GSK Clinical Study Register

In February 2013, GlaxoSmithKline (GSK) announced a commitment to further clinical transparency through the public disclosure of GSK Clinical Study Reports (CSRs) on the GSK Clinical Study Register. The following guiding principles have been applied to the disclosure: Information will be excluded in order to protect the privacy of patients and all named persons associated with the study Patient data listings will be completely removed* to protect patient privacy. Anonymized data from each patient may be made available subject to an approved research proposal. For further information please see the Patient Level Data section of the GSK Clinical Study Register. Aggregate data will be included; with any direct reference to individual patients excluded *Complete removal of patient data listings may mean that page numbers are no longer consecutively numbered

Avodart Establishing Predictors of EP Treatment Adherence: Linking Symptom Improvement to Adherence - A Prospective Patient & Provider Study (8AVO112597) Updated Study Report Revised June 11, 2010 (Revised September 7, 2010) Prepared for: GlaxoSmithKline

Avodart Establishing Predictors of EP Treatment Adherence Please direct comments and questions to: PharmD, PhD Vice President Xcenda 4114 Woodlands Parkway, Suite 500 Palm Harbor, FL 34685 Phone: Fax: PharmD, MS Director Xcenda 4114 Woodlands Parkway, Suite 500 Palm Harbor, FL 34685 Phone: Fax: GlaxoSmithKline Page 2 of 30

Avodart Establishing Predictors of EP Treatment Adherence TABLE OF CONTENTS 1.0 INTRODUCTION... 4 2.0 METHODOLOGY AND PROTOCOL... 4 2.1.1 Study Objectives... 4 2.1.2 Survey Development and Deployment... 4 2.1.3 Steering Committee... 5 2.1.4 Respondent Sample... 5 2.1.5 Screening Criteria... 5 2.1.6 Analytical Approach... 5 2.2 Survey Questions... 9 2.2.1 Patient Survey Questions... 9 2.2.2 Physician Survey Questions... 17 3.0 RESULTS... 21 3.1 Description of Study Sample: Patients... 21 3.1.1 Patient Sample... 21 3.1.2 Compliance and Symptom Relief... 22 3.1.3 Efficacy and Impact on Adherence... 25 3.1.4 Patient IPSS Scores... 27 3.1.5 Patient Quality of Life... 27 3.2 Description of Study Sample: Physicians... 27 3.2.1 Physician Sample... 27 3.2.2 Physician Views on Factors Likely to Increase Patient Compliance... 28 3.2.3 Physician Perceptions on Urinary Symptoms... 29 4.0 CONCLUSIONS... 30 GlaxoSmithKline Page 3 of 30

Avodart Establishing Predictors of EP Treatment Adherence 1.0 INTRODUCTION In collaboration with GlaxoSmithKline, Xcenda evaluated symptom improvement as a driver of medication adherence in men being treated for enlarged prostate (EP) with an alpha reductase inhibitor (5ARI), alpha blocker (AB), or combination therapy. Analyses have shown that medication adherence is strongly associated with improved outcomes in patients with EP. Specifically, data suggest that a 10-point increase in 5ARI adherence may result in a 7% reduction in the need for prostate surgery. Despite the established benefits of EP treatment, factors that drive medication adherence are not well understood. GlaxoSmithKline surveyed both patients and physicians to assess and quantify factors that drive BPH medication adherence, in part to assess the role of symptom improvement on patient adherence. The following report details the process for the completion of the primary research and the results. 2.0 METHODOLOGY AND PROTOCOL 2.1.1 Study Objectives The objective of this study was to evaluate symptom improvement as a driver of medication adherence in patients treated with a 5ARI, AB, or combination therapy for EP from the patient and physician perspectives. Research data were collected by conducting online surveys sent to both patients and physicians (ie, primary care physicians [PCPs], urologists). Two separate web-based surveys were developed to assess and quantify factors relating to medication adherence in a population of men being treated for EP. 2.1.2 Survey Development and Deployment Patient Survey The first survey was administered to patients to assess their opinions and behaviors for taking their prostate medication(s). A minimum of 400 completed surveys was the target threshold. Patients were asked which medications they have been prescribed and are currently taking, how often they take their medications as prescribed (compliance), questions on symptom relief, questions on their beliefs toward taking medications, and questions relating to patient demographics. Two validated survey tools, the International Prostate Symptom Score (IPSS) and the Quality of Life Short Form 12 (SF-12, v.2), were also incorporated into the survey. The survey assessed: Basic demographic characteristics Perceived level of adherence based on a continuum of medication adherence statements Level of agreement with statements about factors related to adherence, such as symptom control, delivery of care, long-term prevention Physician Survey The second survey was administered to physicians to assess their perceptions on factors likely to affect patient adherence. A minimum of 100 completed surveys was the target threshold. Both urologists and PCPs were included. Physicians were asked why they might prescribe one agent over another (5ARI vs AB), how quickly patients exhibit symptom relief, and their perceptions on factors affecting adherence. The patient and physician surveys overlapped in content, both addressing the same broad topics, including adherence levels and factors impacting adherence. GlaxoSmithKline Page 4 of 30

2.1.3 Steering Committee Avodart Establishing Predictors of EP Treatment Adherence A multidisciplinary steering committee was convened to oversee this study. The purpose of the steering committee was to ensure that the study was conducted with high scientific rigor, validity, clinical accuracy, and appropriate statistical methods. The steering committee members were involved in every step of the research process, from conceptualization to interpretation and reporting of results. The members of the steering committee were: PhD, MD, MD, 2.1.4 Respondent Sample Xcenda worked with Harris Interactive, a consumer market research firm, to obtain the necessary patient and physician sample for the surveys. Harris leveraged its database of patients and physicians to obtain the study sample (patient survey: 400; physician survey: 100). Harris coded the survey content into their online tool and launched the survey to their sample through the internet. Harris subsequently provided the raw data to Xcenda. 2.1.5 Screening Criteria The following screening criteria were used for the surveys: Patient Survey Inclusion Criteria Exclusion Criteria Male Receiving medications for EP on a PRN basis 50 years of age No health/prescription insurance Resident of the United States (US) Diagnosis of EP Diagnosis of prostate cancer Medication prescribed to treat EP Initiated medication within the past 12 months* Physician Survey No specific screening questions were applied to the physician survey. The target sample was to consist of 90% PCPs and 10% urologists. *Screening criteria was modified midway through the survey administration time period. Initial criteria were set for medications prescribed within the past 6 months but were relaxed due to an insufficient response rate. 2.1.6 Analytical Approach Responses for the patient and physician surveys were summarized using descriptive statistics, including frequencies, proportions, means, and standard deviations. The association of symptom improvement with compliance, IPSS score, and SF6D (quality of life [QoL] measure) were assessed using multivariate analysis controlling for type of EP treatment, IPSS scores, age, ethnicity, income, family history of prostate cancer, previous surgery for prostate enlargement, presence of acute urinary retention (AUR), total number of medications patient is taking, frequency of these medications, and patient attitude and beliefs variables. The attitude and belief variables included side effects, future, pill attitude, forgetful, habit, lucky, control, science, authority, lifestyle, religiosity, and reactant. The definitions of these variables are provided in Table 1. Symptom improvement was captured in one general question with four sub-domains. The relationship of overall symptom improvement score and each domain of symptom improvement with adherence was assessed separately. Symptom improvement was defined as a score of 4 on the question being assessed. The analytical approach for the compliance analysis is presented in Figure 1 and for the IPSS analysis in Figure 2. GlaxoSmithKline Page 5 of 30

