`
`DOl 10.1111/J.1365-2133.2005.07047.x
`
`Conjoint analysis: a novel, rigorous tool for determining
`patient preferences for topical antibiotic treatment for
`acne. A randomised controlled trial
`
`N. Kellett, F. West* and A.Y. Finlay~
`
`Inveresk Consumer Research Unit, Edinburgh, U.K.
`*Strakan Ltd, Buckholm Mill, Gdashiels TD1 2HB, O.K.
`"~Cardiff University-, Wales College of Medicine, Cardiff, U.K.
`
`Summary
`
`Correspondence
`Fiana West.
`E-mail: fiana.west@prostrnkan.com
`
`Accepted for publication
`? August 2005
`
`Keg words
`ache, clindamycin, compliance, conjoint analysis,
`topical antibiotic, zinc
`
`Conflicts of interest
`This study was funded by Strnkan Pharmaceuticals
`Ltd. A.Y.F. formerly had a paid c0nsultancy
`agreement with Strnkan Pharmaceuticals ttd and
`gives paid consultancy advice to Galdern~a.
`
`Background Ache vulgaris is an extremely common skin disorder that can be trea-
`
`ted effectively with drugs that are currently available. Poor compliance, however,
`
`is a major factor in the high failure rates seen in acne treatment. Compliance
`
`might be enhanced by considering patient preferences for acne medications. Con-
`
`joint analysis is well suited for the study of patient preferences in healthcare, but
`
`is novel to the field of dermatology.
`
`Objectives The study aimed to determine and compare patient preferences for four
`
`topical antibiotics used for 1 week, once or twice daily, to treat acne vulgaris.
`Methods A randomised, phase IV, single-centre, cross-over study was performed
`
`using conjoint analysis and a traditional patient questionnaire. Over 4 weeks, the
`patients used each of four topical antibiotics for 1 week: erythromycin/zinc solu-
`
`tion, clindamycin phosphate lotion, benzoyl peroxide (BP)/erythromycin gel
`
`(each applied twice daily) and clindamycin phosphate gel (applied once daily).
`
`The conjoint analysis examined five different attributes of acne medications:
`
`form, storage, product life once opened, method of application and regimen
`
`(each with two or three possible options). From 108 possible permutations of
`
`the five attributes, 16 hypothetical medications were selected at random and des-
`
`cribed on printed cards. Pre- and post-treatment, the patients ranked the cards in
`
`order of preference and rated each hypothetical product based on their likelihood
`
`to use it. For each patient, product ’utilities’ were then calculated by multiple
`
`regression. The patients also completed a patient acceptability questionnaire, by
`
`which they rated the product acceptability after 1 week of treatment with each
`
`of the four topical antibiotics. The patients later ranked the medications in order
`
`of preference after using all four treatments. Adverse events were recorded in
`
`diary cards to assess tolerability.
`Results Of 67 patients recruited, 64 used all four medications and completed the
`
`study. The conjoint analysis found that a gel formulation, room temperature stor-
`
`age, product life of up to 18 months once opened, application with fingers and
`
`once-daily regimen were the options ranked first for the five product attributes.
`
`According to the ranking order (out of 108) for the combination of attributes
`
`representing the four study medications, clindamycin phosphate gel had the
`highest rankings (6 and 1 pre- and post-treatment, respectively) and BP/erythro-
`
`mycin gel had the lowest rankings (93 and 70 pre- and post-treatment). The
`rankings of clindamycin phosphate lotion and erythromycin/zinc solution wor-
`
`sened from pre- to post-treatment, indicating a shift in patient preference after
`
`they experienced products ’in-use’ during the study. Based on the questionnaire,
`
`52/4
`
`© Z006 British Association of Dermatologists ¯ British Journal of Dermatology Z006 1.54, ppSZ4 53Z
`
`1 of 9
`
`Almirall EXHIBIT 2037
`Amneal v. Almirall
`IPR2018-00608
`
`
`
`Patient preference in acne treatment, N. Kellett et al. 525
`
`clindamycin phosphate gel was liked best by the highest proportion of patients
`(33%). In terms of overall satisfaction, the order of preference was: (i) dindamycin
`phosphate gel, (ii) clindamycin phosphate lotion, (iii) BP/erythromycin gel and
`(iv) erythromycin/zinc solution. Adverse events related to medication occurred
`most frequently with erythromycin/zinc solution and BP/erythromycin gel. Clin-
`damycin phosphate gel was the only product not associated with any episodes
`resulting in a change of medication or dose.
