Training in Pathology Informatics: Implementation at the University of Pittsburgh Open Access

Pathology informatics is an emerging subspecialty of pathology practice that is concerned with the application of data processing and communications technology to problems in pathology research, diagnosis, and information management. The field encompasses development, deployment, management, and evaluation of software and hardware supporting pathology and clinical practice as well as a number of additional activities that are necessary for the implementation of effective information systems. These latter activities range from the definition of standard terminologies, data elements, and communications protocols used in automated information processing to training in the use and maintenance of information systems and management of vendor and hospital/department administration relationships.

The practice of anatomic pathology and laboratory medicine has become closely associated with electronic computing and communications systems. The primary work product of pathology, diagnostic information, is stored and transmitted to clinicians primarily through electronic systems. As these systems begin to support direct interaction with physicians, their design is influencing pathology workflow, clinical workflow, and clinical perception and interpretation of pathology data. Furthermore, new types of data analysis and decision support based on data mining or bioinformatics techniques can add substantial value to standard pathology data.2–4 Because these systems profoundly affect the details of pathology practice and the effectiveness of patient data communication, it is crucial that pathologists participate in their selection, implementation, and management. To contribute effectively to this task, pathologists need to understand basic facts about information system architecture and capabilities, basic information system terminology, and important areas of health-related information system development. We believe that the best way to address these needs is to provide all pathologists with introductory training across a defined range of topics in pathology informatics as a standard component of the pathology residency.

Recently, the Training and Education Committee of the Association for Pathology Informatics (API TEC; available at: http://www.pathologyinformatics.org/) developed curriculum guidelines for residency training in pathology informatics (see Henricks et al5). These guidelines identify the scope of information that might appropriately be included in introductory pathology informatics training. We have developed at the University of Pittsburgh (Pittsburgh, Pa) a 3-week intensive rotation in pathology informatics that implements a subset of these guidelines. This manuscript describes our general approach and resident performance during 3 years of experience.

The important relationship and potential synergy between pathology practice and informatics was recognized and discussed more than 15 years ago. In 1987, Korpman6 proposed the pathologist as an ideal individual to fill the emerging role of the “medical information specialist” and pointed out the need for appropriate training within the pathology residency. Friedman7 subsequently defined the term pathology informatics and proposed the development of a separate division within the pathology department to support this specialized activity. In 1993, Buffone and Beck8 recommended the creation of a subspecialty of pathology informatics encompassing all of the activities required to optimize pathology information systems for pathologist and clinician use.

The need for informatics training in pathology residencies was generally recognized more than a decade ago. The report of the ASCP (American Society of Clinical Pathology) Colorado Spring Conference9 in 1990 identified informatics as a necessary component of training in both anatomic and clinical pathology, and the report of the subsequent Graylyn Conference10 on clinical pathology training labeled informatics a “critically important component” of practice. Gorstein and Weinstein11 recently reiterated the importance of informatics in general pathology training and practice as well as in trainee recruitment. Haber et al12 described a management informatics residency rotation that included basic desktop computing skills, spreadsheet analysis, and entering/extracting data from the laboratory information system (LIS). Peters and Clark1 proposed a general pathology informatics curriculum that would include sections on computer hardware and operating systems, communications and interfaces, word processing, spreadsheets, statistics, graphics, databases, programming, and vendor interactions. As part of their subspecialty proposal, Buffone and Beck8 recommended a basic informatics curriculum in residency training that included general computer literacy and productivity tools (eg, word processing, spreadsheets, graphics), database and statistical analysis tools, and anatomic pathology LIS selection and management.