Table 1. Key Study Variables and Definitions Used in Patient Survey Avodart Establishing Predictors of EP Treatment Adherence Variable Compliance Symptom Improvement Side Effects Future Family History Previous Surgery AUR1 TotalRx FreqMed SF6D_R2 Pill Attitude Forgetful Habit Lucky Control Science Authority Lifestyle Reactant Religiosity Selected Variable Definitions Ordinal on a scale of 1-5, with 5 being "never miss a dose" or dichotomous with a score of 4-5 considered compliant. 5 questions pertaining to symptom improvement since starting medication were assessed. Any urinary symptom relief was the overall symptom improvement score and each of the following 4 domains: 1) using the restroom less at night, 2) trouble starting to urinate, 3) using the restroom less during the day, and 4) ability to empty bladder completely. Experiencing side effects since taking medication (a higher score means more agreement with side effects). Average score from 3 questions surrounding belief that taking medication will help prevent future surgery, AUR, and prostate cancer. Positive family history of prostate cancer. Previous surgery for prostate enlargement. Ever experienced AUR. Number of different prescriptions currently taking. How many times a day taking medications. SF6D utility score. Utility is a quantitative expression of an individual's preference for a particular state of health under the condition of uncertainty. It is assessed on a scale where 0 represents a state of being dead and 1 represents perfect health. Average score from 3 questions about preference for herbals, not liking to take pills, and not liking to take multiple pills. Average score from 2 questions about forgetting to take pills and forgetting to get Rx refilled (a higher score means more forgetful). Average score from 3 questions about set routines and orderly life (a higher score means more orderly). Feeling luckier than other people. Average score of 2 questions about control over life/things (a higher score means seeking more control). Average score from 3 questions about belief in science and "playing God" (a higher score indicates a lesser belief in science). Average score from 3 questions about following the law and doing what others say (a higher score indicates affinity for authority). Average score from 2 questions pertaining to diet and exercise (a higher score indicates more exercise/better diet). Average score from 2 questions pertaining to reactant attitude (a higher score indicates more reactant). Average score from 2 questions pertaining to religious beliefs (a higher score means more religious/spiritual). GlaxoSmithKline Page 6 of 30

Avodart Establishing Predictors of EP Treatment Adherence Figure 1. Analytical Approach for Evaluating Association of Symptom Improvement with Patient Compliance Symptom improvement was evaluated for its relationship to compliance while controlling for demographics, attitudes, and beliefs Linear Regression DV: Compliance (continuous) IV: Symptom improvement (continuous) Linear Regression DV: Compliance (continuous) IV: Symptom improvement (categorical) Logistic Regression DV: Compliance (categorical) IV: Symptom improvement (categorical) Relief of urinary symptoms Relief of urinary symptoms Relief of urinary symptoms Using restroom less at night Using restroom less at night Using restroom less at night Less trouble starting to urinate Less trouble starting to urinate Less trouble starting to urinate Using restroom less during the day Using restroom less during the day Using restroom less during the day Able to empty bladder completely Able to empty bladder completely Able to empty bladder completely GlaxoSmithKline Page 7 of 30

Avodart Establishing Predictors of EP Treatment Adherence Figure 2. Analytical Approach for Evaluating Association of Symptom Improvement with Patient IPSS Scores Symptom improvement was evaluated for its relationship to IPSS while controlling for demographics, attitudes, and beliefs Linear Regression DV: IPSS (continuous) IV: Symptom improvement (continuous) Linear Regression DV: IPSS (continuous) IV: Symptom improvement (categorical) Logistic Regression DV: IPSS (categorical) IV: Symptom improvement (categorical) Relief of urinary symptoms Relief of urinary symptoms Relief of urinary symptoms Using restroom less at night Using restroom less at night Using restroom less at night 46.75% Less trouble starting to urinate Less trouble starting to urinate Using restroom less during the day Using restroom less during the day Using restroom less during the day Able to empty bladder completely Able to empty bladder completely Able to empty bladder completely DV = Dependent variable IV = Independent variable GlaxoSmithKline Page 8 of 30

Avodart Establishing Predictors of EP Treatment Adherence 2.2 Survey Questions 2.2.1 Patient Survey Questions Following are the questions used for the patient survey, developed in collaboration with the steering committee and GlaxoSmithKline. Some of the questions contained branch logic that would change or skip subsequent questions based on the respondent s answer. All responses are captured in the Excel files that serve as an addendum to this report. Each patient was categorized in 5ARI, AB, or Combination drug categories based on the responses for questions 1 through 5. 1. Which medication(s) were you first prescribed for prostate enlargement? Please select all that apply. 1. Terazosin (Hytrin) 2. Doxazosin (Cardura) 3. Tamsulosin (Flomax) 4. Alfuzosin (Uroxatral) 5. Silodosin (Rapaflo) 6. Finasteride (Proscar) 7. Dutasteride (Avodart) 8. I don t know 2. Are you still taking this/these medication(s) for prostate enlargement? 1. Yes 2. No 3. Were you prescribed any of these medications after or in addition to your first prostate enlargement medication? Please select all that apply. 1. Terazosin (Hytrin) 2. Doxazosin (Cardura) 3. Tamsulosin (Flomax) 4. Alfuzosin (Uroxatral) 5. Silodosin (Rapaflo) 6. Finasteride (Proscar) 7. Dutasteride (Avodart) 8. I don t know 9. I was not prescribed any additional medications 4. Why did you stop taking that medication? Please select all that apply. 1. I switched to a different medication 2. I didn t like the way that medication made me feel (side effects) 3. That medication wasn t working 4. That medication was too expensive 5. I no longer needed medication 6. I don t remember / I don t know 7. Other 5. Are you taking any of the following medications now? Please select all that apply. 1. Terazosin (Hytrin) 2. Doxazosin (Cardura) GlaxoSmithKline Page 9 of 30

Avodart Establishing Predictors of EP Treatment Adherence 3. Tamsulosin (Flomax) 4. Alfuzosin (Uroxatral) 5. Silodosin (Rapaflo) 6. Finasteride (Proscar) 7. Dutasteride (Avodart) 8. I don t know 9. I am not currently taking any of these medications 6. When did you receive your prescription for this medication/s? 1. About a month ago 2. 2-3 months ago 3. 4-6 months ago 4. 7-12 months ago 5. I don t remember 7. Have you ever received two drugs at the same time to treat prostate enlargement? 1. Yes 2. No 8. Could you tell if one worked better than the other? 1. Yes 2. No 9. Which drug worked better? 1. Terazosin (Hytrin) 2. Doxazosin (Cardura) 3. Tamsulosin (Flomax) 4. Alfuzosin (Uroxatral) 5. Silodosin (Rapaflo) 6. Finasteride (Proscar) 7. Dutasteride (Avodart) 8. I don t remember the name of the drug Compliance was calculated based on the patient s response for question 10(1). 10. For the following questions, we will ask you how often you take your prostate medication. There is no right or wrong answer for each question, so please answer in the way that best describes you using the select statements. 1 2 3 4 5 I have stopped taking my medication completely I miss 1 or 2 doses of my medication each week I miss 1 or 2 doses of my medication each month I miss 1 or 2 doses of my medication every few months I never miss a dose of my medication 1. In general: 2. In the past 2 weeks, how many times did you miss taking your prostate medication? GlaxoSmithKline Page 10 of 30

