient, and pleasant manner. Patients have the choice to use public hospitals and see wellqualified physicians at little or no cost. Therefore, a key differentiating factor in their choice is time and efficiency. A hospital’s information system is the single most important investment in delivering on this expectation.

Bumrungrad Hospital has closely tracked the before and after operational effects of the implementation of the enterprise software solution. The following are a few highlights of the assessment:

• Time to retrieve a medical record reduced from 25 minutes to virtually instantaneous;
• Patient waiting times reduced by 39 percent;
• Laboratory processing time reducing from 14.5 to three minutes per sample;
• Radiology processing time reduced from 57 minutes to 18 minutes per study;
• $294,000 of radiology film costs reduced to $26,000 in hard disk storage;
• Outpatient pharmacy dispensing time reduced by 50 percent;
• Outpatient bill preparation time reduced from 22 to eight minutes;
• Accounts receivable days reduced from 15 to 10 (quicker third-party billing);
• Forms stocked reduced from 395,616 pieces on average to zero;
• Inventory turns per year increased from 11.89 to 23.52 due to real-time inventory control and automated re-ordering;
• 10,000 square feet of medical records storage space was converted to a revenue-producing pediatric center treating over 110,000 children in 2003; and
• Staff efficiencies: 129 full-time employees savings attributed to Hospital 2000.

“Hospital 2000 has provided Bumrungrad the unique capability to handle a number of patients, which generates the critical mass necessary to maintain very competitive pricing,” adds Schroeder. Hospital prices at Bumrungrad are one-tenth of those in the United States, one-quarter of Europe, and one-third of Singapore.

Another Small Bang

Bumrungrad Hospital has adopted a philosophy that hardware needs to be refreshed periodically to keep a computer system at its peak performance and also keep the vendors interested in the facility as a customer. A computer today has a useful life that seems to get shorter and shorter as newer and faster hardware appears on the market on a daily basis. Desktop PCs can have a useful life of between three and five years, but servers and storage need refreshing every three years to keep up to date with the latest technology and provide users with big performance gains. Silicon-based chips get faster on an exponential scale (density doubling every 12 months according to Moore’s Law).

After four years of operation, Bumrungrad’s new system was running with the desired performance, meeting business objectives. However, great advances had been made in computer performance in that short period of time. It was time to evaluate a hardware upgrade. Reliability and performance improvements were to be gained by simply replacing the existing Dell quad 500 MHz Intel Xeon processor-based server and database storage system. New equipment was specified in January 2003 and was purchased and assembled in July and August. The most difficult part of moving a live system to a new hardware platform is the migration of the data. At this point, Bumrungrad had accumulated 500 GB of database transactions and more than 4 TB of PACS images and scanned medical records. The goal was to have a downtime window of less than 30 minutes while moving from the production system to the new replacement hardware.

The data conversion process was a success, and the server operating system on the new replacement system was also upgraded to Windows 2003 Advanced Server at the same time. The downtime window was actually less than 10 minutes. The performance gains to be had from upgrading the server hardware are:

• Database server: 500 percent performance increase;
• Database storage system: 700 percent performance increase; and
• PACS server storage: No performance increase, but dramatically smaller and easier to manage.

The current Bumrungrad system processes over 1.9 billion transactions per year (SQL statements) and has a peak-time performance load of approximately 25 percent CPU utilization. The system sees peak loads sustaining 250 transactions per second between 8 a.m. and noon. Bumrungrad will keep growing in patient base and computerized medical management and will continually be improving and upgrading PCs, servers, and storage.

With a highly computerized, large tertiary facility, one could imagine the need for a large IT department maintaining the equipment and software. The IT department at Bumrungrad is staffed at 10 FTEs. The department has aroundthe- clock operational responsibility for the system.

The Future of Bumrungrad

Bumrungrad Hospital is continuously expanding both its physical facilities and its range of services. Construction has recently started on yet another expansion building, scheduled for completion in December 2005. The facility has a projected patient volume of 1.6 million per year, nearly doubling the current capacity. Inpatient beds will be expanded to 650, and radiology studies will grow to over 250,000 per year. The entire computer system hardware will be replaced every three years. At these prices, upgrades and system replacements should not create any financial burden for the Bumrungrad facility.

 

 

 

About the Author

Title: 

Chief Executive Officer and Co-founder

Global Care Solutions

Patrick Downing is chief executive officer and one of the founders of Global Care Solutions, and acted as chief designer for the Hospital 2000product line. Before joining Global Care Solutions, he worked for Perot Systems Corporation for 10 years as a software developer, consultantand major account manager.