Patient preferences for acute pain treatment†
†Declaration of interest. Co‐authors Connie Chen, PharmD and Tracy Mayne, PhD are employees of Pharmacia Corporation and Pfizer, Inc. respectively, who funded this project.
Abstract
Background. Optimal treatment for acute pain is a function of an individual’s willingness to make trade‐offs between treatment side effects and pain control. The objective was to investigate the degree to which patients are willing to make these trade‐offs.
Methods. Fifty patients undergoing major abdominal surgery were enrolled and completed interviews before and after surgery. Measures included an experience with pain questionnaire and an adaptive conjoint analysis (ACA) interview.
Results. Percentage of pain relief obtained post‐surgery was between 70 and 80%. Eight‐two per cent reported at least one moderate or severe side effect. ACA results demonstrated that pain efficacy and side effect type/severity have almost equal ‘importance’ scores. Patients varied in their willingness to trade‐off pain efficacy for different or milder side effects.
Conclusions. We conclude that people have different relative preferences for different side effects and are willing to trade‐off pain relief for less upsetting and/or less severe side effects but to different degrees. Thus, physicians should consider offering pain medications with fewer side effects than narcotics as a first choice. Our study indicates the need to balance analgesia and side effects in order for patients to achieve optimal pain control.
Br J Anaesth 2004; 92: 681–8
Key words
Acute pain is still a common occurrence after surgery despite improvements in treating postoperative pain. Pain after surgery can impede recovery and interfere with a patients’ return to normal activities. In a recent study of patients undergoing elective orthopaedic or abdominal surgery, 76% reported moderate to severe postoperative pain and of this group, 5% reported unbearable pain.1 Furthermore, Strassels and colleagues2 found that postoperative pain interfered with patients’ general activity, mood, walking ability, relations with other people, and sleep. Despite these difficulties, most patients do not seek out additional or complete pain relief.3 It is likely that patients are weighing the positive (e.g. pain relief) and negative (e.g. side effects) outcomes of treatment and making trade‐offs between these outcomes when making treatment decisions.
Physicians and patients may have differing perspectives regarding the relative importance of these treatment outcomes and thus, which therapy is the best choice. For example, in a study of anaesthesia, physicians considering both frequency and severity of anaesthesia outcomes ranked the top five least desirable outcomes (in order) as incisional pain, nausea, vomiting, preoperative anxiety, and discomfort from i.v. catheter insertion.4 When patients rated anaesthesia outcomes in a companion study, the five most undesirable were (in order) vomiting, gagging on a tracheal tube, incisional pain, nausea, and recall without pain.5 The higher importance placed on postoperative nausea and vomiting in this study is also confirmed by a previous study6 that demonstrated that patients were willing to pay $40–100 to avoid postoperative nausea and vomiting.
Conjoint analysis is a technique for indirectly eliciting preferences for product’s or service’s features (or attributes) by asking respondents to trade‐off combinations of these features.7–9 The indirect conjoint technique is more effective for identifying true preferences than direct techniques because it forces respondents to make trade‐offs similar to the way individuals make decisions in real life. This technique has been used extensively in market research to estimate consumer preferences for products and to predict consumer choice behaviour. More recently, it has been applied to health care to determine patient preferences for treatment, with the ultimate goal of improving medical decision‐making and health outcomes. Adaptive Conjoint Analysis (ACA) utilizes a computer‐interactive interview that adapts the interview for each respondent, resulting in a less burdensome interview and higher quality data as respondents are more interested and involved in the task.10
Investigating patients’ preferences in the acute pain setting using conjoint analysis has not been reported previously. We sought to quantify the trade‐offs that patients make between pain relief, side effects, and side effect severity. We used adaptive conjoint analysis to quantify patient preferences. Understanding patient preferences allows analgesia to be tailored to an individual. This will maximize the analgesia while also minimizing the side effects.
Methods
After Institutional Review Board (IRB) approval and written informed patient consent, 50 surgical patients were recruited preoperatively. Inclusion criteria were subjects more than/equal to 18 yr old, undergoing inpatient abdominal surgery followed by treatment with patient controlled analgesia (PCA) for acute pain for 2–3 days, English‐speaking, able to read at a 12‐yr‐old level or above, free of cognitive, visual or motor impairments that would preclude completion of the questionnaire booklet and preference interview, and not on opioids during the previous month. Research personnel identified and recruited eligible subjects using a convenience sample approach. Eligible subjects were enrolled at their pre‐operative outpatient visit. To investigate whether actual experience affected patient preferences, each subject participated in four interviews. The first took place at the pre‐surgical screening visit. The second interview took place within 12 h of transitioning from patient‐controlled analgesia to oral analgesia. The third occurred within 48 h of the second interview or immediately before discharge, whichever occurred first. The fourth and final interview was held within 14 (±1) days post‐discharge from the hospital at a follow‐up visit. These interviews were a mixture of written questionnaires and computer assisted interviews.
