Management of Colorectal Cancer in Medicare Health Maintenance Organizations SHELDON M. RETCHIN, MD, MSPH, BARBARA BROWN, PhD Because o f the f i n a n c i a l incentives o f p r e p a i d care, t h e quality o f care f o r Medicare enrollees in Health M a i n t e n a n c e Organizations (HMOs) is a concerto Therefore, t h e care in 150 newly diagnosed cases o f colorectal cancer in eight HM Os was compared with that in 180 similar fee-forservice (FFS) cases. As p a r t o f the diagnostic workup, HMO patients were more likely to have h a d f e c a l occult blood tests (74% vs 52%, p < 0.01) a n d endoscopy o r barium enemas (97% vs 90%, p < 0.05). FFS patients w e r e m o v e likely to h a v e h a d preoperative imaging studies (54% us 38%, p < 0.01). Although there were longer diagnostic delays f o r HMO enrollees with gastrointestinal bieeding~ there were no significant differences in disease stage o r clinical status, a n d postoperative follow-up was similar. The authors conclude that enrollees in Medicare HMOs with colorectal cancer r e c e i v e m e d i c a l a n d surgical care at least equal to that received in FFS settings. Key words: health maintenance organizations; quality o f cave; cancer. J CJENINTERN MED 1990; 5:110--114.
AN EXPANDING ELDERLYPOPULATION has led to a rapid g r o w t h in the rate of health care c o n s u m p t i o n in recent years. ~With medical e x p e n d i t u r e s likely to c o n t i n u e to rise, initiatives a i m e d at constraining costs are receiving m o r e attention. For this reason, the Medicare program has sought to curtail growing costs t h r o u g h m e t h o d s such as p r e p a i d health plans. In 1982, the Health Care FinancingAdministration i n t r o d u c e d a Medicare demonstration p r o g r a m o f prepaid health care. As part of this program, 27 health m a i n t e n a n c e organizations (HMOs) assumed the financial risk of providing medical care for Medicare recipients w h o chose to enroll in one of the plans. In turn, HMOs received m o n t h l y payments for each enrollee. Since the original demonstrations, the c h o i c e of enrollm e n t in HMOs for Medicare recipients has b e e n expanded, and there are n o w over one million enrollees in Medicare HMOs. The major t e c h n i q u e HMOs use to r e d u c e the cost of care is to substitute a m b u l a t o r y care for m o r e expensive inpatient care. Inpatient utilization rates as l o w as half that of traditional Medicare beneficiaries have b e e n observed in HMOs2-4; however, substituting ambulatory for inpatient care may b e deleterious to elderly patients. 5 Furthermore, overzealous attempts to limit access to resource-intensive services may lead to Received from the Departments of Medicine and Health Administration, and the Williamson Institute for Health Studies, Medical College of Virginia, Virginia Commonwealth University, Richmond, Virginia. Supported by a contract from the Health Care Financing Administration, No. 500-83-0047. Presented in part at the annual meeting of the Society of General Internal Medicine, Washington, DC, April 27, 1989. Address correspondence and reprint requests to Dr. Retchin: Box 287, MCV Station, Richmond, VA 23298. 11~
inappropriate delays for n e e d e d t h e r a p e u t i c interventions, such as surgical procedures. While there are data to suggest that rates for elective p r o c e d u r e s are r e d u c e d in p r e p a i d plans, 6 surgical rates for non-elective procedures are a m o r e i m p o r t a n t concern. 7, s The p u r p o s e of the early Medicare demonstrations in prepaid care was to d e t e r m i n e w h e t h e r HMOs c o u l d r e d u c e health care costs to Medicare recipients w i t h o u t c o m p r o m i s i n g quality o f care. The c u s t o m a r y fee-forservice (FFS) m e t h o d for r e i m b u r s e m e n t was used as a referent standard o f care. The evaluation o f quality o f care for b o t h systems i n c l u d e d an assessment o f a condition usually m a n a g e d surgically, colorectal cancer. Colorectal cancer was chosen b e c a u s e o f several characteristics. First, colorectal cancer is the most freq u e n t l y occurring n e o p l a s m in the United States, w i t h an incidence rate of 47 cases per 1 0 0 , 0 0 0 persons per yearg; in addition, it is associated with a heavy b u r d e n o f illness, with less than 50% surviving for five years following diagnosis. 10 Second, since the diagnosis of colorectal cancer is histopathologic, usually b y mucosal biopsy, diagnostic accuracy is high. Thus, n e w cases o f colorectal cancer are unlikely to be misclassified, and most require a surgical p r o c e d u r e . Finally, there is fairly wide a g r e e m e n t on a p p r o p r i a t e surgical managem e n t of the disease a c c o r d i n g to the stage o f the cancer, and this a p p r o a c h has c h a n g e d very little o v e r recent years.t ~ Therefore, a study of the process of care for patients n e w l y diagnosed as having colorectal c a n c e r was undertaken, and the results are r e p o r t e d here.
