- Email: [email protected]
Outcome states of colorectal cancer: identification and description using patient focus groups 1
THE AMERICAN JOURNAL OF GASTROENTEROLOGY Copyright © 1998 by Am. Coll. of Gastroenterology Published by Elsevier Science Inc.
Vol. 93, No. 9, 1998 ISSN 0002-9270/98/$19.00 PII S0002-9270(98)00350-5
Outcome States of Colorectal Cancer: Identification and Description Using Patient Focus Groups Reid M. Ness, M.D., M.P.H., Ann Holmes, Ph.D., Robert Klein, M.S., James Greene, Ph.D., and Robert Dittus, M.D., M.P.H. Division of Gastroenterology and Hepatology, Department of Medicine, Indiana University School of Medicine, Indiana University, Indianapolis; the Regenstrief Institute for Health Care, and the Richard L. Roudebush VA Medical Center, Indianapolis; School of Public and Environmental Affairs, Indiana University, Indianapolis; Bowen Research Center, Indiana University, Indianapolis; Wishard Memorial Hospital, Indianapolis, Indiana; and Division of General Internal Medicine, Department of Medicine, Vanderbilt University Medical Center, Nashville, Tennessee
INTRODUCTION
Objective: Utilities for the outcome states of colorectal cancer must be measured to evaluate the cost-utility of screening and surveillance strategies for this disease. We sought to identify these outcome states, define their associated areas of morbidity, and construct representative descriptions of them for use in a utilities assessment instrument. Methods: We identified candidate colorectal cancer outcome states based on a review of the literature and interviews with health care professionals. We organized patient focus groups from each of the candidate outcome states to examine their homogeneity and define their associated areas of morbidity. After analyzing the focus group transcripts, we identified and described outcome states of colorectal cancer for future incorporation into a utilities assessment instrument. Results: Six candidate outcome states of colorectal cancer were identified based on disease stage and location at diagnosis. Thirty-eight patients then participated in six focus groups. Analysis of the focus group transcripts revealed seven areas of morbidity associated with colorectal cancer. These areas included problems with social interaction and cognition, fear of cancer recurrence, pain, fatigue, changes in bowel habits, and sexual dysfunction. Based on differences in the intensity and frequency of the symptoms reported in each of these areas, seven distinct outcome states of colorectal cancer were identified and described. Conclusion: Clinically distinct outcome states of colorectal cancer are determined by the stage and location of the cancer at the time of diagnosis. Descriptions of these outcome states were created using data collected from patient focus groups. These descriptions can be incorporated into a utilities assessment instrument. (Am J Gastroenterol 1998;93:1491–1497. © 1998 by Am. Coll. of Gastroenterology)
During 1997 in the United States, an estimated 131,200 persons were diagnosed with colorectal cancer (CRC) and 54,900 people died from this disease (1). As there is no effective treatment for advanced disease, efforts to reduce morbidity and mortality associated with CRC have focused on disease prevention and early detection. Indeed, the removal of colonic polyps, a presumed precursor of CRC, has been shown to decrease the incidence of CRC (2, 3). Likewise, screening for CRC using fecal occult blood tests has been proved effective in decreasing mortality secondary to CRC (4 – 8). Which of the proposed CRC screening and surveillance strategies are most cost-effective continues to be debated. A common method for evaluating cost-effectiveness is decision analysis, a methodology that allows the comparison of alternative clinical practices using available data. As currently applied in clinical medicine, decision analysis uses quality-adjusted life-years (QALYs) as its outcome measure (9, 10). To calculate QALYs, the relevant, disease-specific outcome states and individuals’ perceptions of the relative morbidity associated with each state must be determined (11). Scant published data exist as to what constitute the important outcome states of CRC or how patients perceive the relative morbidity of each outcome state. Prior work in this area either focused on only one aspect of these states (e.g., having a colostomy) or sought to combine all of the outcomes of CRC into a single state (12, 13). These data are inadequate for informing a decision-analytic model with multiple outcome states. The construction of such a model to evaluate the relative cost-effectiveness of competing CRC surveillance strategies requires the determination of individuals’ preferences (utilities) for the outcome states of CRC that the model would employ. Therefore, in this project, we sought to: 1) identify the important outcome states of CRC; 2) obtain valid information concerning the important sources of morbidity for each of these states using
Supported by a Fellowship Training Grant from the Regenstrief Institute for Health Care. Work conducted at: the Regenstrief Institute for Health Care and the Indiana University School of Medicine. Received Dec. 5, 1997; accepted June 1, 1998. 1491
1492
NESS et al.
patient focus groups; and 3) construct representative descriptions of these outcome states for use in a utilities assessment instrument. MATERIALS AND METHODS Outcome state identification After reviewing the literature describing the outcomes of CRC and its treatment, we met with health care professionals involved in the treatment of patients with CRC at the Indiana University Medical Center and Wishard Memorial Hospital in Indianapolis. These professionals included oncologists, gastroenterologists, general surgeons, radiation oncologists, and oncology nurses involved in the diagnosis, treatment, and chronic care of patients with CRC. We asked these professionals to list the major determinants of morbidity in patients with CRC and how they would classify patients into important outcome states. Based on our review of the literature and these meetings, we were able to identify and label six distinct candidate outcome states of CRC. Focus groups We then organized patient focus groups. The purpose of the focus groups was to examine the homogeneity of the identified outcome states and to define the areas of morbidity associated with each state. Patient records from the Multi-Disciplinary GI Cancer clinic, the Surgical Out-Patient clinic, and the Radiation Oncology clinic at the Indiana University Medical Center were reviewed looking for patients previously treated for CRC. One hundred and fourteen patients were identified who had previously been treated for CRC. Patients were excluded if they were unable to speak English (N 5 2), had moved out of Indiana (N 5 1), were being, or had been, treated for another primary cancer (N 5 4), had comorbid illnesses likely to be fatal within 5 yr (N 5 4), had metastatic CRC to the liver treated with hepatic resection (N 5 12), or had stage III CRC not treated with adjuvant chemotherapy (N 5 2). The remaining 89 eligible focus group participants were first contacted by mailed invitation, then by phone. Following agreement to participate, respondents were scheduled for focus groups composed entirely of subjects with the same candidate outcome state as themselves. Focus groups in this study were composed of four to nine individuals with CRC, a facilitator, a recorder, and a support person. Each group met at the Regenstrief Institute for Health Care in either June or July 1996 with sessions lasting approximately 90 min. All participants gave informed consent in the manner approved by the Indiana University– Purdue University at Indianapolis (IUPUI) Institutional Review Board and were reimbursed $25 for their participation. The focus group scripts were simple, consisting of broad questions concerning the effects of CRC on several dimensions of health: social functioning, emotional functioning, cognitive functioning, physical pain, and nonpain physical
AJG – Vol. 93, No. 9, 1998 problems. Each session was audiotaped for later transcription. The focus group transcripts were analyzed by two independent reviewers using group-by-question grids (14). The objective of this analysis was to determine what, if any, areas of morbidity were reported in each dimension of health. Each area of morbidity was identified and characterized as to its stated intensity and/or frequency of occurrence. The reviewers then determined how many participants noted each morbidity and the modal level of intensity and/or frequency of occurrence they reported. After independently analyzing the transcripts, the two reviewers met to discuss their conclusions and to reach a consensus concerning the characterization of each identified area of morbidity. If a consensus could not be reached, a third independent reviewer was engaged to arbitrate. Description construction After the focus group transcripts were analyzed, the results were used to redefine the important outcome states of CRC and to help construct their descriptions. Those morbidities reported most frequently by the focus group participants within each dimension of health were incorporated into the descriptions. When the data obtained from a focus group indicated a possibly bimodal response in one or more dimensions of health, the candidate outcome state in question was split into two new states. After the state was split, the published frequency and severity of CRC symptoms and treatment-related morbidities for that candidate state was reviewed in an attempt to confirm the observation. Likewise, when the data indicated that two or more focus groups were nearly identical in their responses, they were combined. Again, after the states were combined, the published literature was reviewed in an attempt to confirm the observation (13, 15–23). The descriptions were constructed in a tabular fashion to allow greater comparability between different outcome states. Each state was described in the short term (first 18 months) and long term (after 18 months), as reported morbidities during and immediately after therapeutic interventions were substantially different from those reported thereafter. Each outcome state was assigned a duration of 30 yr except for those involving metastatic CRC, which were given a duration of only 18 months. Statistical analysis Demographic comparisons between participants and nonparticipants were analyzed using Fisher’s exact test or x2 for categorical data and t tests for continuous data. The statistical package employed was SPSS for Windows, release 7.5.1 (SPSS Inc., Chicago, IL). RESULTS Based on our literature review and discussions with health care professionals, we created a list of six candidate outcome states of CRC (see Table 1) (13, 15–23). We
AJG – September 1998
COLORECTAL CANCER, PATIENT FOCUS GROUPS
TABLE 1 Candidate Outcome States of CRC State 1 2 3 4 5 6
Description Stage I rectal or stage I/II colon Ca treated with resection only Stage III colon Ca treated with resection and chemotherapy Stage II/III rectal Ca treated with resection/chemotherapy/radiation therapy Stage II/III rectal Ca treated with resection/chemotherapy/radiation therapy/permanent ostomy Stage IV metastatic CRC Stage IV locally unresectable CRC
CRC, colorectal cancer; Ca, cancer. TABLE 2 Characteristics of Participants and Nonparticipants in Colorectal Cancer Focus Groups† Characteristic Age* Sex: M/F Race Caucasian African-American Ethnicity Hispanic Non-Hispanic Years since diagnosis* Outcome state 1 2 3 4 5 6
Interviewed (n 5 38)
Not Interviewed (n 5 51)
62 6 13 14/24
65 6 12 23/28
34 4
50 1
1 37 261
0 51 261
6 6 7 4 8 7
8 8 1 5 18 11
* Plus-or-minus values are means 6 SD. † There are no significant differences among the groups for any of the parameters measured.