`Conclusions Conjoint analysis provided a convenient, reliable tool for assessing
`patient preferences for topical antibiotics used to treat acne. The patients clearly
`preferred a gel formulation that could be applied with the fingers once daily and
`stored at room temperature for as long as 18 months. One product (clindamycin
`phosphate gel) combined all five of the preferred attributes, a preference confirmed
`by the simulated product rankings. These findings of the conjoint analysis are con-
`sistent with the safety profiles and the results of the traditional questionnaire.
`
`Acne vulgaris is a common skin disorder that is experienced
`by most people at some stage during their lifetime1 and
`accounts for approximately 25% of patient visits in private
`dermatology practice in the U.S.A.2 During the last 20 years,
`the number of topical and systemic drugs available for the
`treatment of acne vulgaris has increased,3 with many offering
`good efficacy,
`Poor clinical response to acne treatment can result from
`factors such as poor compliance, inadequate duration of
`therapy and resistance of Propionibacterium aches to the antibi-
`otic administered.4 In dermatology, poor compliance with
`treatment is a well-recognized problem, and it is believed
`that 30 40% of patients using topical formulations fail to
`comply with their treatment regimen,s’6 While the reliable
`distinction between noncompliance and nonresponse is a
`new issue for medicine, the high failure rates in acne treat-
`ment appear to be linked directly to the poor compliance
`observed. As with prescribed treatment for any disorder,
`poor outcome adds extensive costs to the healthcare sys-
`tem.7 Therefore, physicians need to make a careful exam-
`ination of treatment compliance before investigating possible
`pharmacological reasons for drug failure or initiating alter-
`native treatments and special diagnostic tests.7
`Noncompliance is an especially important problem in treat-
`ing adolescents, and the success of treatment depends not on
`their parents’ involvement but rather on the patient’s own
`implementation of the regimen.* Motivating patients to adhere
`to treatment is a constant challenge. The greatest success seems
`to result from a multifactorial approach that combines non-
`pharmacological interventions (e.g. patient education and
`rewards, medication reminders, self-monitoring and peer sup-
`port) with effective, well-tolerated and simplified drug regi-
`mens.9 Another key to compliance is the patient physician
`interaction.l°’11 The formation of a ’therapeutic partnership’
`between physician and patient can promote compliance,12 and
`some authors have encouraged clinicians to consider the
`
`importance of patient preference and therefore offer patients a
`choice of acne formulation.1°’11
`Conjoint analysis is a well-established research technique
`used to predict the choices people will make when faced with a
`number of products that vary in terms of specific features or
`attributes.13 This technique has been used extensively in eco-
`nomics, psychology, marketing and statistics,13’14 but it is also
`particularly well suited for the study of patient preferences and
`how patients perceive and respond to the salient features of
`medical treatments,is Established as being both internally con-
`sistent and internally valid, conjoint analysis allows physicians
`to take account of patient preferences and product attributes
`beyond health outcomes.16’17 Ryan and Farrar completed a sys-
`
`tematic review of databases (Medline, Embase, HealthSTAR,
`PsychLIT and EconLIT) between 1989 and 199918 and revealed
`numerous examples of the application of this method in health-
`care. The authors concluded that ’conjoint analysis is a rigorous
`method of eliciting preferences...and allows estimation of the
`relative importance of different aspects of care, the trade-offs
`between these aspects, and the total satisfaction and utility that
`respondents derive from healthcare services’.
`In recent years, conjoint analysis has been used widely and
`successfully to assess patient preferences for treatments ranging
`from haemodialysis19 and anti-inflammatory drugs2° to
`human immunodeficiency virus testing21 and hearing aids.22
`The current paper, however, appears to represent the first
`study applying the method of conjoint analysis in the field of
`dermatology.