Conjoint analysis
Patients were presented scenarios, which include the following attributes: (1) degree of pain relief, (2) type of side effect, (3) severity of side effect, and (4) setting/route of pain control administration. Each attribute was defined by specific levels (see Appendix for full description). Degree of pain relief ranged from poor pain relief to excellent pain relief; type of side effect included no side effects, vomiting, itching, constipation, etc.; severity of side effect included mild, moderate, and severe; setting/route of pain control administration included inpatient receiving PCA, inpatient receiving pain pills and pain pills at home.
An ACA, which utilizes a computer‐interactive interview, was used to administer the trade‐off questions. During the conjoint interview, each respondent was asked to trade‐off two hypothetical pain control scenarios (defined by the set of attributes and levels) and to provide a rating (i.e. a preference index ranging between 1 and 9) indicating which scenario was preferred and the strength of that preference. The resulting responses were used to estimate a utility function using regression techniques. The respondent utility function quantified the relationship between respondent preferences and the specific pain control attributes or features. Although the ‘utility’ associated with each pain control scenario is not directly observable, the rating provided by the respondent is observable and is related to the utility value. This rating is used in the regression analysis as the dependent variable in the utility function.11
Data items
At interview 1, subjects provided information on their age, gender, race, marital status, education level, employment status, and co‐morbid conditions. Information on pain medications prescribed was obtained from medical records for interviews 2, 3, and 4.
Patients provided written information at interviews 2, 3, and 4 regarding their method of pain control, amount of pain experienced, success and satisfaction with pain control, and side effects they experienced.
Computer‐assisted conjoint analysis interview
Subjects completed a computerized interview at each of the four time points using Sawtooth Software’s ACA (Version 5). The four attributes were:
1. Degree of pain control.
2. Type of side effect.
3. Severity of side effect.
4. Setting/route of pain control administration.
Before starting the computer interview, we provided detailed descriptions of each level of each attribute to the respondents (see Appendix). We also told respondents that the time frame for these descriptions was 1 day following surgery. Hence, the patient imagined experiencing the health state defined by each combination of attribute levels on a single day following surgery.
The computer interview first asked respondents to rank their preferences for the various levels of each attribute and to rate the relative importance of each attribute. Examples of these questions are shown in Figure 1. The trade‐off questions subsequently completed were different for each respondent depending upon their responses to these initial questions.
Fig 1 Overview of the conjoint analysis interview.
The computer interview next presented each respondent with a series of ‘trade‐off’ questions each with two pain control scenarios shown on the screen. One set of attributes (defining a pain control scenario) was shown on the left of the screen and another pain control scenario was shown on the right. For each screen, the respondent was asked, ‘Which do you prefer?’ The respondent selected a rating‐score between 1 and 9 with 1 indicating a strong preference for the pain control scenario on the left, 9 indicating a strong preference for the scenario on the right, and 5 indicating indifference or no preference between the two scenarios. Figure1 shows an example of a trade‐off question.
Data analyses
Data collected were analysed using descriptive statistics. Frequencies, means, and standard deviations were calculated for the socio‐demographic characteristics of the sample at interview 1 and the clinical characteristics at interviews 2–4. ANOVAs were used to compare the experience with pain management at interviews 2, 3, and 4.
Using responses to the conjoint interview at interview 4, we used the ACA program to generate a utility function for each respondent based on OLS regression analysis. Individual utilities for each level of each attribute were generated and normalized so that zero is the lowest utility score for each attribute and higher utilities indicate stronger preference. We also used the ACA program to generate mean utilities and standard errors for the total study population. These indicate the ‘value’ or utility that respondents have for each level of a particular attribute.
In addition to the utility values, we estimated relative ‘importance’ scores for each attribute. The ‘importance’ scores were computed by first summing across all the attributes, and the range of the utility values for each attribute. Then the ‘importance’ weight for a single attribute was computed as 100 multiplied by the range of the utility values for that attribute divided by the sum of the ranges for all the attributes. The ‘importance’ scores demonstrate the relative importance of each of the four treatment attributes for determining a patient’s preferences for treatment. ‘Importance’ scores were computed for each patient and then mean values computed for all the patients in the study. A combined ‘importance’ weight of side effect type and side effect severity was also computed and compared with the ‘importance’ weight for pain control for all patients in the study to demonstrate the relative importance to patients of pain relief vs side effects.