Avodart Establishing Predictors of EP Treatment Adherence 11. To better understand your personal feelings toward prostate medication therapy, you will be asked about your level of agreement with several statements. Additionally, some questions will ask you about more general aspects of your life to better understand your personality. There is no right or wrong answer for each question, so please answer in the way that best fits your personal opinion. For these questions, think about the prostate medication(s) you used most recently, even if you no longer take this medication. Do not over-think your response, but answer based on your first or gut response. On a scale from 1 to 5, where 1 means Strongly Disagree and 5 means Strongly Agree, how much do you agree with the following? 1 2 3 4 5 Strongly Disagree Disagree Neither Agree nor Disagree Agree Strongly Agree Four domains of Symptom Improvement variable were based on the responses for questions 1 through 4. 1. Since taking my prostate medication(s), I have to get up to use the restroom less at night. 2. Since taking my prostate medication(s), I have less trouble starting to urinate. 3. Since taking my prostate medication(s), I have to use the restroom fewer times during the day. 4. Since taking my prostate medication(s), I feel like I can empty my bladder completely. Overall urinary symptom relief ( Symptom Improvement ) variable was based on the response for question 5. 5. My medication(s) relieves my urinary symptoms. Side effect variable was based on the response for question 6. 6. Since taking my prostate medication(s), I have experienced unwanted side effects. Future variable was based on the average score of questions 7 through 9. 7. I believe taking my prostate medication(s) will prevent future prostate surgery. 8. I believe taking my prostate medication(s) will help prevent AUR (the sudden inability to urinate), which requires catheterization. 9. I believe taking my prostate medication(s) may help reduce my chances for prostate cancer in the future. Pill attitude variable was based on the average score of questions 10 through 12. 10. I prefer herbal therapies over prescription medications. 11. I do not like to take pills. 12. I do not like taking multiple pills. Forgetful variable was based on the average score of questions 13 and 14. 13. I sometimes forget to take my medication(s). 14. I sometimes forget to get my prescription refilled. GlaxoSmithKline Page 11 of 30

Avodart Establishing Predictors of EP Treatment Adherence Habit variable was based on the average score of questions 15 through 17. 15. My days tend to follow a set routine. 16. I need an orderly environment to feel good. 17. I often form strong habits. Lucky variable was based on the response for question 18. 18. I find that I m often luckier than other people. Control variable was based on the average score of questions 19 and 20. 19. I m easily frustrated when people don t take my advice. 20. I m very comfortable with not having control over parts of my life. (Responses reversed) Science variable was based on the average score of questions 21 through 23. 21. Scientific progress is extremely important. (Responses reversed) 22. The problem with science is that it allows people to play God. 23. There are many aspects of science today that make me uneasy. Authority variable was based on the average score of questions 24 through 26. 24. I believe in the importance of always following the law. 25. I tend to do what the experts tell me. 26. Bossy or opinionated people tend to rub me the wrong way. (Responses reversed) Lifestyle variable was based on the average score of questions 27 and 32. 27. I have a good exercise regimen that I stick to almost every day. Religiosity variable was based on the average score of questions 28 and 29. 28. I don t see myself as a spiritual person. (Responses reversed) 29. I am very active in my religious community. Reactant variable was based on the average score of questions 30 and 31. 30. I probably fight with my family more than other people. 31. I have learned to ignore my wife s (or companion s) nagging. (Responses reversed) 32. I think I eat a healthier diet than most people my age. The next series of questions examines factors related to your prostate medication. Please remember that all answers remain anonymous and that there is no right or wrong answer for each question. General Treatment questions 12-16 12. Please check the urinary symptoms of prostate enlargement you experienced prior to initiating your current medication. Please select all that apply. 1. Difficulty starting a urine stream (hesitancy and straining) 2. Decreased strength of urine stream GlaxoSmithKline Page 12 of 30

3. Dribbling after urination 4. Feeling that the bladder was not completely empty 5. An urge to urinate again soon after urinating 6. Pain during urination (dysuria) 7. Waking at night to urinate (nocturia) 8. Frequent urination 9. A sudden, uncontrollable urge to urinate 10. None of the above describe my symptoms [EXCLUSIVE] Avodart Establishing Predictors of EP Treatment Adherence 13. How soon after initiating your currently prescribed medication(s) for prostate enlargement did you begin to experience urinary symptom relief? 1. I ve had no improvement thus far 2. 1-2 weeks 3. 1 month 4. 3 months 5. 6 months 14. Overall, how would you rate your urinary symptoms of prostate enlargement now as compared to the time period before initiating therapy? 1 2 3 4 5 Symptoms are much worse Symptoms are a little worse Symptoms are about the same Symptoms are a little better Symptoms are much better 15. If you are taking combination therapy (ie, more than 1 medication) for your prostate, how bothersome is it for you to take 2 pills for this condition? 1. Very bothersome 2. Bothersome 3. No feelings about it one way or the other 4. A little bothersome, but I still take my medication 5. Not bothersome at all 6. I don t take combination therapy 16. If you are taking combination therapy for your prostate, how helpful would it be to only have to take 1 pill for this condition instead? 1. Not at all helpful 2. Not very helpful 3. No feelings about it one way or the other 4. A little helpful 5. Very helpful 6. I don t take combination therapy The last series of questions ask about demographic characteristics and baseline medical conditions. Demographic questions 17 & 18 17. Please select the ethnicity that best describes you. 1. Caucasian 2. African American 3. Hispanic 4. Other GlaxoSmithKline Page 13 of 30

Avodart Establishing Predictors of EP Treatment Adherence 18. In what range does your total 2008 annual household income (before taxes) fall? 1. $14,999 or less 2. $15,000 to $24,999 3. $25,000 to $34,999 4. $35,000 to $49,999 5. $50,000 to $74,999 6. $75,000 to $99,999 7. $100,000 or more 8. Decline to answer Family history variable was based on the response for question 19. 19. Do you have a family history of prostate cancer? 1. Yes 2. No Previous surgery variable was based on the response for question 20. 20. Have you had previous surgery for prostate enlargement? 1. Yes 2. No AUR1 variable was based on the response for question 21. 21. Have you ever experienced a condition known as AUR, a sudden inability to urinate which requires catheterization)? 1. Yes 2. No 3. I don t know 22. Please use the scale below to rate the following statements, where 0 means Not at all and 5 means Almost always. 0 1 2 3 4 5 Not at all Less than 1 time in 5 Less than half the time About half the time More than half the time Almost always 1. Over the last month or so, how often have you had a sensation of not emptying your bladder completely after you finished urinating? 2. During the last month or so, how often have you had to urinate again less than 2 hours after you finished urinating? 3. During the last month or so, how often have you stopped and started again several times when you urinated? 4. During the last month or so, how often have you found it difficult to postpone urination? 5. During the last month or so, how often have you had a weak urinary stream? 6. During the last month or so, how often have you had to push or strain to begin urination? 7. During the last month, how many times did you most typically get up to urinate from the time you went to bed at night until the time you got up on the morning? GlaxoSmithKline Page 14 of 30

Avodart Establishing Predictors of EP Treatment Adherence 23. If you were to spend the rest of your life with your urinary condition just the way it is now, how would you feel about that? Delighted Pleased Mostly satisfied Mixed Mostly disappointed Unhappy Terrible 0 1 2 3 4 5 6 Total Rx variable is based on the response for question 24. 24. How many different prescription medications are you currently taking for all of your medical conditions? 1. 0 2. 1 3. 2 4. 3 5. 4 6. 5 or more FreqMed variable is based on the response for question 25. 25. How many times a day do you take medications? 1. 0 2. 1 3. 2 4. 3 5. 4 or more 26. Where do you have most of your prescriptions filled? 1. Retail chain pharmacy (CVS, Walgreen s, Rite- Aid, etc) 2. Grocery store pharmacy 3. Local pharmacy (non-chain/ independent) 4. VA 5. Mail order 6. Other SF6D_R2 variable was calculated by Quality Metrics based on the responses for questions 28 through 33. 27. In general, would you say your health is: Excellent Very good Good Fair Poor 1 2 3 4 5 GlaxoSmithKline Page 15 of 30

Avodart Establishing Predictors of EP Treatment Adherence This section contained Clinical Outcome Assessment data collection questionnaires or indices, which are protected by copyright laws and therefore have been excluded. GlaxoSmithKline Page 16 of 30

