| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Special Review |
1 Department of Radiology, University of Alabama at Birmingham, 619 S 19th St, Birmingham, AL 35233.
Abstract
The year 2000 computer problem arises from a long-standing and often-duplicated computer programming error. Affected programs use only two digits to represent years, which may lead to a variety of computer malfunctions and data errors related to crossing from 1999 (99) to 2000 (00), at which point computers may interpret 00 as 1900 or other incorrect dates. Radiology and medicine may be seriously affected by this problem as it relates to the function of its equipment; business functions such as scheduling, billing and purchasing; the reliability of infrastructure such as power and telecommunications; the availability of supplies; and many other issues. It is crucial that radiologists, as practitioners of one of the most computer-oriented medical specialties, help lead the effort to ensure continuity of operations as the year 2000 boundary approaches and passes. This article provides suggestions for a structured approach, as well as tools and checklists, to guide project leaders attempting to identify and remediate year 2000–associated problems within radiology facilities.
Index terms: Computers • Radiology and radiologists, departmental management
A challenge to radiology threatens to surpass the risks posed by the high-profile problems of managed care, government regulation, third party reimbursement policies, and turf conflicts. The threat is imminent, yet few radiology practices are seriously attending to it. This threat is the year 2000, or Y2K, computer problem. Because radiology is arguably the most technically oriented and computer-intensive of all medical specialties, it would be unfortunate for patient care and the perception of radiology should we not be among the best prepared. Year 2000 preparedness also creates the opportunity for us to serve our profession, to be recognized for it, and perhaps to strengthen our position within the medical community.
NATURE OF THE YEAR 2000 PROBLEM
For many years after the beginning of software programming, computer engineers employed two digits to represent the year (eg, 98 rather than 1998) (1). This causes problems for date records and calculations that cross the century boundary. The year 00 may be interpreted as 1900 or other dates within the 20th century (such as 1980 or 1990). Therefore, the order of records may be sorted incorrectly, financial and other calculations (such as accrued interest) may be incorrect, and future scheduling may not be possible. Logically impossible results may occur such as calculating the age of someone born in 1965 as 00 - 65 = -65 rather than 2000 - 1965 = 35. Some software is programmed to reject impossible results and may shut down and may even be impossible to restart. Furthermore, equipment may contain embedded microprocessors that use date information, and it may be difficult to locate these and, in some cases, impossible to repair them without replacing the equipment.
These programming features have been retained or brought forward into literally billions of devices and programs that are still in use today, leading to the fear of serious disruptions in countless organizations from financial institutions to government agencies to utilities. Although many systems do not have this problem and most others will be fixed, a substantial number of problems may remain.
Describing the complexity of the problem is beyond the intent of this article. Information has been evolving rapidly and numerous references are available on the Internet. Specifically, http://www.rx2000.org is focused on medicine and includes self-help tools, proposed letters to send to vendors, and a fee-based database of vendor compliance. A site established by Peter de Jager at http://www.year2000.com is recognized as one of the most comprehensive sites with detailed articles and links to numerous other sites. The site http://www.yardeni.com is oriented towards economic consequences. Other sites with extensive year 2000 information include http://www.y2ktimebomb.com and http://www.yourdon.com. Additional information can be found at our radiology department Web site: http://www.rad.uab.edu. There are also an increasing number of books (referenced at some of these sites), but to my knowledge there is little in print that is specifically relevant to radiology.
PREDICTED SEVERITY OF THE YEAR 2000 COMPUTER PROBLEM
The risk posed to society at large by the year 2000 problem is controversial. Yet serious and respected professionals, including Edward Yourdon, PhD (1), an expert in software development, Edward Yardeni, PhD (2), Chief Economist for Deutsche Bank Securities, and John Koskinen, Director of the President's Council on Year 2000 Conversion (3), express grave concerns that elements of our basic physical and social infrastructure will be affected by the year 2000 problem. Although estimates vary, there is not enough time left (or money or inclination) to repair or replace more than about 80%–95% of noncompliant computer systems (1).
Those minimizing the potential importance of this problem observe that (a) remediation for this problem is delimited and well described, (b) tools to repair or replace systems are widely available, (c) major institutions usually have the resources to perform such remediation, (d) it is strongly in their interest to do so, and (e) some effects of the year 2000 problem will be positive (4). Some are skeptical of the financial or even religious motives of those who have spoken out (5).