METHODS To provide a reasonable c o m p a r i s o n of the care of patients w i t h colorectal c a n c e r in HMOs, a similar g r o u p of FFS cases was selected for evaluation. Both groups w e r e restricted to incident cases w i t h a hospitalization during the study period. The quality o f care was judged for b o t h groups of patients.
Development of Quality of Care Criteria To provide a c o m p r e h e n s i v e appraisal of the care of patients w i t h colorectal cancer, b o t h o u t p a t i e n t and inpatient care w e r e evaluated. Physician advisory panels w e r e selected to d e v e l o p quality o f care criteria for the o u t p a t i e n t detection, surgical m a n a g e m e n t , and postoperative follow-up of the disease. Physician representation i n c l u d e d specialists in o n c o l o g y and epidemiology. In d e v e l o p i n g criteria for the a p p r o p r i a t e detection and m a n a g e m e n t of colorectal cancer, comm o n norms o f clinical care w e r e emphasized. Several activities w e r e c o n s i d e r e d particularly important: de-
,JOURNALOFGENERALINTERNALMEDICINE, Volume 5 (March/April), 1990
tection, use of time-intensive cognitive evaluations (e.g., medical history, physical examination), appropriate hospitalization, appropriate use of diagnostic tests and procedures, appropriate surgical management, and adequate follow-up after hospital discharge. Since colorectal cancer is usually managed surgically, the delay between presentation and diagnosis was considered particularly important. Because inappropriate delays may lead to patients' being more severely ill during hospitalization, two procedures were used to evaluate the stage of illness at the time of surgery. First, as a means of determining severity of illness, preoperative physiologic function was d e t e r m i n e d by using laboratory tests routinely p e r f o r m e d prior to surgery. Second, depending on local convention, several standard classification schemes were used to determine the histopathologic stage at the time of diagnosis: Dukes' classification, Asler-Coller modifications, and international TNM classifications. These classifications w e r e c o m b i n e d into a four-stage grouping. To determine the appropriateness of the diagnostic workup, patient records were abstracted from the office practices of the primary care physicians. Hospital records were abstracted to evaluate the preoperative care and surgical management. Both records were used to judge the quality of postoperative follow-up. Selection of Patients Study patients w e r e selected from each HMO plan from a list of all hospitalizations for colorectal cancer from the beginning of each plan enrollment date (range,January 1983 to May 1984) to March 31, 1986. For the study design, there w e r e two major issues for selecting a sample of HMOs. First, since the incidence of n e w cases is relatively low, even in an older population, HMOs with fewer than 2,000 enrollees w e r e eliminated because they w o u l d be unlikely to have sufficient cases for statistical power. Further, since medical care often follows conventional practice by area of the country, geographic location was important in the selection of plans for study. Thus, eight HMOs distributed among the four major geographic areas were selected for study. To secure the plans' participation and confid e n c e in the integrity of the study objectives, strict confidentiality was ensured, and specific HMO plans were not identified in the analysis. Since individual patient claims are inconsistently filed for HMO patients, discharge diagnoses were used for drawing the HMO sample. Therefore, HMO plans participating in the study were asked to provide patient lists with the following discharge diagnoses, or equivalent symptoms, for colorectal cancer: cancer of the rectum, sigmoid, cecum, or colon; mass or tumor of the rectum, sigmoid, cecum, or colon; admission for colectomy, hemicolectomy, or polypectomy; rectal or gastrointestinal bleeding; and abdominal pain, vomiting, obstruction, dehydration, obstipation, or severe constipation. All sampled cases were screened by nurse abstractors to ensure that they were incident cases, and
111
most of the ineligible cases represented recurrent disease. In the eight HMOs selected for study, there w e r e approximately 900 colorectal cancer cases screened, from w h i c h 150 eligible cases were successfully abstracted. For seven of the plans, there were 110 eligible cases from 196 drawn (59%). Because one plan was unable to provide a list of incident cases, all diagnoses and symptoms were used. From this one plan the proportion of new cases among the total sample of discharges with colorectal cancer was less than 10%. Samples of medical records of patients with colorectal cancer were drawn from Medicare claims files through the Medicare Automated Data Retrieval System for FFS hospitals located in geographic locales identical to that of the