Installing a new laboratory computer.
Preparing to receive a new laboratory computer system can be more traumatic than selecting one, since the entire laboratory staff is involved in the installation process. Laboratories that have never used a computer must revamp many comfortable, familiar procedures to accommodate the newcomer.
In our 687-bed hospital lab, the countless adjustments took place two and a half years ago. They were a bit easier to accomplish in chemistry and hematology, which had worked with an earlier computer system for 10 years. The other sections totally lacked data processing experience.
Our original system was fading rapidly when we selected its successor. Lengthening downtime-- one episode lasted seven days-- made installation of the new computer a top priority. Even so, considering the extensive laboratory groundwork required, we decided that six months would be more realistic than the computer vendor's proposed four-month schedule.
Figure I shows our installation schedule. Among other things, we had to redesign laboratory space to make room for the computer equipment; seven sections had to codify data for each of their tests; and 260 full-time equivalents had to be trained for the system.
Engineering, carpentry, and other hospital departments render important assistance on such a project. You can't assume their schedules will somehow fit your deadlines. A few preliminary meetings with the support services can prevent many last-minute headaches.
Shortly after signing the contract, we began site planning. As part of the contractual requirements, the vendor specifies that the facility must be ready for installation before the hardware is shipped. (If the hardware sat unused because the lab wasn't prepared for delivery, the vendor's invoices might go unpaid.)
To speed things along, our vendor provided what it euphemistically called a "site kit,' an imposing 170-page manual crammed with specifications on each piece of hardware, and space, power, and environmental requirements. Much of the information was on charts that guided us through the steps we had to take.
Details will vary, but any laboratory must perform the same type of planning to transform part of its space into a facility that can accept a complex computer system. In our case, we had to create a computer room within the lab-- the original system was located in another building.
Planning a computer facility can seem monumental at the outset. Some computers require a raised floor to provide space for cabling and air flow. Environmental controls are almost always necessary. We installed a separate air conditioner and humidity regulator in the comuter room; the temperature had to be a relatively constant 72 F. Other key details include estimated electrical power requirements and plans for cabling runs to all terminals.
Safeguards must also be reviewed. For example, to protect against possible vandalism, we installed a combination lock on the door to the computer room. Other considerations include the potential for a corrosive atmosphere or excessive dust; electromagnetic interference from such sources as nearby television, radio, or telephone company antennas; and fire protection, preferably an inert gas system because water from sprinkers would damage the hardware. Overlooking any of these factors could hamper or incapacitate the system.
You may also have to purchase other products or services as part of the installation process. Your computer vendor will specify the maximum allowable cable lengths from the computer room to the terminals; in our case, it was 2,000 feet. The alternative, for longer distances, is to lease data lines from the phone company.
Plan on at least eight weeks before leased data lines are in place. They are more complicated to specify, install, and test than lines for desk phones.
Some computer installations may require that a number of terminals be located remotely--beyond allowable cable limits--in one or more buildings. This might occur when two or more hospital laboratories in the same community decide for the sake of economy or standardization to buy one large computer to support several sites. While a modem can provide the link between a single terminal and a telephone line, multiplexors are better for several terminals. They concentrate the data on fewer lines at substantial savings in cost.
We used this approach because our cytology, endocrinology, radioimmunoassay, and toxicology labs are located in another building too distant for economical cabling. with multiplexor equipment trimming the need for telephone lines, we were able to project savings at 1980 cost levels of approximately $5,000 over five years. Since telephone costs have increased substantially since then, the actual savings have been even greater.
If you plan to interface any laboratory instruments directly to the computer, your vendor may require that you purchase special microcircuits from the instrument manufacturer. Don't forget to budget this item when estimating installation costs. We paid more than $17,000 just to interface four chemistry analyzers to the computer. There are now seven instruments on-line, including our cell counters.
