Development of the TIMER
TIMER is a guide for pharmacists and pharmacy students to follow when conducting medication reviews. TIMER was developed by 2 of the authors (KF and EC), with input from a consensus panel of 4 regional experts who reviewed the tool and provided feedback. Using a scale ranging from strongly agree to strongly disagree, reviewers rated each section of the TIMER on whether the content was evidence-based, important, helpful/useful, and understandable. Feedback from reviewers resulted in several improvements to TIMER, for example, including drug-drug interactions based on both prevalence and severity rather than just on severity, and reducing the symptom timeframe to several months.
An important assumption made in developing TIMER was that its users have conducted a patient medication history so that a complete medication list is available. TIMER has 4 sections including cost-effective drug selection, adherence, medication safety, and attaining therapeutic goals, and covers the most common medication issues that affect older adults. Specific reference to the 8 DRPs commonly used in practice-based research studies was not included because TIMER was intended to encourage pharmacists to look beyond those DRPs and consider patients’ symptomatology and complications among older adults.
Each of the 4 sections includes points to discuss with patients and suggested recommendations if a DRP is found. The section on cost-effective drug selection suggests generic and therapeutic substitution to ensure that patients are getting the most cost-effective medications. The section of adherence gives examples of how to question patients about adherence and provides specific recommendations. A section on medication safety addresses adverse drug effects, screening for symptomology, inappropriate medications, drug interactions, and Beer's criteria medications. When determining whether patients are attaining their therapeutic goals, TIMER contains guidance on cardiovascular risk reduction and complication management. The section on cardiovascular risk management outlines the major risk factors for coronary heart disease and evaluates treatment goals. The section on complication management identifies common complications seen in the geriatric population, how to screen for them, and recommendations for each complication (Appendix 1
Evaluation of TIMER Via Written Cases
To evaluate TIMER, a 2-part randomized, controlled study was designed that involved practicing pharmacists and pharmacy students using the tool to assess hypothetical patient cases.
Patient Case 1 was developed by one of the authors (MA) and based on a case taken from IowaTeach, a University resource for faculty members to use in developing teaching activities. Both clinical and hypothetical patient cases are available in IowaTeach and many contain additional instructional materials such as test questions and teaching notes. Case 1 was an older adult presenting to a community pharmacist for MTM services. Two expert clinicians reviewed Case 1 and identified 13 DRPs, 6 of which were tool-related and 7 that were non-tool-related (Table ). Tool-related DRPs were those covered in TIMER and non-tool-related DRPs were problems not included in TIMER.The non-tool-related DRPs identified by the experts were not eliminated from the case but were not expected to be affected because TIMER did not contain them.
Drug-related Problems (DRPs) in Three Cases Evaluating TIMER
Cases 2 and 3 were written by 2 of the authors (MC and JR). Both cases involved older adults presenting to pharmacists for MTM services. The authors identified 4 non-tool-related DRPs and 5 tool-related DRPs in their respective case, and this list later served as the key when coding DRPs (Table ). Case 1 from the pharmacists study served as the basis for cases 2 and 3. The DRPs from case 1 were incorporated into cases 2 and 3, with substitution of medications. For example, warfarin interactions in cases 1 and 2 involved levothryoxin, while warfarin interactions in case 3 were caused by bismuth subsalicylate. Another example of modifications was the wrong drug used in the cases. In cases 1 and 2, the wrong drug was propoxyphene and in case 3 it was diphenhydrame: both are Beer's list drugs.
Part 1: Pharmacists’ Use of the TIMER
A randomized controlled study of practicing pharmacists who were members of a regional MTM network (Iowa, Minnesota, Nebraska, North Dakota, South Dakota, Wyoming and Montana) was conducted and half of the pharmacists were randomly selected to receive TIMER. All pharmacists received a printed copy of case 1 and a response form with instructions to identify DRPs in the patient case and write SOAP notes including recommendations. A document of consent to participate was also included in the packet. Participants were asked to return all materials to the investigators and by doing so indicated their informed consent. IRB approval for this study was obtained.