`The primary objective of this study was to determine
`patient acceptability of four topical antibiotic products when
`used for 1 week, once or twice daily, to treat acne vulgaris.
`Product acceptability was determined using a self-administered
`patient acceptability questionnaire by which patients graded
`each treatment in order of preference. Other endpoints of the
`study were a conjoint analysis of the patients’ product prefer-
`ences, the level of product use, safety and tolerability.
`
`© 2006 British Association of Dermatologists ¯ British Journal 0f Dermatolo~y 2006 154, pp524 532
`
`2 of 9
`
`
`
`526 Patient preference in acne treatment, N. Kellett et al.
`
`Patients and methods
`
`Keynes, U.K.), MA No. PL00032/0156 (applied twice daily);
`
`(iii) treatment C: Benzamycin® gel [benzoyl peroxide (BP)
`
`5% w/w and erythromycin 3% w/w; Bioglan Laboratories
`Study design Ltd, now Schwarz Pharma Ltd, Chesham, U.K.], MA No.
`
`This randomized, phase IV, single-centre, cross-over study
`
`PL4438/0063 (applied twice daily); and (iv) treatment D:
`
`was designed to compare the acceptability of four established
`
`Zindaclin® 1% gel (clindamycin phosphate equivalent to clin-
`
`topical antibiotics during 1 week of treatment for acne vul-
`
`damycin 1% w/w; Strakan Ltd, Galashiels, U.K.), MA No.
`
`garis. Three of the treatments are licensed for use twice daily
`
`PL16508/0011 (applied once daily).
`
`and the fourth for use once daily. The study centre (Inveresk
`
`Over a period of 4 weeks, every patient was to use each of
`
`Consumer Research Unit, Edinburgh, U.K.) is a clinical trial
`
`the four products for 1 week. This period was considered suf-
`
`unit experienced in the conduct of phase IV studies and was
`
`ficient time to allow the patients to evaluate product accepta-
`
`able to recruit the required number of patients during the 1-
`
`bility. For each patient, the order of the four study treatments
`
`month recruitment period (by local advertisement). The study
`
`was randomised to reduce bias in the assessments. These top-
`
`was approved by the Inveresk Research Ethics Committee prior
`
`ical treatments have minimal carry-over effect, and the attrib-
`
`to patients entering the study. The cross-over study design
`
`utes being assessed over each 1-week period did not relate
`
`made it possible to use a smaller number of patients for the
`
`directly to efficacy. A washout period between the treatments
`
`treatment comparisons because each patient acted as his/her
`
`was therefore not required.
`
`own comparator.
`
`At visit 2 (week 0), the first randomised medication was
`
`dispensed to the patient, who then applied it for 1 week
`
`Patients according to product-specific dosing instructions from the
`patient information leaflet for the corresponding medication.
`
`Seventy patients were to be recruited in order that 65
`
`In addition, patients were given the following instructions:
`
`patients would be evaluable for the primary variable of
`
`’Gently wash the affected area, rinse with warm water and
`
`assessment of product acceptability. This sample size was
`
`gently pat dry. Do not scrub your skin and use only mild
`
`chosen to ensure that a certain percentage point difference
`
`soaps or cleansing agents. If you do miss an application, reap-
`
`in preference between clindamycin phosphate gel and the
`
`ply the treatment as soon as you remember, but leave about
`
`next preferred product would be significant. The study’s
`
`3 hours before you use it again’.
`
`inclusion criteria specified male and female patients aged
`
`After 1 week of treatment, the patient crossed over to the
`
`16 40 years with mild-to-moderate acne graded between
`
`next randomised medication. This process was continued for a
`
`1"0 and 7"0 on the Leeds revised acne grading system.23
`
`total of 4 weeks, with the medications dispensed to patients
`
`Patients with acne conglobata, acne fulminans, sandpaper
`
`on a weekly basis. Patients were instructed to follow the same
`
`acne, submarine comedonal acne or secondary acne were
`
`skincare routines and to use the same skincare products (e.g.