Utility values and ‘importance’ scores were also estimated using the conjoint interview data from interviews 1, 2, and 3 and compared with those estimated using the data from interview 4. Utility values and ‘importance’ scores were also estimated for different pre‐specified subsets of the population, including gender, pain experience, side effect experience, and type of surgery.
Finally, we conducted simulation analyses to estimate the percentage of patients that prefer one set of treatment attributes over another. These analyses computed the utility (strength of preference) associated with two hypothetical treatments (A and B) with different combinations of attributes for each individual. An individual was assumed to prefer treatment A to treatment B if the utility with treatment A was higher than for treatment B. The proportion of individuals who prefer treatment A to treatment B was estimated based on the results of these calculations in 50 000 hypothetical individuals drawn with replacement from our 50 patient population.12
Results
Sixty‐five patients were enrolled in the study. Of these, 12 withdrew because PCA was not required following surgery, one withdrew because of extreme nausea and drowsiness following surgery, and one was unable to complete visit 4 because of time constraints. Table 1 describes the characteristics of the 50 patients who completed the study. Table 2presents the results of the patients’ experience with postoperative pain and pain relief from medication that they received. Pain intensity decreased over time since surgery for both ‘worst pain in the last 24 h’ and ‘least pain in the last 24 h’. The mean value for the ‘worst pain in the last 24 h’ at interview 2 was 6.7 on a scale of 1–10 where 10 is the worst imaginable pain. The degree to which pain relief was obtained at the three post‐surgery interviews was between 70 and 80% and there was no significant trend over time (Table 2).
Table 1
Patient and clinical characteristics. Percents may not sum to 100% because of rounding
Table 2
Patients’ experience with pain at visits 2, 3, and 4. *0 equals ‘no pain’, 10 equals ‘worst imaginable pain’. †P‐value from repeated measures analysis of variance; visit comparisons significant at P<0.05
All subjects were prescribed pain medications that included opioids at interview 2. At interview 4, all but three patients (94%) were still prescribed opioids for pain relief. Table 3 presents the patients’ experience with side effects of pain medication during the study. Almost every patient, 96% (n=48), reported at least one side effect from pain medication of any level of severity, 82% (n=41) reported at least one side effect that was moderate or severe, and 40% (n=20) reported at least one side effect that was severe.
Table 3
Proportion of patients experiencing side effect at any time during the study by maximum severity
Table 4 presents the mean utilities for each level of each attribute. Respondents assign the highest utility value to ‘excellent’ pain relief and the lowest to ‘very poor’ pain relief. Vomiting, nightmares, hallucinations, and nausea are the least preferred types of side effects. PCA in the hospital has the highest utility value for site/type of administration.
Table 4
Average utility levels at interview 4 for each attribute level. *Higher values indicate greater preference; zero is assigned to the attribute level with the lowest utility
The ‘importance’ scores for the attributes are presented in Figure 2, calculated using the differences between the least and most preferred levels for each attribute. ‘Importance’ scores are shown for the four attributes, site of care/route of administration, pain relief, side effect type, and severity. Combining side effect type and severity, the figure shows a combined ‘importance’ weight of side effect type and severity that is slightly greater than that of pain relief.
Fig 2 Importance weights for setting/route, side effect type and severity, and pain control. Also illustrates the combined side effect type and severity ‘importance’ weight compared with those of pain control and setting/route.
Table 5 presents the ‘importance’ scores for interviews 1–4 and for different population subgroups. The ‘importance’ scores do not change between interview 1 and 4, although the patients are experiencing surgery and the post‐surgery recovery period. The ‘importance’ scores are also similar for all the subgroups analysed. Table 6 presents the results of the simulations of treatment preferences using the results of the conjoint analysis. These results demonstrate that patients have different preferences for different treatments with different combinations of attribute levels, as they do not all make the same choice. Also, the results demonstrate that some patients are willing to give up some pain efficacy for a reduced level of severity of certain side effects.
