- Top of page
Background: Treatment decision making for postoperative Crohn's disease is complex because of the increasing number of maintenance therapies available with competing risk–benefit profiles. The main objective of this study was to determine the distribution of patients' preferences for selected postoperative maintenance therapies.
Methods: The study was a cross-sectional survey in which patients with Crohn's disease completed a standardized interview. Each participant completed 5 tasks that compared: (1) no medication and 5-ASA, (2) fish oil and 5-ASA, (3) metronidazole and 5-ASA, (4) budesonide and 5-ASA, and (5) azathioprine and 5-ASA. For each task, the minimum change in treatment effect size between the 2 treatments that the participant considered worthwhile was determined
Results: The distribution of the participants' preference scores varied widely for each task. When fish oil, metronidazole, budesonide, and azathioprine were considered equally effective to 5-ASA, 92.9%, 28.8%, 38.4%, and 19% of the participants, respectively, preferred these medications relative to 5-ASA. These percentages increased to 98.4%, 54.8%, 61.9%, and 50.8%, respectively, when fish oil, metronidazole, budesonide, and azathioprine were considered to offer a 5% absolute risk reduction relative to 5-ASA. Regression analysis did not identify any clinical or demographic variables predictive of the participants' treatment preferences.
Conclusions: The participants' preferences for postoperative maintenance therapies were widely distributed, and no clinical or demographic factors predicted these preferences. This emphasizes the need for effective communication between physician and patient in order to select the treatment options most consistent with a patient's informed preferences.
In the patient–physician relationship, the role of the physician is to act in the best interests of the patient by making treatment decisions and directing care on a patient's behalf.1, 2 This paternalistic approach is based on the assumptions that patients and providers have the same therapeutic goals, that providers can accurately judge patients' preferences about what therapeutic actions to take, and that it is best to spare patients the worry of decision making.1, 2 However, as the margin between the benefits and risks of available treatment options narrows, treatment decision making becomes more complex, and these assumptions become more questionable. In these circumstances, it is increasingly difficult for physicians to recommend 1 treatment option over another, and the historically paternalistic patient–physician relationship becomes less tenable.3 Accordingly, there is increasing recognition of the suitability of a shared decision-making model for deciding on treatments. This model promotes effective communication between physician and patient, with the goal of selecting the treatment option most consistent with the patient's informed preferences.4, 5 Currently, considerable attention is devoted to studying how best to inform patients about the clinical issues they may need to know about in order to make an informed choice. However, in the context of Crohn's disease, there has been relatively little systematic investigation into: (1) the processes by which patients formulate informed preferences about the available therapeutic options, (2) how strongly patients hold those preferences, (3) the reasons underlying their informed preferences, or (4) the patterned ways in which those preference distributions converge or diverge. This information will assist clinicians in improving communication and focusing on the aspects of the treatment decision that are important to their patients. In this regard, the treatment selected is most likely to be consistent with the patient's values and lifestyle. This kind of “treatment matching” has been shown to improve patient knowledge and satisfaction with the treatment decision and has also been associated with improved clinical outcomes.6, 7
The threshold technique is a preference assessment method that was introduced into the health care setting in 1989.8, 9 The threshold technique can be used to elicit patients' opinions about the minimal clinically important difference (MCID), which is defined as the minimum change in absolute risk reduction between 2 treatment options that a patient considers worthwhile. To date, applications of the threshold technique have involved research questions in different clinical contexts including (1) decision aids, (2) health policy and guidelines, (3) ethics, and (4) clinical trial design, and this technique has been adapted to assess attitudes toward a variety of therapies.10–16 The threshold technique has also been shown to be a reliable and valid assessment method across different patient populations.17
The main objectives of this study were to determine: (1) the distribution of patients' preferences for selected postoperative maintenance therapies for Crohn's disease by eliciting the MCID scores for (i) 5-ASA relative to no medication and (ii) fish oil, metronidazole, budesonide, and azathioprine relative to 5-ASA; (2) the extent to which selected clinical or demographic factors are predictive of the patients' treatment preferences; (3) the qualitative reasons for patients' treatment preferences; and (4) the test–retest reliability of the threshold technique to determine the MCID scores in this setting.