`
`excluded. The use of topical or systemic antibiotics or top-
`
`soap, face-wash or moisturiser) for the entire study. Any
`
`ical antimicrobials within the previous week was not
`
`changes in skincare routine, as well as adverse events and con-
`
`allowed; however, these patients could enter the study fol-
`
`comitant medications, were recorded.
`
`lowing a washout period of 1 week, if all other eligibility
`
`criteria were satisfied. Patients giving informed written con-
`
`sent were screened for eligibility at visit 1 by determining
`
`Blinding
`
`their acne grade. At visit 2 (baseline, week 0), medical his-
`
`The marketed names of the medications were not revealed
`
`tory, including previous acne therapy, and all concomitant
`
`to the patients. For purposes of blinding, each medication
`
`medications were recorded. All women of childbearing
`
`was removed from its outer packaging, relabelled, and
`
`potential were asked to undergo a urine pregnancy test
`
`placed in new outer packaging, which was labelled A, B, C
`
`(which had to be negative), and sexually active women
`
`or D. Because the medications were not identical in presen-
`
`were required to use adequate contraception (combined oral
`
`tation, appearance, frequency or mode of application, com-
`
`contraception, barrier methods, intrauterine device, depot
`
`plete blinding was not possible. However, the study
`
`injection) throughout the study. Eligible patients were rand-
`
`personnel who performed the clinical assessments did not
`
`omised and allocated a unique treatment number,
`
`handle any of the medications and therefore remained
`
`Treatments
`
`blinded to the medications being used by patients. Person-
`
`nel who did not conduct any study assessments dispensed
`
`study medication and instructed patients on the use of
`
`The four medications used in the study were: (i) treatment A:
`
`medication.
`
`Zineryt® topical solution (erythromycin 40 mg and zinc acet-
`
`ate 12 mg mL ]; Yamanouchi Pharma Ltd, West Byfleet,
`U.K.), MA No. PL0166/0109 (applied twice daily); (ii) treat-
`ment B: Dalacin T® topical lotion (clindamycin phosphate
`
`Assessment of product acceptability and preference
`
`Product acceptability and preference were assessed using a
`
`equivalent to clindamycin 10 mg mL ]; Pharmacia Ltd, Milton
`
`self-administered patient acceptability questionnaire, conjoint
`
`© 2006 British Association of Dermatologists ¯ British Journal of Dermat010gy 2006 154, pp524 532
`
`3 of 9
`
`
`
`Patient preference in acne treatment, N. Kellett et al. 527
`
`Table 1 Possible attributes of hypothetical medications for acne
`
`Attributes
`
`Option 1
`
`Option 2.
`
`Form
`Storage
`Product life
`once opened
`Method of
`application
`Regimen
`
`Lotion/solution
`Gel
`Refrigerator Room temperature
`5 weeks
`3 months
`
`Option 3
`
`Cream
`
`18 months
`
`Roller ball
`
`Fingers
`
`Pad
`
`Once daily
`
`Twice daily
`
`analysis, and
`on weighing
`return.
`
`the measurement of product use based
`the medications when dispensed and on
`
`Patient acceptability questionnaire
`
`At the start of the study (visit 1), patients were given a sample
`questionnaire and asked to consider the questions each time
`they applied treatment during the week between visits 2 and
`3. At visit 3, they were asked to complete the questionnaire
`based on the acceptability of the product that they had used
`during the previous week. At visits 4, S and 6, this process
`was repeated for the other three products.
`The questionnaire asked three sets of questions. Firstly,
`
`the patients were asked to assess the acne product they had
`used during the previous week in terms of how closely it
`matched the following six statements: (i) ’The texture/con-
`sistency of the product was acceptable’. (ii) ’The product
`was absorbed easily into the skin’. (iii) ’The product
`smelled OK’. (iv) ’The product made my skin feel uncom-
`fortable’. (v) ’Make-up/facial skincare products were easy to
`put on after application of the product’. (vi) ’Overall, I am
`happy with the product’. For each statement, patients were
`to rate the product on a scale of 0 50, in increments of 1.