Those who argue for greater concern note that (a) the scope of the problem is extreme (and much worse in most nations other than the United States, Canada, United Kingdom, the Netherlands, Australia, and a few others), (b) the peak of the problem will be nearly simultaneous and global, (c) most software projects either fail or are late, (d) there is a high degree of system interconnectedness that markedly increases the complexity of the problems (6), (e) many embedded microprocessors with the flaw are difficult to identify and replace, (f) remediation is expensive, (g) the numbers of people with the ability to fix these problems are limited, and (h) time is short, considering the quantity of work yet to be completed (1,7–11).
Perhaps one of the most compelling arguments for concern has been put forth by Way (12):
The problem is that when something gets to be very much bigger in any dimension, it becomes a `new thing,' and the tools and techniques that work when it was smaller are no longer effective. . . . The reasons for this phenomenon originate . . . in the laws of nature. . . . The laws of chaos or complexity theory tell us that when a system gets to be large. . . [and] has many interconnected elements, then even small changes can have very large and unpredictable, non-linear effects on the system as a whole. . . . As the number of elements gets large, the connections between the elements come to dominate the results. . . . The probability of failure (F) of a system is one minus the product of the reliabilities (R) of ALL of the interdependent elements: F = [1 -(1 - Ri) as I = 1,n]. Even if a particular program has a very high reliability, say 0.99999,. . .by the time [many] programs are interconnected, the fractional reliabilities are multiplied. . . . Hence the probability of failure goes up and becomes very real.
RISKS TO THE HEALTH CARE SYSTEM
The risks posed to medicine include the failure of many devices to function properly or at all, including medical devices, information systems, business systems, and connections with outside organizations such as third party payers and government agencies. An additional risk includes the contamination of data from noncompliant to compliant computers. Timely reimbursement may be threatened by problems in financial and government computers. Some express concern about the viability of basic utilities and the possible long-term consequences, including serious recession (2,11) with consequent changes in health care reimbursement and policy.
AUTHOR'S PURPOSE
The radiologist's interest goes beyond ensuring the ability to perform examinations to ensuring the continuity of the radiology practice. The purpose of this article is to (a) familiarize the reader with the nature and scope of the problem, (b) define the risks to their organization, (c) provide a nontechnical guide to project management, and (d) to reference useful online information.
Because of voluminous, rapidly evolving information, the complexity of the problem and the difficulty of verifying data, I cannot stipulate that this information is complete, that following it will itself not cause damage or data loss, or that the suggestions herein will guarantee compliance. Furthermore, the reader should appreciate that although I have supervised software development projects, I am not qualified to write computer code. I have not tested the specific tools mentioned in this article and thus cannot verify their quality, and I do not recommend any particular product or approach. I also have no financial or personal interest in any product or service mentioned in this article or any of the Web sites noted.
REALMS OF RISK
Managing this problem cannot be delegated, and radiologists will be held accountable. At this late date, it may be difficult to engage a high-quality consultant. Therefore, it is likely that much of the work will fall to current employees. Because of the limitations on time and resources, contingency planning should be performed including manual backup processes for examination ordering, internal communications, report communication, billing, and so forth.
The scope of this problem demands a well-organized approach. The following categories of risk are proposed to help segment planning efforts and ensure that all genuine risks are addressed. Many generic schemes have been proposed (10), but I have adapted this specifically to a radiology environment. The nine categories of risk are (a) medical devices supplied by vendors, (b) radiology-specific information systems, (c) connections with nonradiology systems within the institution, (d) networks and personal computers (PCS), (e) relationships with outside systems and entities, (d) utilities and physical infrastructure (internal and external), (e) societal, health policy, and macroeconomic effects, (f) adequacy of contingency planning, and (g) legal liability. Approaches and special considerations will be more fully discussed in the following sections.
YEAR 2000 PROJECT PLANNING
It is quite late to begin a project of this scope, and success may require persistent effort from top management. A project team should include key technical and management personnel and may be responsible for creating an educational plan to sensitize other management personnel and the institution at large to the nature of the problem and to motivate collaboration. The team should also (a) establish liaisons with other institutional resources, (b) create a method for prioritizing research and remediation, (c) create a process for documenting its actions and findings, and (d) report regularly to senior management. Different phases of the plan will proceed concurrently. The following nine project management components apply best to the first four realms of risk relating to computers, medical devices, and information systems, with some elements appropriate for the remaining five.