HMOs. The study period for drawing the sample was related to the start of Medicare HMOs in the area. Since the HMO sample was drawn from discharge diagnoses, and not Medicare claims, there were slight discrepancies in coding. FFS patients hospitalized for newly diagnosed colorectal cancer were selected if discharge diagnoses included malignant neoplasms of the colon, rectum, rectosigmoid junction, or anus. Patients with carcinoma in situ of either the colon or the rectum were also eligible for the study. The sample was screened to ensure that all cases w e r e incident cases, although the study patients did not have to be n e w to the practice. The FFS sample of cases was selected according to the n u m b e r of colorectal cancer patients discharged from July 1, 1983, to March 31, 1986. The hospitals were selected from a list of hospitals ranked by the n u m b e r of patients discharged with the diagnosis during the study period. Of the 70 hospitals identified for inclusion in the study, one had to be replaced because all patients were HMO patients, and three refused to participate. There were 969 potential colorectal cancer patients from the 66 hospitals left for study. Of the 969 patients identified, 180 (18.5 %) were considered eligible and their charts were abstracted. Sixty per cent of the ineligible cases were not new cases of colorectal cancer. The remaining ineligible cases were related to difficulties e n c o u n t e r e d in the ambulatory setting. Thirteen percent of the patients' physicians refused to participate; 9% of the practices could not find charts, or the sampled patient did not belong to the physician interviewed. Another 18% were not considered eligible for a variety of reasons. Registered nurses, specially trained for this study, were used to abstract both inpatient and outpatient medical records for all cases. Reliability was determined by circulating a sham chart to all nurse abstractors. Interrater reliability was assessed by having all ten abstractors and the principal investigator (SMR) review the sham record. Agreement among raters was determined by the kappa statistic) 2 A kappa value of > 0 . 7 5 is considered excellent agreement, 0.40 to 0.75 good agreement, and < 0 . 4 0 weak agreement. For this study, crude agreement between raters ranged from 0.90 to 0.93, with kappa values of 0.54 to 0.68.
Retchin, Brown, COLORECTALCANCERMANAGEMENTIN HMOs
11 ~.
TABLE 1 Comparisonof Personaland IllnessCharacteristicsfor Colorectal Cancer Patients, Accordingto Type of Plan
Fee-for-Service ( n = 180)
Health Maintenance Organization (n== 150)
Mean age at time of first visit to practice
70.1 years
72.3 years*
Mean length of time between first and most recent visits to practice
78.7 months
29.6 months*
Female
91 (50.6%)
62 (41.3%)
White
156 (86.7%)
143 (95.2%)*
Married
111 (62.7%)
98 (67.1%)
Symptoms at presentationt Anemia Gastrointestinal bleeding Altered bowel habit Weight loss Abdominal pain
33 (18.3°/0) 84 (46.7%) 51 (28.3%) 38 (21.1%) 58 (32.2°/0)
22 83 36 23 40
(14.7%) (55.3%) (24.0%) (15.3%) (26.7%)
*p
AN EXPANDING ELDERLYPOPULATION has led to a rapid g r o w t h in the rate of health care c o n s u m p t i o n in recent years. ~With medical e x p e n d i t u r e s likely to c o n t i n u e to rise, initiatives a i m e d at constraining costs are receiving m o r e attention. For this reason, the Medicare program has sought to curtail growing costs t h r o u g h m e t h o d s such as p r e p a i d health plans. In 1982, the Health Care FinancingAdministration i n t r o d u c e d a Medicare demonstration p r o g r a m o f prepaid health care. As part of this program, 27 health m a i n t e n a n c e organizations (HMOs) assumed the financial risk of providing medical care for Medicare recipients w h o chose to enroll in one of the plans. In turn, HMOs received m o n t h l y payments for each enrollee. Since the original demonstrations, the c h o i c e of enrollm e n t in HMOs for Medicare recipients has b e e n expanded, and there are n o w over one million enrollees in Medicare HMOs. The major t e c h n i q u e HMOs use to r e d u c e the cost of care is to substitute a m b u l a t o r y care for m o r e expensive inpatient care. Inpatient utilization rates as l o w as half that of traditional Medicare beneficiaries have b e e n observed in HMOs2-4; however, substituting ambulatory for inpatient care may b e deleterious to elderly patients. 5 Furthermore, overzealous attempts to limit access to resource-intensive services may lead to Received from the Departments of Medicine and Health Administration, and the Williamson Institute for Health Studies, Medical College of Virginia, Virginia Commonwealth University, Richmond, Virginia. Supported by a contract from the Health Care Financing Administration, No. 500-83-0047. Presented in part at the annual meeting of the Society of General Internal Medicine, Washington, DC, April 27, 1989. Address correspondence and reprint requests to Dr. Retchin: Box 287, MCV Station, Richmond, VA 23298. 11~