concluded that morbidity in patients with CRC is related mainly to the treatment modality employed, which is itself usually determined by the location and stage of the cancer at diagnosis (13, 15, 16). The cancer itself is usually a minimal direct source of morbidity except in the setting of metastatic or locally unresectable disease. We assembled focus groups for each of the candidate outcome states of CRC. Of the 89 patients invited to participate, 30 refused, 11 had died, and 10 could not be reached by mail or telephone. A comparison of the 38 participants to nonparticipants is presented in Table 2. No statistically significant differences were found between participants and nonparticipants. We conducted the analysis of the focus group transcripts by examining participants’ responses to broad, open-ended questions concerning the impact of their disease and its treatment on their social functioning, cognitive functioning, emotional health, pain, and nonpain physical problems. Each of the two reviewers analyzing the transcripts created summary statements for each broad health dimension for
1493
each focus group. The language employed in these summary statements was standardized as much as possible, using the following conventions for the key descriptive words employed. “Mild,” “moderate,” and “severe” were used to describe intensity, “mild” being less intense than “moderate,” which is less intense than “severe.” “Occasional” and “frequent” were used to describe the frequency of occurrence, “occasional” meaning less than half the time and “frequent” meaning more than half the time. Finally, “some” and “most” were used to describe the frequency of a specific response among the focus group participants, “some” meaning less than half and “most” meaning more than half of the participants reported this problem. This analysis uncovered one or more specific areas of morbidity within each of the broad health dimensions explored for each focus group. These areas of morbidity included: 1) problems with social interaction; 2) problems with cognition; 3) fears of cancer recurrence; 4) pain; 5) fatigue; 6) changes in bowel habits; and 7) sexual dysfunction. Although these specific areas of morbidity varied somewhat between groups, considerable overlap was noted. After preparing these summary statements, the reviewers compared their analyses. No significant areas of disagreement were discovered. When the focus group responses were analyzed, it became apparent that outcome state “2” (stage III colon cancer treated with resection and chemotherapy) was bimodal as to the number and intensity of symptoms reported by participants while on chemotherapy. Approximately half of the participants complained of significant nausea, painful mouth sores, and diarrhea that, in some cases, required hospitalization. The other half of the participants reported only minimal symptoms while on chemotherapy. This result is in agreement with the published side effect profiles of both 5FU/levamisole and 5FU/leukovorin therapy that report significant nausea, mucositis, and diarrhea in only 40 – 80% of patients (17, 18). Analysis of the focus groups’ responses revealed little if any difference between the outcome state described for participants dying of metastatic disease (state 5) versus that described for participants dying of unresectable or locally recurrent disease (state 6). The health care professionals we queried proposed that those patients with locally unresectable disease would complain of more pain than those with purely metastatic disease because of tumor invasion of large nerve plexi. Although many participants in both states 5 and 6 noted pain, they all reported its control with medications. States 5 and 6 also appeared homogenous, in that a majority of the participants in both of these groups reported that their most troublesome symptom was severe fatigue that limited their ability to care for themselves and their families. Few of the participants in the other focus groups reported severe chronic fatigue. Because state 2 appeared heterogeneous and states 5 and 6 homogenous, the listed outcome states were revised. State 2 was split between patients having and those not having
1494
NESS et al.
AJG – Vol. 93, No. 9, 1998
TABLE 3 Revised Outcome States of Colorectal Cancer State A B C D E F G
Description Stage I rectal or stage I/II colon Ca treated with resection only Stage III colon Ca treated with resection and chemotherapy without significant side effects Stage III colon Ca treated with resection and chemotherapy with significant side effects Stage II/III rectal Ca treated with resection/chemotherapy/radiation therapy Stage II/III rectal Ca treated with resection/chemotherapy/radiation therapy/permanent ostomy Stage IV metastatic/unresectable disease without ostomy Stage IV metastatic/unresectable disease with ostomy
Ca, cancer.
significant side effects of chemotherapy. Similarly, states 5 and 6 were combined into a single outcome state of patients with terminal (stage IV) CRC. We became concerned about the possible interactive effect on future utilities assessment of having stage IV CRC and a permanent ostomy from previous therapy present in the same outcome state. Even though this outcome was felt to represent a very small patient population, we thought that incorporation of this combination state might provide insight into how an ostomy influences people’s utilities for outcome states of CRC. The revised outcome states of CRC are listed in Table 3. The final task was to incorporate the information gained from the focus groups into brief descriptions of each selected outcome state of CRC. The first objective was to provide enough data to clearly delineate each outcome state while not overwhelming the reader with excessive information. The second objective was to present the data for each outcome state in a form facilitating comparison between the descriptions for different states. Key descriptive words and short phrases were used to convey information on each specific area of morbidity using the conventions presented earlier. The key descriptive words employed to describe each specific area of morbidity or symptom in the short term (0 –18 months) and long term (.18 months) can be found in Tables 4 and 5. Except for the stated morbidities, persons in each described outcome state were considered to be unimpaired. A tabular format eased comparison of one state to another. An example of the final description format appears in Table 6. DISCUSSION We identified clinically important outcome states of CRC, used focus groups to obtain information concerning the important sources of morbidity in each of these states, and constructed descriptions of these outcome states using this information. The outcome state descriptions can be incorporated into a utilities assessment instrument to determine population and/or individual utilities for the identified outcome states of CRC. The importance of identifying the outcome states of CRC
rests on the need for these states by decision analysis, a methodology commonly used to examine the cost-effectiveness of health care (10). A decision analysis combines observational and experimental data into a model that evaluates the expected outcomes of various diagnostic or treatment strategies (10). The primary outcome measure is usually the marginal improvement in cost per quality-adjusted life-year saved (QALYs). QALYs incorporate changes in survival and in health-related quality of life into a single measure (9 –11, 24). To be consistent with this construct, quality weights should be measured by, or transformed into, an interval scale; that is, the method of measurement should be one in which the ratio of differences between values is meaningful. The scale most commonly used employs 0 (immediate death) and 1 (perfect health) as end points. The quality weights used should also be based on utilities for the health outcome states (9 –11). Health status scales that are not utility-based or do not have interval characteristics, such as the Sickness Impact Profile (SIP) or the Medical Outcomes Study Short-Form Health Survey (SF36), are not suitable for cost-effectiveness analysis (9 –11, 25). Methods based on decision theory, such as the standard gamble, are utility-based and have interval characteristics, making them appropriate methods for obtaining quality weightings for economic analysis (9 –12, 24, 25). Although we identified and characterized the important outcome states of CRC, there are other health states associated with CRC screening and surveillance that we did not assess. When considering outcome states for incorporation into a cost-utility model, they should be of sufficient duration and frequency to potentially affect the model’s conclusions. For example, the health state associated with a colonoscopic perforation was felt to be so uncommon that any change in long-term utility would probably not affect the model’s conclusions. Some health states, such as the one associated with an asymptomatic patient after a false-positive screening test, we did not consider to be different enough from the baseline health state to warrant a separate utility assessment. Such assumptions can be tested in the cost-utility model itself through sensitivity analysis. This is accomplished by varying the input utility for the outcome state in question and examining whether the model’s output or conclusions are significantly altered. The source from which utilities for the outcome states of any given clinical model are obtained is dependent on the viewpoint being employed in the cost-effectiveness analysis. Commonly employed viewpoints include that of society as a whole, a third-party payer, a patient decision-maker, and a patient already in a given outcome state. Except when the patient is already in the outcome state being evaluated, utilities must be assessed on a sample of persons who are informed only by descriptions of the outcome states. This makes the construction of a valid, understandable description of the outcome state being evaluated an important step in utilities measurement. Often, physician experts and the medical literature have
AJG – September 1998
COLORECTAL CANCER, PATIENT FOCUS GROUPS
1495
