Promoting the Universal Medication Schedule Via Mobile and EHR Technologies

Predictive probabilities of prescription understanding will be calculated based on patients' ability to correctly dose their prescription medications using unadjusted Generalized Estimating Equation models, specifying the binomial family and logit link, with medication as the unit of analysis. Correct dosing per medication will be scored as yes or no, reflecting having demonstrated all of the following: proper dose (# of pills), spacing (hours between doses), frequency (# of times per day), and total pills per day. Results are presented as predicted probabilities with 95% Confidence Intervals.

Predictive probabilities of medication knowledge will be calculated based on patients' ability to identify each medication's purpose and side effects, risks, warnings, and benefits using unadjusted Generalized Estimating Equation models, specifying the binomial family and logit link, with medication as the unit of analysis. Patients will be asked through a structured questionnaire about each of the above via structured, open-ended items. Each medication will be scored as correct/incorrect if the participant knew the purpose and could name at least 1 side effect, risk, or warning of the medication. Results are presented as predicted probabilities with 95% Confidence Intervals.

Predictive probabilities of medication adherence will be calculated based on telephone pill counts using unadjusted Generalized Estimating Equation models, specifying the binomial family and logit link, with medication as the unit of analysis. Pills taken/pills prescribed will be calculated for each medication and scored as adherent for that medication if that score is between 80% and 120%. Results are presented as predicted probabilities with 95% Confidence Intervals.

Predictive probabilities of medication adherence will be calculated based on a 4-item validated Patient Medication Adherence Questionnaire (PMAQ) using unadjusted Generalized Estimating Equation models, specifying the binomial family and logit link, with medication as the unit of analysis. The PMAQ assesses adherence behaviors by asking patients to self-report missed/wrong doses in past 4 days, scoring each medication as adherent for that medication if no missed doses were reported. Results are presented as predicted probabilities with 95% Confidence Intervals.

Predictive probabilities of primary medication adherence will be calculated based on the proportion of days covered (PDC) with medication obtained from pharmacy records using unadjusted Generalized Estimating Equation models, specifying the binomial family and logit link, with medication as the unit of analysis. PDC is calculated by summing the number of days' supply obtained by a patient during a given time period and dividing by the number of days for which the patient was prescribed the medication. Each medication is scored as adherent if the patient was covered for that medication more than 80% of the time. Results are presented as predicted probabilities with 95% Confidence Intervals.

An exploratory analysis will be conducted to investigate the impact of receiving the UMS in either intervention arm on biologic outcomes for these prevalent conditions. Systolic blood pressure will be obtained from patients' electronic health records. Differences will be measured between the last clinical measurement prior to baseline assessment and the last measured value during the study period (closest to 6 month assessment).

An exploratory analysis will be conducted to investigate the impact of receiving the UMS in either intervention arm on biologic outcomes for these prevalent conditions. Hemoglobin A1c (hbA1c) will be obtained from patients' electronic health records. Differences will be measured between the last clinical measurement prior to baseline assessment and the last measured value during the study period (closest to 6 month assessment).

An exploratory analysis will be conducted to investigate the impact of receiving the UMS in either intervention arm on biologic outcomes for these prevalent conditions. Low-density lipoprotein cholesterol (LDL) will be obtained from patients' electronic health records. Differences will be measured between the last clinical measurement prior to baseline assessment and the last measured value during the study period (closest to 6 month assessment).