Informatics training practices in US and Canadian pathology residencies were surveyed 3 times in the past decade. In 1993, Balis et al13 found that 84% of 142 programs offered at least some informatics education. Informatics training was required in 49% of all programs, and 26% offered a dedicated informatics rotation. Four years later, Goldberg-Kahn and Healy14 carried out a similar survey and found that 90% of 84 programs offered informatics training, which was required in 68% of all programs. Dedicated rotations were offered in 24% of all programs. A more recent survey (2002) by Henricks and Healy15 found similar numbers: 93% of 72 programs offered informatics training, and 72% of the programs required it. Of programs offering informatics training, a dedicated rotation was available in 19%, and the remainder incorporated informatics training into other rotations. Nineteen percent of the programs formally evaluated informatics competency. Virtually all of the programs supply computers for resident use (an average of 6.8 resident computers per program), and 13% provide residents with their own computers. A variety of teaching strategies were found in all surveys, with hands-on, problem-oriented training incorporated into non-pathology informatics rotations as the most common approach. These authors14,15 observed that, while informatics was perceived as an important area for training, the definition of pathology informatics varied widely between training programs and that neither training goals nor competency objectives were well-defined.

If the pathologist of the future is to meet the goals of Peters and Clark1 of effectively communicating with and guiding information technology professionals, participating in meaningful ways in system selection and evaluation, and targeting appropriate problems for information technology solutions, all residency programs must provide pathology informatics training that is appropriate in scope and detail, reasonably standardized, and effectively evaluated. The API TEC has now articulated a set of training and competency goals for pathology informatics in residency.5 The University of Pittsburgh Pathology Informatics Training Program implements a substantial portion of the API TEC goals and provides a practical model of a successful approach to pathology informatics training in a pathology residency.

Required training in informatics in the University of Pittsburgh Pathology residency is carried out in 2 components. New first-year residents are introduced to the University of Pittsburgh computing environment and the anatomic pathology information system during their arrival orientation. Residents are provided shared computer workstations in all work areas (in the future, they will receive individual laptop computers). They become familiar with standard productivity applications, e-mail, Web-based Internet/intranet applications, basic presentation graphics, and patient data access through practical experience during a first year consisting primarily of anatomic pathology training. At the beginning of the second year, which consists primarily of laboratory medicine training, residents participate in a required, 3-week, full-time pathology informatics core rotation. This time is made available in the rotation schedule by having upper-level residents continue their June rotation through these 3 weeks in July. Depending on the numbers of residents between the years and scheduling requirements, we occasionally include a first-year resident in the core rotation or shift a second-year resident's core rotation time to the third year. After residents complete the required core rotation, they may carry out elective, project-oriented rotations in pathology informatics, or they may include informatics as a collaborative component in elective projects carried out in other areas of pathology.

The goal of the pathology informatics core rotation at the University of Pittsburgh is to ensure a minimum level of competency in 5 areas: (1) operation of desktop computers and standard productivity software, (2) general computer-related terminology, (3) standards and techniques important in health care and pathology data processing and research, (4) health care enterprise information system design, and (5) general structure and typical deployment strategies of anatomic pathology and clinical pathology LIS systems, including their relationship to other clinical information systems. We also address developmental areas in pathology informatics that have the potential to affect future pathology practice. By the end of the first year (the beginning of the core rotation), most residents understand standard word processing and spreadsheet techniques as well as the basics of e-mail communication and Web browsing. In these areas, our curriculum attempts to augment preexisting knowledge by focusing on, for example, the use of imaging, multiple software packages, and file sharing to carry out tasks such as presentation development. Teaching in other, less familiar areas such as enterprise system architecture and health care data standards provides an opportunity for the overview and discussion of key terminology.

The curriculum of the core rotation is shown in Table 1, and representative articles from the reading list are shown in Table 2. The curriculum includes about 84% of the 179 key concepts in the API TEC knowledge objectives list.5 The total student contact time in the 3-week curriculum is 63 hours, including 35 hours of lecture, 20 hours of laboratory sessions/demonstrations, 4 hours of field trips, and 2 hours each for pretests and posttests. Lectures are generally presented in the mornings, with laboratory sessions and field trips in the afternoons. Paper handouts of the lecture slides are provided for direct note-taking, and hands-on laboratory sessions are given in a computer lab within the medical library. In 2002, a total of 19 instructors contributed to the rotation, including 7 pathology informatics faculty members (1 faculty member directs the course and contributes 26 contact hours), 5 pathology informatics staff members, 1 non-pathology informatics faculty member, 1 Pittsburgh Supercomputing Center research faculty member, 2 upper-level pathology residents, 1 librarian, 1 hospital system support employee, and 1 hospital administrator.