- Top of page
One hundred and thirty patients were invited to take part in the study, of whom 127 (97.7%) agreed to participate. One of the 127 potential participants was unable to complete the interview because of poor understanding of written English, resulting in 126 completing the interview. The characteristics of the participants are summarized in Table 2. On average, participants reported that they had had CD for approximately 12 years, were young, were in remission or had mild symptoms, and had had previous surgery. Most participants had CD in multiple sites (i.e., small and large bowel), and approximately one quarter of the participants were currently taking 5-ASA, prednisone, and azathioprine and/or ciprofloxacin and/or metronidazole. Most participants were well educated, worked full time, had health coverage provided through work or school, and had an annual household income higher than average. Overall, this group was representative of an outpatient population at a tertiary-care facility for inflammatory bowel disease. The first 70 participants who participated in the interview were asked to return for a second interview. Of these, 64 agreed to participate (5 subjects declined because of subsequent time constraints, and 1 was lost to follow-up). This subgroup's characteristics did not differ significantly from the original cohort of 126 participants.
Table 2. Participant Characteristics
|Mean age, years (minimum–maximum)||35.9 years(19–69 years)|
|Sex (% male)||46.0% (58/126)|
|Current medications (%)|| |
| 5-ASA||26.2% (33/126)|
| Prednisone||27.8% (35/126)|
| Azathioprine||21.4% (27/126)|
| Cipro/metronidazole||31.7% (40/126)|
| Methotrexate||4.0% (5/126)|
| Remicade||2.4% (3/126)|
|Disease location (%)|| |
| Isolated TI/ileocolic disease||22.2% (28/126)|
| Isolated colon (includes rectum)||11.1% (14/126)|
| Multiple sites||66.7% (84/126)|
|Perianal involvement (%)||49.2% (62/126)|
|Previous bowel resection (%)||63.5% (80/126)|
|Mean duration of disease, years (minimum–maximum)||12.9 years (1–39.4 years)|
|Time since last flare-up, years, (minimum–maximum)||2.4 years (0.3–18.4 years)|
|Ileostomy (%)||18.3% (23/126)|
|Mean Harvey Bradshaw score (minimum–maximum)||5.9 (0–35)|
|Mean IBDQ score (minimum–maximum)||166.8 (53–221)|
|Working full-time/attending school (%)||63.5% (80/126)|
|Health insurance paid for ≥80% medications (%)||73.0% (92/126)|
|College degree or higher (%)||53.2% (67/126)|
|Annual household income ≥ $50,000 (%)||57.1% (72/126)|
|Contribute ≥ 50% of annual household income (%)||51.8% (29/56)|
Distribution of Patients' Treatment Preferences
Table 3 shows the cumulative and raw frequencies of the participants' MCID scores for 5-ASA compared to no medication. This distribution ranged across the entire spectrum of possible scalar values. Approximately 10% of the participants preferred 5-ASA to no medication when 5-ASA was equally effective to no medication (i.e., MCID = 0). All these participants reported that taking the medication made them feel they were actively doing something about their disease or that they had more control over it, even though they understood that 5-ASA did not decrease the risk of symptomatic recurrence. Table 3 also shows that if 5-ASA reduced the absolute risk of recurrence by 5%, so that the absolute risk of symptomatic recurrence was 20% 1 year after surgery, approximately one third (34.1%) of the participants would be willing to take 5-ASA. This proportion increased to 61.1% if 5-ASA reduced the absolute risk of symptomatic recurrence by 10% (i.e., absolute risk of symptomatic recurrence 15% 1 year after surgery).
Table 3. Distribution of Participants Preferring 5-ASA at Various Levels of Effectiveness Relative to No Medication (n = 126)*
| ||Absolute reduction in risk of 5-ASA relative to no medication|
|Cumulative % (frequency)||10.3% (13)||34.1% (43)||61.1% (77)||81.7% (103)||88.1% (111)||100.0% (126)|
|Raw % (frequency)||10.3% (13)||23.8% (30)||26.9% (34)||20.6% (26)||6.3% (8)||11.9% (15)|
Table 4 shows the cumulative and raw frequencies of participant MCID scores for the target medications (i.e., fish oil, metronidazole, budesonide, and azathioprine) relative to 5-ASA. Table 4 shows the proportion of participants preferring the target medications relative to 5-ASA when these medications were both less effective and more effective than 5-ASA. With the exception of fish oil, all the distributions ranged across the entire spectrum of possible scalar values.
