Effects of Perceived Treatment on Quality of Life and Medical Outcomesin a Double-blind Placebo Surgery Trial FREE

A relatively new experimental approach to treating Parkinson's disease(PD) is the stereotactic implantation of human embryonic dopamine neuronsinto the brains of persons with severe PD. Unblinded clinical trials^1- 21 haveshown that this surgical procedure ameliorates some of the symptoms of thedisease and that the transplanted dopamine neurons survive. Very little researchhas focused on subjective changes in quality of life (QOL) that may resultfrom the surgery.²¹

To assess the efficacy of fetal transplant surgery, the procedure wassubjected to a double-blind sham surgery–controlled trial, with halfthe patients receiving the transplant and half receiving sham surgery.²² The study design not only included the unusual conditionof sham surgery, but it also maintained the double-blind phase for 1 year,a very long time for placebo-controlled trials.²³ Theseconditions combined to provide unique opportunities to investigate the effectsof sham surgery as well as the benefits of transplant surgery and to determinethe effects of a 12-month double-blind trial. One of the primary goals ofthe "parent" study was to determine whether the signs and symptoms of PD improvedmore in transplant recipients than in those who received the sham surgery.Similarly, one of the goals of this (QOL) study was to determine whether QOLimproved more in the transplant group than in the sham surgery group. A relatedgoal was to investigate outcomes at 1 year based on perceived treatment, orthe type of surgery patients thought they received. The final goal of thisstudy was to examine medical outcomes based on actual as well as perceivedtreatment to determine whether perceived treatment was related to ratingsby medical staff at 12 months.

A complete description of the parent study can be found in the articleby Freed et al.²² The following subsectionsdescribe the methods of the QOL study.

Forty persons from across the United States and Canada were recruitedto participate in this double-blind surgical trial. Twenty patients were randomlyassigned to receive the fetal tissue transplant and 20 patients were randomlyassigned to the sham surgery condition. Patients were told that those in thesham surgery group would have the option of receiving the transplant afterthe blind was lifted. Inclusion criteria for the study were (1) a diagnosisof idiopathic PD of at least 7 years' duration; (2) a continuing responseto the administration of levodopa, the principal drug used to treat PD; and(3) the presence of an intractable problem, such as "off" periods, dyskinesias,or "freezing" not controlled by dopamine agonist therapy. Exclusion criteriaincluded (1) obvious depression or cognitive impairment (as assessed by aneuropsychologist), (2) previous brain surgery, or (3) the presence of diabetesmellitus, severe cardiopulmonary disease, other severe medical disease, ormagnetic resonance imaging evidence of cerebrovascular disease. After rigorousscreening, patients who were accepted into the parent study were invited toparticipate in the QOL portion of the project. Thirty patients agreed to participate.All neurologic examinations were performed at Columbia-Presbyterian MedicalCenter (CPMC).

Quality of life is a multidimensional construct that was measured usinga variety of widely recognized, commonly used instruments focusing on 3 fundamentalaspects of QOL: Physical, Emotional, and Social functioning. Data were collectedby sending questionnaires to patients 1 week after scheduled evaluation visitsto CPMC. Participants were asked to respond to the questionnaire and returnit in the postage-paid envelope provided. Patients were paid $25 for eachcompleted questionnaire. The response rate was 98%. The scales contributingto each aspect of QOL are presented in the following subsections.

Physical Functioning. UnifiedParkinson's Disease Rating Scale (patient version). This scale²⁴ is an adaptation ofthe physician version of the Unified Parkinson's Disease Rating Scale (UPDRS)that is uniformly used to rate the physical functioning abilities and problemsof patients with PD.²⁵ The patient versionwas developed for a health promotion program that was disseminated throughthe mail.²⁴ Four primary subscales were usedin this version of the UPDRS: Activities of Daily Living (ADLs) at "best"and "worst" and Severity of Symptoms at "best" and "worst." Correlations ofthe patient version with a variety of the UPDRS subscale scores rated by medicalstaff range from r = 0.58 to r = 0.71.²⁶ Because previous results²² have shown that scores "off" medications providea more valid assessment of patient status, only the "off" scores (worst) fromthe patient UPDRS were included in the composite variable. Lower scores indicatebetter outcomes. The estimate of reliability (Cronbach's α) for the"worst" scores was .84 at baseline.

