.jpg){kind=logo}
Abstract
Clinical information systems in oncology have gradually been transferred from paper to electronic formats. This paper details the development of the Oncology Information System (OIS) from paper format to a modern unified electronic system, highlighting issues of choice, implementation and use. Various patient outcomes are discussed and the impact of electronic systems within oncology on these outcomes is highlighted.
The widespread implementations of computers within hospital and oncology departments means that all oncology departments will face the challenge of implementing modern electronic information systems. Indeed many departments already possess the software required.
As the changes implemented in one New Zealand centre demonstrate, the OIS is more than just a paper substitute. A clear advantage of the newer system is its capacity to re-use data and to quickly and repeatedly apply quality assurance to prospectively collected data.
Development of Clinical Information Systems in Oncology
Clinical information systems have been used in oncology for many years. In the past, however, disparate systems existed without integration. Almost all electronic development occurred initially within the confines of the Radiation Oncology Departments with staff that included computer literate radiation physicists and radiation oncologists (yes, there are some!), and where the special abilities of computers were easy to use to advantage. The fully integrated OIS is a recent development.
In radiation medicine, more than any other specialty, the design and implementation of electronic information systems have occurred because of the need to reduce and eliminate the element of human error in the delivery of radiation to cancer patients. The concept of risk management, as a process, has only been discovered recently by hospital administrations. However, risk management began within radiotherapy after the effects of the atomic bombs dropped on Nagasaki and Hiroshima in 1945 became evident. Although a series of systematic errors in radiotherapy have been reported (eg, the Manchester SSD incident, [ 1 ]Costa Rica time calibration and Spanish incorrect energy incident,[ 2 ] the Panama planning incident [ 3 ]), the small number of these is testament to radiation professionals’ enormous attention to detail in handling their materials and the extreme care they take to avoid inadvertent exposure.
Almost every electronic clinical system has been adopted because either the process of double- and triple-checking humans has shown them to be unreliable, or the process being checked has largely comprised repetitive calculations which are better done by computers.
In the first instance, the DICOM-RT protocol is used for the electronic transfer of radiation therapy planning (RTP) computer output into the linear accelerators’ "record and verify" (R&V) software, thereby removing the inaccuracy of human transcription. In the second instance, RTP computers now calculate dose deposition in a 3D heterogeneous CAT scan image set, whereas the earlier method of "hand planning" used a single 2D homogeneous slice. Hand planning used a calculator and tables to hand-draw isodose curves, while the most modern RTP computers are now employing the highly accurate Monte Carlo algorithms to determine dose deposition.
Information Systems in Oncology have progressed through three phases.
1. The Paper OIS
While the term "information system" is now tied to computer-based systems, it should be briefly acknowledged that there were information systems in the past. In the period up to the early 1970s, oncology systems were entirely paper-based. Radiotherapy was hand planned. Schedules were written in diaries, and radiotherapy treatment delivery was verified by signature alone. Complex statistical analysis of any sort required a very specialised knowledge, so outcome analysis was difficult.
2. The Hybrid and Disconnected OIS
After the 1970s, computers began to appear in Oncology Departments. Their first routine clinical uses were for RTP and machine R&V. Departments that wished to report prospective or retrospective data after statistical analysis employed stand-alone software. These statistical software suites were entirely separate, and often on completely separate networks or on stand-alone computers.
The disjointed nature of these solutions was the result of the then level of computing expertise and of expense.
To be effective, the OIS has to be available to all users requiring information in the OIS. Every staff member has a need for some portion of the data stored in an OIS and as a result the OIS Palmerston North has 50 user licences. In the period covering the 1970s and 1980s, such distributed access was too expensive for a single department to contemplate even if suitable software had existed.