1. Develop or acquire an inventory tool (Table 1) and
|
|
|
4. Develop fallback (contingency) plans (Table 4). The term "contingency" fails to capture the breadth of effort needed to deal with possible scenarios from the year 2000. Particularly because health care has begun planning later than other economic sectors, more effort must go into thoughtfully planning for working around device and system failures, for planning for external failures (such as utilities outside your control), or planning for meeting supply needs if the usual supply systems become less efficient. The latter have been termed "fallback" plans. As well as such "bottom-up" issues, the "top-down" problems of maintaining business operations must also be considered with regard to billing, cash flow, payroll, purchasing, and so forth. Whereas standard disaster plans or weather-related contingency plans can serve as a model for some of the contingencies that may arise, other plans need to be thought through from the beginning.
|
5. Test to confirm compliance (Table 5). One must avoid the temptation to simply advance the date of the system to determine compliance (Table 5) (16). This may lead to inaccurate assessment of the state of compliance, damage to data, or system malfunction, which might not occur if more controlled testing were performed. For example, inaccurate assessment might occur if the system properly recognizes January 1, 2000, but does not recognize that 2000 is a leap year. One risk of advancing the date is corrupting data in databases and spreadsheets. Some systems may fail if dates are advanced because they are programmed with security provisions that detect tampering. Another possible problem would occur with programmed maintenance schedules that prevent function if maintenance has not been performed within a specified interval. For example, if a system's clock was set forward to 2000 in January of 1999 and the system was scheduled for maintenance every 6 months, it might not function, whereas if allowed to roll over naturally from December 1999 it might function normally.
|
6. Remediate or repair. For each system, determining whether to retire, replace, or repair a system helps permit calculating resource needs. If replacing a noncompliant system is not possible before January 2000, determine if an interim contingency plan is feasible. One may also wish to attempt to work with hospital administrators and your vendors to facilitate the acquisition process if necessary. To prevent interconnected systems from contaminating data, one precautionary procedure is to disconnect from such systems until compliance and compatibility are verified.
7. Test remediated systems. Testing remediated or replaced systems is similar to the testing for compliance. Many software engineers and project managers indicate that this testing is actually much more time-consuming than the remediation itself. At this point, tests of interfaces and interdependencies must also be performed. When systems are verified and validated, they can be put back into production. Personnel training must not be ignored for any changes that have occurred.
8. Return tested systems to production and
9. Apply postremediation plans. Staff should be alerted to recognizing possible date-related malfunctions and must have methods to report errors after remediation and after January 1, 2000. Fallback and contingency procedures are also part of the staff education process. The latter part of the year 2000 project should involve extensively preparing personnel for possible problems. New procedures should be rehearsed in a realistic environment and measures taken to ensure employee compliance with such training. To prepare for the January 1, 2000 event, recovery procedures should be designed in advance and one should be prepared to mobilize any technical resources available. Also, staffing schedules should be re-adjusted, crucial supplies stockpiled, and perhaps elective procedures limited.
ADDRESSING THE NINE REALMS OF RISK
Medical Devices Supplied by Vendors
Devices to inventory include all examination units, ancillary devices, and physiologic monitoring and resuscitation equipment. Many of these have proprietary hardware and software and must be replaced or remediated by the vendor. Other systems with embedded microprocessors may be difficult or impossible to test and are at high risk if date-related information is included (22).
The nature of the risk is for complete system failure, inaccurate tracking of dates, or inaccurate calculations of physiologic parameters when dates are employed. An example of the latter would be incorrect calculation of gestational age with an ultrasonographic unit. Effects of such failures would include delays and incorrect diagnoses and needing to triage patients to other examination modalities or units.
Radiology-specific Information Systems
The patient scheduling, order generation, film tracking, examination reporting, and other functions of radiology information systems are all subject to year 2000 problems. Backup procedures should be designed even if the system is believed to be compliant, because such systems often have complex interactivity with vulnerable outside systems.
Picture archiving and communication systems (PACS) may be sensitive to date-related examination sorting. Such systems also often have complex interactivity with outside systems and data may be corrupted, even if the systems themselves are compliant. Although, most PACS are relatively new, this does not ensure compliance.
Business systems are particularly vulnerable to year 2000 problems because date comparisons and calculations are often performed. Also, some of these systems have used internally developed software that may be poorly documented.