TABLE 4 Key Words for Each Specific Area of Morbidity: Short Term
ATTRIBUTES Fatigue Diarrhea Fecal urgency Fecal incontinence Sexual dysfunction Pain Cognitive problems Social interaction problems Fear of disease recurrence
STATES A
B
C
D
E
F
G
none none mild occasional none none none none
mild none mild occasional none none none some
moderate occasional mild occasional occasional mild-moderate none some
moderate frequent severe frequent frequent moderate none most
moderate occasional ostomy ostomy frequent moderate none most
severe occasional mild occasional frequent severe frequent most
severe occasional ostomy ostomy frequent severe frequent most
none
severe
severe
severe
severe
severe
severe
TABLE 5 Key Words for Each Specific Area of Morbidity: Long Term
ATTRIBUTES Fatigue Diarrhea Fecal urgency Fecal incontinence Sexual dysfunction Pain Cognitive problems Social interaction problems Fear of disease recurrence
STATES A
B
C
D
none none mild occasional none none none none
none none mild occasional none none none none
none none mild occasional none none none none
none none severe frequent occasional none none some
none
moderate
moderate
moderate
E
F
G
none none ostomy ostomy frequent none none some
NA NA NA NA NA NA NA NA
NA NA NA NA NA NA NA NA
moderate
NA
NA
NA, Not applicable. TABLE 6 Description of Person in Outcome State D (Duration of State: 30 yr) Problem
Short Term (first 18 mo.)
Tiredness and weakness Bowel habits
● moderate ● frequent diarrhea ● severe fecal urgency (urgent need to pass a bowel movement) ● frequent fecal incontinence (having a bowel movement in one’s pants) ● frequent ● moderate but controllable with medication ● none ● cutbacks in most social activities ● severe fear of disease recurrence
Sexual problems Pain Thinking problems Social problems Emotional problems
Long Term (rest of life) ● none ● severe fecal urgency (urgent need to pass a bowel movement) ● frequent fecal incontinence (having a bowel movement in one’s pants)
● occasional ● none ● none ● cutbacks in some social activities ● fear of disease recurrence decreases with time
Note: This patient has no other problems (i.e., hearing, speech, mobility, dexterity, etc., are all normal).
provided all the information used to inform descriptions of outcome states for utilities assessment. This approach has been criticized for not truly reflecting the patient’s viewpoint on his or her illness (25). Patient focus groups in which a broad spectrum of individuals are sampled are an economical and relatively quick way of gathering a wealth of patient opinion on a subject. Focus groups have been widely used in market research to ascertain consumer opinion on
various topics and products (14). Focus groups have also been used previously to inform outcome state descriptions for utilities assessment and have been shown to provide invaluable information regarding aspects of quality of life and their relative importance (26). In this study, the assumption was made that the morbidities of importance to the patient were not already known. Focus groups were used to provide this information within a basic framework drawn
1496
NESS et al.
from a literature review and discussions with health care professionals. This process of informing the outcome state descriptions using patient focus groups provides those descriptions with both face and content validity (25, 26). The patient focus groups provided information significantly different than that provided by the literature review and our discussions with local health care professionals. This information prompted several changes in the composition of the revised list of CRC outcome states. The focus groups revealed that the side effects of chemotherapy clustered predominantly in a subpopulation of participants, whereas the remainder had no complaints. This led to the splitting of outcome state 2 on the basis of the presence or absence of chemotherapy side effects. The focus groups also revealed that differences in pain severity between patients with locally unresectable/recurrent disease and those with metastatic disease that had been suggested by our discussions with health care professionals were either nonexistent or were dominated by the severity of other concurrent symptoms. Focus groups are susceptible to both sampling bias and to various systematic biases related to the interview process (14). In this study, the focus groups were drawn from a predominantly white, middle-class population. The results should be generalizable to similar populations but may not be secondary to sampling bias. We have been unable to identify other focus group studies that specifically address race or socioeconomically dependent variation in patients’ assessments of their disease-specific outcome states. However, there are several examples of focus groups uncovering race and/or socioeconomic differences in patients’ attitudes and concerns about such diverse subjects as breast cancer screening (27), cervical cancer screening (28), prostate cancer screening (29), depression (30), and exercise (31). Future research should be done to delineate any possible race or socioeconomic status-related differences in patients’ descriptions of their colorectal cancer outcome states. Focus groups are not surveys and are not meant to represent definitively the views of the population from which they are drawn (14). The strength of focus groups rests in their simplicity, ease, and affordability (14). They are well suited to the task of rapidly obtaining information in a setting in which little or no pre-existing data are either available or assumed. Quality of life is a multidimensional construct. Although the number of components in any given conceptualization of quality of life may vary, the minimum dimensions should include: 1) physical (i.e., symptoms, pain); 2) emotional; 3) social; and 4) mental (32–37). Two additional domains commonly included in this list are the vocational or role performance and the spiritual (33, 34). In general, the generic, utilities-based, health state classification systems (e.g., Health Utilities Index, EuroQol, and the Quality of Well-Being Scale) include physical, social, cognitive, and emotional function dimensions, along with domains that map to role performance (37). Information on patients’ role
AJG – Vol. 93, No. 9, 1998 performance was gathered in the course of the focus group sessions, but because of heterogeneity in the line of work and employment status of our participants, role performance was not included in the state descriptions. Much of the reported vocational morbidity can be accounted for by areas of physical morbidity included in the descriptions. This allows the readers to decide whether or not they could work in a given state. Spiritual function was not included because of the absence of adequate measures. The areas of morbidity listed under the nonpain physical problems section (fatigue, bowel dysfunction, sexual dysfunction) were those most commonly reported across the whole study population, although they were not relevant to every outcome state. These symptoms are also those most often reported in clinical studies of CRC treatment and outcome. Bowel dysfunction is described by a majority of the patients treated with distal colonic resection and/or chemotherapy. Likewise, limiting fatigue is reported by a majority of patients with metastatic disease (15, 16, 23). There is currently little consensus as to how health state descriptions should be structured for use in utilities assessment. The length and amount of information conveyed by a health state description has been shown by some to affect assessed utility values, although this finding has been inconsistent (25, 32). Knowing that research has shown that humans can process only about five to nine pieces of information concurrently, it may be hypothesized that long narrative descriptions could overload the cognitive capacity of the participant (32, 38). Alternatively, although a brief point form description may improve retention of information, it may provide insufficient detail to characterize a disease state adequately (25). In this study, the outcome state descriptions used key descriptive words and short phrases organized in a tabular fashion. This decision was made for three reasons: 1) to simplify data presentation; 2) to ease comparisons between states; and 3) to avoid issues of labeling and framing inherent to narrative descriptions (25, 32). We believe the chosen format allows and facilitates comparisons among the described outcome states. To limit the number of state descriptions, we chose to describe only the “typical” health state experienced by a person in each outcome state defined by stage at diagnosis and treatment. Specifically, we incorporated into each outcome state description only the symptom levels most commonly reported for each area of morbidity by the participants in the focus group(s) associated with that state. Furthermore, the “typical” health state derived from each focus group for the first 12–18 months following CRC diagnosis (short term) was substantially different from the state derived for the period thereafter (long term). We chose to combine these two states into a single, holistic, health state description for each outcome. These constructs are valid because utilities measured for them incorporate the person’s actual preferences for both the short-term and long term aspects of the described outcome states (39). We conclude that, in patients without terminal illness, the
AJG – September 1998
COLORECTAL CANCER, PATIENT FOCUS GROUPS
majority of the short and long term morbidity associated with CRC is a consequence of disease treatment. Because therapy is dependent on the stage and location of the cancer at the time of diagnosis, the important outcome states of CRC are also determined by the stage and location of the cancer at the time of diagnosis. Descriptions of these outcome states have been created for incorporation into a utilities assessment instrument that can be used to measure individuals’ utilities for these states. The measured utilities can then be incorporated into a decision-analytic model to determine the cost-utility of various diagnostic and therapeutic strategies for CRC.