"Free or Restricted". This single, global itemmeasures how free or restricted the person feels "in doing what you want todo." A Likert scale ranging from 1 (I still do everything I want to do) to7 (I can no longer do the things I want to do) is used. Lower scores indicatebetter outcomes.

Emotional Functioning.Parkinson'sDisease Stress Scale. This scale was developed for use with Germanpatients with PD and is available to interested readers on request (J. H.Ellgring, PhD, M. Macht, PhD, R. Schwarz, MA, unpublished data, 1993). Thescale has 19 items, with lower scores indicating less stress. Cronbach's αfor the scale was .77.

Center for Epidemiologic Studies–DepressionScale. This scale²⁷ is a 20-item self-reportmeasure of depressive symptoms. Each item is rated on a 4-point scale relatedto frequency; for example, 0 indicates "less than 1 day" and 3 indicates "5to 7 days." Lower scores indicate fewer depressive symptoms. The estimateof reliability (Cronbach's α) was .89.

State-Trait Anxiety Inventory. The "State"version²⁸ of this well-validated instrumentwas designed to assess the current, situational aspects of anxiety. The 20items are rated on a 4-point scale ranging from 1 (not at all) to 4 (verymuch). Lower scores indicate less anxiety. Cronbach's α for the scalewas .89.

Intrusiveness of Illness Scale. This scale²⁹ is an adaptation of one designed to measure the degreeto which a chronic illness interferes with usual life activities. A scaleranging from 1 (very little) to 7 (a great deal) was used for each item. Lowerscores on this 17-item scale indicate less intrusiveness of illness. The estimatedreliability of the scale (Cronbach's α) was .88.

Social Functioning.SocialProvisions Scale. This 24-item scale³⁰ assessesperceived social support. Each item is rated on a 4-point scale ranging from1 (strongly disagree) to 4 (strongly agree). Lower scores indicate less perceivedsupport. The reliability estimate (Cronbach's α) at baseline was .91.

Social Contact. A measure of social contactwas developed for this study to assess the amount of social interaction oractivity experienced by each participant. Three items assessing frequencyof socializing with friends, telephone communication, and participation inpublic activities were scored on a 6-point scale ranging from 1 (not at all)to 6 (every day). Lower scores indicate less social contact. Estimated reliability(Cronbach's α) was .69 at baseline.

Patients were evaluated at the Irving Center for Clinical Research atCPMC twice before surgery and 4, 8, and 12 months after surgery. Each assessmentlasted 3 to 4 days, allowing for evaluations on and off medications. Eachpatient was followed by a neurologist and by 1 of 2 research nurses (S.D.and H.W.) throughout the study. Medical staff were unaware of treatment groupassignment for the duration of the study.

Global Rating Scale. The primary outcome variableof the parent study was a single item representing a subjective global ratingof change in severity of disease.²² Possiblescores ranged from –3 (much worse since surgery) to +3 (much improvedsince surgery). One week before the follow-up evaluations at 4, 8, and 12months, patients filled out global ratings of their health status relativeto their condition just before surgery. Medical staff completed the same evaluationat each patient visit. No reliability or validity information is availablefor this scale.

Perceived Surgery. Seven days after surgeryand 1 week before the follow-up evaluations at 4, 8, and 12 months, patientswere asked to indicate whether they thought they received the transplant orsham surgery. These evaluations were mailed to patients by the staff at CPMC.Evaluations were returned to the biostatistician at CPMC and were not revealedto the medical staff.

Unified Parkinson's Disease Rating Scale. TheUPDRS²⁵ is a standard instrument used to assessvarious aspects of PD, including motor performance, muscle rigidity, tremor,speech, and gait. Lower scores indicate better performance. The UPDRS hashigh interrater³¹ and test-retest³² reliability.