inappropriate delays for n e e d e d t h e r a p e u t i c interventions, such as surgical procedures. While there are data to suggest that rates for elective p r o c e d u r e s are r e d u c e d in p r e p a i d plans, 6 surgical rates for non-elective procedures are a m o r e i m p o r t a n t concern. 7, s The p u r p o s e of the early Medicare demonstrations in prepaid care was to d e t e r m i n e w h e t h e r HMOs c o u l d r e d u c e health care costs to Medicare recipients w i t h o u t c o m p r o m i s i n g quality o f care. The c u s t o m a r y fee-forservice (FFS) m e t h o d for r e i m b u r s e m e n t was used as a referent standard o f care. The evaluation o f quality o f care for b o t h systems i n c l u d e d an assessment o f a condition usually m a n a g e d surgically, colorectal cancer. Colorectal cancer was chosen b e c a u s e o f several characteristics. First, colorectal cancer is the most freq u e n t l y occurring n e o p l a s m in the United States, w i t h an incidence rate of 47 cases per 1 0 0 , 0 0 0 persons per yearg; in addition, it is associated with a heavy b u r d e n o f illness, with less than 50% surviving for five years following diagnosis. 10 Second, since the diagnosis of colorectal cancer is histopathologic, usually b y mucosal biopsy, diagnostic accuracy is high. Thus, n e w cases o f colorectal cancer are unlikely to be misclassified, and most require a surgical p r o c e d u r e . Finally, there is fairly wide a g r e e m e n t on a p p r o p r i a t e surgical managem e n t of the disease a c c o r d i n g to the stage o f the cancer, and this a p p r o a c h has c h a n g e d very little o v e r recent years.t ~ Therefore, a study of the process of care for patients n e w l y diagnosed as having colorectal c a n c e r was undertaken, and the results are r e p o r t e d here.
METHODS To provide a reasonable c o m p a r i s o n of the care of patients w i t h colorectal c a n c e r in HMOs, a similar g r o u p of FFS cases was selected for evaluation. Both groups w e r e restricted to incident cases w i t h a hospitalization during the study period. The quality o f care was judged for b o t h groups of patients.
Development of Quality of Care Criteria To provide a c o m p r e h e n s i v e appraisal of the care of patients w i t h colorectal cancer, b o t h o u t p a t i e n t and inpatient care w e r e evaluated. Physician advisory panels w e r e selected to d e v e l o p quality o f care criteria for the o u t p a t i e n t detection, surgical m a n a g e m e n t , and postoperative follow-up of the disease. Physician representation i n c l u d e d specialists in o n c o l o g y and epidemiology. In d e v e l o p i n g criteria for the a p p r o p r i a t e detection and m a n a g e m e n t of colorectal cancer, comm o n norms o f clinical care w e r e emphasized. Several activities w e r e c o n s i d e r e d particularly important: de-
,JOURNALOFGENERALINTERNALMEDICINE, Volume 5 (March/April), 1990
tection, use of time-intensive cognitive evaluations (e.g., medical history, physical examination), appropriate hospitalization, appropriate use of diagnostic tests and procedures, appropriate surgical management, and adequate follow-up after hospital discharge. Since colorectal cancer is usually managed surgically, the delay between presentation and diagnosis was considered particularly important. Because inappropriate delays may lead to patients' being more severely ill during hospitalization, two procedures were used to evaluate the stage of illness at the time of surgery. First, as a means of determining severity of illness, preoperative physiologic function was d e t e r m i n e d by using laboratory tests routinely p e r f o r m e d prior to surgery. Second, depending on local convention, several standard classification schemes were used to determine the histopathologic stage at the time of diagnosis: Dukes' classification, Asler-Coller modifications, and international TNM classifications. These classifications w e r e c o m b i n e d into a four-stage grouping. To determine the appropriateness of the diagnostic workup, patient records were abstracted from the office practices of the primary care physicians. Hospital records were abstracted to evaluate the preoperative care and surgical management. Both records were used to judge the quality of postoperative follow-up. Selection of Patients Study patients w e r e selected from each HMO plan from a list of all hospitalizations for colorectal cancer from the beginning of each plan enrollment date (range,January 1983 to May 1984) to March 31, 1986. For the study design, there w e r e two major issues for selecting a sample of HMOs. First, since the incidence of n e w cases is relatively low, even in an older population, HMOs with fewer than 2,000 enrollees w e r e eliminated because they w o u l d be unlikely to have sufficient cases for