14. 15.
16.
17.
18. 19.
ACKNOWLEDGMENTS 20.
The authors would like to acknowledge and thank Drs. Douglas Rex and David Crabb for their help in the planning of this project. We would also like to thank Dr. Patrick Loehrer and all of the physicians and nurses associated with the Multi-Disciplinary GI Cancer clinic at the Indiana University Medical Center for their cooperation and assistance in patient recruitment.
21.
22.
23. Reprint requests and correspondence: Reid M. Ness, M.D., M.P.H., Wishard Memorial Hospital, OPW 2005, 1001 West Tenth Street, Indianapolis, IN 46202.
REFERENCES 1. Parker SL, Tong T, Bolden S, et al. Cancer Statistics, 1997. CA Cancer J Clin 1997;47:5–27. 2. Winawer SJ, Zauber AG, Ho MN, et al. Prevention of colorectal cancer by colonoscopic polypectomy. N Engl J Med 1993;329:1977– 87. 3. Winawer SJ, Zauber AG, O’Brien MJ, et al. Randomized comparison of surveillance intervals after colonoscopic removal of newly diagnosed adenomatous polyps. N Engl J Med 1993;328:901– 6. 4. Mandel JS, Bond JH, Church TR, et al. Reducing mortality from colorectal cancer by screening for fecal occult blood. Minnesota colon cancer study. N Engl J Med 1993;328:1365–71. 5. Kronborg O, Fenger C, Worm J, et al. Causes of death during the first 5 years of a randomized trial of mass screening for colorectal cancer with fecal occult blood test. Scand J Gastroenterol 1992;27:47–52. 6. Winawer SJ, Flehinger BJ, Schottenfeld D, et al. Screening for colorectal cancer with fecal occult blood testing and sigmoidoscopy. J Natl Cancer Inst 1993;85:1311– 8. 7. Hardcastle JD, Armitage NC, Chamberlain J, et al. Fecal occult blood screening for colorectal cancer in the general population. Results of a controlled trial. Cancer 1986;58:397– 403. 8. Kewenter J, Brevinge H, Engaras B, et al. Results of screening, rescreening, and follow-up in a prospective randomized study for detection of colorectal cancer by fecal occult blood testing. Results for 68,308 subjects. Scand J Gastroenterol 1994;29:468 –73. 9. Russell LB, Gold MR, Siegel JE, et al. The role of cost-effectiveness analysis in health and medicine. JAMA 1996;276:1172–7. 10. Weinstein MC, Siegel JE, Gold MR, et al. Recommendations of the panel on cost-effectiveness in health and medicine. JAMA 1996;276: 1253– 8. 11. Kaplan RM, Feeny D, Revicki DA. Methods for assessing relative importance in preference based outcome measures. Quality Life Res 1993;2:467–73. 12. Boyd NF, Sutherland HJ, Heasman KZ. Whose utilities for decision analysis? Med Decis Making 1990;10:58 – 67. 13. Sprangers MAG, Taal BG, Aaronson NK, et al. Quality of life in
24. 25. 26. 27.
28. 29.
30.
31. 32.
33. 34.
35. 36.
37. 38. 39.
1497
colorectal cancer: Stoma vs. nonstoma patients. Dis Colon Rectum 1995;38:361–9. Stewart DW, Shandasani PN. Focus groups theory and practice. Newbury Park, CA: Sage Publications, 1990. Cohen AM, Minsky BD, Schilsky RL. Cancer of the colon. In: De Vita VT Jr, Hellman S, Rosenberg SA, eds. Cancer: Principles and practice of oncology. Philadelphia: Lippincott-Raven, 1997:1144 –97. Cohen AM, Minsky BD, Schilsky RL. Cancer of the rectum. In: De Vita VT Jr, Hellman S, Rosenberg SA, eds. Cancer: Principles and practice of oncology. Philadelphia: Lippincott-Raven, 1997: 1197–1234. Moertel CG, Fleming TR, MacDonald JS, et al. Levamisole and fluorouracil for adjuvant therapy of resected colon carcinoma. N Engl J Med 1990;322:352– 8. IMPACT Investigators. Efficacy of adjuvant fluorouracil and folinic acid in colon cancer. Lancet 1995;345:939 – 44. Enker WE. Potency, cure, and local control in the operative treatment of rectal cancer. Arch Surg 1992;127:1396 –1402. van Driel MF, Weymar Schultz WCM, van de Weil HBM, et al. Female functioning after radical surgical treatment of rectal and bladder cancer. Euro J Surg Oncol 1993;19:183–7. Krook JE, Moertel CG, Gunderson LL, et al. Effective surgical adjuvant therapy for high-risk rectal carcinoma. N Engl J Med 1991;324: 709 –15. O’Connell MJ, Martenson JA, Wieand HS, et al. Improving adjuvant therapy for rectal cancer by combining protracted-infusion fluorouracil with radiation therapy after curative surgery. N Engl J Med 1994;331: 502–7. Glimelius B, Hoffman K, Graf W, et al. Quality of life during chemotherapy in patients with symptomatic advanced colorectal cancer. Cancer 1994;73:556 – 62. Weeks J. Assessing utilities for health states. Proc Am Soc Clin Oncol 1997;16:353–5. Smith R, Dobson M. Measuring utility values for QALY’s: Two methodologic issues. Health Economics 1993;2:349 –55. Hall J, Gerard K, Salkeld G, et al. A cost-utility analysis of mammography screening in Australia. Soc Sci Med 1993;34:993–1004. Danigelis NL, Roberson NL, Worden JK, et al. Breast screening by African-American women: Insights from a household survey and focus groups. Am J Prev Med 1995;11:311–7. Clark DO. Age, socioeconomic status, and exercise self-efficacy. Gerontol 1996;36:157– 64. Robinson SB, Ashley M, Haynes MA. Attitude of African-Americans regarding prostate cancer clinical trials. J Comm Health 1996;21:77– 87. Cooper-Patrick L, Powe NR, Jenckes MW, et al. Identification of patient attitudes and preferences regarding treatment of depression. J Gen Intern Med 1997;12:431– 8. Jennings KM. Getting a Pap smear: Focus group responses of African American and Latina women. Onc Nurs Forum 1997;24:827–35. Froberg DG, Kane RL. Methodology for measuring health state preferences—I: Measurement strategies. J Clin Epidemiol 1989;42:345– 54. Eberst RM. Defining health: A multidimensional model. JOSH 1984; 54:99 –104. Rowland JH. Outcomes assessment: Cancer-specific quality-of-life measures— beyond the research setting. Proc Am Soc Clin Oncol 1997;16:342–9. Fletcher A. Quality-of-life measurements in the evaluation of treatment: proposed guidelines. Br J Clin Pharmac 1995;39:217–22. Fitzpatrick R, Fletcher A, Gore S, et al. Quality of life measures in health care. I: Applications and issues in assessment. Br Med J 1992; 305:1074 –7. Patrick DL, Erickson P. Health status and health policy. New York: Oxford University Press, 1993. Miller GA. The magical number seven plus or minus two: Some limits on our capacity to process information. Psychol Rev 1956;63:81–97. Froberg DG, Kane RL. Methodology for measuring health state preferences—II: Scaling methods. J Clin Epidemiol 1989;42:459 –71.