Connections with Nonradiology Systems within the Institution
As noted above, connectivity with noncompliant systems strongly affects the reliability of the networked systems as a whole. Such connections with noncompliant devices may lead to input of incorrect data or data that is improperly formatted. In addition to obvious malfunctions, inconspicuous data errors may be difficult to identify. Examples of problems include the incorrect passage of dates between a radiology information system and a billing system, which may lead to rejection of claims. Also, billing and clinical systems databases may require cross-checking for date reconciliation before billing can proceed. If this cannot be performed, large numbers of procedures may not be billed or claims approved.
Networks and PCs
PCs are often under widely dispersed control and are composed of diverse hardware and software. A surprising percentage of PCs (for example, almost one-half of PCs sold in 1997) are noncompliant. Spreadsheet and database programs may be most affected. Unfortunately, PCs may represent a surprisingly difficult problem in 2000 because so many people use PCs, many are ill-prepared to deal independently with problems, and PC remediation is usually a low priority, (21,23,24).
Particularly in academic institutions, there is much customized software for research for which source code is usually poorly documented, if at all. All crucial data should be backed up, with testing occurring to minimize the risk to these data. Unfortunately, networks are rarely under the control of radiology practices, network expertise is limited, and networks depend heavily on software at risk.
Relationships with Outside Systems and Entities
All medical facilities depend on effective data interchange with third-party payors and government agencies. Unfortunately, the U.S. government is among the most poorly prepared groups at this time for the year 2000 (25), and there is concern about the effectiveness of the Medicare payment systems. However, connections with private payers are also at risk. Careful auditing of results will be required after January 1, 2000. Thorough computerized and paper backup of clinical and business data late in 1999 is often recommended. While computerized backup should be routine, the need for paper backup might be arduous and decisions must be based on the confidence one has in the thoroughness of your testing and remediation of affected systems.
The recent wave of restructuring and institutional mergers has created connections among institutions that may be fragile and susceptible to problems. Relationships with other entities outside the medical institution that are most crucial to radiologists include those with vendors supplying film and chemicals, contrast material, catheters and other interventional devices, emergency medications, parts to maintain equipment, and other general hospital supplies. While savings in inventory management have been gained in recent years from using fewer vendors, a better strategy for the year 2000 may be to diversify sources of goods.
Utilities and Physical Infrastructure (Internal and External)
Among the greatest fears of all sectors of society is disruption to utilities and basic physical infrastructure. This is among the most hotly debated of all concerns. While most thoughtful observers doubt massive utility failures within the United States, regional or local failures, brownouts, and voluntary usage cutbacks are predicted by some, and few are willing to preclude more severe scenarios (11). All utilities are potentially affected, including water, electrical power, natural gas, telecommunications, and transportation.
Because of the core importance of power and despite the inability of the radiology practice to control remediation, there are important aspects of this problem that can be addressed by radiology. If emergency power is supplied in selected locations in your institution, confirm that your most critical systems are connected. You may also attempt to determine from your institution whether your emergency back-up systems themselves have been evaluated and remediated and how long emergency power will be available if necessary. One might also communicate with local utilities to determine whether, if any power is available from local utilities, it is likely to be preferentially provided to hospitals.
Even if power is available, power quality may be poor with spikes, surges, and brownouts that can damage computerized equipment such as computed tomographic scanners and magnetic resonance imagers. Devices at risk should be supplied with uninterruptible power supplies and other power-conditioning devices.
Even if regional telecommunications function properly, computerized PBX (private branch exchange) telephone systems in institutions may not. Some institutions have separate emergency phone systems in selected locations. Confirm that these are in locations you consider most important. Computer networks are controlled by numerous routers, bridges, hubs, and software-mediated communications protocols, which must be evaluated and remediated. Insofar as such systems are used to communicate medical information (such as radiology reports) and perform daily business operations, such communications and operations are at risk.
Such items as heating and air conditioning systems and elevators are susceptible to year 2000 problems. Despite their superficial indifference to dates, these systems often contain embedded microprocessors that require periodic maintenance to continue operation or involve scheduled operations. An affected system may not function properly because it calculates in the year 2000 that no maintenance has been performed in 99 years. This is possible if the year 00 minus year 99 = -99 is programmed to be interpreted as the absolute value of -99, which is 99. A more apparent risk is to timed security systems, which might misinterpret the day of the week or simply not function at all, for example, leaving all doors either locked or unlocked.