`A score of 0 indicated that the patient completely disagreed
`with the statement, and a score of 50 indicated that they
`completely agreed with the statement. Secondly, patients
`were also asked to comment on what they liked and dis-
`liked about each of the products. Thirdly, after completing
`the four treatment periods, each patient was asked to reflect
`on the four different acne products and to write down the
`product codes in order of preference, starting with
`the
`product they liked best.
`
`Conjoint analysis assessment of product preferences
`
`Conjoint analysis assumes that a product can be broken down
`into various characteristics (’component attributes’) and that
`the overall value (’utility’) that individuals place on any prod-
`uct is equal to the sum of the values (or ’utilities’) of all the
`product’s attributes)4 Product preferences are ascertained by
`having individuals rank and/or rate the products based on the
`component attributes, or in some cases individuals choose
`their preferred product (’discrete choice’).]*
`In this study, five different attributes of acne medications
`were identified: form, storage, product life once opened,
`method of application and regimen. Two or three possible
`options (or levels) were assigned to each attribute (Table 1).
`From the 108 possible permutations of the five attributes, 16
`hypothetical medications were selected at random. A descrip-
`tion of each hypothetical medication, including the selected
`option or level for each of the five different attributes, was
`printed on to cards (16 in total),
`At visit 2 of the study (before the start of treatment),
`patients were given the 16 cards corresponding to the 16
`hypothetical products. Patients were asked to read each card
`
`carefully and to consider how easy they would find each
`product to use and what they liked and disliked about it.
`Then they were asked to: (i) sort (rank) the cards in order
`of preference and (ii) rate each product (out of 100) based
`on how likely they would be to use it. A score of 100 rep-
`resented the ideal product, and a score of 0 represented the
`worst.
`To determine how experience of using the products might
`affect results, the patients repeated this conjoint assessment at
`the end of the study, after all four study medications had been
`used (visit 6).
`For each patient, product utilities were calculated by mul-
`tiple regression. The rating scores are the dependent variable
`for each regression, while the attribute levels (options) are the
`independent/predictor variables. The resultant regression coef-
`ficients are the utilities, and they were reported for the study
`populations by averaging across all patients.
`Conjoint analysis enables the investigator to simulate the
`’likelihood to use’ any combination of product attributes
`not actually shown to the respondent. Simulations for all
`108 possible permutations of acne medication were calcula-
`ted using the regression equation: Score Constant + Utility
`for ’Form is a gel or lotion/solution or cream’ + Utility for
`’Storage is in refrigerator or at room temperature’ + Utility
`for ’Product life is S weeks or 3 months or 18 months’
`+ Utility for ’Method of application is roller ball or fingers
`or pad’ +Utility for ’Regimen is once daily or twice
`daily’.
`For example, to simulate the score for erythromycin solu-
`tion, the utilities for the attributes that make up its product
`description were summed together with the constant.
`
`Diary card
`
`At visits 2, 3, 4- and 5, patients were given a diary card on
`which to record the time of each treatment application.
`Each patient was also asked to record information about
`any additional medication taken during the study, details of
`the skincare regimen used in the morning and evening, and
`any illness or unusual symptoms experienced. The comple-
`ted diary cards were returned at the following visit, and
`the key information was transcribed into the case report
`form.
`
`© 2006 British Association of Dermatologists ¯ British Journal of Dermatolo~y 2006 154, pp524 532
`
`4 of 9
`
`
`
`528 Patient preference in acne treatment, N. Kellett et al.
`
`Treatment compliance
`
`Product acceptability was also assessed based on patient com-
`pliance with treatment, which was determined using: (i) time
`of each treatment application recorded on the diary card; (ii)
`number of missed doses of study medication, as recorded in
`the case report form; and (iii) estimation of product usage
`over time (weight when dispensed minus weight when
`returned after use).
