ACADEMIA AND CLINIC
A Consultant-Extender System for Breast Cancer Adjuvant Chemotherapy DAVID WIRTSCHAFTER, M.D.; JOHN T. CARPENTER, M.D.; and EMMANUEL MESEL, M.D.; Birmingham, Alabama
We have implemented a consultant-extender system that enables community physicians, in cooperation with regional specialists, to deliver adjuvant chemotherapy to patients with node-positive breast cancer. The system employs computer-generated care protocol forms that indicate the data to be collected and the drug dose(s), with the appropriate rules for their administration. This continuous process of monitoring and modifying therapy assures protocol compliance and facilitates quality of care assessment. Seventy-three physicians throughout Alabama delivered appropriate chemotherapy at nearly 9 7 % of 2 6 1 2 visits by 195 patients. Disease-free intervals of 149 of those patients treated in a prospective clinical trial are indistinguishable from those of comparable patients treated largely within academic centers. This system provides a mechanism for decentralizing speciality care, incorporating community physicians into clinical trials, and improving continuing medical education techniques. A D J U V A N T CHEMOTHERAPY has reduced the early relapse rate in women with operable node-positive breast cancer, particularly in premenopausal women (1, 2). In an effort to transfer this treatment advance to patients in the community, surgeons and physicians in Alabama have joined in a network (the Alabama Breast Cancer Project) to deliver similar treatment to their patients; the early clinical results of that trial have been reported elsewhere (3). To assure optimal chemotherapy, we have developed a consultant-extender system; the system is based on a clinical algorithm and produces visit-by-visit, patient-specific treatment advice. The system also includes continuous data monitoring to allow frequent analysis both of the consultant-extender system and of the clinical results. We describe here our first 36 months' experience. Methods THERAPY
Details of the Alabama Breast Cancer Project and its adjuvant chemotherapy program have been previously described (36). Briefly, practicing board-certified surgeons throughout the state enter informed and consenting patients into a surgical protocol (radical versus modified radical mastectomy); if the patient has pathologically proven metatasis to one or more axillary nodes without distant metatasis, the doctor may also enter her onto an adjuvant chemotherapy protocol. The chemotherapy consists either of eight cycles of melphalan, given by mouth for 5 consecutive days every 6 weeks, or 24 cycles of cyclophosphamide, methotrexate, and 5-fluorouracil, given intravenously in a single dose every 2 weeks. If marrow toxicity necessitates dosage adjustment, then it is done according to the flowchart shown in Figure 1. A medical oncologist (JTC) manages ambiguous or unusual clinical situations on an individual basis. • From the Clinical Information Systems Group, the Departments of Pediatrics, Medicine, and Information Science, and the Comprehensive Cancer Center, University of Alabama; Birmingham, Alabama. 396
Annals of Internal Medicine 90:396-401, 1979
Downloaded from https://annals.org by Karolinska Institute user on 01/18/2019
C O N S U L T A N T - E X T E N D E R SYSTEM
We monitor the response to therapy and modify it on a visitby-visit basis, using the consultant-extender system (7) developed by the Clinical Information Systems Group, University of Alabama in Birmingham. This system is analogous to the physician-extender models for augmenting and distributing healthcare delivery expertise. The system includes consulting specialists from the University of Alabama in Birmingham, primary physicians within the community, and the information system. The consulting specialist is the hematologist-oncologist who supervises the medical program; he has defined the recommended chemotherapy regimen and has helped develop the clinical algorithms. He is also responsible for consultations with community physicians should serious problems arise. The consultant-extender is the community-based physician who, with the aid of care protocol forms generated by the information system, provides services usually associated with a consulting specialist. The information system integrates the general rules of the chemotherapy regimen with the specific needs of individual patients on a visit-by-visit basis. The discursive chemotherapy regimen was translated into a clinical algorithm, a step-by-step procedure of managing therapy. Figure 1 shows the clinical algorithm. It specifies both the data collection requirements and the plan for managing treatment. The general requirements and recommendations, the visit-specific advice rule, and administrative data are all incorporated into a patient-specific, visit-specific form (see Figure 2). This document, termed a care protocol form, is patterned after those developed for the Ambulatory Care Project, Beth Israel Hospital, Boston, Massachusetts (8). It both advises the consultant-extender of the protocol rules for that upcoming visit and provides a place for recording the physician's findings and actions at the time of that visit; that is, it serves as a "progress" note for that visit. CLINICAL ALGORITHM