Once facility planning gets under way, the laboratory can tackle the next task: preparing the data base. This consists of information that must be stored in the computer to insure that the system will meet the needs of your particular lab. The computer functions as an electronic filing cabinet, storing data in logical groupings or files. Patient room numbers, test prices, organisms (for microbiology), and doctors' names, numbers, or codes are a few of the relevant files placed in the computer.
A complete file--it's sometimes called a methodology file-- is devoted to each test in the lab. This file is made up of all data relevant to that test and normally includes the test's name for reporting purposes, test code, reference values by age and sex, critical and technical values, test price, accounting codes, and quality control criteria.
Preparing the test descriptions can become quite time-consuming. We had every section draft its own methodologies. This spread the work around considerably and also put the responsibility for this important step at the level where it truly belonged. Each section selected a computer representative who then obtained and recorded the necessary data.
Our vendor's standard methodology forms simplified the task considerably by asking the right questions and defining the lengths and types of data the system would accept. Our maximum character count per entry for test names was 12, so a number of them had to be abgreviated. Once a section had completed its methodology forms, the laboratory's computer supervisor--a former chemistry technologist--reviewed all the data for content and uniformity. Figure II shows part of the methodology for BUN.
The entire process took three months. Nearly 9k per cent of our tests were computerized via 330 methodologies running from six to 10 pages in length. We ultimately shipped a stack of forms four feet high to the vendor's plant, where everything was entered into the system. When our computer operators traveled to the plant shortly thereafter for training, the entire data base was printed out and ready for proofreading. These operators, by the way, do not necessarily have laboratory backgrounds.
After the data base has been entered, the laboratory can begin planning its patient reports. Report software from the vendor, combined with the data base, enable the lab to enter test results and print out sample reports in whatever formats are available with the system.
At this point, it's worthwhile to involve the medical staff, since clinicians are the main users of patient reports. Our laboratory sent out a questionnaire soliciting their views on the most desirable report format. We also held several meetings with clinicians who had expressed opinions in the past about our chemistry and hematology computer reports.
We ultimately adopted a vertical format with the test dates across the top of the page and the test names down the left-hand margin. , indicated a preference for this approach because it was consistent with the reports generated by our previous computer.
Unfortunately, medical staff recommendations don't always reflect what will please the staff in practice. Within six months, we began hearing complaints that the reports were "busy' and difficult to read. Our software permitted a switch to a horizontal format, which we introduced this past summer. That hasn't proved entirely popular, either. We're taking clinicians' comments under advisement, but won't make any more changes until they reach a consensus.
Several other seemingly mundane matters must not be overdane looked during the final planning stages. For example, what supplies will the new computer require? The vendor can help you prepare a shopping list of standard items as well as items peculiar to your system.
Place orders early enough to insure that the supplies are on hand when the hardware arrives, and keep track of inventory levels and delivery periods for timely reordering in the future. Some items that typically must be ordered include disk packs, magnetic tape for systems with a tape drive, labels for the collection list, printer ribbons, 8 1/2 11 report paper, and 11 14 "stock' paper.
Also review insurance requirements with the hospital's risk management or legal department. The type and amount of insurance needed may depend on whether you buy or rent, whether fire protection has been designed into the computer room, and so forth.
Security measures are another important consideration--primarily, controlling use of the computer room and the computer itself. Computer access is restricted by establishing "authority' codes and limiting their distribution to employees who actually will work with the controlled programs. Most systems also have ID codes to identify the technologists entering test results. Some carry similar codes for individuals who enter test requests and verify the collection list. Apart from security, the ID codes fix accountability for entries, and help track errors to their source.
You may want to locate a vault or secure room outside the laboratory for backup programs or data. These are generally kept on portable disk packs, which are not excessively bulky. We stored our backup packs and tapes in another building at the university, well be-yound reach of, say, a fire in the computer room.
Once the computer facility is completed and the hardware installed. it's time to educate the laboratory staff. Most suppliers of laboratory systems offer in-depth training for computer operators, and they may also train some other personnel, such as supervisors. These laboratory representatives in turn oversee the training of most of the technologists.