The packet of materials was mailed to 136 pharmacists in mid-April 2007 and a follow-up postcard was sent to non-responders 2 weeks later. A second packet with the same materials was mailed again in mid-June to nonresponders and a follow-up postcard was sent 1 month later. Responses obtained by mid-August 2007 were included in the analysis. Demographic information obtained when the MTM network was formed was linked to the responses in this study. These data included age, practice years, gender, average hours spent in pharmacy per week, and state in which they practiced.
An investigator (AS) coded the pharmacists’ responses as either correctly identifying each of the 13 DRPs or not, using a pre-established set of coding rules. Only the DRPs were considered when coding the responses, and not the actions the pharmacist recommended. If a DRP was unclear, the study team reviewed it and consensus was reached. The data were entered into a spreadsheet and analyzed using SPSS. The 2 groups of pharmacists (those who received the TIMER and those who did not) were compared for age, years in practice, and hours worked per week using t tests, and for gender using the chi-square test. The numbers of tool-related and non-tool-related DRPs were summed for each respondent. The total number of tool-related and non-tool-related DRPs identified per study group were compared, controlling for gender and practice years. A chi-square test also was used to compare each tool-related DRP identified with whether the pharmacist had received the TIMER.
Part 2: Pharmacy Students’ Use of the TIMER
In the second part of the study, third-year pharmacy students enrolled in the Pharmacy Practice Laboratory course at University of Iowa College of Pharmacy were asked to identify DRPs in 2 cases, 1 using and 1 not using TIMER. IRB approval for this study was obtained. As seating was randomly assigned at the start of the course, students were already assigned to 54 groups of 2. Students were informed that their answers would not affect their grade for the course. Students reviewed a study information sheet containing all elements of informed consent. Their submission of answers indicated informed consent. For the first 30 minutes of class time, each group of 2 students was assigned 1 of 2 patient cases and asked to identify drug-related problems. For the next 30 minutes, each group was given a second case along with the TIMER and again asked to identify DRPs. Groups of 2 were randomly assigned to receipt of TIMER for 1 of the 2 cases.
The students were asked to provide their age, pharmacy grade point average (GPA), gender, laboratory section, and a unique identifier, which allowed for statistical comparisons to be made later without compromising students’ anonymity. Groups of 2 students were asked to list all DRPS that they could find in 30 minutes and state the action that would be taken for each DRP. As the objective was to determine whether TIMER improved students’ ability to identify DRPs, only the DRPs were considered during coding and not students’ proposed solutions.
Coding of responses was done by an investigator (SL) trained to examine the SOAP notes and identify the presence or absence of the DRPs. Each DRP was coded as either correctly identified or not correctly identified for each group. A set of coding rules for each case was developed by the investigator and reviewed by other investigators before coding was completed. The classification of a correct versus incorrect DRP was based on their written identification of a DRP, not necessarily how the student described it. For example, for the presence of nonadherence to verapamil in case 2, any mention of poor compliance with verapamil was coded as a “yes, the DRP was identified” whether the cause of noncompliance was attributed to side effect, inability to swallow, or patient thinking the drug did not work. Also, if poor compliance with another drug besides verapamil was identified, subjects were not given credit for identifying noncompliance with verapamil. Simply identifying the ADR of verapamil and constipation was not sufficient if compliance or patient education was not mentioned. Proposed actions or recommendations were not used to determine DRPs—the DRP had to be stated. If a DRP was unclear, the study team reviewed it and consensus was reached.
Results were entered into a spreadsheet and analyzed using SPSS. Subjects were divided into 2 study groups: those who received case 2 first and those who received case 3 first. Age, gender, and GPA of the student pharmacists in the 2 study groups were analyzed using chi-square and t tests. The number of tool and non-tool related DRPs identified by student groups were summed across both cases and used as dependent variables. Tool-related DRPs were those that were included in TIMER. The independent variables used in analyses were laboratory section of the student, whether TIMER was used, and patient case. A one-way ANOVA was used to calculate the difference in dependent variables between the 3 Pharmacy Practice Laboratory sections. Then the effect of TIMER and case on total tool-related and total non-tool-related DRPs identified was examined using 2-way ANOVA. We also analyzed the cases separately for significance of TIMER using 2-tailed t tests, as case was significant in the primary analysis. A p value <0.05 was considered significant.