Schwab and England Activities of Daily Living Scale. The Schwab and England³³ assesses abilityto perform activities of daily living on a scale ranging from 0% to 100%,with higher scores indicating more normal performance. Descriptive anchorsfor the scale are presented in increments of 10, for example, 100, 90, 80,etc. This is a standard assessment instrument in PD and has well-establishedreliability and validity.³¹

Hoehn and Yahr. The Hoehn and Yahr³⁴ is the standard disease staging scale for PD. Scoresrange from 0 (no signs of disease) to 5 (wheelchair bound or bedridden unlessaided). Lower scores indicate fewer signs of disease. High interrater reliabilitybetween neurologists, patients, and caregivers has been found.³⁵

For the purpose of this study, "off" medications was defined as beforeadministration of the first morning dose of levodopa and at least 12 hoursafter the last administration of levodopa the previous day.³⁶ Asnoted earlier, only "off" scores were included in data analyses. Patientsand medical personnel completed the Global Rating Scale, the Schwab and England,and the Hoehn and Yahr at each assessment period. The standard version ofthe UPDRS was completed only by medical personnel.

All patients had a stereotactic frame attached to the head for magneticresonance imaging to establish the coordinates for the needle tracks for implantationof the tissue. Four burr holes were drilled bilaterally in the forehead toaccommodate the needles. Patients were awake during the procedure. Scalp incisionsand twist drill holes were done under local anesthesia. Both groups receivedidentical preoperative evaluation and intraoperative sedation and pain control.For patients in the transplant group, implants were made into the putamenwith embryonic mesencephalic tissue containing dopamine neurons. Each needlepass into the putamen contained tissue from a single embryo. Four embryoswere used for the 4 sites implanted. Needles for participants in the placebocondition remained empty and did not penetrate the brain. No patient receivedimmunosuppressive therapy. The dialogue during the surgery was limited andunscripted. Additional information related to the surgery can be found inthe article by Freed et al.²²

Preliminary analyses were performed to determine relationships amongthe variables predicted to form the 3 aspects of QOL. Correlations for thePhysical functioning variables ranged from r = 0.30to r = 0.75. Relationships among the Emotional functioningvariables ranged from r = 0.63 to r = 0.77. Reliabilities of the linear composites for these constructsat baseline and at 4, 8, and 12 months ranged from 0.72 to 0.89. The 2 variablespredicted to constitute Social functioning did not correlate highly enoughto be combined into one composite measure. Neither was the reliability ofthe linear composite high enough to justify combining the variables. Therefore,measures of perceived Social Support (Social Provisions Scale) and SocialContact were analyzed separately. For the composite variables (Physical andEmotional functioning), scores for each measure were standardized at eachperiod based on the mean and standard deviation of the measure at baseline.Standard scores for each measure were then combined to create a total z score representing the composite variable.

Phi coefficients were calculated to determine the relationship betweenactual and perceived treatment at 7 days and at 4, 8, and 12 months. Independentsamples t tests were used to examine differencesbetween the transplant and sham surgery groups and between the perceived treatmentgroups on the 3 QOL constructs at each time. t Testswere also used to compare the same groups on the medical outcome variables.The repeated-measures generalized estimating equation was used to examinechange over time. Based on the assumption that characteristics of each personare correlated over time, all observations (baseline and 4, 8, and 12 months)were entered simultaneously into these analyses. Analysis of variance techniqueswere used to examine differences in change scores. All P values are 2-tailed. No adjustments were made for multiple comparisonsbecause of the exploratory nature of the analyses and because there are nocomparable data from other studies.

The demographic information for the 30 participants in the QOL portionof the study is presented in Table 1.There were no statistically significant differences between the transplantand sham groups on demographic characteristics or QOL variables at baseline.