Table 4. Distribution of Participants Preferred Target Medications When Less Effective, As Effective, and More Effective Than 5-ASA
| ||Preferred target medication when less effective than 5-ASA (%)||Preferred target medication when as effective as 5-ASA (%)||Preferred target medication when more effective than 5-ASA (%)|
|Fish oil MCID score|| || || || || || |
| Cumulative % (frequency)||36.5%(46)||72.2%(91)||92.9%(117)||98.4%(124)||100.0%(126)||—|
| Raw % (frequency)||36.5%(46)||35.7%(45)||20.6%(26)||5.6%(7)||1.6%(2)||—|
|Metronidazole MCID score|| || || || || || |
| Cumulative % (frequency)||12.7%(16)||24.6%(31)||28.8%(36)||54.8%(69)||90.5%(114)||99.2%(125)|
| Raw % (frequency)||12.7%(16)||11.9%(15)||4.0%(5)||26.2%(33)||35.7%(45)||8.7%(11)|
|Budesonide MCID score|| || || || || || |
| Cumulative % (frequency)||17.5%(22)||31.0%(39)||38.4%(48)||61.9%(78)||81.0%(102)||99.2%(125)|
| Raw % (frequency)||17.5%(22)||13.5%(17)||7.1%(9)||23.8%(30)||19.4%(24)||18.3%(23)|
|Azathioprine MCID score|| || || || || || |
| Cumulative % (frequency)||5.6%(7)||14.3%(18)||19.0%(24)||50.8%(64)||88.1%(111)||100.0%(126)|
| Raw % (frequency)||5.6%(7)||8.7%(11)||4.8%(6)||31.7%(40)||37.3%(47)||11.9%(15)|
When the target medications were considered equally as effective as 5-ASA (i.e., 15% absolute risk of symptomatic recurrence 1 year after surgery), 92.9% of the participants indicated they would prefer fish oil, 28.8% metronidazole, 38.4% budesonide, and 19.0% azathioprine. More than one third of the participants would prefer fish oil even if fish oil were 10% less effective than 5-ASA. On the other hand, when metronidazole, budesonide, and azathioprine reduced the absolute risk of symptomatic recurrence by 5% (i.e., 10% absolute risk of symptomatic recurrence 1 year after surgery), only 54.8%, 61.9%, and 50.8% of the participants, respectively, preferred these target medications relative to 5-ASA. Furthermore, metronidazole, budesonide, and azathioprine had to offer a treatment benefit of 10% (i.e., virtually prevent recurrence) before more than 80% of the participants would prefer these medications relative to 5-ASA.
Determination of Clinical/Demographic Variables Predictive of Participants' Treatment Preferences
Linear regression (with the variables listed in Table 2) did not reveal any demographic or clinical variables to be predictive of the participants' MCID scores.
Qualitative Reasons for Participants' Treatment Preferences
Most participants reported that the type and severity of side effects rather than the likelihood of developing the side effect or the treatment protocol (i.e., number of pills per day, the dosing schedule, and required blood work) were the primary factors affecting their initial choices (Table 5). These results are consistent with the results of the debriefing questionnaire, which indicated that 55.2% of the participants felt that the type and severity of the side effects was the most important aspect regarding postoperative maintenance therapy, whereas only 36% felt effectiveness was the most important.
