Development of an optically scanned Consultation-Liaison data base

Development of an Optically Scanned Consultation-Liaison Data Base F. Patrick McKegney, M.D., Charles E. Schwartz, M.D., Mary Alice O’Dowd, M.D., Itamar Salamon, M.D., and Robert Kennedy, M.A.

Abstract: Every clinical service must record certain data about its patient care activities. In low-volume services, such as Psychiatric lnpatient Services, with perhaps 20-30 admissions per month, and an average patient stay of 2-4 weeks, a large amount of data on each patient can be obtained during the contact time with the patient. On the other hand, very large-volume psychiatric services, such as Emergency and Consultation-Liaison (C-L) Services, may not need or be able to gather such a large amount of data on every patient seen. This article describes the development of a brief, optically scannable, and computerized minimal data base form for patients seen by a very large division of C-L Psychiatry. The system is feasible and easily auditable for completeness and reliability. This data base has already served many important functions beyond providing an administrative statistical summary of services rendered. It is presented as a model for the development of similarly efficient data collection methods for other highvolume psychiatric services.

Background The Division of Consultation-Liaison (C-L) Psychiatry of the Albert Einstein College of Medicine/ Montefiore Medical Center consists of psychiatric consultation services in four separate hospitals: Montefiore Medical Center, Bronx Municipal Hospital, North Central Bronx Municipal Hospital, and the Weiler Hospital of the Albert Einstein College of Medicine. Each service needs, of course, to provide the usual annual report of patient care statistics. In addition, after the integration of these four From the Department of Psychiatry, Albert Einstein College of MedicineIMontefiore Medical Center, Bronx, New York. Address reprint requests to: F. Pa&k McKegney, M.D., Department of Psychiatry, Montefiore Medical Center, 111 East 210th Street, Bronx, NY 10467. General Hospital Psychtity 12, 71-76, 1990 0 1990 Elsevier Science PublishingCo., Inc. 655AvenueoftheAmericas,NewYork,NYlOOlO

C-L services within the division in 1983, it was apparent that an unusual opportunity existed for examining a large volume of patient care data from 4,000 consultations per year in different clinical settings by using the same data base. However, this large number of consultations required limiting the amount of data that would be gathered, and developing a time and cost efficient method of data collection and storage. Other factors determining data system design were the large number of residents, fellows, and attendings who would be contributing to this data base, and the goal that all completed data bases should be audited by attendings for completeness and reliability. Together with characteristically overworked secretarial staffs and no extra funding, these constraints led to development decisions being made according to the principle of simplicity for maximal feasibility, reliability, and auditability. The content of this data base, called the Patient Consultation Record (PCR) was drawn from instruments used by the authors and from published reports of consultation service activities, recently summarized by Hengeveld et al. [l]. Because of the above constraints, the data base was not meant to be a research instrument, i.e., to test hypotheses that might be developed in the future. It was envisioned that such hypotheses would warrant additional data bases, to be collected in conjunction with the PCR data base. Thus, the PCR data form would need to be short and convenient to use, in order to maximize completion rate. The content of the data items needed to be clearly defined, in order to maximize relia71 ISSN 0163~8343/90/$3.50

F. P. McKegney

et al.

Table 1. Initial cost of PRC system (1986) Scanner NCS-3000 Maintenance contract Batch feeder NCS Form development NCS program to scan data Printing of 4,000 forms @ $125/1,000

$3,500 400 1,000 600 100 500

Total

$6,100

Subsequent

redesigned

revisions

$500 each

bility. The large volume of patient consultations suggested computerization of data processing, and perhaps even data entry. In 1985, optical scanning had reached a sufficient level of technologic development to be economically feasible at a local, institutional level. Recording information by filling in “bubbles” (open circles on a printed page) with a number 2 pencil was a process familiar to residents and fellows who had often been tested by multiple-choice examinations using computer scanned answer sheets. The use of such an optically scanned data form would eliminate the need for secretaries or other persons to type data into the computer. Finally, once computerized, the data collected by such a scannable form could then be very rapidly manipulated for any purpose.