Pretests and posttests are given on the first and last days of the rotation, respectively. These exams include 30 short-answer questions with a 2-hour time limit. The questions are intended to be answered with a phrase or a few sentences, a list, or a diagram. Partially correct answers receive half credit. The pretest and posttest questions are similar in complexity and cover a range of issues from desktop computing problems to system architecture, data formats and coding, general terminology, database management, short pathology informatics practice scenarios, and other issues. Several representative questions are shown in Table 3. (Editor's Note: the full list of questions used during the past 3 years, with answers, has been submitted to the Archives for possible publication.) A grade of 50% or more on the posttest is required to successfully complete the rotation. Residents also rank the individual presentations and instructors on a 1 to 5 (5 = best) scale and can submit comments and suggestions on an evaluation form that is completed immediately after the posttest.

Twenty-one residents took the rotation from 2000 to 2002 (6 in 2000, 8 in 2001, and 7 in 2002). In addition, a visiting pathologist and a medical informatics fellow participated in the course in 2001, and a second medical informatics fellow participated in 2002 (their scores are excluded from this analysis). The average pretest score during the 3 years was 20.0%, and the average posttest score was 69.7%, for an average increase in score of about 50%. A relatively high pretest average seen in 2000 (Table 4) was the result of a very knowledgeable resident in the rotation that year. The average self-scored computer skill level (1–5 scale, novice to expert) during the 3 years was 2.1. Table 4 lists pretest scores, posttest scores, pretest/posttest increases, and skill levels for each of the years 2000–2002.

Pretest scores were relatively low, and they correlated with self-scored computer skill level (r = 0.80, Table 5). In contrast, posttest scores did not correlate with skill levels (r = 0.06), and there was a negative correlation between the skill level and the increase in scores between the pretest and posttest (r = −0.64). Similarly, the correlation between pretest and posttest scores was relatively low (r = 0.19, Table 5), and there was a substantially higher negative correlation between the pretest scores and the pretest/posttest increase (−0.77). In aggregate, these data suggest that residents with prior computer knowledge did not have a particular advantage in the course and that residents who were initially less computer knowledgeable tended to have a larger score increase and thus “catch up” to their initially more knowledgeable peers during the rotation (Table 5).

In detailed evaluations of the individual instructors' presentations, which were collected anonymously during the 3 years, residents gave the course an average rating of about 3.7 on a 1 to 5 scale (5 = high; see Table 4 for individual-year scores). The average rating per presentation for all 3 years ranged from 2.0 to 4.5. We did not formally assess differences between presentations, though our impression is that ratings reflected the degree of preparation and communication effectiveness of the instructor rather than the intrinsic complexity or technical nature of the topic. Thus, residents did not appear to be averse to technical informatics topics if they were well presented. Of the residents in this study, 5 (about 24%) have carried out or are planning independent projects in informatics, and 2 (to date) plan further informatics work after completion of their residency.

The interest in pathology informatics training that developed during the 1990s coincided with a general increase in attention to informatics training in medicine. This focus of attention ultimately culminated in the report of the Informatics Panel of the Medical School Objectives Project by the Association of American Medical Colleges (AAMC MSOP).16 This report confirmed the importance of medical informatics training in general medical education and identified informatics learning objectives associated with broad areas of physician practice comprising lifelong learning, clinical practice, teaching/communication, research, and management. The report also recommended that formal informatics training be implemented in all medical training programs in a manner that ultimately embedded informatics teaching in existing medical courses throughout all years of training.

