Effectiveness of electronic reminders to improve medication adherence in tuberculosis patients: a clusterrandomised

Size: px

Start display at page:

Download "Effectiveness of electronic reminders to improve medication adherence in tuberculosis patients: a clusterrandomised"

Erick Holmes
6 years ago
Views:

1 Effectiveness of electronic reminders to improve medication adherence in tuberculosis patients: a clusterrandomised trial Katherine Fielding on behalf of: Xiaoqiu Liu, James Lewis, Hui Zhang, Wei Lu, Shun Zhang, Guilan Zheng, Liqiong Bai, Jun Li, Xue Li, Hongguang Chen, Mingming Liu, Rong Chen, Junying Chi, Jian Lu, Shitong Huan, Shiming Cheng, Lixia Wang, Shiwen Jiang, Daniel P Chin

2 TB in China Ranked 2 nd in the world in number of TB cases; accounts for 11% of the ~9 million cases DOTS covered whole country in 2005 Difficult to implement directly observed therapy in some areas; selfadministered treatment / treatment monitored by family members allowed Systematic review in China: 52% receive self-administered treatment; 27% observed by family members; 20% observed by health workers 20% of TB patients had defaulted or were not taking their medications regularly WHO Global TB report 2014 Hou et al. IJTLD 2012 Wang et al. Lancet 2014

3 Intervention components 1. Remind patient to take their medication every other day 2. Remind patient of upcoming monthly follow-up visits 3. Collect data to enable doctor to assess patient s adherence each month 1-2 doses missed: counsel on importance of adherence 3-6 doses missed for 1 st time: switch to intensive management 3-6 doses missed for 2 nd time: switch to DOT 7+ doses missed: switch to DOT intensive management township and village/community doctors visited the patient twice a month or once a week, respectively, for rest of treatment DOT each dose of treatment supervised by the township or village/community doctor

4 Medication monitor Stores one month of TB medication in FDC blister packs Records date/time whenever opened onto chip in monitor Data downloaded to computer via cable at monthly visits Can be set to remind to take medication (beeps) First beep of day at patient preferred time If not opened, up to eight more beeps from 5 mins to 8 hours after agreed time Can be set to remind patient of follow-up visit Human voice 4,3,2,1 days before scheduled monthly follow-up visit

SMS Patients sent an SMS with standard text to remind them to take their tablets every other day They send a reply when they take their medication, otherwise receive up to two more messages that day

5 SMS Patients sent an SMS with standard text to remind them to take their tablets every other day They send a reply when they take their medication, otherwise receive up to two more messages that day SMS reminder sent 4,3,2,1 days before the scheduled monthly follow-up visit SMS replies stored in database to enable doctor to assess numbers of missing doses (based on lack of reply) at monthly visit

6 Trial design Pragmatic cluster randomised trial 36 clusters; 4 provinces Cluster defined as rural counties or urban districts Very few exclusion criteria; ~34% smear +ve 4 arms: control, SMS, medication monitor (MM) & combined Outcomes Primary: adherence outcome measured using medication monitor % of months with at least 20% doses missed (3/15 doses) based on lack of box opening and pill count Secondary: end of treatment outcome as per NTP

7 Four trial arms Control SMS only MM only Combined Medication Monitor Records date/time of opening Reminds to take meds Reminds monthly visit SMS Reminds to take meds Reminds monthly visit Information available to doctor Patient interview Pill count Lack of opening of MM Lack of SMS reply

8 Four trial arms Control SMS only MM only Combined Medication Monitor Records date/time of opening Reminds to take meds Reminds monthly visit SMS Reminds to take meds Reminds monthly visit Information available to doctor Patient interview Pill count Lack of opening of MM Lack of SMS reply

9 Four trial arms Control SMS only MM only Combined Medication Monitor Records date/time of opening Reminds to take meds Reminds monthly visit SMS Reminds to take meds Reminds monthly visit Information available to doctor Patient interview Pill count Lack of opening of MM Lack of SMS reply

10 Sample size & randomisation 36 clusters randomised to four arms Stratified by urban/rural Restricted for balance by arm for province 110 TB patients/cluster 40% reduction in poor adherence, power 90%, coefficient of variation of 0.25

11 Demographics 4,020 TB patients recruited and followed up 70.8% of participants were male Median age was 43 years (inter-quartile range: 29 to 56 years) Some imbalance by intervention arm for occupation, local residency, distance from nearest TB clinic, education level, income category

12 Results - adherence Study arm Number of patients % of months with 20% doses missed Mean ratio (95% CI) p-value Control % 1 Text messaging % 0.91 (0.68, 1.23) Medication monitor % 0.54 (0.39, 0.76) Combined % 0.41 (0.22, 0.79) Very similar results from analysis adjusted for individual level variables of gender, age category, occupation, living in household registration place or not, distance from nearest TB clinic, education level, income category, smear result at start of treatment and cluster level variable of pre-randomisation strata (rural/urban).

13 Results - default Study arm Number of patients % of patients who defaulted Mean ratio (95% CI) p-value Control % 1 Text messaging % 0.35 ( ) Medication monitor % 0.54 ( ) Combined % 0.76 ( ) Very similar results from analysis adjusted for individual level variables of gender, age category, occupation, living in household registration place or not, distance from nearest TB clinic, education level, income category, smear result at start of treatment and cluster level variable of pre-randomisation strata (rural/urban).

14 % of patients with technology problem Technology problems 60% 50% 40% 30% 20% 10% 0% Control SMS only MM only Combined (n=1067) (n=971) (n=969) (n=1013) trial arm Problem with SMS Problem with MM

15 Discussion Non-adherence substantially lower in medication monitor arms; possible reduction in default (underpowered) SMS alone did not impact on non-adherence; reduced default Problems with both technologies Switching to intensive management or DOT rarely happened, possibly due to doctor losing faith in the technology Designing a second trial now with improved medication monitor technology and a clinical endpoint

16 Acknowledgments China CDC and TB control programme Bill & Melinda Gates Foundation

INDEPTH Scientific Conference, Addis Ababa, Ethiopia November 11 th -13 th, 2015

The relationships between structure, process and outcome as a measure of quality of care in the integrated chronic disease management model in rural South Africa INDEPTH Scientific Conference, Addis Ababa,