Avodart Establishing Predictors of EP Treatment Adherence This section contained Clinical Outcome Assessment data collection questionnaires or indices, which are protected by copyright laws and therefore have been excluded. 2.2.2 Physician Survey Questions Following are the questions included in the physician survey developed in collaboration with GlaxoSmithKline. 1. Approximately how many years have you been in practice? 1. <2 years 2. 2-15 years 3. 15-30 years 4. 30-50 years 5. 50 years 2. Approximately how many patients with prostate enlargement do you see per month? 1. <10 2. 10-20 3. >20 3. Please estimate the percentages of patients in your practice that fall into the following ethnic categories. (Must add to 100%.) % Caucasian % African American % Hispanic % Other 4. For approximately what percentage of patients with prostate enlargement with urinary symptoms in your practice do you prescribe medication? % GlaxoSmithKline Page 17 of 30

Avodart Establishing Predictors of EP Treatment Adherence 5. For approximately what percentage of patients with prostate enlargement with urinary symptoms in your practice do you prescribe ABs (eg, doxazosin, terazosin, tamsulosin, alfuzosin, or silodosin)? % 6. For approximately what percentage of patients with prostate enlargement with urinary symptoms in your practice do you prescribe 5ARIs (eg, finasteride, dutasteride)? % 7. Which of the following is the most important reason that you might decide to prescribe an AB instead of a 5ARI for prostate enlargement? 1. ABs provide faster relief of symptoms 2. ABs provide better overall control of symptoms 3. ABs have fewer sexual side effects 4. ABs have the added benefit of lowering blood pressure 5. I never prescribe an AB without simultaneously providing a 5ARI 6. Other 8. Which of the following is the most important reason you might decide to prescribe a 5ARI instead of an AB for prostate enlargement? 1. 5ARIs shrink the prostate; therefore, I use them preferentially in men with large prostates 2. 5ARIs shrink the prostate, giving better long-term results 3. 5ARIs are less expensive 4. I never prescribe a 5ARI without simultaneously providing an AB 5. Other 9. What proportion of the time do you prescribe monotherapy and combination therapy for these patients? 1. % initial monotherapy 2. % initial combination therapy 3. % initial monotherapy then combination therapy later 4. % watchful waiting 10. Please indicate what proportion of the time you prescribe the following agents for prostate enlargement. (May add to >100% to indicate combination therapy.) 1. % Terazosin (Hytrin) 2. % Doxazosin (Cardura) 3. % Tamsulosin (Flomax) 4. % Alfuzosin (Uroxatral) 5. % Silodosin (Rapaflo) 6. % Finasteride (Proscar) 7. % Dutasteride (Avodart) 11. Please rank the frequency for which your patients report the following urinary symptoms (1 = most common, 9 = least common). 1. Difficulty starting a urine stream (hesitancy and straining) 2. Decreased strength of urine stream 3. Dribbling after urination 4. Feeling that the bladder was not completely empty 5. An urge to urinate again soon after urinating 6. Pain during urination (dysuria) 7. Waking at night to urinate (nocturia) GlaxoSmithKline Page 18 of 30

Avodart Establishing Predictors of EP Treatment Adherence 8. Frequent urination 9. A sudden, uncontrollable urge to urinate 12. How soon after initiating therapy for prostate enlargement do patients receiving an AB typically begin to experience urinary symptom relief? 1. Within 1-2 weeks 2. Within 1 month 3. Within 3 months 4. Within 6 months 5. Longer than 6 months 6. ABs don t relieve urinary symptoms 13. How soon after initiating therapy for prostate enlargement do patients receiving a 5ARI typically begin to experience urinary symptom relief? 1. Within 1-2 weeks 2. Within 1 month 3. Within 3 months 4. Within 6 months 5. Longer than 6 months 6. 5ARIs don t relieve urinary symptoms 14. After the first 6 months of therapy, how do most patients receiving an AB describe their urinary symptom control? 1. Symptoms are worse 2. Symptoms are the same 3. Symptoms have improved somewhat 4. Symptoms have improved drastically 15. After the first 6 months of therapy, how do most patients receiving a 5ARI describe their urinary symptom control? 1. Symptoms are worse 2. Symptoms are the same 3. Symptoms have improved somewhat 4. Symptoms have improved drastically 16. On a scale of 1 to 5, where 1 is not at all effective at relieving urinary symptoms and 5 is very effective at relieving urinary symptoms, how would you rate the following medications? 1 2 3 4 5 No effect on relieving any urinary symptoms Very little effect on relieving any urinary symptoms Some effect on relieving some urinary symptoms Good effect on relieving most urinary symptoms Completely relieves all urinary symptoms 1. Terazosin (Hytrin) 2. Doxazosin (Cardura) 3. Tamsulosin (Flomax) 4. Alfuzosin (Uroxatral) 5. Silodosin (Rapaflo) 6. Finasteride (Proscar) 7. Dutasteride (Avodart) GlaxoSmithKline Page 19 of 30

Avodart Establishing Predictors of EP Treatment Adherence 17. The next few questions are focused on patient adherence and medication-taking behavior. Please answer the questions based on your perceptions of your patients behaviors. 18. Based on the following scale, where would you say the typical patient falls with regard to taking his prostate medications? 1 2 3 4 5 Stopped taking medication completely Miss 1 or 2 doses of medication each week Miss 1 or 2 doses of medication each month Miss 1 or 2 doses of medication every few months Never miss a dose of medication 19. What percentage of patients would you estimate fall into each category? 1. % Take medication every day 2. % Miss 1 or 2 doses each month 3. % Miss 1 or 2 doses of medication each week 4. % Miss more than 2 doses of medication each week 5. % Have stopped taking medication altogether 20. On a scale of 1 to 10 (1 = non-compliant and 10 = 100% compliant with prostate medication), where do most of your patients with prostate enlargement fall? (1 to 10) 21. We are interested in your opinion of the relative influence of different patient traits or attitudes on adherence. For the following individual patient factors, please indicate whether you think that each characteristic of the patient increases, decreases, or has no effect on the likelihood of adherence. -2-1 0 1 2 Very likely to decrease adherence No effect on adherence Very likely to increase adherence 1. If the patient experiences urinary symptom relief from their prostate medication 2. If the patient experiences side effects from their prostate medication 3. If the patient has had previous prostate surgery 4. If the patient has experienced AUR in the past 5. If the patient believes his prostate medication will prevent future prostate surgery 6. If the patient believes his prostate medication will prevent future AUR 7. If a patient receiving combination therapy could take 1 pill for his prostate instead of 2 separate pills 8. If the patient has insurance coverage 9. If the patient incurs out-of-pocket costs for medications 10. If the patient has an orderly daily life 11. If the patient has a positive attitude toward science in general 12. If the patient has a positive attitude toward authority figures 13. If the patient is an active member of a religious community 14. If the patient has a traditional (tendency to male-dominant) view of gender roles in society 15. If the patient has a healthy lifestyle in terms of diet and exercise GlaxoSmithKline Page 20 of 30

Avodart Establishing Predictors of EP Treatment Adherence 3.0 RESULTS 3.1 Description of Study Sample: Patients 3.1.1 Patient Sample Patient Demographics A total of 400 patients were surveyed based on the screening criteria data. The majority of the patients surveyed were Caucasian (97.75%), with 0.5% African American, 0.25% Hispanic, and 1.5% classified as other. All patients were male 50 years of age with health prescription medication coverage. The geographic region of residence varied across the US; however, most patients were located in the South (35.25%), while 24.25% were located in the West, 21% in the East, and 19.5% in the Midwest. Of the 86.75% respondents who provided their annual household income (13.25% did not report), 55.5% had a household income of $50,000 before taxes in 2008, while 13.25% reported annual household income between $49,999 and $35,000, 9.75% reported income between $34,999 and $25,000, 5.5% reported income between $24,999 and $15,000, and 2.75% reported $14,999. Medication History All patients included in the survey received a prescription for EP within the past 12 months. Most received their first prescription in the last 7-12 months (34.25%), while 27% received a prescription 4 to 6 months ago, 23.25% 2 to 3 months ago, and 15.5% within the last month. The majority of patients (63%) were first prescribed an alpha blocker over a 5ARI (19%) to treat EP-related symptoms, with only 11% of patients first prescribed combination therapy (Figure 3). Figure 3. First Prescribed Medication Type for EP Based on 400 Patients Surveyed Most of the patients received tamsulosin (48.75%) as their first prescribed medication within the last 12 months. Interestingly, the percentage of patients first being prescribed other medications besides tamsulosin for EP was much lower (Figure 4). GlaxoSmithKline Page 21 of 30