Membership in the perceived treatment groups changed at each periodas patients' perceptions of which treatment they received changed over time. Table 2 indicates the type of surgery patientsthought they received at each period as well as the surgery they actuallyreceived. Results show that 22 (76%) of 29 participants thought they had receivedthe transplant 7 days after the procedure. Based on presurgical interviews,this finding seems to reflect the belief of most patients that they wouldbe among those who received the transplant initially. This number was reducedto 10 (33%) of 30 participants 12 months later. Phi coefficients, which measurethe relationship between dichotomous variables, ranged from 0.00 to 0.15 andwere not statistically significant, indicating no relationship between thetype of treatment patients actually received and what they thought they receivedat any follow-up period.

The results in this section are presented in 2 parts. First, resultsrelated to actual transplant and sham surgery groups are presented, followedby results related to perceived treatment groups, or the type of surgery patientsthought they received.

Results from analyses at baseline and 4, 8, and 12 months indicatedthat the only significant difference between the transplant and sham surgerygroups was in Social Contact at 4 months, with the sham surgery group reportingmore social contact (P = .03).

Using repeated-measures analyses to examine change over time, therewas a significant improvement in Physical functioning in both groups over12 months (P = .01) (Figure 1). There was also a significant time × treatment interactionfor Physical functioning between the 2 groups over the 4 assessment periods(P = .04). As can be seen in Figure 1, marked improvement from baseline to 4 months was seenin the transplant group. There was a downward linear trend indicating improvementover time in the sham surgery group.

When analyses were conducted based on perceived treatment groups, orthe treatment patients thought they received at each assessment period, therewere several differences between the groups, all favoring the perceived transplantgroup (Table 3). At 8 and 12 months,those who believed they received the transplant reported better Physical functioningthan those who thought they received the sham surgery (P = .02 and P = .03, respectively). SocialSupport at 8 months (P = .01) and 12 months (P = .03) was better for those who thought they receivedthe transplant vs the sham surgery.

Using repeated-measures analyses and based on perceived treatment at12 months, there was improvement in Physical functioning over time in theperceived groups (P = .01) (Figure 2). A significant interaction between time and perceivedgroup was found for Social Support (P = .05), withthose who thought they received the sham surgery reporting less support at8 and 12 months than those who thought they received the transplant.

To reduce the number of comparisons, and thus the possibility of typeI error, only statistical analyses using the baseline and 12-month assessmentswere conducted.

On ratings by medical staff, there were no differences between the transplantand sham surgery groups at 12 months on the Global Rating Scale, UPDRS andits subscales, Schwab and England, or Hoehn and Yahr ratings. Likewise, therewere no differences between groups on medical variables rated by patients(ie, Global Rating Scale, Schwab and England, and Hoehn and Yahr).

Among medical staff, statistically significant changes over time (baselineto 12 months) were found for 2 variables. First, ratings for transplant andsham surgery groups improved for UPDRS tremor scores in the "off" state (P = .01). Second, an interaction effect was found on theUPDRS rigidity "off" scores (P = .04), with the transplantgroup improving and the sham surgery group getting worse. There were no changesover time in patients' self-report ratings of the medical variables.

Although blind as to the type of treatment patients actually received,medical personnel reported several differences based on the type of treatmentpatients thought they received at 12 months. Results presented in Table 4 show statistically significantdifferences on several clinical ratings based on patients' perceived treatment.Patient ratings of medical variables also showed statistically significantdifferences (Table 4). In allcases, results were better for those who thought they received the transplantvs the sham surgery.

In medical staff ratings, there were several changes over time (baselineto 12 months) based on perceived treatment or the treatment patients thoughtthey received at 12 months (Table 5).Similar to the actual treatment groups, medical personnel rated both perceivedgroups as improving on the UPDRS tremor "off" (P =.02). On all other measures listed in Table5, interaction effects indicated that the perceived transplant groupwas getting better and that the perceived sham surgery group was getting worse.Regarding patient ratings given in Table5, patients who believed they received the transplant at 12 monthsshowed improvement in global rating scores over time, whereas those in theperceived sham group showed a decline in scores (P<.001).