Development The first scannable PCR form was developed in the spring and summer of 1986 with the assistance of the National Computer Systems Inc. (NCS) staff, which wrote the computer program for scanning and entering the data recorded on the paper PCR forms. The approximate initial costs of hardware, software, design consultation and printed forms in 1986 are shown in Table 1. Subsequent costs involve only the number of forms used. The original version of the PCR was used for 1 year. Improvements needed in its layout and content were apparent immediately, and revised versions were developed in 1987 and again in 1989. Figures 1 and 2 show the two sides of the current (Revision II) version of the PCR. Since the NCS system scans both sides of a form at a single “pass,” the revised form uses both sides of one sheet, rather than the original PCR’s two single-sided sheets. Each revision cost approximately $500 for

72

design work by NCS. Computer programs for scanning the forms were originally written by NCS. The most recent Revision II will be scanned by a program written by the authors, using Scantools, a program produced by NCS. Space will not permit discussion of all the factors determining all of the content areas. However, e.g., Axis I lists only major diagnostic categories and not all subcategories of all diagnoses. On C-L Services, the limited time of contact with most patients makes precise subcategory diagnosis problematic. An advantage of this recording system is that the residents do not have to look up DSM-III-R code numbers but instead can designate the diagnoses directly. Inpatient and outpatient psychiatric services might opt for a more detailed coding scheme. Another interesting aspect of the current form is the matrix on page 2 for coding a variety of drug recommendations by the consultant. Other psychiatric services might want the specific drugs and doses recorded as well. Another feature of Revision II of the PCR is the provision of “utility fields” A-F on the lower-righthand portion of page 2. By using a compressed binary coding system, each column of 4 bubbles (for the numbers 1, 2, 4, and 8) can be coded for up to 15 mutually exclusive numbers or items. Each block of three columns can be used for test scores or other numbers up to 999. Thus, this PCR can be put to new uses without requiring another revision.

Procedures A 3-page, single-spaced typewritten set of guidelines for using the PCR was developed, which is available from authors upon request. These guidelines, plus descriptors of the three scales on the PCR (DSM-III-R Axes IV and V, and the Kamofsky scale), are reviewed with all C-L attendings, fellows, and residents. Figure 3 illustrates the process of using the PCR. The initial entries on the PCR are made by the person receiving the consultation, usually a secretary via telephone. When the PCR is given to the consultant assigned the patient, a 3 x 5 card with the patient’s name is placed on a large bulletin board under the consultants name. This 3 x 5 card serves as a visual reminder that the PCR has not been returned. When the consultation has been completed, the PCR is returned to the secretary, and the 3 x 5 card is removed from the board and

Computerized Psychiatric C-L Data Base

0Ml.I

OMS

ONUR

OSW

OOTH

-Renotw - Quoatlotts for

_

tonsullstlon:

W=lUWl - msdlal ststus) I

-

AlcohoM)ng~-

hxletyn-

Blzarro Bohavkwnian-shlsbnm conlualorl

0

m-

:

~spodtl~ NW3

l+Jm@)

DNR Evaluatbn FarnHy/Sodal Pm&ma

:

lmpah~Judonmtc~arw) ManagemmtProbbn Paln Patht relualng pmcduruVoatmwt PatbntFiequost Psydlhtrk dlsudw by history F%ychogu-lr of wmatk nhwu PsyctKbtJoPkmdcauon advke ReacUonb~Rhmss(mabdaph) RoutlndGeneral

: : $ : :

slonlna~~ I I

stanhbkma

I sukMalBobvlom 0

-ONlJfSS -0Povchokora I 0 AltwxXngMD -OFdbw -OSodalWorku

-OW@@kd

Figure 1. Side one of patient consultation

record.

73

F. P. McKegney et al.

‘@3/A

gT* ‘ONi 0Pallth&oh 0-

0cj--’

MAKE NO STRAY MARKS IN THIS AREA

Figure 2. Side two of patient consultation record.

74

Computerized Psychiatric C-L DataBase

3X5 Card to Board -

Consultation called in PCR initiated by secretary 1 PCR to resident who begins consultation

3x5 Card ( Filed

I When consultation finished Completed PCR returned to secretary 1 PCR reviewed by attending (audit for completeness, reliability) I t

PCR's Filed

=

PCR's batch scanned

1. IBM XT 2. NCS 3000 Scanner 3. NCS Scan Program

t

Scan output to floppy (ASCII file)

Floppy backup

as + file

1. Custom

Program

ASCII

file to dBase File 1. dBase Programs 2. SYSTAT Programs 3. Other Programs I dBASE Analysis

OUTPUTS

filed. This completed PCR is then auditedby an attending psychiatrist, who returnsitto the consultant if there are data that are obviously missing or inconsistent. At intervals of 3-4 weeks, a batch of perhaps 100 or 150 PCRs are then optically scanned and the data recorded on the hard disk of an IBM compatible computer. Using a stack feeder, the scanner has the capability of scanning one form (both sides) in 7 seconds, a rate of approximately 300 forms per hour. This means that data from all the patients seen in the AECOM/MMC C-L Division in 1 month can be scanned in a little more than 1 hour. The data from the scanning are copied onto a floppy disk (ASCII format) and are then transferred to dBase IV files, using programs developed by one of the authors (I.S.). The floppy disk containing the scan data serves as a backup for the dBase IV file. The data can then be managed using any program written for dBase IV.