Vol. 93, No. 9, 1998 ISSN 0002-9270/98/$19.00 PII S0002-9270(98)00350-5
Outcome States of Colorectal Cancer: Identification and Description Using Patient Focus Groups Reid M. Ness, M.D., M.P.H., Ann Holmes, Ph.D., Robert Klein, M.S., James Greene, Ph.D., and Robert Dittus, M.D., M.P.H. Division of Gastroenterology and Hepatology, Department of Medicine, Indiana University School of Medicine, Indiana University, Indianapolis; the Regenstrief Institute for Health Care, and the Richard L. Roudebush VA Medical Center, Indianapolis; School of Public and Environmental Affairs, Indiana University, Indianapolis; Bowen Research Center, Indiana University, Indianapolis; Wishard Memorial Hospital, Indianapolis, Indiana; and Division of General Internal Medicine, Department of Medicine, Vanderbilt University Medical Center, Nashville, Tennessee
INTRODUCTION
Objective: Utilities for the outcome states of colorectal cancer must be measured to evaluate the cost-utility of screening and surveillance strategies for this disease. We sought to identify these outcome states, define their associated areas of morbidity, and construct representative descriptions of them for use in a utilities assessment instrument. Methods: We identified candidate colorectal cancer outcome states based on a review of the literature and interviews with health care professionals. We organized patient focus groups from each of the candidate outcome states to examine their homogeneity and define their associated areas of morbidity. After analyzing the focus group transcripts, we identified and described outcome states of colorectal cancer for future incorporation into a utilities assessment instrument. Results: Six candidate outcome states of colorectal cancer were identified based on disease stage and location at diagnosis. Thirty-eight patients then participated in six focus groups. Analysis of the focus group transcripts revealed seven areas of morbidity associated with colorectal cancer. These areas included problems with social interaction and cognition, fear of cancer recurrence, pain, fatigue, changes in bowel habits, and sexual dysfunction. Based on differences in the intensity and frequency of the symptoms reported in each of these areas, seven distinct outcome states of colorectal cancer were identified and described. Conclusion: Clinically distinct outcome states of colorectal cancer are determined by the stage and location of the cancer at the time of diagnosis. Descriptions of these outcome states were created using data collected from patient focus groups. These descriptions can be incorporated into a utilities assessment instrument. (Am J Gastroenterol 1998;93:1491–1497. © 1998 by Am. Coll. of Gastroenterology)
During 1997 in the United States, an estimated 131,200 persons were diagnosed with colorectal cancer (CRC) and 54,900 people died from this disease (1). As there is no effective treatment for advanced disease, efforts to reduce morbidity and mortality associated with CRC have focused on disease prevention and early detection. Indeed, the removal of colonic polyps, a presumed precursor of CRC, has been shown to decrease the incidence of CRC (2, 3). Likewise, screening for CRC using fecal occult blood tests has been proved effective in decreasing mortality secondary to CRC (4 – 8). Which of the proposed CRC screening and surveillance strategies are most cost-effective continues to be debated. A common method for evaluating cost-effectiveness is decision analysis, a methodology that allows the comparison of alternative clinical practices using available data. As currently applied in clinical medicine, decision analysis uses quality-adjusted life-years (QALYs) as its outcome measure (9, 10). To calculate QALYs, the relevant, disease-specific outcome states and individuals’ perceptions of the relative morbidity associated with each state must be determined (11). Scant published data exist as to what constitute the important outcome states of CRC or how patients perceive the relative morbidity of each outcome state. Prior work in this area either focused on only one aspect of these states (e.g., having a colostomy) or sought to combine all of the outcomes of CRC into a single state (12, 13). These data are inadequate for informing a decision-analytic model with multiple outcome states. The construction of such a model to evaluate the relative cost-effectiveness of competing CRC surveillance strategies requires the determination of individuals’ preferences (utilities) for the outcome states of CRC that the model would employ. Therefore, in this project, we sought to: 1) identify the important outcome states of CRC; 2) obtain valid information concerning the important sources of morbidity for each of these states using
Supported by a Fellowship Training Grant from the Regenstrief Institute for Health Care. Work conducted at: the Regenstrief Institute for Health Care and the Indiana University School of Medicine. Received Dec. 5, 1997; accepted June 1, 1998. 1491
1492
NESS et al.
patient focus groups; and 3) construct representative descriptions of these outcome states for use in a utilities assessment instrument. MATERIALS AND METHODS Outcome state identification After reviewing the literature describing the outcomes of CRC and its treatment, we met with health care professionals involved in the treatment of patients with CRC at the Indiana University Medical Center and Wishard Memorial Hospital in Indianapolis. These professionals included oncologists, gastroenterologists, general surgeons, radiation oncologists, and oncology nurses involved in the diagnosis, treatment, and chronic care of patients with CRC. We asked these professionals to list the major determinants of morbidity in patients with CRC and how they would classify patients into important outcome states. Based on our review of the literature and these meetings, we were able to identify and label six distinct candidate outcome states of CRC. Focus groups We then organized patient focus groups. The purpose of the focus groups was to examine the homogeneity of the identified outcome states and to define the areas of morbidity associated with each state. Patient records from the Multi-Disciplinary GI Cancer clinic, the Surgical Out-Patient clinic, and the Radiation Oncology clinic at the Indiana University Medical Center were reviewed looking for patients previously treated for CRC. One hundred and fourteen patients were identified who had previously been treated for CRC. Patients were excluded if they were unable to speak English (N 5 2), had moved out of Indiana (N 5 1), were being, or had been, treated for another primary cancer (N 5 4), had comorbid illnesses likely to be fatal within 5 yr (N 5 4), had metastatic CRC to the liver treated with hepatic resection (N 5 12), or had stage III CRC not treated with adjuvant chemotherapy (N 5 2). The remaining 89 eligible focus group participants were first contacted by mailed invitation, then by phone. Following agreement to participate, respondents were scheduled for focus groups composed entirely of subjects with the same candidate outcome state as themselves. Focus groups in this study were composed of four to nine individuals with CRC, a facilitator, a recorder, and a support person. Each group met at the Regenstrief Institute for Health Care in either June or July 1996 with sessions lasting approximately 90 min. All participants gave informed consent in the manner approved by the Indiana University– Purdue University at Indianapolis (IUPUI) Institutional Review Board and were reimbursed $25 for their participation. The focus group scripts were simple, consisting of broad questions concerning the effects of CRC on several dimensions of health: social functioning, emotional functioning, cognitive functioning, physical pain, and nonpain physical
AJG – Vol. 93, No. 9, 1998 problems. Each session was audiotaped for later transcription. The focus group transcripts were analyzed by two independent reviewers using group-by-question grids (14). The objective of this analysis was to determine what, if any, areas of morbidity were reported in each dimension of health. Each area of morbidity was identified and characterized as to its stated intensity and/or frequency of occurrence. The reviewers then determined how many participants noted each morbidity and the modal level of intensity and/or frequency of occurrence they reported. After independently analyzing the transcripts, the two reviewers met to discuss their conclusions and to reach a consensus concerning the characterization of each identified area of morbidity. If a consensus could not be reached, a third independent reviewer was engaged to arbitrate. Description construction After the focus group transcripts were analyzed, the results were used to redefine the important outcome states of CRC and to help construct their descriptions. Those morbidities reported most frequently by the focus group participants within each dimension of health were incorporated into the descriptions. When the data obtained from a focus group indicated a possibly bimodal response in one or more dimensions of health, the candidate outcome state in question was split into two new states. After the state was split, the published frequency and severity of CRC symptoms and treatment-related morbidities for that candidate state was reviewed in an attempt to confirm the observation. Likewise, when the data indicated that two or more focus groups were nearly identical in their responses, they were combined. Again, after the states were combined, the published literature was reviewed in an attempt to confirm the observation (13, 15–23). The descriptions were constructed in a tabular fashion to allow greater comparability between different outcome states. Each state was described in the short term (first 18 months) and long term (after 18 months), as reported morbidities during and immediately after therapeutic interventions were substantially different from those reported thereafter. Each outcome state was assigned a duration of 30 yr except for those involving metastatic CRC, which were given a duration of only 18 months. Statistical analysis Demographic comparisons between participants and nonparticipants were analyzed using Fisher’s exact test or x2 for categorical data and t tests for continuous data. The statistical package employed was SPSS for Windows, release 7.5.1 (SPSS Inc., Chicago, IL). RESULTS Based on our literature review and discussions with health care professionals, we created a list of six candidate outcome states of CRC (see Table 1) (13, 15–23). We
AJG – September 1998
COLORECTAL CANCER, PATIENT FOCUS GROUPS
TABLE 1 Candidate Outcome States of CRC State 1 2 3 4 5 6
Description Stage I rectal or stage I/II colon Ca treated with resection only Stage III colon Ca treated with resection and chemotherapy Stage II/III rectal Ca treated with resection/chemotherapy/radiation therapy Stage II/III rectal Ca treated with resection/chemotherapy/radiation therapy/permanent ostomy Stage IV metastatic CRC Stage IV locally unresectable CRC
CRC, colorectal cancer; Ca, cancer. TABLE 2 Characteristics of Participants and Nonparticipants in Colorectal Cancer Focus Groups† Characteristic Age* Sex: M/F Race Caucasian African-American Ethnicity Hispanic Non-Hispanic Years since diagnosis* Outcome state 1 2 3 4 5 6
Interviewed (n 5 38)
Not Interviewed (n 5 51)
62 6 13 14/24
65 6 12 23/28
34 4
50 1
1 37 261
0 51 261
6 6 7 4 8 7
8 8 1 5 18 11
* Plus-or-minus values are means 6 SD. † There are no significant differences among the groups for any of the parameters measured.