Much of the information necessary for formulating therapy advice recommendations is available before the patient's next visit for therapy. The clinical situations that require withholding therapy or consultation, for example, progression of disease, life-threatening infection, life-threatening bleeding, or nonmarrow toxicity, are constant; the associated recommendations always appear in the "Assessment and Plan" section of the care protocol form (see Figure 2). The consultant-extender is responsible for examining and assessing the patient for these conditions. The physician should consider chemotherapy that day only if those conditions are absent. The visit-specific protocol recommendations are generated by following the medical logic indicated in Figure 1. Most decisions (regular bordered hexagonal figures) can be made before the specific visit based on the patient's previous response to therapy. Those decisions dependent on the patient's clinical status, leukocyte count, and platelet count at the specific visit are indicated by the heavy-bordered hexagonal figures. Since only the latter data are unknown before the visit, alternative courses of management can be delineated for situations of normal or low blood counts. Appropriate recommendations are printed for each alternative (see advice rule and dosages below the heading "Consider for Rx today" [Figure 2]). For example, if a patient was scheduled for the first visit of the third therapy cycle and had tolerated the previous ©1979 American College of Physicians
full (100%) melphalan dosage without any marrow toxicity, then the recommendation would be either to give a 100% dosage (normal leukocyte and platelet counts) or wait 2 weeks and re-evaluate (low leukocyte or platelet counts). A more complicated example would be that of a woman who was found to have low counts on the first day of therapy cycle and was then instructed to return in 2 weeks. Following the algorithm, one determines that this will be the second attempt to give therapy; it was delayed once because of low counts. Depending on the leukocyte and platelet counts at the upcoming visit, the recommendation would be to repeat the last dosage (normal counts), give 50% of the full dosage (persistently low but stable counts), or withhold therapy and await further special instructions from the consulting oncologist (low and falling counts). SYSTEM OPERATION
When a physician enrolls a new patient into the program, we send him a detailed operations manual. The network's nurse specialist visits with the physician before the first treatment and explains the procedures for using the care protocol forms and handling the antineoplastic drugs. For instance, if the physician checks a "no" box with an asterisk in it, then he need not answer the subsequent questions related to that topic. Clinical criteria for recognizing the therapy contraindications are also described in the manual. The subsequent cycle of events is as follows. We prepare a care protocol form at the center before each of the patient's visits. In the manual mode of operation, a staff member analyzes the patient's previous record and then marks out the possible courses of action on a copy of the clinical algorithm (see Figure 1). The visit-specific advice rule is typed onto the general form, along with the drug dose(s). We then mail the form to the consultant-extender, who uses it as his "progress note" for that visit. He returns a carbon copy of the completed form to the center, where the information is added to the data base and used in preparing the next visit's form. Computer programs have been implemented that emulate the manual process for selecting therapy recommendations and printing the care protocol forms. To date, nearly one half of the encounter forms have been automatically generated. The programs operate on our university's IBM 370/158 (White Plains, New York), under the Time-Share Option, are written in PL/I, and use the general purpose relational medical database and time-series search subsystems developed by the Clinical Information Systems Group (9). Operating cost is subject to arbitrary adjustment of three major factors: variation in the volume of workload (many of the operating costs are fixed and vary little with volume); the sharing of entered data between the care assurance function and the clinical research function (direct interface to the statistical analysis programs to the Alabama Breast Cancer Project); the sharing of operating and supervisory personnel between the care assurance and clinical research segments of the Alabama Breast Cancer Project; and the exclusion of development costs. At our current volume, with all data entry charged to the system and with personnel apportioned by percentage of effort, we estimate the combined care assurance and clinical research costs to be about $10 per visit. Physician supervisory and quality control time has been substantially reduced since the computer programs were implemented and now is less than 1 h per week. The validity of this clinical algorithm rests primarily on the clinical experience, both published (3) and unpublished, of medical oncologists who have given adjuvant chemotherapy. Many of the medical decision rules incorporated into the algorithm are formulations of their "good clinical judgment." These rules represent their judgment on how best to administer antineoplastic drugs safely and practically. We recommended convenient drug dosages when feasible to minimize error in mixing and waste. For example, any dose of 5-fluorouracil from 450 to 650 mg was rounded off to 500 mg, the size of the commercially available vial. Other drug dosages were rounded off to the nearest 100-mg dose level, or to the nearest 10 mg for methotrexate (40 mg, 50 mg, and so forth). Since we could not directly monitor the amount of each drug given in the physician's office, we
Figure 1 . Breast cancer adjuvant chemotherapy algorithm. Mpl = melphalan; CMF = cyclophosphamide, methotrexate, and 5-fluo rouracil; Rx = therapy. * * Six weeks f r o m the date o f last treatment.