Our vendor supplied computeraided instructions (CAI), an extensive programmed introduction to the new system. The CAI package had 20 separate lessons covering such topics as request entry, patient inquiry, quality control procedures, and management reporting.
With CAI, technologists work at their own pace and need only take the lessons that apply to their function in the lab. Each lesson is followed by a quiz. The system retains scores, allowing the supervisor to monitor the progress of individuals and the section as a whole.
This approach is very helpful, especially in laboratories getting their first exposure to data processing. The learning process it-self gives employees the chance to punch the keys and overcome their initial computer wariness. But if CAI is not available, introductory training can readily be provided through lectures to the staff. No matter which method you use, it is important to reach all employees on all shifts, including part-time and weekend staff members.
Even when CAI is available, however, don't rely on it alone to provide comprehensive training. "Parallel runs' are even more important. The idea is to continue reporting results with your current system and simultaneously begin entering data in the new computer as if it were in actual use.
Of course, operating two complete systems is a strenuous undertaking. It's more practical to have employees use the computer for a small percentage of their workload. This lets them get used to the programs they will each run once the system goes "live.' Thus, practical experience augments the more theoretical knowledge edge that is gleaned from CAI or lectures.
There's no need to schedule everyone in the lab on the computer at the same time as would be the case with a 100 per cent parallel run. We completed all of our partial parallel runs in just four or five days.
The training period, particularly the parallel-run portion, gives the laboratory a good opportunity to evaluate the system's performance and make sure the vendor has provided everything promised. Substandard components may fail somewhere in the system, but they can be ferreted out and replaced without disrupting laboratory routine. The lab should not proceed beyond this point until the system operates as advertised and technologists are reasonably
competent in using it.
Before installation, consider centralizing some of the functions previously handled by the individual sections, notably test sorting and specimen receiving. When a computerized laboratory receives specimens, efficiency dictates that most of the requested tests be entered into the system at one time. Specimens that require centrifugation are spun and divided among the sections that need serum, and urines are similarly distributed. The computer prints out specimen labels for each test requested.
I strongly recommend centralizing the service functions as soon as possible, preferably before the computer arrives. The laboratory has enough to worry about during the initial days of computer usage without wrestling with major or-ganizational changes.
During our planning for installation, we paused over including microbiology among the central services. With the former computer. chemistry and hematology work was processed centrally, and now serology, endocrinology, toxicology, radioimmunoassay, and urinalysis specimens would be, too. However, microbiology did not share the same specimen processing techniques, and it required other computer entry methods.
We canvassed a number of computerized hospitals in our area and found that a clear majority processed their microbiology specimens separately. For one thing, computer entry in that section requires more specialized clerical training. Second, microbiology specimens are not normally shared with other lab sections. Third, the test request slip may include pertinent specimen information that must be conveyed to the technologist responsible for the culture. Finally, since we planned to use card readers to enter microbiology results into the computer, card preparation could begin with the secretary who entered the test request.
We thus decided to hold microbiology test requests apart from the rest of the system. This has worked out well, especially now that microbiology has moved to another floor.
When the laboratory has grown reasonably comfortable with the system's capabilities, has implemented necessary organizational changes, and has prepared technologists to cope with computerization, you can set a date to begin using the new system. The question is, should all laboratory sections come in at once or should they be phased in gradually? Putting your toes in the water first sounds more prudent than a plunge, but there are better reasons to have the entire lab go online at about the same time.
First of all, a number of sections are interrelated by specimen --where samples are split-- and it's simplest to enter all tests for all sections in the computer at one time. The alternative would be to enter tests for only one section and prepare paperwork manually for the others.
In addition, most vendors will support a start-up blitz by stationing their personnel in the laboratory during the early days of operation. They will not, however, remain for the several weeks it might take to phase in the system. The lab would then be forced to bring up the final sections without vendor experts on hand if problems arise.
We had three representatives on site. One went home after the first week; the other two stayed three weeks. Frankly, we could have used them longer.
