(TD50 = 30); insulins (TD50 = 45); antidepressants (TD50 = 67); bisphosphonates (TD50 = 82); cardiovascular agents (TD50 = 85); statins (TD50 = 90); oral antidiabetic agents (TD50 = 90); and glaucoma medications (TD50 = 120). Rates of medication dis- continuation among patients who had filled a prescription for an in-class medication in the prior 180 days were: inhaled steroids (TD50 = 60); insulins (TD50 = 130); glaucoma medica- tions (TD50 = 180); antidepressants (TD50 = 187); cardiovascular agents (TD50 = 240); oral antidiabetic agents (TD50 = 270); bisphosphonates (TD50 = 272); and statins (TD50 = 308). CONCLUSIONS: Patients new to therapy faced a far higher likelihood of medication discontinuation compared to those with prior in-class medication experience among all the medication classes studied. Providers and health care systems may want to re-engineer care to provide more frequent follow-up and educa- tion to this commonly treated, high-risk group. PRS20 SYSTEMATIC REVIEW OF FIXED-DOSE COMBINATIONS AND UNIT-OF-USE PACKAGING IN PATIENTS WITH HYPERTENSION, DYSLIPIDEMIA,AIDS,ASTHMA AND DIABETES TYPE 2 Paladio N, Almazan C, Moharra M Catalan Agency for Health Technology Assessment and Research, Barcelona, Spain OBJECTIVES: Poor compliance is the primary reason for sub- optimal clinical benefit, especially in patients with chronic dis- eases. Fixed-dose combinations (FDC) and unit-of-use packaging (UUP) are strategies designed to simplify medication regimen and potentially improve compliance. The aim of our study is to systematically analyze the effect of FDC and UUP on compliance and effectiveness in patients with hypertension (HTA), dyslipi- demia, AIDS, asthma and diabetes type 2 (DMII). METHODS: Systematic review (SR) of studies that compare medications com- bined in a single pill or within a UUP with the same free-drug combinations in HTA, AIDS, asthma and DMII. Bibliographic databases were searched from inception to February 2008 with no date limits. Two reviewers scrutinized retrieved references and full publications sought of potentially relevant studies were iden- tified on the basis of title and/or abstract. Additional relevant studies were identified from manual searches. Only studies pub- lished in English or Spanish were included. Quality of studies was assessed using specific instruments according to their design. RESULTS: A total of 11 original studies (HTA 5; DMII 3; AIDS 2; dyslipidemia 1), 1 SR and 1 metaanalysis of varying quality were included in the SR. Most originals were retrospective (72%), based on medical claims (72%) and used medication possession ratio (RPM) as a proxy of compliance (54%). Some studies also included self-reported compliance (27%) and mea- sures of persistence (18%). Only 3 studies reported comparative effectiveness. Follow-up periods ranged from 16 weeks to 1 year. Overall, results show a tendency for better compliance and less missed doses in the FDC and UPP groups compared to free-drug combination groups. Limited data support noninferiority of FDC and UUP in terms of effectiveness. CONCLUSIONS: FDC and UUP seem to improve compliance with no major effects on effectiveness in patients with chronic diseases. However, evidence is weak and large randomized controlled trials should be carried out. PRS21 CAN ASTHMA QUALITY OF LIFE QUESTIONNAIRE (AQLQ) DATA FROM DIFFERENT COUNTRIES BE COMBINED FOR ANALYSES? Williams AE 1 , Frith L 2 , Gul N 2 , Agier L 3 , Juniper E 4 1 GlaxoSmithKline R&D, Uxbridge, UK, 2 GlaxoSmithKline R&D, London, UK, 3 ENSAI Engineering School, Bruz, France, 4 McMaster University, Hamilton, Ontario, Canada OBJECTIVES: Health-related quality of life data are frequently collected in clinical trials from different countries and combined in analyses. This analysis compared AQLQ data across 16 coun- tries (17 languages) to evaluate suitability to combine data in subsequent analyses. METHODS: AQLQ data from the Gaining Optimal Asthma controL (GOAL) study was used for the analy- ses. Of 3416 patients treated, 1973 had an overall AQLQ score at baseline, 1850 at week 12 and 1832 at both. The original language (North American English) was considered the reference language (RL). Values within 0.5 of the RL were considered comparable given that 0.5 is considered a clinically meaningful difference. RESULTS: Number of patients completing each AQLQ language varied from 27 (Canadian French) to 256 (Man- darin Chinese). Mean age ranged from 31.7 years (Spain) to 52.9 years (Norway) and percent of males ranged from 30.3% (Brazil) to 74.4% (Norway). Mean overall AQLQ score [s.d.] at baseline in the RL was 4.59 [0.94]. Of the 16 languages all but three, Chile (3.58 [1.05]), Denmark (5.10 [0.82]) and Spain (5.19 [0.93]), reported mean baseline AQLQ overall scores within 0.5 of the RL. Similar findings were reported for AQLQ domain scores, with few countries reporting baseline values outside 0.5 of the RL. Mean change from baseline in the overall AQLQ score in the RL was 0.89 [1.06], with all translations reporting values within 0.5 of the RL indicating similar results in all languages. For the AQLQ domains, only the emotional function domain of the Norwegian, Canadian French and French for France trans- lations were outside 0.5 of the RL. CONCLUSIONS: The con- sistency of baseline and change from baseline scores comparing 16 translations with the original language version supports the validity of translations used in this study and the combining of data across countries for analyses. PRS22 ARE THE MEASUREMENT PROPERTIES OF THE ASTHMA QUALITY OF LIFE QUESTIONNAIRE (AQLQ) CONSISTENT ACROSS LANGUAGES? Juniper E 1 , Agier L 2 , Williams AE 3 , Frith L 4 , Gul N 4 1 McMaster University, Hamilton, Ontario, Canada, 2 ENSAI Engineering School, Bruz, France, 3 GlaxoSmithKline R&D, Uxbridge, UK, 4 GlaxoSmithKline R&D, London, UK OBJECTIVES: Health-related quality of life data are frequently collected in clinical studies from a number of countries and combined for analysis. This study compared AQLQ data across 16 countries (17 languages) to investigate if the measurement properties of the AQLQ are sufficiently consistent to support combining data in multinational trials. METHODS: The AQLQ data from the Gaining Optimal Asthma controL (GOAL) study was used for the analyses. Of 3416 patients treated, 1973 had an overall AQLQ score at baseline, 1850 at week 12 and 1832 at both. The original language (North American English) was considered the reference language (RL). Internal consistency, construct validity and responsiveness were evaluated for each translation. RESULTS: Number of patients completing each AQLQ language varied from 27 (Canadian French) to 256 (Man- darin Chinese) with 81 patients completing the RL. Mean age ranged from 31.7 years (Spain) to 52.9 years (Norway) and percent of males ranged from 30.3% (Brazil) to 74.4% A452 Abstracts