Table 5. Primary Reasons for Participants' Initial Treatment Choice
|Task||Options available in PTO task||Initially choosing option when both medications equally effective (%)|| ||Primary reason for initial choice when both medications equally effective|
|Fish oil relative to 5-ASA||5-ASA||7.1%(9/126)||77.8%||The minor side effects caused 5-ASA do not seem as bad as the minor side effects caused by fish oil|
| || || ||11.1%||I think that 5-ASA will work better for me than fish oil|
| ||Fish oil||92.9%(117/126)||58.1%||Fish oil does not cause any major side effects|
| || || ||29.9%||The minor side effects caused by fish oil do not seem as bad as those caused by 5-ASA|
|Metronidazole relative to 5-ASA||5-ASA||71.2%(89/125)||30.3%||The minor side effects caused by 5-ASA do not seem as bad as the minor side effects caused by metronidazole|
| || || ||21.3%||The chances of getting an effect from 5-ASA are less than getting a side effect from metronidazole|
| ||Metronidazole||28.8%(36/125)||63.9%||The major side effects caused by metronidazole do not seem as bad as the major side effects caused by 5-ASA|
| || || ||11.1%||Metronidazole does not cause any side effects that would require me to stay in the hospital|
|Budesonide relative to 5-ASA||5-ASA||62.2%(77/125)||44.2%||The minor side effects caused by 5-ASA do not seem as bad as the minor side effects caused by budesonide|
| || || ||32.5%||The chances of getting a side effect from 5-ASA are less than the changes of getting a side effect from budesonide|
| ||Budesonide||38.4%(48/125)||66.7%||Budesonide does not cause any major side effects|
| || || ||16.7%||The minor side effects caused by budesonide do not seem as bad as those caused by 5-ASA|
|Azathioprine relative to 5-ASA||5-ASA||81.0%(102/126)||42.2%||The major side effects caused by 5-ASA do not seem as bad as the major side effects caused by azathioprine|
| || || ||42.2%||The chances of getting a major side effect from 5-ASA are less than the chances of getting a major side effect from azathioprine|
| ||Azathioprine||19.0%(24/126)||54.2%||The minor side effects caused by azathioprine do not seem as bad as those caused by 5-ASA|
| || || ||16.7%||The major side effects caused by azathioprine do not seem as bad as those caused by 5-ASA|
Effect of Cost of Medication on Participants' Treatment Preferences
Table 6 shows that between 4.7% and 35.5% of the participants reversed their initial treatment choice to the less expensive medication when they had to pay for the medications with their own money. And as the price difference between the target medication and the reference medication increased, so did the number of participants who reversed their initial treatment choice.
Table 6. Effect of Cost on Participants' Initial Treatment Choice
|Task||Options available in task||Price differential/year (Canadian dollars)*||Percentage initially choosing option when medications equally effective AND not paying for medication (%)||Percentage initially choosing option when medications equally effective AND paying for medication (%)||Participants changing initial choice to less expensive option (%)|
|Fish oil relative to 5-ASA||5-ASA||Fish oil $870 less||7.1%(9/126)||2.4%(3/126)||4.7%|
| ||Fish oil|| ||92.9%(117/126)||97.6%(123/126)|| |
|Metronidazole relative to 5-ASA||5-ASA||Metronidazole $1290 less||71.2%(89/125)||35.7%(45/126)||35.5%|
| ||Metronidazole|| ||28.8%(36/125)||64.3% (81/126)|| |
|Budesonide relative to 5-ASA||5-ASA||Budesonide $560 more||61.6%(77/125)||73.8%(93/126)||12.2%|
| ||Budesonide|| ||38.4%(48/125)||26.2%(33/126)|| |
|Azathioprine relative to 5-ASA||5-ASA||Azathioprine $825 less||81.0%(102/126)||57.1%(72/126)||23.9%|
| ||Azathioprine|| ||19.0%(24/126)||42.9%(54/126)|| |
The mean length of time between interviews was 43.5 days (minimum–maximum: 13–161 days), and over this time period the HB symptom severity and IBDQ scores remained stable (Table 7). The intraclass correlation coefficients (ICCs) for the MCID scores in interview 1 and interview 2 ranged from 0.41 to 0.78, indicating fair to good test–retest reliability (Table 8). In addition, between 75% and 98% of the participants chose the same initial treatment option at interviews 1 and 2, and the percentage agreement for the MCID scores ranged from 30% to 41% (Table 8). The mean difference in the MCID scores ranged from −1.06 to +0.87, which represented less than a 5% change in score, given the maximum 25 units available on the scale for each task (Table 8). Furthermore, paired t tests did not show any significant differences between the MCID scores obtained at interviews 1 and 2.