Avodart Establishing Predictors of EP Treatment Adherence Figure 4. Stratification of First Prescribed Medication for EP Based on 400 Patients Surveyed Percent 60 40 20 0 48.75 14 13 12.5 7.25 7.25 7.25 1.75 Terazosin Doxazosin Tamsulosin Alfuzosin Silodosin Finasteride Dutasteride Don't know Drug When patients were asked if they were still taking the same medications first prescribed for EP, 73% said yes, with the remaining 27% saying they had changed or stopped therapy. Of the 73% of patients still taking the same medication, 30.14% were prescribed additional medication after or in addition to their first EP medication, while 67.81% reported no additional medications were prescribed (Figure 5). Figure 5. Medication Prescribed in Addition to or After First EP Medication 80% 70% 67.81% 60% Percentage 50% 40% 30% 20% 10% 0% 5.14% 1.37% 10.27% 2.40% 0.34% 5.48% 5.14% 4.11% Terazosin Doxazosin Tamsulosin Alfuzosin Silodosin Finasteride Dutasteride Don't know No additional meds 3.1.2 Compliance and Symptom Relief Based on the patient sample, 59.75% of patients reported that they never missed a dose, while 11.25% reported missing 1 or 2 doses of medication every few months. The other 11% of patients reported being less compliant, with 8.5% missing 1 or 2 doses of medication each month and 2.5% missing 1 or 2 doses each week (Figure 6). When asked how many times they missed their medication in the past 2 weeks (at the time surveyed), 75.5% of patients did not miss taking their EP medication at all. GlaxoSmithKline Page 22 of 30

Avodart Establishing Predictors of EP Treatment Adherence Figure 6. Reported Patient Compliance with EP Medication Percent 70 60 50 40 30 20 10 0 18% 1. I have stopped taking my medication completely 2.5% 2. I miss one or two doses of my medication each week 8.5% 3. I miss one or two doses of my medication each month 11.25% 4. I miss one or two doses of my medication every few months 59.75% 5. I never miss a dose of my medication Compliance All patients reported high compliance rates regardless of therapy regimen. Differences between the therapy groups were not statistically significant. (Figure 7). Figure 7. Impact of Therapy Type on Adherence (N = 371) With the majority of patients compliant with their medication, almost half (46.75%) agreed or strongly agreed that their medications improved their urinary symptoms (Figure 8). GlaxoSmithKline Page 23 of 30

Avodart Establishing Predictors of EP Treatment Adherence Figure 8. Symptom Relief with EP Medications Percent 50 40 30 20 10 0 6.75% 14.25% 32.25% 1. Strongly Disagree 2. Disagree 3. Neither Agree nor Disagree 38.75% 8% 4. Agree 5. Strongly Agree Category There was a significant difference in medication adherence for patients reporting EP-related side effects (Figure 9). Patients who reported having no side effects were more likely to be compliant compared to patients who reported having side effects, (83.65% vs 64.77%; p=0.0001). Figure 9. Relationship of Compliance and Side Effects (N = 400) Of the symptoms assessed, having less trouble starting to urinate since taking prostate medication had the highest percentage of patients in agreement / strong agreement (59.5%) that their medication gave them symptom relief, followed by having to get up to use the restroom less at night (56.25%) (Figure 10). GlaxoSmithKline Page 24 of 30

Avodart Establishing Predictors of EP Treatment Adherence Figure 10. Patients View of Symptom Relief 70 60 56.25% 59.5% 50 43.75% 14% 15% 47.75% 48% Percent 40 8% 10.5% 10.5% 30 20 38.75% 42.25% 44.5% 37.25% 37.5% 10 0 Relief urinary symptoms Get up to use the restroom less at night Less trouble starting to urinate Symptom Relief Use the restroom less during the day Feel like empty bladder completely Agree Strongly Agree 3.1.3 Efficacy and Impact on Adherence Questions pertaining to symptom improvement since starting medication were averaged using a scale from 1 to 5. Patients with a higher average compliance score were classified as more adherent. Compliance with therapy was defined as having an average score 3, with symptom control defined as having an average score 4. Approximately 91% of patients indicated they were adherent to their medication and experienced symptom control (Figure 11). GlaxoSmithKline Page 25 of 30

Figure 11. Percentage of Compliant Patients Based on Symptom Control Avodart Establishing Predictors of EP Treatment Adherence Patients who reported symptom improvement were more likely to be compliant. When analyzing the domains of symptom improvement separately as categorical variables, patients who had a relief of urinary symptoms (P=0.0363), had less trouble staring to urinate (P=0.0069), had to use the restroom fewer times during the day (P=0.015), and had a feeling of emptying their bladder completely (P=0.0049) were more likely to be adherent, whereas patients who had to get up to urinate less at night had no significant association with adherence (P=0.8619). The odds ratios, P values, and 95% confidence intervals (CIs) for the analysis presented in Figure 11 is provided in Table 2. Table 2. Logistic Regression Comparing Efficacy Domains Analyzed Separately as Categorical Variables Patient Survey Efficacy Domains OR (95% CI) Pr > ChiSq Relief of urinary symptoms 2.399 (1.057-5.404) 0.0363 Having to get up less to use the restroom at night 1.078 (0.463-2.511) 0.8619 Less trouble starting to urinate 3.117 (1.366-7.111) 0.0069 Having to use the restroom less during the day 2.858 (1.220-6.693) 0.015 Feel like I can empty bladder completely 3.352 (1.443-7.786) 0.0049 GlaxoSmithKline Page 26 of 30

3.1.4 Patient IPSS Scores Avodart Establishing Predictors of EP Treatment Adherence Patients reporting symptom improvement had lower IPSS scores than patients without symptom improvement, indicating a lower symptom burden. Across all symptom domains, patients with symptom improvement had lower IPSS scores. Figure 12 below demonstrates patients with symptom control have lower IPSS scores and less severity of symptoms. Figure12. Symptom Control Impact on Mean IPSS Scores and Severity of Symptoms 16 14 13.588 12.705 12.983 13.785 12 11.748 Mean IPSS Score 10 8 6 8.8121 10.187 9.2556 9.439 8.7529 4 2 0 Relief of urinary symptoms Having to get up less to use the restroom at night* Less trouble starting to urinate Having to use the restroom less during the day Can empty bladder completely * P=0.0013, P=0.0003, P<0.0001 Improvement Mean IPSS Symptoms / Efficacy No improvement Mean IPSS As previously noted in the patient survey results, patients with symptom improvement are more adherent to their medications. 3.1.5 Patient Quality of Life When evaluating the QoL data based on the patient surveys, symptom improvement had no effect on QoL as measured by the SF-6D utility scores. Additional information regarding the QoL data is reported in the Appendix Excel files. 3.2 Description of Study Sample: Physicians 3.2.1 Physician Sample Physician Demographics Of the 100 targeted physicians, the survey data captured 101 total physician responses. Of the 101 physicians, 90% were PCPs, and the other 10% were urologists. The PCPs indicated they see <10 patients (4.4%) and between 10 to 20 patients (30.77%) per month for EP, while all urologists from GlaxoSmithKline Page 27 of 30