Figure 3. Patient consultation

record system.

Outcomes To date, the PCR data base has proved far more useful than originally conceived. The December dread of annual report data generation is now a thing of the past. Compiling data from 1 year’s consultations (1,725 at Montefiore in 1988) took 15 minutes, including the report printing by computer. In addition to providing very accurate annual reports for all services, more frequent reports can also be generated, such as monthly reviews of service loads. Even more important uses of the computerized data have emerged. The recent Residency Review Committee requirement that each resident must maintain a log of certain characteristics of patients seen can now be met by using the computerized data base. At the end of their 3-month, full-time

7s

F. P. McKegney et al.

rotation on a consultation service, each resident is given a computer-generated summary of the major characteristics of all of the patients he or she has seen during the 3 months. The time saved for each resident and the reliability of these logs are obvious advantages. The provision of these reports also enhances the residents’ investment in the accurate completion of the PCRs for their patients. The PCR data have been surprisingly useful in other areas as well. For example, questions were raised by the Nursing Administration of Montefiore about the incidence of constant observation orders by the C-L Service. The computerized data base immediately yielded all necessary information about all patients put on constant observation, which was compared to data for patients who were not put on constant observation. These data became a baseline for comparison with data gathered after the criteria for ordering constant observation were tightened. Constant observation became a quality assurance monitor of the C-L Service, and the periodic reports generated by the computer have demonstrated a steady decline in the number of patients put on constant observation since monitoring began. This exemplifies how the timeconsuming, extra effort often required by a quality assurance program is reduced by the use of PCR data as QA monitors. Other questions regarding patient care have also been equally rapidly answered. Published reports about the high incidence of suicidal behavior in AIDS patients suggested an examination of data pertaining to suicide gathered with the PCR. Sixtyseven AIDS patients under 55 seen in consultation on the medical service over 1 year were compared with 121 non-AIDS patients seen on the medical service in the same year and with the same age range, a process that took 10 minutes. The incidence of suicidal behaviors was found to be almost exactly the same, approximately 12%, in both AIDS and non-AIDS patients, contrary to some published reports [2]. Thus, important findings can be obtained almost instantaneously with computerized PCR data. While these projects are not research in the sense of testing hypotheses, the PCR certainly provides pilot data that can suggest directions and feasibilities for subsequent, more in-depth study. Furthermore, the PCR can be used in its present form

76

to record other types of data for pilot research projects, with use of the “utility fields” described earlier. Educational benefits to students, residents, fellows, and even attendings from the use of the PCR have not been measured. However, the use of the PCR obviously requires a certain understanding of the DSM-III-R multiaxial system by residents. The inclusion of the Karnofsky scale should at least remind the rater that severity of physical illness must be taken into account in psychiatric assessment. Checklists have been demonstrated to be effective in enhancing the completeness of psychiatric consultation notes [3,4]. Whether the routine use of the PCR has had that benefit, compared to another case reporting method, has not been addressed. Another unanswered question concerns the optimal length and degree of detail of the reporting form. These parameters balance the amount of information obtained about each patient against the reliability, completeness, and the rate of return of the forms. In conclusion, it should be emphasized that the major responsibility for the development of this computerized system was carried out by people naive about computers at the beginning of this process. Assistance from colleagues and from NCS made the task manageable, and even exciting, to those with the primary responsibility for developing and using this PCR system. It is hoped that the success of this project will encourage others to develop similarly efficient systems for handling large amounts of clinical data.

References Hengeveld MW, Huyse FJ, van der Mast RC, Tunstra CL: A proposal for standardization of psychiatric consultation-liaison data. Gen Hosp Psychiatry 10: 410-422, 1988 O’Dowd, MA, McKegney, FP: AIDS patients compared with others seen in psychiatric consultation. Gen Hosp Psychiatry, in press Huyse FJ, Strain JJ, Hengeveld MW, Hammer J, Zwaan T: Interventions in consultation-liaison psychiatry: The development of a schema and a checklist for operationalized interventions. Gen Hosp Psychiatry 10:88-101, 1988 Small GW, Fawzy FI: Data omitted from psychiatric consultation notes. J Clin Psychiatry 49:307-309,1988