| a) Machine R&V | |
| As linear accelerators (linacs) became increasingly dependent on printed circuit boards (PCBs) and modern electronics, linac manufacturers designed and implemented their own R&V systems. These software systems were peculiar to the linac and linked to no other software. As a result, all data entry into the R&V system was manual. One of the early attempts at an R&V system has the ignominious position of being used as a case study for how NOT to manage software: (see http://www.computingcases.org/case_materials/therac/therac_case_intro.html). These Therac-25 incidents of the early 1980s have been reported elsewhere.[ 4 ] | |
| b) Paper Clinical Record / Electronic Analogue Repository, and | |
| c) Paper / Electronic Schedule and Charging functions | |
| The early attempts at these two functionalities were often based on out-dated but functioning VAX mini-mainframe computers or were tied to large mainframe projects. These systems possessed a range of functions including providing clinical staff with the ability to look at text and occasionally images, as well as serving scheduling and charging functions. The databases were static, ie, the database structure could not be altered except by the designers of the software who needed to be engaged to, eg, add a new field. Many of the attempts to build systems arose from isolated collaborations of clinicians and personnel with computing expertise. When these two operations were incorporated into large mainframe solutions, the needs of the individual oncology department were often poorly met. | |
| d) Radiotherapy Planning Computer | |
| Computerised planning was preferred to manual planning for several reasons including: the reduction in error rates; the ability to provide much more accurate predictions of dose deposition; and the ability to permit clinicians to describe anatomical and disease parameters within a radiation plan. A planning system has a 5- to 10-year life span with a corresponding technological leap on replacement. The commercial planning computer purchased in 1989 by the Department of Radiotherapy in Palmerston North was a DEC PDP11/77 computer with an array processor to accelerate calculations and a separate graphics processing system. It used a variant of the VMS/RT11 operating system called STX. It did not allow fragmentation of files on disk, so a disk drive "repack" was required occasionally to optimise system performance. The planning software called "Theraplan" was written in FORTRAN. This computer was free-standing and its output was manually transferred to the machine R&V software [ a ]. This system only permitted 2D planning, that is, the dose calculated in each plane ignored dose contribution from the planes above and below. Over its lifetime, the ability to display dose superimposed over a CT image was developed, as well as the ability to use Hounsfield CT numbers for dose deposition calculations. The planning computer was replaced in 1999 by another commercial system. The replacement system was Unix based and able to transfer output electronically to the R&V software. The new system permits delineation of structures on a PC, visualisation of target geometry and dose deposition in 3D space, and more accurate 3D calculation of dose. The ability to "inverse plan" and calculate "dose-volume histograms" was included. Inverse plan consists of describing an intended dose distribution as a series of rules, following which the computational algorithm would deliver a solution that fit the parameters. Dose-volume histograms are a graphical representation of dose deposition in defined structures that aids in the interpretation of plan adequacy and side effect prediction. | |
| e) Computerised Outcomes Analysis | |
| Specialised software for the storage and formal analysis of patient outcomes was used in the generation of conventional literature. Analysis such as Kaplan-Meier survival plots, log-rank comparisons and multivariate analysis was usually DOS-based, expensive and non-intuitive. Substantial amounts of training were required, and the idea of doing a small project using this software was daunting to say the least. In situations where a clinical review was being undertaken, data were extracted from paper review notes, reports and letters. Thus, the process of data gathering for outcome reports was separate from the clinical process. Quality assurance of this outcome data was usually infrequent and when performed it revealed substantial inaccuracies. The human resources required for data collection, collation and entry were substantial, and the data collected did not come to reside within the clinical system if a subsequent overlapping study was undertaken. Even systems with electronic report and letter repositories typically followed the same process. This position still describes many radiotherapy departments today where the predominant data format is paper and electronic data capture is only for special circumstances. The move from an un-integrated OIS to an integrated OIS existence requires money and expertise. Unfortunately, both are in short supply in current health systems. The cost of progression to the next phase is considerable as a modern OIS would cost a minimum of $NZ 400,000, before adding in the options and user-licence configuration required for full implementation. In Palmerston North we experienced a credibility gap when asking to purchase more options after the initial purchase. At initial purchase, one option was not selected because the software did not work and its usefulness was not apparent. Once the software became operational and the need for the option was apparent, gaining approval proved to be more difficult than the original purchase. The subsequent implementation of a modern OIS is a major stumbling block because there is a dearth of the implementation expertise required. While many staff members have computing experience, implementation requires several abilities including: a position within the departmental power structures that can influence decision-making and lead an implementation; an intimate understanding of the processes being replaced; knowledge of the software’s capabilities usually only present after a long period of use; time to investigate and develop the software’s customisable features to improve efficiency and meet peculiar local requirements; a personality able to conceptualise the processes, and enthuse a group to achieve a common goal. Such expertise is not taught in any medical, nursing or radiation therapy course. | |
3. The Integrated Computerised OIS
After the early 1990s, the general availability of PCs and relational databases of intermediate size made tailor-made OISs a practical possibility. The reducing cost of PCs, storage and networks has allowed the proliferation of workstations that in turn enable multiple access points throughout an oncology department. The integration of wireless access will further enhance the introduction of this software as the PC and network costs will be reduced. In early 2003, the cost of a single Tablet PC for each mobile clinician was approximately $NZ3,500 at the time of writing, and the cost of a radio network covering the Palmerston North Radiotherapy Department and Oncology Ward was $NZ24,000.