`
`Results
`
`Study patients
`
`Of 67 patients (25 males and 42 females) recruited, 64 used
`all four of the medications and completed the study. Three
`patients withdrew from the study, two because of adverse
`events unrelated to the study treatments and one as a result of
`voluntary withdrawal. Two additional patients discontinued
`use of one of two study medications (clindamycin phosphate
`
`Safety/adverse events lotion or BP/erythromycin gel) due to adverse events, but
`they both remained in the study. The median age at study
`entry was 23 years (range 16 40), and the median age at
`acne onset was 14 years (range 8 34). The median duration
`of acne was 8 years (range < 1 24), and the median acne
`grade at visit 1 was 1"75 (range 1 5"5).
`
`All local and systemic events were recorded. At each visit,
`patients were asked if there had been any problem with the
`treatment and if they had felt different since the last visit,
`Adverse events reported in the patient diary cards during
`the study treatment period were recorded in the case report
`form.
`
`Product acceptability results from questionnaire
`
`Analysis methods Order of preference
`
`The intention-to-treat (ITT) population, consisting of all
`randomised patients, was to be the main population for the
`evaluation of product acceptability. However, the primary
`variable could be assessed only in those patients who had
`used all four study medications (i.e. in the per protocol
`population). This population represented more than 90% of
`the ITT population and was used in all analyses, apart from
`the calculation of mean scores from the patient acceptability
`questionnaire,
`Paired comparison t-tests were conducted to establish
`whether the differences between clindamycin phosphate gel
`and each of the other three medications were significant in
`the analysis of data from the patient acceptability question-
`naire. Differences that were significant at the 90% and 95%
`levels were highlighted. For the pre- and post-treatment
`conjoint assessments, multiple regression was used to ana-
`lyse the likelihood to use each of the product attributes,
`Analyses of data from the patient acceptability questionnaire
`and the two conjoint assessments were performed using the
`statistical package SPSS version 6.1 (SPSS Inc., Chicago, IL,
`U.S.A.; April 1996).
`
`The primary variable for the evaluation of product accepta-
`bility was the order of preference of the four study prod-
`ucts, based on the patient acceptability questionnaire
`completed by patients following 1 week of treatment with
`each product. As shown in Table 2, clindamycin phosphate
`gel was the medication liked best by the highest proportion
`of patients (33%), while erythromycin/zinc solution was
`liked least by the highest proportion of patients (36%). The
`differences in order of preference did not reach statistical
`significance.
`
`Product scores
`
`For the patient ratings of the strength of their agreement with
`the six statements about the products, the scores (out of 50)
`were converted to percentages by multiplying by 2. The mean
`scores are presented in Table 3. There were no statistically sly-
`nificant differences between the four medications in terms of
`the acceptability of their texture/consistency. Clindamycin
`phosphate lotion was significantly less easily absorbed than
`the other three products. The smell of clindamycin phosphate
`
`Table Z Patient acceptability questionnaire: order of preference
`
`Number of patients (%) (n
`
`64)
`
`Preference Erythromycin/zinc solution
`
`Clindamycin phosphate lotion
`
`BP/erythromycin gel
`
`Clindamycin phosphate gel
`
`Liked best
`
`Liked 2nd best
`
`Liked 3rd best
`
`Liked least
`
`Total
`
`13 (20%)
`
`16 (28%)
`
`12 (19%)
`
`23 (36%)
`
`64 (100%)
`
`17 (27%)
`
`13 (20%)
`
`18 (28%)
`
`16 (28%)
`
`64 (100%)
`
`13 (20%)
`
`20 (31%)
`
`18 (28%)
`
`13 (20%)
`
`64 (99%)~
`
`21 (33%)
`
`18 (23%)
`
`16 (28%)
`
`12 (19%)
`
`64 (100%)
`
`BP, benzoyl peroxide. ~As a result of rounding to no decimal places, 1% is effectively ’lost’.
