OBJECTIVE: The primary objective of this study was to investigate whether nonresponders to a medication-beliefs survey exhibited different adherence and persistence patterns than survey responders. METHODS: A medication-beliefs survey was mailed to 7795 adults aged from 40 to 88 years, who filled a qualifying index prescription (cardiovascular, dyslipidemia, oral-antihyperglycemic, oral-bisphosphonate, and asthma-controller medications) in June 2008 at 1 national and 2 regional retail pharmacies. Adherence and persistence to the index drug class was measured using pharmacy-claims data over 12 months. A multivariate generalized linear model with a negative binomial distribution and log-link function was used to determine whether response status was a significant predictor of adherence. Kaplan-Meier estimates of survival curves were used to assess the time to discontinuation (persistence). Differences between nonresponders and responders were assessed using the log-rank test. RESULTS: The survey response rate was 24.25%. The final analytic sample size after exclusions was 6740 patients (5044 nonresponders and 1696 responders). On the basis of multivariate generalized linear model analysis, survey nonresponders had 11% lower medication adherence compared with responders (P < 0.01; goodness-of-fit=1.09 as defined by deviance/df statistics). The proportion of nonresponders deemed nonpersistent at day 305 was 66.3% compared with 58.1% of responders (P < 0.001). The Kaplan-Meier persistence curves were significantly different for nonresponders and responders as assessed by the log-rank test (χ statistic=49.38; P < 0.001). CONCLUSIONS: Our study found that the responders and nonresponders to a medication-beliefs survey differed significantly in their subsequent adherence and persistence, suggesting that biased survey results are likely to accompany low response rates in surveys of medication use. The use of modest monetary incentives had a small effect on survey response; multiple survey levers are recommended to reduce nonresponse and the potential for biased results.
OBJECTIVE: The primary objective of this study was to investigate whether nonresponders to a medication-beliefs survey exhibited different adherence and persistence patterns than survey responders. METHODS: A medication-beliefs survey was mailed to 7795 adults aged from 40 to 88 years, who filled a qualifying index prescription (cardiovascular, dyslipidemia, oral-antihyperglycemic, oral-bisphosphonate, and asthma-controller medications) in June 2008 at 1 national and 2 regional retail pharmacies. Adherence and persistence to the index drug class was measured using pharmacy-claims data over 12 months. A multivariate generalized linear model with a negative binomial distribution and log-link function was used to determine whether response status was a significant predictor of adherence. Kaplan-Meier estimates of survival curves were used to assess the time to discontinuation (persistence). Differences between nonresponders and responders were assessed using the log-rank test. RESULTS: The survey response rate was 24.25%. The final analytic sample size after exclusions was 6740 patients (5044 nonresponders and 1696 responders). On the basis of multivariate generalized linear model analysis, survey nonresponders had 11% lower medication adherence compared with responders (P < 0.01; goodness-of-fit=1.09 as defined by deviance/df statistics). The proportion of nonresponders deemed nonpersistent at day 305 was 66.3% compared with 58.1% of responders (P < 0.001). The Kaplan-Meier persistence curves were significantly different for nonresponders and responders as assessed by the log-rank test (χ statistic=49.38; P < 0.001). CONCLUSIONS: Our study found that the responders and nonresponders to a medication-beliefs survey differed significantly in their subsequent adherence and persistence, suggesting that biased survey results are likely to accompany low response rates in surveys of medication use. The use of modest monetary incentives had a small effect on survey response; multiple survey levers are recommended to reduce nonresponse and the potential for biased results.
Authors: Joanna E Bulkley; Maureen O'Keeffe-Rosetti; Christopher S Wendel; James V Davis; Kim N Danforth; Teresa N Harrison; Marilyn L Kwan; Julie Munneke; Neon Brooks; Marcia Grant; Michael C Leo; Matthew Banegas; Sheila Weinmann; Carmit K McMullen Journal: Qual Life Res Date: 2019-12-06 Impact factor: 4.147
Authors: A Sieben; S J H Bredie; J C H B M Luijten; C J H M van Laarhoven; S van Dulmen; H A W van Onzenoort Journal: BMC Med Res Methodol Date: 2019-05-09 Impact factor: 4.615
Authors: Yiman Wang; Denise M J Veltkamp; Paul J M van der Boog; Marc H Hemmelder; Friedo W Dekker; Aiko P J de Vries; Yvette Meuleman Journal: Transpl Int Date: 2022-02-07 Impact factor: 3.782