Bringing up all sections of the laboratory together gets the job done with one large initial shock. Thereafter, computer processing becomes easier with each passing day, and the turmoil wanes--assuming reasonably intelligent planning and adequate computer performance. Many labs in our informal survey chose a one-day start, and most were reasonably successful.
Another recommendation is to begin computer operations early in the week. The majority of our sections did it on a Thesday, because Mondays are always busy with added work left over from the weekend. This left us the balance of the week with a full staff to work out the problems that inevitably crop up.
Our new computer system went live in June 1981, beating the sixmonth deadline by a few days. Because microbiology does not report out final results for several days, it went on the computer the weekend before our official starting date. On Tuesday, all sections with specimens in common began using the system with the start of the first shift. On Wednesday, we brought up cytology and surgical pathology.
The first few days of computerization are taxing for the entire laboratory. Vacations and time off should be held to a minimum for the first four to six weeks so that a reasonably complete complement of employees is available. This also insures maximum training for everyone at the same time and while the vendor's installation team is on site.
Be prepared for staffing and procedural changes to accommodate the needs of the computer process in the first few days or weeks. These are difficult to predict beforehand, more so if you have never had a computer before. A number of the modifications introduced in the early days of computer usage speeded up the flow of specimens on both the day and evening shifts. We also reorganized the specimen receiving department.
Individual users will benefit from a set of instructions devoted to the computer. This should not attempt to duplicate the vendor's manuals, but rather provide specialized data applicable to your lab. It might, for example, include unusula codes, entry procedures for special requests on particular tests, patient room numbers and locations in the hospital, how outpatients are handled in the computer, billing data, and so forth. Our 35-page manual wasn't actually completed until six months after we went live and all the procedures had jelled.
If you purchase considerable amounts of software written specifically cifically for your site, you may need additional assistance from the vendor. Newly written software invariably has bugs or error, and several weeks may elapse before all runs smoothly.
Another word of advice: Go slow in purchasing special software. It is expensive, and sorting it out takes a lot of time. Spend a few months getting to know the computer and then decide what software you really need. You will probably find that some standard programs work just as well or better. (Some more basic dos and don't are outlined in Figure III.)
Despite all our planning, we experienced severs response problems the first few days. The system would "lock up' in the afternoons at the peak of laboratory activity. We later learned that the files weren't located optimally on the disk packs, forcing the computer to jump around too much to locate data. This led to a traffic jam when entries multiplied.
We called the vendor, and the service manager arrived from the West Coast the next day. He promptly corrected the problem by rearranging the files. Any new installation can be ambushed by difficulties. However, if the lab has done its homework and selected a reliable vendor, reinforcements should arrive quickly.
So, with the cooperation of our employees and extensive support from our vendor, we were able to stay on schedule for installation and even gain a few days. It was a good thing, too. About the time the new system went live, our old computer was on the critical list and going down for the final count.
Figure III; Dos and don'ts for installing a lab computer
Begin facility pianning immediately after execution of the computer contract.
Allow sufficient lead time in ordering telephone lines.
Appoint computer liaison personnel in each section of the lab.
Solicit comments from the medical staff to assist in format of reports.
Implement computer operator staffing as early as possible. These personnel must be trained before the system goes into operation.
Forget supplies (paper, label stock, printer ribbons).
Put off any expected lab reorganizations. There will be enough confusion in going "live' with the new computer.
Overlook parallel runs before going into production use with the system.
Table: Figure I; A lab computer installation schedule
Photo: Figure II; Preparing a computer methodology for BUN testing
Specimen and processing information
|Printer friendly Cite/link Email Feedback|
|Title Annotation:||1983 MILO article awards contest; first prize|
|Author:||Stadler, Stephen L.|
|Publication:||Medical Laboratory Observer|
|Date:||Jan 1, 1984|
|Previous Article:||Hunt for an AIDs market key to resolving transfusion fears.|
|Next Article:||Realistic staffing via workload recording.|