Avodart Establishing Predictors of EP Treatment Adherence the sample (100%) indicated that they evaluate >20 patients with EP per month, compared to 64.84% of PCPs respectively. The sample included responses from physicians residing in geographic regions evenly distributed across the US. Most of the physicians surveyed (53.47%) had been practicing for 15 to 30 years. Prescribing Practices When asked for reasoning behind prescribing an AB instead of 5ARI, the majority of physicians (65.35%) cited that ABs provide faster relief of EP symptoms. Slightly more than half of the physicians (52.48%) indicated that the reason for prescribing a 5ARI instead of an AB is based on the belief that5aris shrink the prostate, giving better long-term results. When asked what percent of time physicians prescribe monotherapy vs combination therapy to patients for EP, 10% to 15% of the time, physicians will initially prescribe combination therapy. Approximately 60% of the time, PCPs prescribe monotherapy as either a 5ARI or alpha blocker, with ~19% of physicians prescribing monotherapy initially, followed by combination therapy later. Figure 13. Physician Prescribing Practices by Provider Type Urologist Primary Care Physician 70.0 60.0 50.0 40.0 30.0 20.0 10.0 0.0 48.0 59.3 Initial monotherapy 15.0 19.5 18.6 17.5 10.5 Initial combination therapy Initial monotherapy then combination therapy later 11.6 Watchful waiting 3.2.2 Physician Views on Factors Likely to Increase Patient Compliance On a scale of 1 to 10 (1 = 0% compliant; 10 = 100% compliant), 59.4% of physicians indicated they felt most of their patients with EP were at an 8 or above in terms of being compliant with their medication. Interestingly, this is consistent with the percentage of patients (59.75%) from the patient sample who reported they were 100% compliant with their medication, never missing a dose. Of the factors thought to affect medication adherence, urinary symptom relief had the highest percentage of physicians agreeing it most likely increases adherence (92.08%). Results indicated that 87.12% of physicians felt that patients who believe their prostate medication will prevent future prostate surgery would be likely or very likely to have improved adherence. Interestingly, if a patient is receiving combination therapy for his prostate and could take 1 pill instead of 2 separate pills, 79.2% of physicians indicated that it would likely or very likely increase adherence (Figure 14). GlaxoSmithKline Page 28 of 30

Avodart Establishing Predictors of EP Treatment Adherence Figure 14. Physician Views of Factors Likely to Increase Patient Compliance \ 3.2.3 Physician Perceptions on Urinary Symptoms Physicians reported that waking at night to urinate (nocturia) was the most commonly reported urinary symptom (37.62%), followed by patients having difficulty starting a urine stream (hesitancy and straining) (26.73%). Physicians indicated pain during urination (dysuria) was ranked as the least commonly reported symptom, followed by a sudden, uncontrollable urge to urinate (Figure15) Figure 15. Physician Perceptions on Urinary Symptoms GlaxoSmithKline Page 29 of 30

4.0 CONCLUSIONS Avodart Establishing Predictors of EP Treatment Adherence The results indicate that symptom improvement is associated with better adherence in men being treated with a 5ARI, AB, or combination therapy for EP. Both the patient and physician assessments of compliance were in agreement: 59.75% of patients stated they never missed a dose of their medication, while 59.4% of physicians indicated their patients are compliant. Physicians rated urinary symptom relief as likely to increase adherence compared to any other factor. The statistical assessment confirmed that relief of urinary symptoms was significantly related to adherence. The only symptom not associated with improved adherence was having to get up to use the restroom at night; interestingly, physicians rated this symptom as the most commonly reported symptom by patients. GlaxoSmithKline Page 30 of 30

640 George Washington Highway, Suite 201 Lincoln, RI 02865 CONFIDENTIAL QualityMetric s Certified Scoring Service Data Quality Evaluation

Data Quality Evaluation (DQE): Data problems can have a major impact on any study and the ability of that study to reach its intended goals. In the past, data problems with studies utilizing the SF-12v2 Health Survey have impacted regulatory decisions, delayed the availability of new treatments in the marketplace, and even eliminated a researcher s ability to compare study results with those from other participating sites. Most importantly, erroneously scored data impacts one s ability to appropriately interpret scores. Some of these errors include: administration of a survey that has inaccurately formatted questions, i.e. questions are omitted or presented in the wrong order, pages of the survey are either not included or are missing during administration, and keying response values into a database incorrectly during the data entry phase. To identify these types of errors, QualityMetric developed Data Quality Evaluation (DQE) algorithms, which are available through our Outcomes Insight Consulting Team. After being provided with a file containing raw item level response data, the Outcomes Insight Consulting Team then performs a thorough review and inspection of the quality of data using DQE algorithms and implements QualityMetric s proprietary scoring algorithms to score the data. Once complete, the Outcomes Insight Consulting Team informs the client of any problems with the data via a comprehensive DQE Summary Report. Please see below for a more detailed description of this process and examples of this detailed report. DQE Tests: The DQE service offered by QualityMetric includes an investigation of missing data rates for each SF-12v2 item and also a series of scaling tests which provide empirical evidence that your data are consistent with the measurement model developed for the SF-12v2. This information is necessary for proper interpretation of scales and summary measure scores. Missing Data: Each SF-12v2 item is investigated for missing data and out of range responses. An extraordinarily high rate of missing data for any one item is an indication that respondents may not understand the question being asked, or that there may be a problem in the way that the survey items were formatted (if the standard survey format for the SF-12v2 Health Survey was not used). High missing data rates for a block of questions could also be an indication that a page of the survey may have been missing during the administration of the survey. Out of range responses are typically an indication that there was a problem during data entry, for example, errors in keying item response values into the database. Scaling Tests: The scales of the SF-12v2 Health Survey are scored using the method of summated rating scales. To compute a scale score using the method of summated rating scales, scores assigned to responses options of each survey item are simply summed (assuming item response values are all in the same direction, either positive or negative). The simplicity of this method is based on a number of assumptions that must be tested. These tests determine the appropriateness of including an item in a particular scale and whether it is appropriate to simply sum item scores to estimate a scale score. Further, these tests lend support to the validity of the measurement model developed for a survey instrument. QualityMetric tests the assumptions underlying the scoring SF-12v2 scales and summary measures using algorithms originally developed for the QualityMetric.com CONFIDENTIAL Page 2 of 9

Multitrait Analysis Program (MAP), which have now been incorporated into the Health Outcomes Scoring Software 2.0 (Ware, Jr., Harris, Gandek, Rogers, & Reese, 1997; Ware, Jr. et al., 1997). The MAP program tests the following assumptions necessary to support the scoring and interpretation of SF-12v2 scales and summary measures: 1. Items measuring the same summary measures have approximately equal variances; otherwise item standardization may be required. 2. Items are substantially, linearly related (greater than 0.40 corrected for overlap) to the underlying concept being measured (item internal consistency). 3. Items correlate significantly (higher) with its hypothesized scale than with scales measuring other concepts (item-discriminant validity). 4. SF-12v2 physical functioning (PF), role-physical (RP), bodily pain (BP), and general health (GH) scales correlate higher with the physical summary measure (PCS) than with the mental summary measure (MCS). 5. SF-12v2 mental health (MH), role-emotional (RE), social functioning (SF), and vitality (VT) scales correlate higher with MCS than with PCS. DQE Results: Results in this report are based on records provided to QualityMetric Incorporated by Xcenda, pertaining to QM001889. 400 records were received. Table 1 presents an overall summary of the results of the scaling tests conducted on the SF-12v2 data for 400 subjects. Overall, the quality of data was good. A full 100% of the sample had complete data, and thus scale scores were computable for 100% of the sample without having to use any methods of missing data estimation. Tests of item convergent validity were satisfactory for 100% of the tests conducted, and 100% of the item discriminant validity tests were satisfied. Table 2 presents the number of subjects from which each scale score and component score were computed. Given the 100% complete data rate, all computations were based upon all 400 responses without the use of MDE. Tables 3-5 present descriptive statistics for each of the scales and items. Table 3 presents the summary statistics for scale and component scores using norm-based scoring. Table 4 presents the means and standard deviations for each item. In general, item means and standard deviations within each scale are roughly equal, satisfying the first scaling assumption tested (approximately equal item variances). Table 5 presents the response frequency for each item. Table 6 presents the correlations between each item and both SF-12v2 summary measures (PCS and MCS). In support of item convergent validity (scaling assumption 2), all items showed substantial (r>0.40) correlations with their hypothesized summary measure. In support of item discriminant validity (scaling assumption 3), all items showed a significantly higher correlation with their hypothesized summary measure than any competing measure. Lastly, the correlations between items and their hypothesized summary measure were roughly equal within each SF- 12v2 scale, supporting the 4 th and 5 th scaling assumption. QualityMetric.com CONFIDENTIAL Page 3 of 9

