JOB SUBMISSION MENU (JSM): A PROGRAMMING TOOL TO STANDARDIZE CLINICAL TRIAL SUMMARIZATION Iraj Mohebalian, Parke-Davis Pharmaceutical Research Mark Becker, STATPROBE, Inc. Lora H. Schwab, STATPROBE, Inc. The standardized SAS programs within JSM do not include batch Job Control Language (JCl), TSO commands, or any SAS code specific to a given drug. Instead unique information about each drug is stored wtthin the JSM environment. During submission of reports, JCl, CLlST, or SAS codes are dynamically constructed and passed to programs for execution. ABSTRACT A Job Submission Menu (JSM) is described which aids in standardizing the SAS programming that lists, summarizes, and analyzes the data from clinical trials. Design objectives of the system include: selfdocumentation of clinical trial programming, simplified batch and interactive job submission of SAS programs within the IBM/MVS operating system, a library system to document all clinical drug projects, minimization of duplicate programs across clinical studies, potential Computer-Assisted New Drug Application (CANDA) uses. The system was designed wtth flexibiltty to handle both standardized and nonstandardized programming. After over six years of use the system has gained dramatically in users and is credited with facilitating the evolution of standardization. JSM is compared to one other approach to standardization. STANDARDIZATION OF CLINICAL TRIAL MANAGEMENT Standardization of clinical trials is the process of making efficient use of similarities and repetition in any aspect of clinical trials. Aspects of clinical trial management where standardization can occur are: • • • • • INTRODUCTION protocols case report forms (CRFs) database design and variable names format decodes and dictionaries computer programs to list and summarize data • research reports The Job Submission Menu (JSM) system aids in standardizing clinical trail summarization. It is written in SAS, IBM ISPF Dialog Manager, and TSO CLlST, and tt works wtthin the IBM/MVS operating system in conjunction wtth the CA-LiBRARIAf118 program source code management system and the CA-'!'" automated scheduling system. After approximately 200 pages of programming code and four different releases over the past six years of use, JSM has gained dramatically in users and is credited with facilttating standardization of programming. Advantages of standardization include reduced time spent reinventing methods and systems already developed; consistent handling of data and presentations, more easily understood because users and readers are accustomed to the methods; ability to quickly absorb a mass of data and provide timely reporting of results; and promotion of good scientific research by demanding careful planning across a series of stUdies. However, there are problems in implementing standardization. Time and effort are required to achieve and to maintain standard programs. Insufficient foresight combined with a lack of flexibiltty can quickly ruin a system of standardization. The necessary cooperation from the various groups involved in each clinical trial can be difficult to obtain. Presently the JSM system allows data processors, statisticians, and programmers to work wtthin an integrated environment in the preparation of reports for clinical trials. Users can locate and process any SAS programs assigned to an investigational drug. 954 APPENDIX B.2 (PAGE 1 OF 1) DOUBLE-BLIND, MULTICENTER" PLACEBO-CONTROLLED STUDY OF TEST DRUG VERSUS CONTROL DRUG IN PATIENTS WITH DISEASE X PROTOCOL 9999-135-OVERALL N SEX MALE FEMALE PLACEBO TEST DRUG 7S •• 52 23 (69.3"10) (30.7%) 38 31 7. (55.1%) (44.9%) AGE (VRS) MEDIAN (RANGE) 34.0 12l-53} 35.0 (19-50) RACE: 57 ,. (78.0%) (18.7%) (5.3%) (0.0%) 57 (52.2%) (30.-4%) (13.0'%) (4.3%) 15 (76.0%) (9.3%) (14.7%) 59 CAUCASIAN BLACK OTHER NOT STATED • 0 REPRODUCTIVE POTENTIAL (FEMALE): BIRTH CONTROL HYSTERECTOMY INFERTILEITUBAL MENOPAUSE/POST GRADE OF DISEASE: STAGE 1 STAGE 2 STAGE 3 12 7 57 7 11 ,. 2 • 11 • ,. TREATMENT GROUP CONTROL DRUG TOTAL ., 2' 21. (58.6%) (41.4%) 34.5 (21-49) ,. (82.6%) (14.5%) 56 (2.9"10) 3 (0.0%) (48.-4%) (35.5%) (12.9%) (3.2%) "8 (78.3%) (7.2%) (14.5%) 52 8 5 2 ,. (80.0"10) (14.3%) (4.3%) (1.4%) (48.3%) (27.6%) (17.2"10) (6.9%) (74.3"0) (11.4%) (14.3%) 131 83 (61.2%) (38.8%) 34.5 (2l-53) 17. 3' • "2. • 12 163 2. 