* Administrative delays, for example, scheduling or intercurrent illnesses. t Low counts: leukocyte
A Consultant-Extender System for Breast Cancer Adjuvant Chemotherapy DAVID WIRTSCHAFTER, M.D.; JOHN T. CARPENTER, M.D.; and EMMANUEL MESEL, M.D.; Birmingham, Alabama
We have implemented a consultant-extender system that enables community physicians, in cooperation with regional specialists, to deliver adjuvant chemotherapy to patients with node-positive breast cancer. The system employs computer-generated care protocol forms that indicate the data to be collected and the drug dose(s), with the appropriate rules for their administration. This continuous process of monitoring and modifying therapy assures protocol compliance and facilitates quality of care assessment. Seventy-three physicians throughout Alabama delivered appropriate chemotherapy at nearly 9 7 % of 2 6 1 2 visits by 195 patients. Disease-free intervals of 149 of those patients treated in a prospective clinical trial are indistinguishable from those of comparable patients treated largely within academic centers. This system provides a mechanism for decentralizing speciality care, incorporating community physicians into clinical trials, and improving continuing medical education techniques. A D J U V A N T CHEMOTHERAPY has reduced the early relapse rate in women with operable node-positive breast cancer, particularly in premenopausal women (1, 2). In an effort to transfer this treatment advance to patients in the community, surgeons and physicians in Alabama have joined in a network (the Alabama Breast Cancer Project) to deliver similar treatment to their patients; the early clinical results of that trial have been reported elsewhere (3). To assure optimal chemotherapy, we have developed a consultant-extender system; the system is based on a clinical algorithm and produces visit-by-visit, patient-specific treatment advice. The system also includes continuous data monitoring to allow frequent analysis both of the consultant-extender system and of the clinical results. We describe here our first 36 months' experience. Methods THERAPY
Details of the Alabama Breast Cancer Project and its adjuvant chemotherapy program have been previously described (36). Briefly, practicing board-certified surgeons throughout the state enter informed and consenting patients into a surgical protocol (radical versus modified radical mastectomy); if the patient has pathologically proven metatasis to one or more axillary nodes without distant metatasis, the doctor may also enter her onto an adjuvant chemotherapy protocol. The chemotherapy consists either of eight cycles of melphalan, given by mouth for 5 consecutive days every 6 weeks, or 24 cycles of cyclophosphamide, methotrexate, and 5-fluorouracil, given intravenously in a single dose every 2 weeks. If marrow toxicity necessitates dosage adjustment, then it is done according to the flowchart shown in Figure 1. A medical oncologist (JTC) manages ambiguous or unusual clinical situations on an individual basis. • From the Clinical Information Systems Group, the Departments of Pediatrics, Medicine, and Information Science, and the Comprehensive Cancer Center, University of Alabama; Birmingham, Alabama. 396
Annals of Internal Medicine 90:396-401, 1979
Downloaded from https://annals.org by Karolinska Institute user on 01/18/2019
C O N S U L T A N T - E X T E N D E R SYSTEM
We monitor the response to therapy and modify it on a visitby-visit basis, using the consultant-extender system (7) developed by the Clinical Information Systems Group, University of Alabama in Birmingham. This system is analogous to the physician-extender models for augmenting and distributing healthcare delivery expertise. The system includes consulting specialists from the University of Alabama in Birmingham, primary physicians within the community, and the information system. The consulting specialist is the hematologist-oncologist who supervises the medical program; he has defined the recommended chemotherapy regimen and has helped develop the clinical algorithms. He is also responsible for consultations with community physicians should serious problems arise. The consultant-extender is the community-based physician who, with the aid of care protocol forms generated by the information system, provides services usually associated with a consulting specialist. The information system integrates the general rules of the chemotherapy regimen with the specific needs of individual patients on a visit-by-visit basis. The discursive chemotherapy regimen was translated into a clinical algorithm, a step-by-step procedure of managing therapy. Figure 1 shows the clinical algorithm. It specifies both the data collection requirements and the plan for managing treatment. The general requirements and recommendations, the visit-specific advice rule, and administrative data are all incorporated into a patient-specific, visit-specific form (see Figure 2). This document, termed a care protocol form, is patterned after those developed for the Ambulatory Care Project, Beth Israel Hospital, Boston, Massachusetts (8). It both advises the consultant-extender of the protocol rules for that upcoming visit and provides a place for recording the physician's findings and actions at the time of that visit; that is, it serves as a "progress" note for that visit. CLINICAL ALGORITHM
