Mobile phone intervention for improving ART adherence

A randomised control trial was done to determine the effect of mobile phone intervention on adherence in a two groups (Intervention and Control). Respondents (n = 362) age 18-60 years, HIV seropositive, with access to mobile phone were recruited and followed-up for six months. The Control group received standard care while the Intervention group received standard care, alarm prompting, weekly text messages and monthly voice calls. Primary (overall adherence: Self-report, visual analogue, pill identification, pill count) and secondary (CD4 count and Body Mass Index) outcomes were measured at baseline (B), month three (F1) and month six (F2). Series of data transformation and computations were done to derive final outcome. Data comprises nominal, ordinal and scale computations. -BMI= Weight in kilogram divided by height in meter square Scoring multimethod adherence tool 1.Self-reporting (Sr) There are four questions with Yes/No responses. Response Interpretation Code % Score No to all items Highly Adherent 3 95 or more Yes to 1 item Moderately adherent 2 75-94 Yes to 2 or more items Low adherence 1 75 or less 2.Visual Analogue Scale (VAS) Responses 0 1 2 3 4 5 6 7 8 9 10 Score (%) 0 10 20 30 40 50 60 70 80 90 100 3.Pill identification test (PIT) Response Interpretation % Score Ability to remember dose, time and instruction Highly Adherent 95 or more Ability to remember dose and time Moderately adherent 75-94 Ability to remember dose only Low adherence 75 or less 4. Pill Count (PC) % Adherence = Dispensed –Returned x 100 divided by Expected to be taken 5. Overall Adherence scores (OAS)More than 95 % (High) 75-95% (Moderate) Less than 75% (Low) Other Variables such as demographic profile, use of mobile phone and factors influencing adherence were also measure and categorised in data set. Mixed modeling requires data restructuring and use of the following syntax (replace variable to estimate other relationships. MIXED OAS BY Index1 id GROUP WITH age sex Ledu Mstatus employment Rel /CRITERIA=CIN(95) MXITER(100) MXSTEP(10) SCORING(1) SINGULAR(0.000000000001) HCONVERGE(0, ABSOLUTE) LCONVERGE(0, ABSOLUTE) PCONVERGE(0.000001, ABSOLUTE) /FIXED= time GROUP age sex Ledu Mstatus employment Rel time*GROUP /REPEATED = time | SUBJECT(id) COVTYPE(arh1) /emmeans = tables(GROUP*time) compare(GROUP) /emmeans = tables(GROUP*time) com MIXED BMIres BY id GROUP time /CRITERIA=CIN(95) MXITER(100) MXSTEP(10) SCORING(1) SINGULAR(0.000000000001) HCONVERGE(0, ABSOLUTE) LCONVERGE(0, ABSOLUTE) PCONVERGE(0.000001, ABSOLUTE) /FIXED= time GROUP time*GROUP /REPEATED = time | SUBJECT(id) COVTYPE(arh1) /emmeans = tables(GROUP*time) compare(GROUP) /emmeans = tables(GROUP*time) compare(time).

本研究采用随机对照试验（randomised control trial）设计，旨在探究手机干预措施对服药依从性的影响，试验分为干预组与对照组两个组别。共招募362名年龄介于18-60岁的人类免疫缺陷病毒（HIV）血清阳性且可使用移动电话的受试者，随访时长为6个月。对照组仅接受标准护理，干预组则在标准护理基础上，额外增加警报提示、每周短信及每月语音通话干预。研究在基线（B）、第3个月（F1）及第6个月（F2）三个时间点，采集主要结局指标与次要结局指标：主要结局指标为整体服药依从性，涵盖自我报告、视觉模拟量表（Visual Analogue Scale, VAS）、药物识别测试（Pill identification test, PIT）与药片计数（Pill Count, PC）；次要结局指标为CD4计数与身体质量指数（Body Mass Index, BMI）。研究经一系列数据转换与计算后得到最终结局数据，数据涵盖名义型、有序型与等距型计量资料。——身体质量指数（BMI）计算公式为：体重（千克）除以身高（米）的平方。

### 多方法依从性评分工具
1. 自我报告量表（Self-reporting, Sr）：共包含4道是非题，应答结果与评分规则如下：
- 所有题目均应答“否”：高度依从，赋值3分，对应95%及以上依从率
- 仅1道题目应答“是”：中度依从，赋值2分，对应75%-94%依从率
- 2道及以上题目应答“是”：低度依从，赋值1分，对应75%及以下依从率
2. 视觉模拟量表（VAS）：应答分值与对应依从率（%）对应关系为：0 → 0%，1 → 10%，2 → 20%，3 → 30%，4 → 40%，5 → 50%，6 → 60%，7 → 70%，8 → 80%，9 → 90%，10 → 100%
3. 药物识别测试（PIT）：应答结果与评分规则如下：
- 可记住服药剂量、时间与相关说明：高度依从，对应95%及以上依从率
- 可记住服药剂量与时间：中度依从，对应75%-94%依从率
- 仅可记住服药剂量：低度依从，对应75%及以下依从率
4. 药片计数（PC）：依从率计算公式为：(发放药量 - 剩余药量) × 100% / 预期服用药量

整体依从性评分（Overall Adherence scores, OAS）判定标准为：>95%为高度依从，75%-95%为中度依从，<75%为低度依从。

数据集同时纳入并分类整理了人口统计学特征、移动电话使用情况及影响服药依从性的相关因素等其他变量。本研究采用混合效应模型进行统计分析，需对数据进行重构，使用以下SPSS语法（替换变量即可估计其他关联关系）：
MIXED OAS BY Index1 id GROUP WITH age sex Ledu Mstatus employment Rel
/CRITERIA=CIN(95) MXITER(100) MXSTEP(10) SCORING(1) SINGULAR(0.000000000001) HCONVERGE(0, ABSOLUTE) LCONVERGE(0, ABSOLUTE) PCONVERGE(0.000001, ABSOLUTE)
/FIXED= time GROUP age sex Ledu Mstatus employment Rel time*GROUP
/REPEATED = time | SUBJECT(id) COVTYPE(arh1)
/emmeans = tables(GROUP*time) compare(GROUP)
/emmeans = tables(GROUP*time) com
MIXED BMIres BY id GROUP time
/CRITERIA=CIN(95) MXITER(100) MXSTEP(10) SCORING(1) SINGULAR(0.000000000001) HCONVERGE(0, ABSOLUTE) LCONVERGE(0, ABSOLUTE) PCONVERGE(0.000001, ABSOLUTE)
/FIXED= time GROUP time*GROUP
/REPEATED = time | SUBJECT(id) COVTYPE(arh1)
/emmeans = tables(GROUP*time) compare(GROUP)
/emmeans = tables(GROUP*time) compare(time).