31 (79.4%) (15.9%) (4.2%) (0.5%) (49.9%) (31.3%) (19.5%) (9.8%) (76.2%) (9.3%) (14.5%) FIGURE 1. AN EXAMPLE OF A CLINICAL TRIAL REPORT A LOGICAL APPROACH TO STANDARDIZATION • a time-saving system which "self-documents" the computer programs which list, summarize, and analyze the results of a clinical trial. JSM was originally developed to facilitate standardization of clinical trial report generation. The task of standardizing all aspects of clinical trial data management and reporting was overwhelming so the JSM system began with one part of the process and expanded standardization from this base. The part selected was the generation of reports Oisting and summary tables) of clinical trial results. As it developed, JSM provided a logical approach to standardization. • a time-saving system which provides for easy submission of SAS computer programs in the environment of an unfriendly operating system (e.g., JCL requirements of TSO). • a system which has the potential to assist in achieving a currently popular result, ComputerAssisted New Drug Applications (CANDA). EXAMPLE: STANDARDIZING A "REPORT" To be readily accepted, a system for standardizing programs across stUdies must be capable of evolving over time. The system must be flexible to handle both standard and non·standard input. A system which handles subgroups of standard input will provide the seed offuture widespread standardization. Finally, to be more readily accepted, the system should offer an advantage which is of interest to users who don't care about standardization. JSM satisfied this last ideal by providing the following: An example ·report· is given in Figure 1 for a ·Summary of Patient Characteristics". The report shown is typical of a clinical trial in which there are three treatment groups (placebo, test drug, and control drug). The report is a summary of patient characteristics over all investigative sites or centers in a multicenter study; a similar summary would also be generated for each center. Assume that the same report is wanted for many clinical trials and assume 955 that the databases for these clinical trials are "standard" wtth respect to variable names and codes for patient characteristics. special PROe FORMAT statements). Table 1 is a sample of how JSM handles some non-standard ttems. Non-standard items, that is, ttems which change from one clinical trial to the next, in the text of the report in Figure 1 include the following: TABLE 1. How JSM Handles Non-Standard Items • • • • • • • • the source of data the study tftle(s) the test drug identffication the protocol identification the study center identifications the report numbers (e.g., Table 4) the treatment group names the number of treatment groups Method C B B B A B E D Non-Standard Items The source of data The study tttle(s) The test drug identffication The protocol identffication The study center identffications The report ttties (tablenames & numbers) The treatment group names The number of treatment groups Non-standard ttems in processing the report include the following: JSM PROFILES • computer system user to notffy where to route job job priortty on the computer system The study profile, saved in a portion of JSM called JSM Profile, is the aggregate of non-standard ttems and default job submission (i.e., JCL) options available when requesting reports for a specffic study. In addttion to constructing JCL or a CLiST based on each user profile, JSM passes 131 SAS macro variables to each selected SAS program based on a user profile and the user-constructed study profile. Table 2 shows a few selected SAS macro variables which are passed to each program. • submtt to print or to held queue number of copies • choice of paper (3-hole, no punch, portratt, landscape) • choice of printer Qaser or not) • SAS mode (batch, interactive) program name and location Standard items in the report include: TABLE 2. Some JSM Macro Variables • the form of data (names of variables, codes) the format of the report Variable Name Purpose • the title of the report itself &JSMBASE &JSMDDBAS &JSMDECOD &JSMDDECO &JSMEXEC There are at least five methods for providing standard treatment of non-standard ttems. Method A is to include non-standard items as variables in the database. Method B is to treat non-standard items as macro variables passed to the SAS report program at the time of job submission. Method C is to treat nonstandard ttems as variables passed to job control at the time of job submission. Method D is to handle the ttem through programming (perhaps keying on a variable passed at the time of job submission or computed from data within the same program, or by providing two or three versions of the "standard" program). Method E is to treat non-standard items as source code statements to be inserted in the SAS report program at the time of job submission (such as &JSMTITLl &JSMOPT1 &JSMTRIAL &JSMLlBRA &JSMNUM1 956 Dataset name of primary SAS library DDname of primary SAS library Dataset name of SAS decode file DDname of primary SAS decode file Specifies which SAS catalog procedure to execute First line of title wtthin a report Used to specffy optional SAS logic wtthin report source code Used to specify the subset of trial numbers (Le., various study centers) Name of LIBRARIAN file containing report SAS source code Name/number of table used within title 8UBJOB MENU COMMAND ~ ,---- JCL Drug.ID =:::::::::> eon 8tudr-1D c:::::::::::> 0& JOBNAMEI REPORT NOTIFY, JDOW COPIE8. 