The modern OIS seeks to manage the unique oncological processes involved in delivering radiotherapy and chemotherapy by providing software options that address all component processes within an oncology department. Where a necessary function is not provided by the OIS, formalised transfer protocols (eg, HL7, DICOM, DICOM-RT) are implemented to ensure that data transfer occurs electronically and predictably. The necessary functions are listed below.
i. A scheduling system able to book repeating appointments. Radiotherapy needs to be able to book between one and 39 consecutive daily treatments. Chemotherapy needs to be able to book between three and eight consecutive three-to -four weekly appointments.
ii. A billing system able to specify codes for any number and combination of professional actions. In New Zealand, radiotherapy is funded on the basis of patient attendance (ie, when a patient walks into the unit for a procedure), however, in Australia, eg, different sites and field numbers have different codes and remuneration. In the US, insurers will accept electronic downloads of charges. In addition, the billing system must be well integrated so that the process of charge capturing is as seamless as possible. In Palmerston North, when a machine has finished a treatment, the software automatically asks whether the charge should be captured, and a simply pressing "Enter" is sufficient to achieve this.
iii. A quality assurance/workflow logging system which enables notification and confirmation of workload through the personnel chain. When a patient has been seen by an oncologist and accepted for treatment, the OIS must store their treatment specification as well has notify the next set of professionals that their input is now required. In particular the radiation oncologist needs a system to inform the radiation therapist that a radiation planning appointment is required. The medical oncologist needs a system to inform the chemotherapy nurses that an appointment for chemotherapy is required. As an increasingly important efficiency, the OIS should support interchange of important appointment and task information with personal data assistants (PDAs).
iv. A clerical document handling system that includes notification of dictation, correction of typing, printing of documents and document archiving and display. Such a system should incorporate premium word processing software such as MS Word and should be enabled for database field merging, scanner input and voice recognition software.
v. A radiotherapy R&V that has been expanded to include the radiation prescription for the individual patient, monitoring of the linacs (the traditional R&V function), the automation of the linac (where multiple fields will be delivered in sequence without the requirement for radiation therapist intervention, analogous to a DOS batch file), the specification of target volumes in 3D image sets (CT scan, MRI scan) with oncological regions of interest (ROIs), the design of field geometry (field numbers, direction and shape) and interfacing with a RTP computer. Until recently, the RTP software contained the voluming and field geometry design tools, however, there are advantages to placing these components within the OIS.
vi. A chemotherapy R&V that includes the description of chemotherapy regimens, their prescription for individual patients and support for drug delivery, preparation and ordering in the oncology pharmacy.
vii. A system for assessing clinical data provided by manual entry or automated download by HL7 protocol from pathology laboratories. The clinical data can consist of such items as blood biochemistry and blood cell counts but might also include histopathology reports and imaging reports. Standardised gradings are legion in oncology. Oncologists will require the ability to enter assessments in a range of toxicities and performance measures. This part of the OIS, therefore, must be expandable and configurable to meet the needs of each particular department, and control over the software needs to be within the oncology department.
viii. A security system that permits different professional groups to maintain control over relevant parts of the system. The radiation oncologist is responsible for the radiation prescription, which must be complete before a patient can be treated. In Palmerston North, only radiation oncologists can approve radiation prescriptions and the linac will not deliver any radiation until the prescription is approved. However, the log-in for use of the linac can only be provided by radiation therapists, not radiation oncologists.
There are available systems that have consolidated the disparate systems into one platform. Currently, there are three major OISs available - Siemens’ LANTIS (which is similar to IMPAC’s Impac), Varian’s Varis and Nucletron’s OnCentra. While some of their functionalities overlap, they are not equivalent. The author has an intimate knowledge of LANTIS, which has been implemented in the Regional Cancer Treatment Service at Palmerston North.
The major issue for OISs in oncology is no longer to do with hardware and networks, or even operating systems. It is theoretically possible for the Windows-based Lantis at Palmerston North to be used on a PC running Linux across an ADSL VPN connection from the UK. In fact, Lantis at the Palmerston North site has been accessed from Australia, USA and Singapore on a Windows 2000 PII laptop across an ISP VPN connection.