concluded that morbidity in patients with CRC is related mainly to the treatment modality employed, which is itself usually determined by the location and stage of the cancer at diagnosis (13, 15, 16). The cancer itself is usually a minimal direct source of morbidity except in the setting of metastatic or locally unresectable disease. We assembled focus groups for each of the candidate outcome states of CRC. Of the 89 patients invited to participate, 30 refused, 11 had died, and 10 could not be reached by mail or telephone. A comparison of the 38 participants to nonparticipants is presented in Table 2. No statistically significant differences were found between participants and nonparticipants. We conducted the analysis of the focus group transcripts by examining participants’ responses to broad, open-ended questions concerning the impact of their disease and its treatment on their social functioning, cognitive functioning, emotional health, pain, and nonpain physical problems. Each of the two reviewers analyzing the transcripts created summary statements for each broad health dimension for
1493
each focus group. The language employed in these summary statements was standardized as much as possible, using the following conventions for the key descriptive words employed. “Mild,” “moderate,” and “severe” were used to describe intensity, “mild” being less intense than “moderate,” which is less intense than “severe.” “Occasional” and “frequent” were used to describe the frequency of occurrence, “occasional” meaning less than half the time and “frequent” meaning more than half the time. Finally, “some” and “most” were used to describe the frequency of a specific response among the focus group participants, “some” meaning less than half and “most” meaning more than half of the participants reported this problem. This analysis uncovered one or more specific areas of morbidity within each of the broad health dimensions explored for each focus group. These areas of morbidity included: 1) problems with social interaction; 2) problems with cognition; 3) fears of cancer recurrence; 4) pain; 5) fatigue; 6) changes in bowel habits; and 7) sexual dysfunction. Although these specific areas of morbidity varied somewhat between groups, considerable overlap was noted. After preparing these summary statements, the reviewers compared their analyses. No significant areas of disagreement were discovered. When the focus group responses were analyzed, it became apparent that outcome state “2” (stage III colon cancer treated with resection and chemotherapy) was bimodal as to the number and intensity of symptoms reported by participants while on chemotherapy. Approximately half of the participants complained of significant nausea, painful mouth sores, and diarrhea that, in some cases, required hospitalization. The other half of the participants reported only minimal symptoms while on chemotherapy. This result is in agreement with the published side effect profiles of both 5FU/levamisole and 5FU/leukovorin therapy that report significant nausea, mucositis, and diarrhea in only 40 – 80% of patients (17, 18). Analysis of the focus groups’ responses revealed little if any difference between the outcome state described for participants dying of metastatic disease (state 5) versus that described for participants dying of unresectable or locally recurrent disease (state 6). The health care professionals we queried proposed that those patients with locally unresectable disease would complain of more pain than those with purely metastatic disease because of tumor invasion of large nerve plexi. Although many participants in both states 5 and 6 noted pain, they all reported its control with medications. States 5 and 6 also appeared homogenous, in that a majority of the participants in both of these groups reported that their most troublesome symptom was severe fatigue that limited their ability to care for themselves and their families. Few of the participants in the other focus groups reported severe chronic fatigue. Because state 2 appeared heterogeneous and states 5 and 6 homogenous, the listed outcome states were revised. State 2 was split between patients having and those not having
1494
NESS et al.
AJG – Vol. 93, No. 9, 1998
TABLE 3 Revised Outcome States of Colorectal Cancer State A B C D E F G
Description Stage I rectal or stage I/II colon Ca treated with resection only Stage III colon Ca treated with resection and chemotherapy without significant side effects Stage III colon Ca treated with resection and chemotherapy with significant side effects Stage II/III rectal Ca treated with resection/chemotherapy/radiation therapy Stage II/III rectal Ca treated with resection/chemotherapy/radiation therapy/permanent ostomy Stage IV metastatic/unresectable disease without ostomy Stage IV metastatic/unresectable disease with ostomy
Ca, cancer.
significant side effects of chemotherapy. Similarly, states 5 and 6 were combined into a single outcome state of patients with terminal (stage IV) CRC. We became concerned about the possible interactive effect on future utilities assessment of having stage IV CRC and a permanent ostomy from previous therapy present in the same outcome state. Even though this outcome was felt to represent a very small patient population, we thought that incorporation of this combination state might provide insight into how an ostomy influences people’s utilities for outcome states of CRC. The revised outcome states of CRC are listed in Table 3. The final task was to incorporate the information gained from the focus groups into brief descriptions of each selected outcome state of CRC. The first objective was to provide enough data to clearly delineate each outcome state while not overwhelming the reader with excessive information. The second objective was to present the data for each outcome state in a form facilitating comparison between the descriptions for different states. Key descriptive words and short phrases were used to convey information on each specific area of morbidity using the conventions presented earlier. The key descriptive words employed to describe each specific area of morbidity or symptom in the short term (0 –18 months) and long term (.18 months) can be found in Tables 4 and 5. Except for the stated morbidities, persons in each described outcome state were considered to be unimpaired. A tabular format eased comparison of one state to another. An example of the final description format appears in Table 6. DISCUSSION We identified clinically important outcome states of CRC, used focus groups to obtain information concerning the important sources of morbidity in each of these states, and constructed descriptions of these outcome states using this information. The outcome state descriptions can be incorporated into a utilities assessment instrument to determine population and/or individual utilities for the identified outcome states of CRC. The importance of identifying the outcome states of CRC
rests on the need for these states by decision analysis, a methodology commonly used to examine the cost-effectiveness of health care (10). A decision analysis combines observational and experimental data into a model that evaluates the expected outcomes of various diagnostic or treatment strategies (10). The primary outcome measure is usually the marginal improvement in cost per quality-adjusted life-year saved (QALYs). QALYs incorporate changes in survival and in health-related quality of life into a single measure (9 –11, 24). To be consistent with this construct, quality weights should be measured by, or transformed into, an interval scale; that is, the method of measurement should be one in which the ratio of differences between values is meaningful. The scale most commonly used employs 0 (immediate death) and 1 (perfect health) as end points. The quality weights used should also be based on utilities for the health outcome states (9 –11). Health status scales that are not utility-based or do not have interval characteristics, such as the Sickness Impact Profile (SIP) or the Medical Outcomes Study Short-Form Health Survey (SF36), are not suitable for cost-effectiveness analysis (9 –11, 25). Methods based on decision theory, such as the standard gamble, are utility-based and have interval characteristics, making them appropriate methods for obtaining quality weightings for economic analysis (9 –12, 24, 25). Although we identified and characterized the important outcome states of CRC, there are other health states associated with CRC screening and surveillance that we did not assess. When considering outcome states for incorporation into a cost-utility model, they should be of sufficient duration and frequency to potentially affect the model’s conclusions. For example, the health state associated with a colonoscopic perforation was felt to be so uncommon that any change in long-term utility would probably not affect the model’s conclusions. Some health states, such as the one associated with an asymptomatic patient after a false-positive screening test, we did not consider to be different enough from the baseline health state to warrant a separate utility assessment. Such assumptions can be tested in the cost-utility model itself through sensitivity analysis. This is accomplished by varying the input utility for the outcome state in question and examining whether the model’s output or conclusions are significantly altered. The source from which utilities for the outcome states of any given clinical model are obtained is dependent on the viewpoint being employed in the cost-effectiveness analysis. Commonly employed viewpoints include that of society as a whole, a third-party payer, a patient decision-maker, and a patient already in a given outcome state. Except when the patient is already in the outcome state being evaluated, utilities must be assessed on a sample of persons who are informed only by descriptions of the outcome states. This makes the construction of a valid, understandable description of the outcome state being evaluated an important step in utilities measurement. Often, physician experts and the medical literature have
AJG – September 1998
COLORECTAL CANCER, PATIENT FOCUS GROUPS
1495