1 M8GCLAS9. iii. CLASSI U BINI 124 SiloS MODE -==c> lATCH LASER: Y (YIN) Option. CONTROLI 8U.MIT 0:;::::::(> IF 8TArUS - 2 I <==<> Tableb) SpeCllloatlon: 1. R9iGA.Oll Appendix E.1 /I LIST OF ADVERSE EVENTS 2. Rg90A12 I Table B.20 /I SUMMARY OF I.E SEVERITY .. .. 8. RSl911H02 I Appendix F.2 1/ LIST MEDICAL HISTORY 5. 7. FIGURE 2. EXAMPLE OF JSM REPORT SUBMISSION SCREEN Then, at the time of job submission (as seen in Figure 3), JSM will prompt the user with a complete list of programs assigned to the specific clinical trial project. One can see at a glance which programs generate which report numbers, and one can quickly change report numbers by modifying the area in JSM Profile rather than the individual programs. JSM has security features which allow only authorized personnel to change a profile. The information in JSM Profile is passed to the SAS report program at the time of job submission. For an example of the JSM job submission screens, see Figures 2 and 3. Figure 2 shows the screen used to submit a report using JSM. The screen permits the user to choose a specific study ("Drug-ID" and "Study-ID") and various JCL options (default values appear so that the set-up may not need to be changed). The user fills in the 'Table(s) Specification" lines with specific report requests. If the user prefers to select from a list of reports, the screen illustrated in Figure 3 appears if no request is entered on the "Table(s) Specification" lines. The user then selects reports by entering "S" to the left of all reports desired. COMPARISON TO ANOTHER APPROACH TO STANDARDIZATION There are other approaches to standardization. In the AUTO REPORT GENERATOR system, non-standard information is fed to a standard program which in turn generates non-standard SAS programs to report a specific clinical trial. In contrast, in the JSM system, standard SAS programs are written with macros which receive non-standard information from the profile set up for each clinical trial. Table 3 compares these two systems briefly. The features of JSM encourage users to document the programs used to generate reports for a dinical trial. First, a special area in the JSM Profile is reserved for listing program names, passwords (assigned when programs are permanently stored in the LIBRARIAN system), and report numbers which are passed as macro variables (e.g., APPENDIX B.2). 957 REPORT 8ELECTION MENU COMM"ND -==<> SOROLL ~ PAGr: Drug_to: 8908 8tudy.IDI DB LINE CMO Table(a} epecltlcatlon A"'A01 I Appel\dlx E.1 II LlaT OF ADVERSE EVENTS R8IUIA11 I Tabl. B.18 1/ SUMMARY RELEATED AE·.ODY BY8TEM RaSUtAtl I TABLE 8.20 /I 8UMMARY OF AE SEVERITY RoaaAiS J APpendix E.7 /I LIST PTa DISCONTINUED . s .. RIIDD10 I rable E.10 II aUMWARY OF PATIENTS DISCONTINUED RIgIE10 I rabl. B •• " SUMMARY OF ETT BY WEEK RlltlaHat I Appendl:a F.1 /I LIST PATIENT CHARACTERISTICS AI.GH02 I Ap,endlx F.2 II LlaT MEDICAL HI8TORY RIShlH03 I App.ndlx F.3 II LlaT PRIOR MEDICATION R9UH04 I Tabl. B.1 1/ SUMMARY PATIENT CHARACTERlaTICa FIGURE 3. EXAMPLE OF JSM REPORT SELECTION SCREEN TABLE 3. Comparison of Two Standardization Systems JSM Easy to set-up Easy to modify Handles non-standard Promotes gradual standard Promotes documentation NO YES YES YES YES ACKNOWLEDGEMENTS SAS is a registered trademark of SAS Institute Inc. Cary, NC. AUTO REPORT IBM is a registered trademark of International Business Machines Corporation. NO NO NO NO CA-UBRARIAN and CA-7 are trademarks of Computer Associates International., Inc., Garden City, NY. ? For further Information regarding this paper, please contact: CONCLUSION Mark Becker. PhD STATPROBE. Inc. 5340 Plymoulh Road Ann Arbor. Michigan 48105 leI. (313) 769-5000 The Job Submission Menu (JSM) has been successful in simplifying submission of report requests (SAS programs) for clinical trials, facilitating standardization of report requests, handling both standard and nonstandard reports, and improving documentation of reports of clinical trials. STATPROBE, Inc. is a contract research organization serving the pharmaceutical industry. 958