Despite the general interest, there is little consensus on the appropriate content or form for informatics training and minimal objective evaluation of existing programs. One study surveyed 80 pediatric residency programs and found that 37% provided formal lectures on informatics and that 68% planned an expansion of informatics training.17 Residents commonly used computers for accessing patient results, e-mail communications, and word processing; relatively few (less than 10%) used more advanced techniques such as decision analysis. A meta-analysis18 of 154 published reports relevant to informatics activities during family medicine or medicine residency found few descriptions of the curriculum, schedule, goals, or outcomes of existing informatics training programs. Program descriptions that do exist show substantial variability in teaching style, scope, and emphasis. For example, family medicine residents received basic computer training in two 3-hour sessions plus independent assignments in the first year; then, a project and workshops were offered in the second year.19 A pediatrics program developed a self-study informatics module, organized around core tasks, that is offered for 2 to 3 weeks in combination with a noninformatics subspecialty rotation.17 A program in medicine offered a 1-week course composed of ½-day sessions covering local network services, Internet communications and the Web, electronic presentation, and evidence-based medicine.20 A 1-month elective rotation in medical informatics for residents and medical students offered directed project work combined with three 2-hour lecture/discussions per week that included such topics as human factors engineering, human error, outcomes management, home care and rural outreach, patient decision making, system evaluation, usability, data integrity, and continuous quality improvement.21 A survey22 conducted in 1998 by the AAMC found that 75% of all medical schools taught informatics in required undergraduate medical courses and that an additional 20% offered informatics training in elective courses. However, the nature and goals of these courses varied widely, and less than 10% of schools taught more advanced medical informatics topics such as medical decision support and effective use of computers in clinical care.

Overall, these reports yield results that are qualitatively similar to the surveys of informatics training in pathology residency programs discussed previously,13–15 though pathology residency programs appear to be more likely to offer informatics training than do other types of residency programs for which data are available. Unfortunately, there is little objective information to use in comparing existing programs with respect to training outcomes or even resident or faculty perceptions of the value and appropriateness of training. Without such data, a critical evaluation of the success of various approaches to training is difficult.

It is important to recognize that the demands placed on pathologists in the area of informatics differ fundamentally from those placed on physicians in most other specialties. The AAMC MSOP learning objectives16 and informatics training in most residencies appropriately focus on developing skills in using technology tools to accomplish the medical goals of individual practitioners. In contrast, pathology informatics focuses on the management of pathology data and data systems,5 the evaluation of technology, and the development or selection of informatics tools for general use both within and outside the field of pathology.8 The responsibility for providing both information and appropriate tools to work with that information distinguishes informatics in pathology from that in most other specialties. Thus, pathology informatics requires a different scope of training that includes, for example, an introduction to LIS and enterprise system architectures and vendor relationship management as well as a broad exposure to medical informatics topics and information technology vocabulary. Reviews of existing pathology informatics training programs13–15 do not clearly indicate that this distinction is generally appreciated in the pathology community, but it is reflected in the concept list proposed by the API TEC.5 

For these reasons, our core rotation includes a significant emphasis on the LIS and enterprise systems, system interfaces, multiuser databases, and information representation and coding in addition to personal productivity and desktop computing (Tables 1 and 2). Although we support and encourage elective project-oriented informatics training after the required core rotation, we feel that it would be difficult to provide individual residents with adequately broad exposure to the field of pathology informatics using that instructional approach alone. Our strategy of progressive lecture/discussions and readings combined with hands-on laboratory experiences allows us to build understanding of complex topics during a series of presentations and to cross-reference related topics while these topics are still fresh in the trainees' minds.