TABLE 4 Key Words for Each Specific Area of Morbidity: Short Term
ATTRIBUTES Fatigue Diarrhea Fecal urgency Fecal incontinence Sexual dysfunction Pain Cognitive problems Social interaction problems Fear of disease recurrence
STATES A
B
C
D
E
F
G
none none mild occasional none none none none
mild none mild occasional none none none some
moderate occasional mild occasional occasional mild-moderate none some
moderate frequent severe frequent frequent moderate none most
moderate occasional ostomy ostomy frequent moderate none most
severe occasional mild occasional frequent severe frequent most
severe occasional ostomy ostomy frequent severe frequent most
none
severe
severe
severe
severe
severe
severe
TABLE 5 Key Words for Each Specific Area of Morbidity: Long Term
ATTRIBUTES Fatigue Diarrhea Fecal urgency Fecal incontinence Sexual dysfunction Pain Cognitive problems Social interaction problems Fear of disease recurrence
STATES A
B
C
D
none none mild occasional none none none none
none none mild occasional none none none none
none none mild occasional none none none none
none none severe frequent occasional none none some
none
moderate
moderate
moderate
E
F
G
none none ostomy ostomy frequent none none some
NA NA NA NA NA NA NA NA
NA NA NA NA NA NA NA NA
moderate
NA
NA
NA, Not applicable. TABLE 6 Description of Person in Outcome State D (Duration of State: 30 yr) Problem
Short Term (first 18 mo.)
Tiredness and weakness Bowel habits
● moderate ● frequent diarrhea ● severe fecal urgency (urgent need to pass a bowel movement) ● frequent fecal incontinence (having a bowel movement in one’s pants) ● frequent ● moderate but controllable with medication ● none ● cutbacks in most social activities ● severe fear of disease recurrence
Sexual problems Pain Thinking problems Social problems Emotional problems
Long Term (rest of life) ● none ● severe fecal urgency (urgent need to pass a bowel movement) ● frequent fecal incontinence (having a bowel movement in one’s pants)
● occasional ● none ● none ● cutbacks in some social activities ● fear of disease recurrence decreases with time
Note: This patient has no other problems (i.e., hearing, speech, mobility, dexterity, etc., are all normal).
provided all the information used to inform descriptions of outcome states for utilities assessment. This approach has been criticized for not truly reflecting the patient’s viewpoint on his or her illness (25). Patient focus groups in which a broad spectrum of individuals are sampled are an economical and relatively quick way of gathering a wealth of patient opinion on a subject. Focus groups have been widely used in market research to ascertain consumer opinion on
various topics and products (14). Focus groups have also been used previously to inform outcome state descriptions for utilities assessment and have been shown to provide invaluable information regarding aspects of quality of life and their relative importance (26). In this study, the assumption was made that the morbidities of importance to the patient were not already known. Focus groups were used to provide this information within a basic framework drawn
1496
NESS et al.
from a literature review and discussions with health care professionals. This process of informing the outcome state descriptions using patient focus groups provides those descriptions with both face and content validity (25, 26). The patient focus groups provided information significantly different than that provided by the literature review and our discussions with local health care professionals. This information prompted several changes in the composition of the revised list of CRC outcome states. The focus groups revealed that the side effects of chemotherapy clustered predominantly in a subpopulation of participants, whereas the remainder had no complaints. This led to the splitting of outcome state 2 on the basis of the presence or absence of chemotherapy side effects. The focus groups also revealed that differences in pain severity between patients with locally unresectable/recurrent disease and those with metastatic disease that had been suggested by our discussions with health care professionals were either nonexistent or were dominated by the severity of other concurrent symptoms. Focus groups are susceptible to both sampling bias and to various systematic biases related to the interview process (14). In this study, the focus groups were drawn from a predominantly white, middle-class population. The results should be generalizable to similar populations but may not be secondary to sampling bias. We have been unable to identify other focus group studies that specifically address race or socioeconomically dependent variation in patients’ assessments of their disease-specific outcome states. However, there are several examples of focus groups uncovering race and/or socioeconomic differences in patients’ attitudes and concerns about such diverse subjects as breast cancer screening (27), cervical cancer screening (28), prostate cancer screening (29), depression (30), and exercise (31). Future research should be done to delineate any possible race or socioeconomic status-related differences in patients’ descriptions of their colorectal cancer outcome states. Focus groups are not surveys and are not meant to represent definitively the views of the population from which they are drawn (14). The strength of focus groups rests in their simplicity, ease, and affordability (14). They are well suited to the task of rapidly obtaining information in a setting in which little or no pre-existing data are either available or assumed. Quality of life is a multidimensional construct. Although the number of components in any given conceptualization of quality of life may vary, the minimum dimensions should include: 1) physical (i.e., symptoms, pain); 2) emotional; 3) social; and 4) mental (32–37). Two additional domains commonly included in this list are the vocational or role performance and the spiritual (33, 34). In general, the generic, utilities-based, health state classification systems (e.g., Health Utilities Index, EuroQol, and the Quality of Well-Being Scale) include physical, social, cognitive, and emotional function dimensions, along with domains that map to role performance (37). Information on patients’ role
AJG – Vol. 93, No. 9, 1998 performance was gathered in the course of the focus group sessions, but because of heterogeneity in the line of work and employment status of our participants, role performance was not included in the state descriptions. Much of the reported vocational morbidity can be accounted for by areas of physical morbidity included in the descriptions. This allows the readers to decide whether or not they could work in a given state. Spiritual function was not included because of the absence of adequate measures. The areas of morbidity listed under the nonpain physical problems section (fatigue, bowel dysfunction, sexual dysfunction) were those most commonly reported across the whole study population, although they were not relevant to every outcome state. These symptoms are also those most often reported in clinical studies of CRC treatment and outcome. Bowel dysfunction is described by a majority of the patients treated with distal colonic resection and/or chemotherapy. Likewise, limiting fatigue is reported by a majority of patients with metastatic disease (15, 16, 23). There is currently little consensus as to how health state descriptions should be structured for use in utilities assessment. The length and amount of information conveyed by a health state description has been shown by some to affect assessed utility values, although this finding has been inconsistent (25, 32). Knowing that research has shown that humans can process only about five to nine pieces of information concurrently, it may be hypothesized that long narrative descriptions could overload the cognitive capacity of the participant (32, 38). Alternatively, although a brief point form description may improve retention of information, it may provide insufficient detail to characterize a disease state adequately (25). In this study, the outcome state descriptions used key descriptive words and short phrases organized in a tabular fashion. This decision was made for three reasons: 1) to simplify data presentation; 2) to ease comparisons between states; and 3) to avoid issues of labeling and framing inherent to narrative descriptions (25, 32). We believe the chosen format allows and facilitates comparisons among the described outcome states. To limit the number of state descriptions, we chose to describe only the “typical” health state experienced by a person in each outcome state defined by stage at diagnosis and treatment. Specifically, we incorporated into each outcome state description only the symptom levels most commonly reported for each area of morbidity by the participants in the focus group(s) associated with that state. Furthermore, the “typical” health state derived from each focus group for the first 12–18 months following CRC diagnosis (short term) was substantially different from the state derived for the period thereafter (long term). We chose to combine these two states into a single, holistic, health state description for each outcome. These constructs are valid because utilities measured for them incorporate the person’s actual preferences for both the short-term and long term aspects of the described outcome states (39). We conclude that, in patients without terminal illness, the
AJG – September 1998
COLORECTAL CANCER, PATIENT FOCUS GROUPS
majority of the short and long term morbidity associated with CRC is a consequence of disease treatment. Because therapy is dependent on the stage and location of the cancer at the time of diagnosis, the important outcome states of CRC are also determined by the stage and location of the cancer at the time of diagnosis. Descriptions of these outcome states have been created for incorporation into a utilities assessment instrument that can be used to measure individuals’ utilities for these states. The measured utilities can then be incorporated into a decision-analytic model to determine the cost-utility of various diagnostic and therapeutic strategies for CRC.