Table 5
Variation in average importance by interview number, gender, pain experience, and side effect experience. *Average importance scores for each attribute are presented for various samples; thus, for the total sample at visit 4, side effect type and severity had the greatest influence on preferences for treatment, with an importance score of 47.24%, followed by pain control (40.69%) and setting and route (12.07%)
Table 6
Preference shares for treatments with different attributes (two hypothetical treatments with different attributes are compared and the per cent of the sample preferring each treatment is reported; in the first example, 69.8% of the sample would prefer treatment B, while 30.2% would prefer treatment A)
Discussion
Patients’ relative preferences for pain vs type and severity of side effects are heterogeneous. Overall, patients placed almost equal importance in analgesic efficacy and the type and severity of the side effects when determining the desirability of the outcome of pain management. However, the relative preferences within attributes may be different for each patient as is shown by the standard errors for the utility estimates. The simulation results also demonstrate that people vary in their willingness to trade‐off pain efficacy for different or milder side effects.
The ‘importance’ scores were stable for the four conjoint interviews given pre‐surgery and post‐surgery. This result was expected as the conjoint survey asked patients about their preferences for hypothetical scenarios and not about their actual experiences. This stability supports the generalizability of the results from the conjoint estimation.
Results from this study are similar to the findings from a similar study investigating the relative preferences for chronic pain relief.13 Both studies demonstrated that site and delivery mode (PCA in the hospital vspills at home in this acute pain study; patch vs oral drugs in the chronic pain study) are of less importance to patients than pain relief and side effects. Also, both found that side effect type and severity and pain relief are equally important in determining the utility of the medication outcomes.
This study has several limitations. First, the sample size was small which limited our ability to test for differences in relative preferences between different subpopulations. Secondly, the Sawtooth software assumes that the utility associated with each attribute is independent of the level of the other attributes. For example, this means that the magnitude of the loss in utility from experiencing a side effect is assumed to be the same no matter what level of pain relief is experienced. Thirdly, this study has taken an ex post perspective and investigated hypothetical levels of pain relief and side effect type and severity. This approach has the advantage of generating clear estimates about the utility of different health states, but it does not allow us to estimate the impact of uncertainty in the outcomes. In addition, our study assumes that only one type of side effect is experienced at one time.
In summary, we demonstrated that to optimize patient care, it is necessary for physicians and other healthcare professionals to better understand how patients value these outcomes. Patients have different relative preferences for different types of side effects. They are also willing to trade off pain relief for less upsetting and/or less severe side effects but to different degrees. Such information may guide physicians in selecting pain medications that will best meet the patient’s needs and expectations and improve his/her compliance, treatment outcomes, and ultimately patient satisfaction.
Appendix
Attribute definitions
Pain relief
• Excellent
• Good
• Fair
• Poor
• Very poor
Side effects
• Constipation. You feel bloated and full. It is difficult to have a bowel movement, if you have one at all.
• Itching. You feel the need to scratch all or parts of your body.
• Mental cloudiness/dizziness. It is difficult to focus. You may be confused, disoriented, or have difficulty concentrating. You or your surroundings may appear to be moving or spinning. If you were to stand up, you might feel unsteady on your feet.
• Mood changes/alterations. You may have a rapid increase in your rate of breathing. If you already have symptoms of depression they may become worse. You may have periods of anxiousness, panic, excitability, restlessness, irritability, or mood swings. You may have decreased appetite.
• Nausea. You have an upset stomach and you feel like you might throw up.
• Nightmares/hallucinations. You may have frightening, strange, or vivid dreams. During waking hours, you may have hallucinations (you think things are happening when they are not).
• No side effects. You have no pain medication side effects.
• Sleep disorders. You have difficulty falling asleep or staying asleep.
• Vomiting. You have an upset stomach and are throwing up.
Route of pain medication
• PCA in the hospital. You are receiving i.v. pain medication following surgery. You can control the amount of pain medication you receive by pushing a button. When you push the button, you get more medication through a needle that has already been inserted in your arm.
• Pain pills in the hospital. You are taking pain pills by mouth following surgery while in the hospital. You take the pills as often as needed.
• Pain pills at home. You have been discharged from the hospital and are taking pain pills by mouth at home. You take the pills as often as needed.
Side effect severity
• Mild side effect. Your are not bothered very much by the side effect from your pain medication. You can easily cope with the side effect and do not need any additional medication to treat it. The side effect does not interfere with your daily functioning.
• Moderate side effect. You are somewhat bothered by the side effect from your pain medication. It is somewhat difficult to cope with the side effect and you may need additional medication to treat it. The side effect interferes with some of your daily functioning.
• Severe side effect. You are very bothered by the side effect from your pain medication. It is difficult to cope with the side effect and you need additional medication to treat it. The side effect interferes with most or all of your daily functioning.
- The Board of Management and Trustees of the British Journal of Anaesthesia
Nenhum comentário:
Postar um comentário