Modern OISs have emerged and expanded from linac R&V software. Frequently, the OIS is changed when the linacs are replaced, with the inevitable consequence of having the choice of linac mandate the choice of OIS. Since the cost of the linac is larger ($NZ 3 million), this eventuality seems inevitable.
However, the functionality and features of modern linacs are converging and are very similar at this time, making the choice of hardware less relevant. Therefore the real issue is the degree to which the attached OIS improves the efficiency of the linac hardware, permits implementation of complex techniques, enhances workflow efficiency and reduces workload levels within the oncology department. An OIS that provides all of this improved functionality will improve the efficiency of the whole system with tangible savings.
The selection of linacs in Palmerston North for the period 2000-2010 has followed this logic. The complexity of treatments has been increased; however throughput on the modern linac is now consistently 25% greater than the target throughput of 160 fractions per week.
Patient Outcomes
The effectiveness and efficiency of an OIS is difficult to quantify. Not all staff are keen to consider the issue. When the efficiency benefits of technology are highlighted, there can be a backlash of opinion pointing to the necessary human elements of care.] 5 ] Administrators are very keen to receive the benefits, but can also be reticent to provide the project leadership and sufficient resources for adequate levels of software and licence purchase.
The outcomes that most medical professionals and the general public are interested in relate to the medical outcomes of morbidity and mortality. The questions - How sick and I going to get? What are my chances of treatment success? - are constants in the minds of doctors and patients. The implementation of computers in radiotherapy systems has improved outcomes by enabling doctors to minimise side effects, while maximising the probability of cure.
The most obvious example is the advent of 3D conformal radiotherapy (3DCRT) and intensity modulated radiation therapy (IMRT) - both of which have been successfully trialled at Palmerston North. The major improvement derived from 3DCRT is that a diagnostic image set (eg, CT, MRI) is used to specify the shape of the radiation portals. The typical 3DCRT field tends to be made of convex shapes. The superiority of IMRT rests in its ability to match concave shapes. Both techniques allow accurate definition of both cancer-affected tissues and the unaffected tissues that can be excluded from significant radiation doses. There is sufficient data to be confident that the side-effects for patients are reduced and cure rates are improved when the techniques are applied to appropriate cases.[ 6,7 ]
Apart from these obvious benefits in conventional measures, there are others that are less obvious until one appreciates the volume and type of routinely collected clinical data stored in the OIS. In Palmerston North, we record weekly rectal toxicity during radiotherapy according to the grading scale of the Radiation Therapy Oncology Group (RTOG). Similar assessments are undertaken on paper by almost every radiotherapy department each week.
As a proof of concept, the data for acute rectal side effects for early stage prostate cancer over a 4-month period was extracted from the database. The Kaplan-Meier survival curve showing the loss of Grade 0 toxicity is shown below (Figure 1). The log-rank comparison of the two curves reveals that the difference is highly significant (p<0.0001). The two curves represent patients of two of the oncologists.
Such comparative data is likely to cause a high degree of anxiety in medical professionals, as individual audits have not previously been presented with this comparative aspect. The difference in side effects is not simple, as there is no comparison of cure rates, prostate volumes, methods of marking up CT scan, or a host of other factors that could lead to a difference in side effects.
However, IF all possible confounding factors were accounted for AND the difference persisted, the public would expect reasonable professionals, whose techniques were resulting in more significant side effects, to undertake further training to reduce side effects without loss of cure. In this way, the OIS can lead to small but real improvements in treatment delivery by enabling concerned departments to objectively identify superior techniques.
Two further improvements derive from the implementation of a modern OIS. The modern OIS is a culmination of developments in modern software and process engineering. Medical systems may be somewhat idiosyncratic but, nevertheless, they have similarities to most other systems. While the transition to an OIS will involve profound changes in a department’s organisation, the ability to customise a modern OIS to specific requirements means it is difficult to conceive of circumstances where a department’s needs could not be met by the OIS.
Properly implemented, the modern OIS will have two immediate benefits. Patient throughflow will be improved and staff workload will be reduced.
Processing patients using an electronic OIS is more efficient as the various tasks required in the treatment process can be notified simultaneously to all concerned. The notification process can be reviewed and quality assured to establish whether resource use is efficient. Certainly, in our experience, the process of redesigning existing systems to match the new OIS has allowed us to‚ "clean" the procedures accumulated over the last two decades. For example, by utilising the abilities of the OIS we have been able to notify booking staff of the need for appointments before a patient has left the consultation room at a peripheral clinic in Gisborne 420 km from Palmerston North, so that patients can be notified of appointments within 10 minutes. The OIS permits a booking clerk to undertake all bookings for a patient’s care at one time, thereby resolving many of the co-ordination issues inherent in paper systems.