Table 1. Data quality indicators Data Quality Indicators: Satisfactory Norms 1. Completeness of Data... Items with 5% or more missing values: NONE 2. Responses within Range... Items with 5% or more out-of-range values: NONE 3. Estimable Scale Scores... 100.0% YES 90 100.0% YES 100 Estimable without MDE 100.0% YES 90 Estimable with MDE 100.0% 90 4. Convergent Validity Items that failed convergent validity test: NONE 5. Discriminant Validity Items that failed discriminant validity test: NONE 100.0% YES 100 100.0% YES 100 Definition of Data Quality Indicators: 1. Percentage of completed responses (within range) divided by the total possible number of responses (items*n). 2. Percentage of item responses within the range of response codes printed on the questionnaire. 3. Percentage of subjects for whom all scales are computable with and without application of the SF- MDE. 4. Percentage of items that correlated 0.40 or higher with their hypothesized component summary score. 5. Percentage of items that correlated higher with their hypothesized component summary score than with the alternative component summary score. QualityMetric.com CONFIDENTIAL Page 4 of 9

Table 2. SF-12v2 Missing Data Estimation Report Scales Summaries Total Summary PF RP BP GH VT SF RE MH PCS MCS Total Number of Records in Data File 400 400 400 400 400 400 400 400 400 400 400 Total Number of Records Scored with Complete Data 400 400 400 400 400 400 400 400 400 400 400 Total Number of Records Scored with Full MDE 400 400 400 400 400 400 400 400 400 400 400 Table 3. SF-12v2 Scale and Summary Measure Scores, Norm-Based Scoring Scales Summaries PF RP BP GH VT SF RE MH PCS MCS Mean 46.87 45.35 46.54 44.78 48.71 49.50 48.35 51.27 44.64 51.17 25th Percentile 39.29 38.75 37.06 44.74 37.69 46.47 44.90 46.25 37.30 45.74 50th Percentile (median) 47.88 47.96 47.25 44.74 47.75 56.57 56.08 52.35 47.04 53.67 75th Percentile 56.47 57.18 57.44 55.52 57.81 56.57 56.08 58.45 53.46 58.15 Standard Deviation 11.57 10.97 11.01 11.40 9.88 10.19 10.19 10.13 11.13 9.81 Min 22.11 20.32 16.68 18.87 27.62 16.18 11.35 15.77 11.43 15.58 Max 56.47 57.18 57.44 61.99 67.88 56.57 56.08 64.54 64.64 69.62 N 400 400 400 400 400 400 400 400 400 400 QualityMetric.com CONFIDENTIAL Page 5 of 9

Table 4. Item means and standard deviations ITEM MEAN SD PF PF02 (Item 2) 2.55 0.67 PF04 (Item 3) 2.34 0.77 RP RP02 (Item 4) 3.72 1.19 RP03 (Item 5) 3.72 1.30 BP BP02 (Item 8) 2.07 1.08 GH GH01 (Item 1) 2.88 0.96 VT VT02 (Item 10) 2.91 0.98 SF SF02 (Item 12) 4.30 1.01 RE RE02 (Item 6) 4.20 1.07 RE03 (Item 7) 4.42 0.91 MH MH03 (Item 9) 2.33 0.90 MH04 (Item 11) 4.16 0.96 QualityMetric.com CONFIDENTIAL Page 6 of 9

Table 5. Item Frequency Statistics - Raw Data Scale = PF - Physical Functioning Item Allowed Range Missing Response Value Frequency Freq. % 1 2 3 4 5 PF02 1-3 0 0.0 41 99 260 0 0 PF04 1-3 0 0.0 75 116 209 0 0 Missing 0 0.0 Out-of- Range 0 0.0 Scale = RP - Role Physical Item Allowed Range Missing Response Value Frequency Freq. % 1 2 3 4 5 RP02 1-5 0 0.0 18 53 90 102 137 RP03 1-5 0 0.0 31 49 79 85 156 Missing 0 0.0 Out-of- Range 0 0.0 Scale = BP - Bodily Pain Item Allowed Range Missing Response Value Frequency Freq. % 1 2 3 4 5 BP02 1-5 0 0.0 149 136 62 44 9 Missing 0 0.0 Out-of- Range 0 0.0 Scale = GH - General Health Item Allowed Range Missing Response Value Frequency Freq. % 1 2 3 4 5 GH01 1-5 0 0.0 23 118 166 70 23 Missing 0 0.0 Out-of- Range 0 0.0 Scale = VT - Vitality Item Allowed Range Missing Response Value Frequency Freq. % 1 2 3 4 5 VT02 1-5 0 0.0 16 140 135 84 25 Missing 0 0.0 Out-of- Range 0 0.0 LIST CONTINUED ON THE NEXT PAGE QualityMetric.com CONFIDENTIAL Page 7 of 9

Table 5. Item Frequency Statistics - Raw Data (Continued) Scale = SF - Social Functioning Item Allowed Range Missing Response Value Frequency Freq. % 1 2 3 4 5 SF02 1-5 0 0.0 8 17 62 73 240 Missing 0 0.0 Out-of- Range 0 0.0 Scale = RE - Role Emotional Item Allowed Range Missing Response Value Frequency Freq. % 1 2 3 4 5 RE02 1-5 0 0.0 8 30 58 84 220 RE03 1-5 0 0.0 4 16 45 77 258 Missing 0 0.0 Out-of- Range 0 0.0 Scale = MH - Mental Health Item Allowed Range Missing Response Value Frequency Freq. % 1 2 3 4 5 MH03 1-5 0 0.0 50 222 84 33 11 MH04 1-5 0 0.0 6 19 65 127 183 Missing 0 0.0 Out-of- Range 0 0.0 Notes: 1. Scale Percent Missing = Number Missing / Total Possible 2. Scale Percent Out-of-Range = Number Outside Allowed Range / Total Possible QualityMetric.com CONFIDENTIAL Page 8 of 9

Table 6. Item Component Correlation Matrix Component Summaries Items PCS MCS Scale = PF - Physical Functioning PF02 0.78 0.10 PF04 0.82 0.06 Component Summaries Items PCS MCS Scale = RP - Role Physical RP02 0.78 0.24 RP03 0.85 0.21 Component Summaries Items PCS MCS Scale = BP - Bodily Pain BP02 0.66 0.28 Component Summaries Items PCS MCS Scale = GH - General Health GH01 0.69 0.33 Component Summaries Items PCS MCS Scale = VT - Vitality VT02 0.51 0.61 Component Summaries Items PCS MCS Scale = SF - Social Functioning SF02 0.45 0.61 Component Summaries Items PCS MCS Scale = RE - Role Emotional RE02 0.26 0.70 RE03 0.21 0.70 Component Summaries Items PCS MCS Scale = MH - Mental Health MH03 0.18 0.76 MH04 0.08 0.78 Notes: 1. Poor convergent validity is indicated when items do not correlate.40 or higher with their hypothesized component summary score. 2. Poor discriminant validity is indicated when items correlate higher with the alternative component summary score than with their hypothesized component summary score. QualityMetric.com CONFIDENTIAL Page 9 of 9