statistical power. Further, since medical care often follows conventional practice by area of the country, geographic location was important in the selection of plans for study. Thus, eight HMOs distributed among the four major geographic areas were selected for study. To secure the plans' participation and confid e n c e in the integrity of the study objectives, strict confidentiality was ensured, and specific HMO plans were not identified in the analysis. Since individual patient claims are inconsistently filed for HMO patients, discharge diagnoses were used for drawing the HMO sample. Therefore, HMO plans participating in the study were asked to provide patient lists with the following discharge diagnoses, or equivalent symptoms, for colorectal cancer: cancer of the rectum, sigmoid, cecum, or colon; mass or tumor of the rectum, sigmoid, cecum, or colon; admission for colectomy, hemicolectomy, or polypectomy; rectal or gastrointestinal bleeding; and abdominal pain, vomiting, obstruction, dehydration, obstipation, or severe constipation. All sampled cases were screened by nurse abstractors to ensure that they were incident cases, and
111
most of the ineligible cases represented recurrent disease. In the eight HMOs selected for study, there w e r e approximately 900 colorectal cancer cases screened, from w h i c h 150 eligible cases were successfully abstracted. For seven of the plans, there were 110 eligible cases from 196 drawn (59%). Because one plan was unable to provide a list of incident cases, all diagnoses and symptoms were used. From this one plan the proportion of new cases among the total sample of discharges with colorectal cancer was less than 10%. Samples of medical records of patients with colorectal cancer were drawn from Medicare claims files through the Medicare Automated Data Retrieval System for FFS hospitals located in geographic locales identical to that of the HMOs. The study period for drawing the sample was related to the start of Medicare HMOs in the area. Since the HMO sample was drawn from discharge diagnoses, and not Medicare claims, there were slight discrepancies in coding. FFS patients hospitalized for newly diagnosed colorectal cancer were selected if discharge diagnoses included malignant neoplasms of the colon, rectum, rectosigmoid junction, or anus. Patients with carcinoma in situ of either the colon or the rectum were also eligible for the study. The sample was screened to ensure that all cases w e r e incident cases, although the study patients did not have to be n e w to the practice. The FFS sample of cases was selected according to the n u m b e r of colorectal cancer patients discharged from July 1, 1983, to March 31, 1986. The hospitals were selected from a list of hospitals ranked by the n u m b e r of patients discharged with the diagnosis during the study period. Of the 70 hospitals identified for inclusion in the study, one had to be replaced because all patients were HMO patients, and three refused to participate. There were 969 potential colorectal cancer patients from the 66 hospitals left for study. Of the 969 patients identified, 180 (18.5 %) were considered eligible and their charts were abstracted. Sixty per cent of the ineligible cases were not new cases of colorectal cancer. The remaining ineligible cases were related to difficulties e n c o u n t e r e d in the ambulatory setting. Thirteen percent of the patients' physicians refused to participate; 9% of the practices could not find charts, or the sampled patient did not belong to the physician interviewed. Another 18% were not considered eligible for a variety of reasons. Registered nurses, specially trained for this study, were used to abstract both inpatient and outpatient medical records for all cases. Reliability was determined by circulating a sham chart to all nurse abstractors. Interrater reliability was assessed by having all ten abstractors and the principal investigator (SMR) review the sham record. Agreement among raters was determined by the kappa statistic) 2 A kappa value of > 0 . 7 5 is considered excellent agreement, 0.40 to 0.75 good agreement, and < 0 . 4 0 weak agreement. For this study, crude agreement between raters ranged from 0.90 to 0.93, with kappa values of 0.54 to 0.68.
Retchin, Brown, COLORECTALCANCERMANAGEMENTIN HMOs
11 ~.
TABLE 1 Comparisonof Personaland IllnessCharacteristicsfor Colorectal Cancer Patients, Accordingto Type of Plan
Fee-for-Service ( n = 180)
Health Maintenance Organization (n== 150)
Mean age at time of first visit to practice
70.1 years
72.3 years*
Mean length of time between first and most recent visits to practice
78.7 months
29.6 months*
Female
91 (50.6%)
62 (41.3%)
White
156 (86.7%)
143 (95.2%)*
Married
111 (62.7%)
98 (67.1%)
Symptoms at presentationt Anemia Gastrointestinal bleeding Altered bowel habit Weight loss Abdominal pain
33 (18.3°/0) 84 (46.7%) 51 (28.3%) 38 (21.1%) 58 (32.2°/0)
22 83 36 23 40
(14.7%) (55.3%) (24.0%) (15.3%) (26.7%)
*p