The time that medical staff members spend on the recording of routine data has reduced. The data entry pattern is structured so that data are recorded as they become available; it is also available at all times later and to all staff with access. Previous paper-based systems had oncologists recording the same data two or three times, with the data only available to those who actually had the paper record. At the end of a patient’s treatment in Palmerston North, all their relevant details have been recorded in the OIS so that a standardised report has been written to generate a discharge summary. This saves time as no input is required from the oncologist.
The OIS will lighten clerical workloads. For example, the time spent retrieving files stored at another location is regained. Clerical staff members become responsible for entry of data into the OIS rather than filing and collating paper.
Finally the time spent by all staff in inefficient systems is time not available for caring for patients. When the implementation of new systems improves efficiency and reduces workloads, the best use of this extra time is worthy of some discussion. OISs should free staff to spend as much time as possible serving the needs of their cancer patients.[ 8 [As a system architect, my desire is to have a system that smoothes patient flow and frees staff to spend time in their caring role. The caring role cannot be placed above the system, or vice versa. Both need to be operating at maximum efficiency if we are to provide our cancer patients with our best care.
Terms
| OIS Oncology Information System, an integrated suite of software tools to support the operations of an oncology department | |
| Radiation Oncologist A cancer doctor who uses radiotherapy as the treatment modality | |
| Medical Oncologist A cancer doctor who uses chemotherapy as the treatment modality | |
| Radiation Physicist A physicist responsible for the maintenance, safe use and calibration of radiation machines | |
| R&V "Record & Verify"; this software measures hardware intercepts to verify that the field settings are as intended, and then records the actual treatment delivered | |
| Linear accelerator A medical device for generating high energy x-rays that are used in the treatment of cancer patients | |
| RTP Radiotherapy Planning computer which uses software algorithms to emulate the radiation deposition pattern of a defined set of radiation beams and parameters | |
| DICOM A protocol for the electronic storage and transfer of medical images (CT, MRI, nuclear medicine, PET) | |
| DICOM-RT A protocol for the electronic storage and transfer of radiotherapy machine settings | |
| RTOG Radiation Therapy Oncology Group; a large collaborative group of radiation oncologists predominantly in the USA | |
| EORTC European Organisation for Research and Treatment of Cancer; a large collaborative group of radiation oncologists predominantly in Europe | |
| LENT/SOMA Late Effects in Normal Tissues/Subjective, Objective, Management, Analytical - a toxicity scale devised by RTOG and EORTC |
Footnotes
a. K Croft, Chief Physicist, personal communication, 2003
References
- Ash D, Bates T. Report on the clinical effects of inadvertent radiation underdosage in 1045 patients. Clin Oncol 1994; 6(4):214-226.
- ICRP Recommendation no 86. Prevention of accidents in radiotherapy. Oxford: Pergamon Press; 2001.
- Vatnisky S, Ortiz-Lopez P, Izewska J, Meghzifene A, Levin V. The radiation overexposure of radiotherapy patients in Panama. Radiother Oncol 2001; 60:237-238
- Leveson N, Turner CS. An investigation of the Therac-25 accidents. IEEE Computer 1993; 25(7):18-41.
- Smith D. Shadows 2003 Mar 6;46(1).
- Fenwick JD, Khoo VS, Nahum AE, Sanchez-Nieto B, Dearnaley DP. Correlations between dose-surface histograms and the incidence of long-term rectal bleeding following conformal or conventional radiotherapy treatment of prostate cancer. Int J Radiat Oncol Biol Phys 2001 Feb 1; 49(2):473-80.
- Lee N, Xia P, Fischbein NJ, Akazawa P, Akazawa C, Quivey JM. Intensity-modulated radiation therapy for head-and-neck cancer: the UCSF experience focusing on target volume delineation. Int J Radiat Oncol Biol Phys 2003 Sep 1; 57(1):49-60.
- Menke JA, Broner CW, Campbell DY, McKissick MY, Edwards-Beckett JA. Computerized clinical documentation system in the pediatric intensive care unit. BMC Medical Informatics and Decision Making 2001; 1:3.