Table 7. Changes in HB Symptom Severity and IBDQ Scores
| ||Interview 1||Interview 2||Mean difference|
|Mean length of interview, minutes (range)||45.8 (25–-66)||40.7 (26–61)||4.9|
|Mean number of days between interviews (range)||—||—||43.5 days (13–161 days)|
|Mean Harvey-Bradshaw score (range)||4.9 (0–14)||4.6 (0–15)||0.3|
|Mean IBDQ score (range)||169.4 (82–219)||175.6 (90–221)||6.2|
Table 8. Test–Retest Reliability Results
| ||Percentage who initially chose the same option (when both medications equally effective)||Percentage agreement in MCID score||Mean difference in MCID score (SD)||ICC|
|5-ASA relative to no medication||98.4%||41.3%||0.87 (5.86)||0.78|
|Fish oil relative to 5-ASA||84.3%||38.1%||−1.06 (6.24)||0.45|
|Metronidazole relative to 5-ASA||84.1%||32.8%||−1.05 (6.76)||0.67|
|Budesonide relative to 5-ASA||76.2%||29.7%||0.45 (7.27)||0.75|
|Azathioprine relative to 5-ASA||75.0%||31.3%||−0.98 (8.06)||0.41|
Overall, 74% of the participants indicated that they were either very or quite certain of their responses during the interview. Ninety-three percent of the participants felt the interview was either very or quite easy to understand, and 89% felt that the interview would be very or quite helpful for making treatment decisions regarding postoperative maintenance therapy.
- Top of page
Because CD is a chronic disease that occurs in young patients and patients are often treated with a variety of medication throughout their lifetime, we thought it was important to investigate the pattern of patient preferences for treatment and identify important predictors of these preferences. We believed that this was especially important in the postoperative setting because patients are asymptomatic and must consider taking medication indefinitely, which may lead to increased nonadherence to treatment. Although this may seem simplistic, this information is critical to highlighting to physicians not only the pattern of patients' preferences, but also the specific reasons for these preferences. We thought this was an important starting point for improving communication between patient and physician, with the overall goal of selecting a treatment that incorporates the physician's medical opinion and the patient's lifestyle, values, and preferences for treatment, as in other clinical settings this has been shown to lead to improved adherence to treatment and better clinical outcomes.6, 7, 30
With this in mind, we designed the clinical scenario, target and reference medications, probabilities of side effects, and recurrence rates as realistically as possible. At the time this study was initiated, several trials had shown a decreased risk of recurrence with 5-ASA, and therefore 5-ASA was chosen as the reference medication. However, since then there has been increasing evidence for the use of azathioprine and other biologicals such as infliximab for postoperative maintenance therapy.23, 27, 31, 32 We also did not include the risk of developing lymphoma from azathioprine, as we believed this was extremely rare and there was uncertainty about its quantitative risk.33 It is quite likely that including this information in the threshold task would have changed the distribution of the patients' MCID scores; however, we would still expect the overall distribution of these scores to be widely distributed and range across the entire spectrum of available scalar values. In another study for chronic active CD in which methotrexate was compared to infliximab, our group found that 64.4% of the participants initially preferred methotrexate relative to infliximab when both treatments were considered equally effective. Of these participants, 33.8% indicated that the primary reason for this choice was that the long-term side effects of infliximab were still not known.34 Despite this, the overall distribution of the MCID scores for methotrexate were widely distributed and ranged across the entire spectrum of available scalar units for the task.
Although there may be some limitations to these data, we feel that these are minor and unlikely to change the more important “big picture” finding that the patients' treatment preferences were widely distributed and that there were no clinical or demographic variables that were predictive of these treatment preferences. These findings suggest that the participants valued the competing risks and benefits associated with each of the treatment options differently based on their individual needs and lifestyles. Interestingly, the participants reported that the type of side effect was more important than the likelihood of the side effect occurring or the effectiveness of the medication. This was supported by the results of this study because the participants were more accepting of fish oil than of any other medication studied. Adding to the complexity of the participants' treatment preferences was the effect of cost, which in this study was responsible for reversing the initial treatment preference in up to 35% of participants.
These results suggest that physicians are likely to be unable to select the most appropriate treatment option for their patients based on demographic and clinical data alone and highlights the importance of effective communication between patient and physician. This interaction needs to involve a discussion of the available treatment options and their associated benefits (i.e., effectiveness) and risks (i.e., side effects) in order to ensure that the treatment the patient actually receives is most consistent with their values and lifestyle.