`
`© 2006 British Association of Dermatologists ¯ British Journal of Dermatology 2006 154, pp824 832
`
`5 of 9
`
`
`
`Table 3 Patient acceptability questionnaire: mean product scores (intention-to-treat population)
`
`Patient preference in acne treatment, N. Kellett et al. 529
`
`Erythromycin/
`
`zinc solution (k)
`
`Statements about products Mean scorea
`
`n Mean scorea
`
`BP/erythromycin Clindamycin
`Significance: clindamycin
`Clindamycin
`phosphate lotion (B) gel (C)
`phosphate gel (D) phosphate gel vs.
`treatment A, B, C
`P-valueb
`
`n Mean scorea
`
`n Mean scorea
`
`n
`
`Texture/consistency
`acceptable
`Easily absorbed
`Smelled OK
`Made skin feel
`uncomfortable
`Make-up/
`skincare products
`easy to apply
`afterwards~
`
`Overall, happy
`with product
`
`70"9 (1)
`
`65
`
`66"1 (3)
`
`66 65"5 (4)
`
`65 68"5 (2.)
`
`65
`
`72"1 (1)
`33"9 (4)
`50"1 (4)
`
`68
`.51"9 (4)
`68 80"8 (1)
`68 31"7 (1)
`
`66 68"i (2.)
`66 88"3 (3)
`66 49"2 (3)
`
`68 64".5 (3)
`68 64"7 (2)
`68 33"2 (2)
`
`68 a (P 0"098) B (P 0"018)
`68 a (P < 0"001) B (P < 0"001)
`68 A (P 0"003) C (P 0"001)
`
`62"4 (4)
`
`88 64"9 (2)
`
`88 64"0 (3)
`
`87 66"8 (1)
`
`83
`
`56"0 (4)
`
`68 60"6 (2)
`
`66 57"7 (3)
`
`68 67"0 (1)
`
`68 A (P 0"029) C (P 0"028)
`
`BP, benzoyl peroxide. ~Scores are presented out of a maximum score of 100. The higher the score, the more agreement with the statement.
`The numbers in parentheses (1, 2, 3, 4) indicate the corresponding order of preference, bDitTerences that are significant at the 90% and
`98% levels are highlighted (bold scores). ~The total number of patients for ’Make-up/skincare products easy to apply afterwards’ is lower
`than the total numbers for the other statements because some male patients did not respond to this question.
`
`lotion was the most acceptable and that of erythromycin/zinc
`
`was second (+ 0"22), and cream ranked top in likelihood
`
`solution the least acceptable. Clindamycin phosphate lotion
`
`to use (+ 0"32). However, in the post-treatment assessment
`
`and clindamycin phosphate gel were significantly more accept-
`
`following patients’ experience of using the medications, this
`
`able in not making the skin feel uncomfortable. In terms of
`
`ranking changed: gel became the top-ranked ’Form’ attrib-
`
`the ease of applying make-up or skincare products after topical
`
`ute (+ 2"11), cream was second (- 0"09), and lotion/solu-
`
`antibiotic use, clindamycin phosphate gel was the most
`
`tion ranked worst in likelihood to use (-2"02).
`
`acceptable, although the differences were not significant.
`
`Pretreatment, pad was the most preferred ’Method of appli-
`
`For overall satisfaction with the product, clindamycin phos-
`
`cation’ (+ 1"12), fingers were second (+ 0"91), and roller
`
`phate gel received the top score of 67"0, compared with 60"6
`
`ball ranked third in likelihood to use (- 2"03). Post-treat-
`
`for clindamycin phosphate lotion, 57"7 for BP/erythromycin
`
`merit, fingers took over the top-ranked position (+ 3"66),
`
`gel (P 0"028) and 56"0 for erythromycin/zinc solution
`
`pad was second (+ 2"42), and the score for the roller ball
`
`(P
`
`0"029).
`
`fell even further in its ranking as the least preferred method
`
`of application (-6"08). The rankings of the other three
`
`Results of conjoint analysis of preferences attributes (storage, product life once opened and regimen)
`remained similar pre vs. post-treatment. After the patients
`
`For the conjoint analysis, all 67 patients completed the pre-
`
`used the medications, the following options were ranked
`
`treatment assessment of the 16 hypothetical products, and 65
`
`first for the five attributes: gel formulation, room tempera-
`
`completed the post-treatment assessment, although one of
`
`ture storage, product life of 18 months once opened, appli-
`
`these 65 patients had used only three of the four study medi-
`
`cation with fingers and once-daily regimen.