14. 15.
16.
17.
18. 19.
ACKNOWLEDGMENTS 20.
The authors would like to acknowledge and thank Drs. Douglas Rex and David Crabb for their help in the planning of this project. We would also like to thank Dr. Patrick Loehrer and all of the physicians and nurses associated with the Multi-Disciplinary GI Cancer clinic at the Indiana University Medical Center for their cooperation and assistance in patient recruitment.
21.
22.
23. Reprint requests and correspondence: Reid M. Ness, M.D., M.P.H., Wishard Memorial Hospital, OPW 2005, 1001 West Tenth Street, Indianapolis, IN 46202.
REFERENCES 1. Parker SL, Tong T, Bolden S, et al. Cancer Statistics, 1997. CA Cancer J Clin 1997;47:5–27. 2. Winawer SJ, Zauber AG, Ho MN, et al. Prevention of colorectal cancer by colonoscopic polypectomy. N Engl J Med 1993;329:1977– 87. 3. Winawer SJ, Zauber AG, O’Brien MJ, et al. Randomized comparison of surveillance intervals after colonoscopic removal of newly diagnosed adenomatous polyps. N Engl J Med 1993;328:901– 6. 4. Mandel JS, Bond JH, Church TR, et al. Reducing mortality from colorectal cancer by screening for fecal occult blood. Minnesota colon cancer study. N Engl J Med 1993;328:1365–71. 5. Kronborg O, Fenger C, Worm J, et al. Causes of death during the first 5 years of a randomized trial of mass screening for colorectal cancer with fecal occult blood test. Scand J Gastroenterol 1992;27:47–52. 6. Winawer SJ, Flehinger BJ, Schottenfeld D, et al. Screening for colorectal cancer with fecal occult blood testing and sigmoidoscopy. J Natl Cancer Inst 1993;85:1311– 8. 7. Hardcastle JD, Armitage NC, Chamberlain J, et al. Fecal occult blood screening for colorectal cancer in the general population. Results of a controlled trial. Cancer 1986;58:397– 403. 8. Kewenter J, Brevinge H, Engaras B, et al. Results of screening, rescreening, and follow-up in a prospective randomized study for detection of colorectal cancer by fecal occult blood testing. Results for 68,308 subjects. Scand J Gastroenterol 1994;29:468 –73. 9. Russell LB, Gold MR, Siegel JE, et al. The role of cost-effectiveness analysis in health and medicine. JAMA 1996;276:1172–7. 10. Weinstein MC, Siegel JE, Gold MR, et al. Recommendations of the panel on cost-effectiveness in health and medicine. JAMA 1996;276: 1253– 8. 11. Kaplan RM, Feeny D, Revicki DA. Methods for assessing relative importance in preference based outcome measures. Quality Life Res 1993;2:467–73. 12. Boyd NF, Sutherland HJ, Heasman KZ. Whose utilities for decision analysis? Med Decis Making 1990;10:58 – 67. 13. Sprangers MAG, Taal BG, Aaronson NK, et al. Quality of life in
24. 25. 26. 27.
28. 29.
30.
31. 32.
33. 34.
35. 36.
37. 38. 39.
1497
colorectal cancer: Stoma vs. nonstoma patients. Dis Colon Rectum 1995;38:361–9. Stewart DW, Shandasani PN. Focus groups theory and practice. Newbury Park, CA: Sage Publications, 1990. Cohen AM, Minsky BD, Schilsky RL. Cancer of the colon. In: De Vita VT Jr, Hellman S, Rosenberg SA, eds. Cancer: Principles and practice of oncology. Philadelphia: Lippincott-Raven, 1997:1144 –97. Cohen AM, Minsky BD, Schilsky RL. Cancer of the rectum. In: De Vita VT Jr, Hellman S, Rosenberg SA, eds. Cancer: Principles and practice of oncology. Philadelphia: Lippincott-Raven, 1997: 1197–1234. Moertel CG, Fleming TR, MacDonald JS, et al. Levamisole and fluorouracil for adjuvant therapy of resected colon carcinoma. N Engl J Med 1990;322:352– 8. IMPACT Investigators. Efficacy of adjuvant fluorouracil and folinic acid in colon cancer. Lancet 1995;345:939 – 44. Enker WE. Potency, cure, and local control in the operative treatment of rectal cancer. Arch Surg 1992;127:1396 –1402. van Driel MF, Weymar Schultz WCM, van de Weil HBM, et al. Female functioning after radical surgical treatment of rectal and bladder cancer. Euro J Surg Oncol 1993;19:183–7. Krook JE, Moertel CG, Gunderson LL, et al. Effective surgical adjuvant therapy for high-risk rectal carcinoma. N Engl J Med 1991;324: 709 –15. O’Connell MJ, Martenson JA, Wieand HS, et al. Improving adjuvant therapy for rectal cancer by combining protracted-infusion fluorouracil with radiation therapy after curative surgery. N Engl J Med 1994;331: 502–7. Glimelius B, Hoffman K, Graf W, et al. Quality of life during chemotherapy in patients with symptomatic advanced colorectal cancer. Cancer 1994;73:556 – 62. Weeks J. Assessing utilities for health states. Proc Am Soc Clin Oncol 1997;16:353–5. Smith R, Dobson M. Measuring utility values for QALY’s: Two methodologic issues. Health Economics 1993;2:349 –55. Hall J, Gerard K, Salkeld G, et al. A cost-utility analysis of mammography screening in Australia. Soc Sci Med 1993;34:993–1004. Danigelis NL, Roberson NL, Worden JK, et al. Breast screening by African-American women: Insights from a household survey and focus groups. Am J Prev Med 1995;11:311–7. Clark DO. Age, socioeconomic status, and exercise self-efficacy. Gerontol 1996;36:157– 64. Robinson SB, Ashley M, Haynes MA. Attitude of African-Americans regarding prostate cancer clinical trials. J Comm Health 1996;21:77– 87. Cooper-Patrick L, Powe NR, Jenckes MW, et al. Identification of patient attitudes and preferences regarding treatment of depression. J Gen Intern Med 1997;12:431– 8. Jennings KM. Getting a Pap smear: Focus group responses of African American and Latina women. Onc Nurs Forum 1997;24:827–35. Froberg DG, Kane RL. Methodology for measuring health state preferences—I: Measurement strategies. J Clin Epidemiol 1989;42:345– 54. Eberst RM. Defining health: A multidimensional model. JOSH 1984; 54:99 –104. Rowland JH. Outcomes assessment: Cancer-specific quality-of-life measures— beyond the research setting. Proc Am Soc Clin Oncol 1997;16:342–9. Fletcher A. Quality-of-life measurements in the evaluation of treatment: proposed guidelines. Br J Clin Pharmac 1995;39:217–22. Fitzpatrick R, Fletcher A, Gore S, et al. Quality of life measures in health care. I: Applications and issues in assessment. Br Med J 1992; 305:1074 –7. Patrick DL, Erickson P. Health status and health policy. New York: Oxford University Press, 1993. Miller GA. The magical number seven plus or minus two: Some limits on our capacity to process information. Psychol Rev 1956;63:81–97. Froberg DG, Kane RL. Methodology for measuring health state preferences—II: Scaling methods. J Clin Epidemiol 1989;42:459 –71.