Much of the information necessary for formulating therapy advice recommendations is available before the patient's next visit for therapy. The clinical situations that require withholding therapy or consultation, for example, progression of disease, life-threatening infection, life-threatening bleeding, or nonmarrow toxicity, are constant; the associated recommendations always appear in the "Assessment and Plan" section of the care protocol form (see Figure 2). The consultant-extender is responsible for examining and assessing the patient for these conditions. The physician should consider chemotherapy that day only if those conditions are absent. The visit-specific protocol recommendations are generated by following the medical logic indicated in Figure 1. Most decisions (regular bordered hexagonal figures) can be made before the specific visit based on the patient's previous response to therapy. Those decisions dependent on the patient's clinical status, leukocyte count, and platelet count at the specific visit are indicated by the heavy-bordered hexagonal figures. Since only the latter data are unknown before the visit, alternative courses of management can be delineated for situations of normal or low blood counts. Appropriate recommendations are printed for each alternative (see advice rule and dosages below the heading "Consider for Rx today" [Figure 2]). For example, if a patient was scheduled for the first visit of the third therapy cycle and had tolerated the previous ©1979 American College of Physicians
full (100%) melphalan dosage without any marrow toxicity, then the recommendation would be either to give a 100% dosage (normal leukocyte and platelet counts) or wait 2 weeks and re-evaluate (low leukocyte or platelet counts). A more complicated example would be that of a woman who was found to have low counts on the first day of therapy cycle and was then instructed to return in 2 weeks. Following the algorithm, one determines that this will be the second attempt to give therapy; it was delayed once because of low counts. Depending on the leukocyte and platelet counts at the upcoming visit, the recommendation would be to repeat the last dosage (normal counts), give 50% of the full dosage (persistently low but stable counts), or withhold therapy and await further special instructions from the consulting oncologist (low and falling counts). SYSTEM OPERATION
When a physician enrolls a new patient into the program, we send him a detailed operations manual. The network's nurse specialist visits with the physician before the first treatment and explains the procedures for using the care protocol forms and handling the antineoplastic drugs. For instance, if the physician checks a "no" box with an asterisk in it, then he need not answer the subsequent questions related to that topic. Clinical criteria for recognizing the therapy contraindications are also described in the manual. The subsequent cycle of events is as follows. We prepare a care protocol form at the center before each of the patient's visits. In the manual mode of operation, a staff member analyzes the patient's previous record and then marks out the possible courses of action on a copy of the clinical algorithm (see Figure 1). The visit-specific advice rule is typed onto the general form, along with the drug dose(s). We then mail the form to the consultant-extender, who uses it as his "progress note" for that visit. He returns a carbon copy of the completed form to the center, where the information is added to the data base and used in preparing the next visit's form. Computer programs have been implemented that emulate the manual process for selecting therapy recommendations and printing the care protocol forms. To date, nearly one half of the encounter forms have been automatically generated. The programs operate on our university's IBM 370/158 (White Plains, New York), under the Time-Share Option, are written in PL/I, and use the general purpose relational medical database and time-series search subsystems developed by the Clinical Information Systems Group (9). Operating cost is subject to arbitrary adjustment of three major factors: variation in the volume of workload (many of the operating costs are fixed and vary little with volume); the sharing of entered data between the care assurance function and the clinical research function (direct interface to the statistical analysis programs to the Alabama Breast Cancer Project); the sharing of operating and supervisory personnel between the care assurance and clinical research segments of the Alabama Breast Cancer Project; and the exclusion of development costs. At our current volume, with all data entry charged to the system and with personnel apportioned by percentage of effort, we estimate the combined care assurance and clinical research costs to be about $10 per visit. Physician supervisory and quality control time has been substantially reduced since the computer programs were implemented and now is less than 1 h per week. The validity of this clinical algorithm rests primarily on the clinical experience, both published (3) and unpublished, of medical oncologists who have given adjuvant chemotherapy. Many of the medical decision rules incorporated into the algorithm are formulations of their "good clinical judgment." These rules represent their judgment on how best to administer antineoplastic drugs safely and practically. We recommended convenient drug dosages when feasible to minimize error in mixing and waste. For example, any dose of 5-fluorouracil from 450 to 650 mg was rounded off to 500 mg, the size of the commercially available vial. Other drug dosages were rounded off to the nearest 100-mg dose level, or to the nearest 10 mg for methotrexate (40 mg, 50 mg, and so forth). Since we could not directly monitor the amount of each drug given in the physician's office, we
Figure 1 . Breast cancer adjuvant chemotherapy algorithm. Mpl = melphalan; CMF = cyclophosphamide, methotrexate, and 5-fluo rouracil; Rx = therapy. * * Six weeks f r o m the date o f last treatment.
* Administrative delays, for example, scheduling or intercurrent illnesses. t Low counts: leukocyte