`
`cations. This patient was included in the analysis, as she/he
`
`Of the four study medications, only clindamycin phos-
`
`had completed the post-treatment conjoint assessment. Figure 1
`
`phate gel combined all five of these top-ranked attribute
`
`shows the utilities of the options (or levels) for the five attrib-
`
`options. This finding is consistent with the patient scores
`
`utes (i.e. form, storage, product life once opened, method of
`
`(out of 100) and ranking order (out of 108 simulated
`
`application and regimen) both pre- and post-treatment. Utilit-
`
`products) for the combination of attributes representing the
`
`ies always sum to zero across the options for a particular
`
`four study medications, which are summarized in Table 4.
`
`attribute. For example, in the post-treatment assessment of
`
`Of these four medications, clindamycin phosphate gel had
`
`’Form’, the sum of the utilities is +2"11 + (-2"02)+
`
`the highest rankings (6 and 1) and BP/erythromycin gel
`
`(-0"09) 0.
`
`had the lowest rankings (93 and 70) both pre- and post-
`
`The most striking changes in the attribute utilities
`
`treatment, respectively. The ranking of clindamycin phos-
`
`occurred for the attributes of ’Form’ and ’Method of appli-
`
`phate lotion worsened from 23 pretreatment to 59
`
`cation’. In the pretreatment assessment of ’Form’, gel
`
`post-treatment, as did the ranking of erythromycin/zinc
`
`ranked worst in likelihood to use (- 0"55), lotion/solution
`
`solution (from 48 to 58).
`
`© 2006 British Association of Dermatologists ¯ British Journal of Dermatdo~y 2006 154, pp824 832
`
`6 of 9
`
`
`
`53o Patient preference in acne treatment, N. Kellett et al.
`
`1
`E] UTILITIES PRE (n = 67) UTILITIES POST (n = 65)J
`-4
`-2
`+0
`+2
`+4
`
`+6
`
`+8
`
`-8
`
`-6
`
`Gel -0.55 [~
`
`+2.11
`
`E +0.22
`Lotion/solution
`
`Cream -0.09
`
`+0.32
`
`Fridge -6.491 -5.04
`
`o
`
`Room
`temperature
`
`o~
`
`18 months
`
`.,=_
`-J ~
`
`3 months
`
`] +6.49
`
`+5.04
`
`1+5.15
`+4.90
`
`5 weeks
`
`Fingers
`
`Roller ball -2.03 ~
`-6.08
`
`Pad
`
`Once daily
`
`Twice daily
`
`._~
`
`._E
`
`~c
`
`~ +2.42
`
`*Total for pre-treatment ’Form’ equals -0.01 due to rounding figures to 2 decimal points
`
`Pretreatment
`
`Post-treatment
`
`Medication
`
`(out of 100)
`
`(out of 108)
`
`(out of 100)
`
`(out of 108)
`
`Score
`
`Rank
`
`Score
`
`Rank
`
`Erythromycin/zinc solution
`
`88"4
`
`Clindamycin phosphate lotion
`
`64"3
`
`BP/erythromycin gel
`
`Clindamycin phosphate gel
`
`47"0
`
`70"9
`
`48
`
`23
`
`93
`
`6
`
`88"7
`
`88"6
`
`83" 1
`
`74"8
`
`88
`
`89
`
`70
`
`1
`
`Fig 1. Conjoint analysis: graphical
`representation of utilities for the attribute
`options pre- and post-treatment.
`
`Table 4 Conjoint analysis: pre- and post-
`
`treatment scores and ranking order for the
`
`four study medications
`
`BP, benzoyl peroxide.
`
`Safety results
`
`phosphate lotion (six). Other frequently reported related
`
`events were an aggravation of acne (29 events), pain (27) and
`
`During the study, 283 adverse events were reported by 56
`
`sk