These results are somewhat different from those reported for the parentstudy²² because of the difference in samplesize (n = 39 in the parent study, with 19 persons receiving transplants, vsn = 30 in the QOL study, with only 12 receiving transplants). In the parentstudy,²² there was a significant differencein improvement (change scores) in UPDRS motor "off" scores between the actualtransplant and sham surgery groups (P = .04). Asshown in Figure 3, differences inimprovement for the perceived treatment groups were also significant (P<.05). In the parent study, scores for Schwab and England"off" improved significantly more for the actual transplant group than forthe actual sham group over 12 months (P = .04), asdid scores for the perceived transplant group relative to the perceived shamsurgery group (P<.01). Because no statisticallysignificant interactions were found in these analyses, results indicate thatactual and perceived treatment had independent effects on the outcomes.

One of the primary goals of this study was to determine whether QOLimproved more in the transplant group than in the sham surgery group duringthe 1-year period of the double-blind study. There was only one statisticallysignificant difference between the transplant and sham surgery groups in termsof QOL. Results of comparisons at 4, 8, and 12 months revealed a differencein Social Contact at 4 months, with the transplant group reporting less socialinteraction than the sham surgery group. Both groups showed statisticallysignificant improvement in Physical functioning during 12-month follow-up(Figure 1). These results, therefore,do not indicate statistically significant improvement in QOL in the transplantgroup over the sham surgery group by the end of the initial phase of the study.

The reasons for the transplant group reporting less social interactionare unclear. This result may reflect a type I error, or it may be relatedto the subtle but prolonged recovery period reported by some patients whoreceived the transplant.³⁷ Regarding both groupsreporting improvement in Physical functioning (Figure 1), it is possible that hope or optimism contributed to thoseresults. Individuals desperate enough for relief from symptoms to risk theconditions of this study may have been inclined to imagine themselves gettingbetter over time, particularly when they did not really know what "gettingbetter" might look or feel like.

Because the study involved the unusual element of sham surgery and thedouble-blind phase lasted approximately 13 months, one of the goals of thestudy was to investigate the effect of perceived treatment on QOL outcomes.Results indicated that the sustained improvement in Physical functioning from4 to 12 months among the perceived transplant group (Figure 2) was accompanied by relatively stable scores in Emotionalfunctioning and Social Support over time (both of these remained in the samerange as normative scores throughout the 12 months) (Table 3). Conversely, the perceived sham surgery group improvedonly slightly in Physical functioning over time (Figure 2) and subsequently declined in Emotional functioning andSocial Support at 8 months, with a slight improvement by 12 months (Table 3). Overall, results indicated thatperceived treatment was more strongly related to outcome for the durationof the 12-month double-blind study than was the actual treatment they received,with perceived transplant patients consistently reporting better QOL thanthe perceived sham surgery group. Thus, it seems that the placebo effect wasvery strong.

The final goal of this study was to examine differences and changesover time in medical outcomes based on actual and perceived treatment. Therewere no differences on the medical variables as rated by medical staff orpatients at 12 months based on actual treatment in the subset of 30 patients.By contrast, the parent study showed statistically significant improvementin UPDRS motor "off" scores and Schwab and England "off" scores in transplantpatients, with no change in sham surgery patients.²²

When analyses were performed based on perceived treatment, several differenceson medical variables at 12 months were found (Table 4). All differences, as rated by medical staff and patients,indicated that those who thought they received the transplant were doing betterthan those who thought they received the sham surgery. Analyses using datafrom the total group of 39 patients (1 patient died in a car accident beforethe 12-month evaluation) produced similar results, wherein medical staff ratedthe perceived transplant group as performing statistically significantly betterthan the perceived sham group.

Regarding changes over time, most were the result of interaction effectswherein those who thought they received the transplant improved and thosewho thought they underwent sham surgery got worse (Table 5). Similar results were found in analyses including the totalgroup of participants in the parent study, that is, interaction effects werefound.