Societal, Health Policy, and Macroeconomic Effects
If major problems arise from the year 2000, such as an economic recession, as predicted with a 70% chance by Yardeni (2), then the health care system is likely to be adversely affected. Higher unemployment will lead to a lower percentage of insured patients, an increase in uncompensated care, and more pressure on insurers to limit payments to providers, exerting further pressures on institutional finances. This may further heighten the government's interest in health care controls.
Internally, the institution may find that it has become perceived as a more attractive site of care based on year 2000 problems or the reverse may occur. Changes in staffing may be required to respond to these shifts. However, adding staff takes time for recruitment and is financially stressful, while laying-off staff is distasteful at best. Therefore, restructuring to maximize productivity and flexibility in advance may be worthwhile to minimize such problems.
Adequacy of Contingency Planning
Even if well prepared, the thoroughness of contingency planning may be the most important factor in the ability to continue operations smoothly through systems disruptions. The risks for which contingencies must be planned in addition to the above are those that relate to supplies. The move towards just-in-time supply policies to save money on inventory management jeopardizes the availability of supplies when transportation or financial systems are disturbed. Contingencies related to personnel availability should also be considered. Although automobiles and trucks are unlikely to be seriously affected, if power is affected, many employees may choose to attend to urgent personal and family priorities in midwinter rather than coming to work. Another scenario to consider may be that a hospital with emergency power may be one of the few warm places in a region affected by power problems, leading to the need to provide for an influx of people seeking refuge.
Even if major problems such as these do not occur but many computer systems are affected, manual procedures will increase labor needs and decrease productivity. Procedures should be considered to encourage overtime in an orderly and equitable way. Much may be expected of us and those with whom we work, and broadly, honestly, and openly sharing information will encourage cooperation.
As has been stated online:
Organizations need to move Y2K from the domain of technology experts into the entire organization. . . . The trust and loyalty developed through these strategic conversations and joint planning will pay enormous dividends later on, even if projected breakdowns don't materialize. . . . In developing scenarios, information is sought from all over. People think together about its implications and thus become smarter as individuals and as teams. Whatever future then materializes is dealt with by people who are more intelligent and who know how to work well together (26).
Legal Liability
The incontrovertible primary mission of any health care institution to care for its patients in the highest quality manner will be severely stressed if the medical care system is more than minimally affected. Many believe that legal liability risk will lead to suits totaling $1–2 trillion in American society at large. Health care would be among the greatest at risk. Methods to minimize this risk include specially attending to centralizing documentation of remediation efforts and avoiding internal e-mail communication that may be damaging during the legal discovery process. Disclosures that are required for fiduciary or legal obligations should be constructive but carefully worded.
One must appreciate the limited protection offered by vendors' statements of compliance and appreciate that legal recovery of damages from vendors is unlikely. Other risks to the institution are from an inability to monitor regulatory compliance because of year 2000 system failures; from malpractice, personal injury, and wrongful death actions; from accusations of failure by directors and officers to apply due diligence and reasonable business judgment; and from difficulties in maintaining policies and procedures that comply with licensing and accreditation rules (27).
CONCLUSION
Regardless of expectations concerning the year 2000, from minor inconvenience to global catastrophe, the problems are not trivial and correcting them is time- and resource-consuming. The process can be most effective if well organized. Health care has been widely reported as being well behind most other economic sectors regarding year 2000 awareness, probably because of its highly fragmented nature. As one of the core services of health care with which most clinical specialties have frequent contact, and as one of the services that is most conscious of the nature of technology and computerization, radiology should endeavor to lead and assist in this important but arduous project to maintain health care services as the clocks roll over to the 21st century.
Footnotes
Address reprint requests to the author.
Received September 2, 1998; revision requested October 14, 1998; revision received October 22, 1998; accepted October 23, 1998. References
This article has been cited by other articles:
|
|
 |
|
  A. V. Proto Radiology 1999: New Feature-Viewpoint Radiology, October 1, 1999; 213(1): 1 - 1. [Full Text]
|
|
|
|
|
|
A. V. Proto Radiology 1999—New Features Radiology, January 1, 1999; 210(1): 1 - 1. [Full Text]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| RADIOLOGY | RADIOGRAPHICS | RSNA JOURNALS ONLINE |