One strategy to improve communication and foster clarification of values for both the patient and the physician is the use of patient decision aids. A Cochrane review of 35 randomized controlled trials evaluating the effectiveness of patient decision aids was recently published.5 The overall results of this study showed that when compared with the usual care, patient decision aids significantly increase patients' knowledge and decrease decisional conflict about treatment. One of these trials also showed that when given the opportunity to use a decision aid, patients with benign prostatic hypertrophy reported significantly higher quality of life in the physical functioning domain of the SF-36 than did those patients who received the usual care.6
With respect to Crohn's disease, Schreiber et al. have shown that “information giving” may improve clinical outcomes of patients with CD. In this trial, 84 patients were randomized to receive group presentations and discussion or the usual care. The results showed that compliance with their treatment regimen of participants in the intervention group had significantly improved and that there were significantly fewer clinic visits.35
In 2000, an international survey was performed to obtain information on the clinical experience of azathioprine/6-mercaptopurine, methotrexate, and cyclosporine in the treatment of patients with inflammatory bowel disease.36 The results showed that of these 3 medications, azathioprine was prescribed most often, by 83.4% of the 300 clinicians surveyed. However, in the same year, a national, cross-sectional survey of more than 1700 patients with Crohn's disease in Canada found that only 24% of patients reported using azathioprine or 6-mercaptopurine. In this study, the use of azathioprine was not associated with sex, income, or region of residence; however, it was associated with age and markers of disease activity.37 These results are certainly in keeping with the results of the current study, which showed that only 50% of patients considered a 5% increase in effectiveness with azathioprine relative to 5-ASA worthwhile. The results of the survey may have been a result of its retrospective nature and the population sampled; however, it may also represent patient nonadherence with treatment. In a recent prospective study of 153 patients with inflammatory bowel disease, 41.2% of patients reported nonadherence to prescribed medications only 2 weeks after their index visit with their physician.38 Thirty-three percent of the patients reported unintentional nonadherence to the medication regimen, 15% reported intentional nonadherence, and 7.2% reported both unintentional and intentional nonadherence. Similarly, during our interviews, patients often commented that they would stop taking their prescribed medication or seek an alternate prescription with a different physician when they were not satisfied with the treatment they were initially prescribed. During the interviews, the patients were also particularly interested in herbal medicines and dietary treatments. In fact, many stated that they were taking herbal medicines such as aloe vera and cat's claw. These findings again highlight the importance of effective communication between physician and patient about available treatment options in order to optimize treatment matching, which may lead to improved satisfaction with and adherence to treatment.
Although the threshold technique involves a considerable number of procedural tasks, it was our group's experience that in real time, the interviews ran very smoothly, with few logistical problems. This was shown by the entire interview including the covariates and debriefing questionnaires and the 5 threshold tasks taking an average 45 minutes (range 25–66 minutes). In addition, more than 90% of the participants felt that the interview was easy to understand (i.e., score of 6 or 7 on the 7-point Likert scale). The participants also indicated that they appreciated knowing all the available treatment options, and almost 90% found the threshold task format helpful for making treatment decisions. This is supported by the test–retest reliability results, which indicated fair to good reliability.
The participants in this study were a convenience sample. Although it would have been preferable to have participants making decisions in a real-life situation (i.e., following ileocolic resection), this was not feasible given the sample size required and the time constraints of this study. However, the clinical situations depicted seemed highly salient (i.e., realistic) to participants because most felt the interview was easy to understand and more than 60% of the participants had had a previous bowel resection and therefore had experience with acute flare-ups and long-term therapy. However, the participants were recruited from a tertiary referral center with a known interest in the management of inflammatory bowel disease and the patients who made up the cohort were well educated and economically well off.
In summary, we believe that this study is of value because it is the first to formally assess patient preferences for postoperative maintenance therapy for CD. Given that CD is a chronic illness that affects young people, multiple treatment options are available, and maintenance therapy is given long term, we believe it is mandatory to consider patient preferences for treatment. This may be an important reason for not adhering to treatment and affecting clinical outcomes.