SF-12v2 Scored Dataset The scored dataset contains all of the fields provided in the original dataset, plus the following additional fields: Field Label PF RP BP GH VT SF RE MH PF_NBS RP_NBS BP_NBS GH_NBS VT_NBS SF_NBS RE_NBS MH_NBS PCS MCS SF6D_R2 Description Physical Funtioning-0-100 Score Role Physical-0-100 Score Bodily Pain Scale-0-100 Score General Health Scale-0-100 Score Vitality Scale-0-100 Score Social Functioning Scale-0-100 Score Role Emotional Scale-0-100 Score Mental Health Scale-0-100 Score Physical Functioning Scale-Norm Based Scores Role Physical Scale-Norm Based Scores Bodily Pain Scale-Norm Based Scores General Health Scale-Norm Based Scores Vitality Scale-Norm Based Scores Social Functioning Scale-Norm Based Scores Role Emotional Scale-Norm Based Scores Mental Health Scale-Norm Based Scores Physical Component Summary Mental Component Summary SF6D Utility Score Revision 2 (if requested) Page 1 of 1 QualityMetric Incorporated

Establishing Predictors of EP-Treatment Adherence: Linking Symptom Improvement to Adherence Avodart Study Protocol February 2009 Prepared for: PharmD, MPH Director, USP Health Outcomes GlaxoSmithKline

Establishing Predictors of EP-Treatment Adherence Study Protocol Please direct comments and questions to: PharmD, PhD Senior Director Xcenda 4114 Woodlands Parkway, Suite 500 Palm Harbor, FL 34685 Phone: Fax: Email: PharmD, MPH Director Xcenda 4114 Woodlands Parkway, Suite 500 Palm Harbor, FL 34685 Phone: Fax: Email: GlaxoSmithKline 2

TABLE OF CONTENTS Establishing Predictors of EP-Treatment Adherence Study Protocol 1.0 INTRODUCTION... 4 2.0 METHODOLOGY... 4 2.1 STUDY OBJECTIVE... 4 2.2 STEERING COMMITTEE... 4 2.3 SURVEY DEVELOPMENT... 4 2.3.1 Approach... 4 2.3.2 Quality Assurance... 5 2.4 SURVEY DEPLOYMENT... 5 2.4.1 Respondent Sample... 5 2.4.2 Screening Criteria... 5 2.5 DATA ANALYSIS... 6 2.6 PUBLICATION... 6 3.0 APPENDIX: FINAL SURVEY CONTENT... 6 GlaxoSmithKline 3

Establishing Predictors of EP-Treatment Adherence Study Protocol 1.0 Introduction Analyses have shown that medication adherence is strongly associated with improved outcomes in patients with an enlarged prostate (EP). Specifically, data suggest that a 10 point increase in adherence may result in a 7% reduction in the need for prostate surgery. Furthermore, these medications have demonstrated effectiveness in the prevention of prostate cancer. Despite the established importance and chemoprophylaxis benefits of adherence to EP mediation, the factors that drive medication adherence are not well understood. The following protocol details the process for completion of primary research aimed at understanding the drivers of EP medication adherence from the provider and patient perspectives. A specific area of focus will be on the use of EP medication for the prevention of prostate cancer and how this relates to adherence. This research will be conducted through on-line surveys and the results are planned for presentation at scientific meetings and publication in peer-reviewed journals. 2.0 Methodology 2.1 Study Objective The objective of this study is to evaluate the drivers of medication adherence in patients treated with a 5ARI (alpha reductase inhibitor), alpha-blocker, or both for EP. 2.2 Steering Committee A multi-disciplinary steering committee will be convened to oversee this study. The purpose of the steering committee is to ensure that the study is conducted with high scientific rigor, validity, clinical accurateness, and appropriate statistical methods. The steering committee members will be involved in every step of the research process, from conceptualization to interpretation and reporting of results. The members of the steering committee are planned to be as follows: PhD MD MD MD 2.3 Survey Development 2.3.1 Approach The survey is anticipated to be approximately 25-35 questions in length and have few or no openended questions. This survey will be distributed separately to a patient and to a physician audience. Xcenda will develop the first draft of the survey and work jointly with GSK and the steering committee to finalize the survey content. Patient Survey First, the survey will assess basic demographic characteristics of the respondents. Next, the survey will assess the level of or perceived level of adherence based on a continuum of medication adherence statements. The ensuing survey questions will assess the respondents level of agreement with statements about factors related to adherence. These questions are expected to cover topics GlaxoSmithKline 4

Establishing Predictors of EP-Treatment Adherence Study Protocol such as symptom control, delivery of care, long term prevention, etc, and will be answered with a Likert scale. The last section will include specific questions related to patient s perception of prostate cancer and prostate cancer prevention. Provider Survey The patient and provider survey are anticipated to largely overlap in content. They will address the same broad topics, including demographics, adherence level, factors impacting adherence, and prostate cancer prevention. Generally, the questions will be re-phrased from the patient audience to be suitable for providers. 2.3.2 Quality Assurance Prior to launch, the survey questions will be pilot tested with five sample subjects to ensure question validity. Additionally, a quality control question will be included in the survey, asking the respondent to select a specific answer to validate that they are reading the question. Respondents who do not answer this question correctly will not be included in the results analysis. Response times will be collected for all questions, if possible. 2.4 Survey Deployment 2.4.1 Respondent Sample Xcenda will work with Harris Interactive to obtain the necessary patient and physician sample for the survey. Harris will draw upon its database of patients and physicians to obtain the study sample. The target sample for the patient survey is 400, while the target sample for the physician survey is 200 or less (to be determined). Xcenda will provide a word document of the survey questions to Harris. Harris will code the survey content into their on-line tool and launch the survey to their sample through the internet. Harris will provide Excel spreadsheets of the results in a cross-tabular format within 3 to 4 weeks of receipt of the survey content from Xcenda. 2.4.2 Screening Criteria The patient survey will be submitted to all men aged 50 years or greater. Those with self-reported EP who are currently taking a 5ARI, alpha-blocker, or both will be eligible for this survey. A subset of men with prostate cancer will be considered for inclusion by the steering committee. Additional screening questions will be considered by the steering committee, such as income level, health insurance coverage, and amount of time on medication (new starts). These additional screening criteria can provide for a cleaner respondent sample; however, it may also decrease the ability to achieve the desired sample size. The physician survey will be submitted to practicing urologists who have been practicing a prespecified number of years (range) and see a pre-specified number of patients (per week). These variables will be determined by the steering committee. GlaxoSmithKline 5

Establishing Predictors of EP-Treatment Adherence Study Protocol 2.5 Data Analysis Xcenda recognizes that there are a variety of statistical techniques that may be appropriate to answer a specific research question. Therefore, the final statistical approach will be determined after reviewing the crude study results (sample sizes, compliance category distributions, etc). It is anticipated that descriptive statistics will be analyzed in Excel and reported for each of the study questions. Descriptive statistics are expected to be the primary method of analysis for the provider survey. To assess the impact of various factors on medication adherence in the patient survey, regression models may be utilized. Regression models would include a number of independent variables (survey questions) to predict the respondent s medication adherence (dependent variable). This approach would result in Beta scores for each independent variable, indicating the direction and magnitude of an effect on medication adherence. Additionally, the independent variables may be categorized/grouped and their impact assessed accordingly, e.g., symptom control, long-term prevention, delivery of care, etc. Factors influencing patient adherence as determined by the regression analysis will be compared to physician s perceptions of drivers of adherence. 2.6 Publication The results of this study are intended to be submitted to scientific conferences for presentation. The specific conferences will be determined based on the timing of the results and the deadlines for various conferences. The results are also planned to be submitted to peer-reviewed journals for publication. 3.0 Appendix: Final Survey Content This section of the protocol will be updated once the content of the survey is finalized by the steering committee. GlaxoSmithKline 6