These results suggest that expectancy regarding which type of treatmentpatients received had a statistically significant effect not only on subjectiveparameters (Emotional functioning and Social Support) but also on motor symptoms(Physical functioning). The effects on motor symptoms were also recognizedby the medical staff, as shown in their clinical ratings (Table 4 and Table 5 and Figure 3). Whether staff ratings of motorsymptoms were affected by mood and other subjective expressions of the patients(similar to a "halo" effect) cannot be determined in this study. An alternativeexplanation is that actual changes in physical functioning led patients tobelieve that they got the transplant (or not) and resulted in changes in ratingsby the medical staff.

Similar results related to the placebo effect have been found in otherstudies with patients with PD.^38- 40 Theseresults underscore the need for placebo controls in studies evaluating treatmentfor PD as the placebo effect seems to be very strong in this disease.

In terms of the larger issue of the expectancy or placebo effects, resultssuggest that the possible effectiveness of double-blind conditions in somestudies may be longer than previously believed. The average length of timeof a double-blind trial is 8 weeks.²³ The double-blindin this study was effectively maintained for 12 months. It seems plausiblethat because effects of the surgery were not predictable or definitive overthe period of the double-blind, patients did not have clear cues as to whichsurgery they had received. Research^41- 43 indicatesthat the more extreme the placebo treatment is in a clinical trial, the moresusceptible participants are to the placebo effect, or believing that theyare being helped by the sham medication or condition. Because this study involvedbrain surgery, arguably an extreme placebo treatment, and results were notclearly discerned by patients or staff for the double-blind period, it seemsthat conditions were favorable for evoking a strong placebo response in thisstudy. Although this study did not include a standard care group that didnot intend to receive the surgery, which would have provided the most rigoroustest of the effects of transplant surgery, these results are consistent witha strong placebo effect.

The lack of relationship between actual and perceived treatment throughoutthe study underscores several important points. First, it indicates the successof the conditions of the sham surgery. Surgical staff performing the surgeriesseem to have been convincing in their presentation of the placebo treatment(Table 2). Second, the shift from22 persons (76%) believing that they got the transplant at 7 days to only10 (33%) at 12 months indicates that optimism or hope may have influencedperceptions early in the study. The percentage of persons thinking that theyreceived the transplant was similar in the parent study: 28 (72%) at 7 daysand 14 (36%) at 12 months. What contributed to the change in ratings duringthe 12-month period is beyond the scope of this article. We speculate thatit may have been perceived changes in physical condition, lack of perceivedchanges, or diminished optimism as results of the study failed to meet expectations.

In summary, there were more differences and changes over time in QOLamong perceived treatment groups than among actual treatment groups. Medicalstaff, who did not know which treatment each patient received, also reportedmore differences and changes at 12 months based on patients' perceived treatmentthan on actual treatment. These results suggest that patients' perceptionsinfluenced their behavior, which in turn resulted in ratings among medicalstaff that paralleled patients' self-ratings. It is also possible that subtlechanges in physical functioning influenced patients' perceptions, which theninfluenced behavior. These data cannot be used to determine causal relationshipsamong these indicators of improvement. However, the results clearly show thatimprovement, or lack of improvement, was not solely determined by actual treatmentat this point in the study. Analyses based on the type of surgery patientsthought they received produced results that have implications for researchrelated to double-blind placebo-controlled trials and the length of time ablind can be maintained.

Although the sham surgery research design is currently regarded as somewhatcontroversial and has raised some important ethical concerns,⁴⁴ theinvestigators of the parent study determined that the scientific benefitsof this design outweighed potential ethical considerations. The results ofthis study demonstrate the importance of a double-blind design to distinguishthe actual and perceived values of a treatment intervention.

Corresponding author: Cynthia McRae, PhD, College of Education, Universityof Denver, Denver, CO 80208 (e-mail: cmcrae@du.edu).

Submitted for publication September 30, 2002; final revision receivedOctober 21, 2003; accepted October 29, 2003.

This study was supported in part by grants R29 NS32009-05 (Dr McRae)and R01 NS32368 (Dr Freed) from the National Institute of Neurological Disordersand Stroke, Bethesda, Md.