Residents typically scored low (less than 20%, Table 4) on the pretests given at the start of our core rotation, but within this low range, performance correlated with the self-scored computer skill level (r = 0.80, Table 5). The pretests cover a range of topics including desktop computing and enterprise information systems (essentially the topics outlined by our curriculum, Table 1). Residents who are experienced desktop computer users or computer enthusiasts scored well on the desktop computing components of the pretest but not on the topics related to enterprise systems and data management, which are important in pathology informatics practice. Furthermore, residents score low on the pretest after 1 year of experience in our program with the e-mail and Internet software, patient data management systems, and the anatomic pathology LIS. Our results highlight the fact that user experience with standard desktop computing software and an LIS does not equate with informatics expertise. Task-oriented training with desktop computers and medical information systems in a setting that does not include a broader presentation of medical informatics and enterprise computing topics may lead to rote learning that is poorly generalizable and thus not well suited to the needs of pathologists.

Our residents show substantial increases in performance (averaging 50%) on the posttest compared with the pretest (Table 4). The posttest scores do not correlate with the pretest scores or the self-reported computer skill levels (Table 5), indicating that we successfully teach a range of informatics topics irrespective of a resident's prior knowledge of informatics. The amount of increase between the pretests and posttests negatively correlates with the pretest score and the self-reported computer skill level (Table 4), further suggesting that we effectively teach both desktop and enterprise information system topics to residents with lower prior computer knowledge and that we narrow the knowledge gap between residents with and without previous computing experience.

We do not currently test informatics knowledge of residents subsequent to our core rotation posttest, and thus, we do not know the degree to which this knowledge is retained. It is likely that opportunities to work with informatics in later training would aid in consolidating and reinforcing knowledge. Project-oriented informatics training, which allows additional in-depth learning in a particular area, is available as an elective rotation, and residents regularly participate in such projects. However, this is not a general solution for all trainees. We feel that elective rotations and projects carried out in non-pathology informatics subspecialties should routinely have an identified informatics component (these projects generally do have an informatics component that is not identified as such) that includes collaboration with informatics faculty or staff. These project components would allow the reinforcement of informatics principles later in training and a clear demonstration of the importance of informatics in pathology practice. Our program needs to establish the internal relationships that would allow these collaborations.

Our core rotation includes about 84% of the 179 specific concepts listed in the API TEC knowledge objectives,4 with coverage in all major categories. Our core rotation in combination with the first year of residency also includes 94% of the 52 proficiency objectives suggested by the API TEC.5 Concepts and proficiencies not covered include some that are not applicable to our environment, some for which we use alternative nomenclature or concepts, and some that we will review for inclusion in the future. A number of the more routine proficiencies related to anatomic pathology and laboratory medicine LIS systems are carried out during noninformatics rotations in the first and second years of residency training. These proficiencies are currently evaluated on a relatively informal basis. It would probably be beneficial to formalize these activities as a component of informatics training by developing a checklist of proficiency objectives that should be completed and signed off before an established deadline in training.

Our training program provides a model for implementation of the API TEC knowledge and proficiency objectives in pathology informatics. Although other approaches may also be effective, our combination of pathology training in an informatics-rich practice environment with a progressive, full-time didactic rotation for 3 weeks offers a well-rounded program with pedagogic advantages and demonstrated success. Pathology informatics has distinct requirements that are connoted in the quotation from Peters and Clark at the start of this article. To extend their assertion, future pathologists will communicate with, and critically evaluate responses from, information technology workers, administrators, and clinicians; they will understand inherent strengths and weaknesses in information technology tools and use this knowledge to guide planning; and they will contribute to decision making in the purchase, development, and management of these tools. Because modern information systems will bind pathologists and clinicians in new, shared workflows, the informed evaluation and management of these systems by pathologists will be crucial both to patient care and to the progression of the field of pathology. Our program is designed to produce pathologists who will be able to carry out this work.

We thank Cindy Gadd, PhD, for manuscript review and helpful comments. This study was supported by grant 5 T15 LM/DE07059 from the National Library of Medicine to the Center for Biomedical Informatics, University of Pittsburgh, Pittsburgh, Pa (Dr Stewart).