Background: The use of oral antineoplastic agents has increased in cancer medicine. However, the convenience of oral medication carries the risk of nonadherence and dosing errors, which could jeopardize therapeutic benefits and patient safety. This is a quality improvement project to investigate reasons for nonadherence and medication errors in patients receiving oral investigational treatment at the Phase 1 Department at MD Anderson Cancer Center, Houston, Texas. Early-phase clinical trials have an enormous impact on drug development and patient safety, not much has been done to evaluate adherence in patients receiving investigational oral antineoplastic agents. Materials and Methods: We examined our clinic dynamic including the initial encounter, follow-up phone calls, medication administration, and patient adherence the following visit. Then we explored and classified the main possible reasons for nonadherence and medication error across the workflow. Results: When examining potential deficiencies in the clinic flow, which are actionable and carry high impact, we found the initial encounter had a significant room for improvement and errors happened when instructions provided were unclear, not correct, or contradicted with the prescription or the label. Furthermore, the follow-up calling was also an important step to monitor and improve compliance. However, it was not a consistent practice and lacked a standardized format. Lastly and although the multistep reconciliation process for oral medication is important to monitor compliance, it was complex, had multiple manual aspects, and added substantial burden on the research staff. Conclusion: In this project, our goal was to shed light on the possible causes of oral medication errors and nonadherence in clinical trials. We proposed feasible measures including educational, training, and adherence monitoring tools. We will continue to monitor and evaluate our data to see any positive or negative impact from our interventions.
The use of oral antineoplastic agents has increased in cancer medicine, especially with the approval of many oral targeted agents since 2001.[1,2] Multiple studies have shown that patients with cancer prefer oral antineoplastic agents over intravenous chemotherapy.[3–6] However, the convenience of oral medication carries the risk of nonadherence and dosing errors, which could jeopardize therapeutic benefits and patient safety.[7–9]
Furthermore, pharmacokinetic and pharmacodynamic studies are highly dependent on the intake of oral medication and noncompliance, including falsely reporting of adherence, could negatively impact drug development.[10–12]
In phase 1 clinical trials, nonadherence and dosing errors have a lasting effect as many protocol end points including the recommended dosage will be decided based on a limited number of patients.[10,11] However, not much has been done to evaluate or improve adherence in early clinical trials setting. This represents a significant knowledge gap, considering its impact on patient safety and drug development.
Acquiring this knowledge and developing improvement assessment measures may improve the current practice in clinical trials and general oncology and it will open the door for in-depth research studies in this area.
In this paper, we followed a quality improvement model to find potential causes of errors, pitfalls at each step of patient care, and proposed recommendations to implement.
Materials and Methods
This is an observational quality improvement project to investigate reasons for nonadherence and medication errors in patients receiving oral investigational treatment at the Phase 1 Department at MD Anderson Cancer Center, Houston, Texas.
We have over 160 active therapeutics protocols for advanced solid tumors, 12 faculty members and 1200 patients enrolled in 2017 alone with 818 therapeutics protocols approved by the institutional review board since 2004. The number of oral investigational agents dispensed reached 112 between February 2017 and April 2018. Those oral medications are part of 124 different therapeutic protocols. During this period, 719 patients received oral investigational (INV), and the investigational pharmacy has generated 3878 oral INV agents' prescriptions.
To conduct this quality improvement study, we formed a team of physicians, clinical research staff, and experts involved in patient education, data management, and quality control. Then, we reviewed the workflow within our clinical center for targeted therapy (CCTT) for patients receiving oral cancer treatment. Then, we categorized all potential causes of error using a fishbone chart. Further, we narrowed down the list to a few actionable items by prioritizing areas of low effort and high impact using the effort–impact chart.
Clinic dynamics Initial encounter
The workflow begins in the CCTT clinic with the initial encounter between the physician and the patient as shown in Figure 1. The physician explains treatment risks and benefits and reviews labs and comorbidities to ensure the patient meets the eligibility criteria and the clinical safety measures. The research study coordinator or research nurse as delegated by the principal investigator is the liaison between the patients, health-care providers, pharmacists, and sponsors of the trials. The role of study coordinator and research nurse is summarized in three steps, starting with the initial encounter when explaining the protocol schedule, reviewing prescription instructions and prohibited medications, and handing out the pill diary, study calendar, and any other related study materials to the patient. They also assure that the patient has a full understanding of the study protocol and they do follow up with the patient regarding scheduling or to address protocol-related questions.
When examining the potential deficiencies in the initial encounter, we found out that errors happen when instructions provided are unclear, not correct, or contradict with the prescription or the label in addition to the inconsistent physical or visual demonstration of how to take the pills.
After the patient finishes with the initial in-clinic drug administration, the research coordinators occasionally conduct a follow-up phone call as a reminder for the patient to take the medication and to answer any questions the patient might have. However, this practice is not consistent and occasionally done. The factors that limit this practice are the time constraint, the lack of standardization, and the significant burden on the staff. The only exception is when conducting phone calls is mandated by the protocol. Furthermore, even with calling, we observed a lack of specific format to conduct the phone call, including refraining from asking open-ended questions or late calling after the patient received the medication.
The patient self-administers the oral medication based on the instructions received in the clinic. These instructions could be verbal, written, or both. The patient in many cases has to deal with a frequent and complex treatment schedule with supplementary information regarding meals timing and a list of prohibited medication and food.
We asked a group of physicians and research staff about the reasons of nonadherence of their patients based on their experience in our phase 1 clinic. The most frequent answers were as follows: not understanding the instructions, forgetting, confusion with other medication, specifically for patients with polypharmacy, demanding treatment schedule, late filing or inconsistency of the pill diaries, or lack of information regarding the contact person from the research or clinical staff.
Next clinic visit
The last step comes when the patient presents to the clinic for a safety checkup or subsequent treatment cycle clearance. At this stage, the research staff will receive the pill diary from the patient and complete the medication reconciliation. This process consists of multiple manual steps including counting the pills, reviewing dispensation records to match the bottle number, and review the pill diary and cosign it with the patient. In a busy clinic environment, having a complex manual process increases the incident for errors. At the same time, failing to follow all these reconciliation steps might result in missing medication errors or nonadherence issue by the patient that could recur in the following treatment cycles.
Categories of errors bases
The second step included categorizing the causes of nonadherence or medication errors using a fishbone diagram. After we explored the main possible reasons for nonadherence across the workflow, we classified the specific causes and ranked them as primary (#1), secondary (#2), or tertiary (#3) as shown in Figure 2.
Effort and impact chart
As we identified multiple items as potential causes of medication errors, we plotted all the reasons of medication errors on an effort and impact chart to prioritize areas of highest impact and less effort as shown in Figure 3. By doing so, we were aiming to achieve improved outcome without creating a burden on the workflow that could hinder any sustainable solution. The main areas that we considered as areas of high impact and low efforts were as follows: patient did not follow or comprehend the instructions, providing contradicted information to the patient, not providing a template for dosing strength, the patient not aware of the appropriate contact person, no follow-up phone call, and not asking open-ended questions.
The errors in cancer treatment might occur at any stage including prescription, labeling, dispensation, and drug administration. The rate of those errors is approximately one to four per 1000 orders, which impact at least 1%–3% of cancer population. Generally, in clinical trials and phase 1 in particular, health-care providers and research staff are expected to follow detailed instruction as written in the protocols. This instruction consists of dosage schedule, dosage modification parameters, holding or discontinuation criteria, prohibited medication, and sometimes specific instructions on how to take the medication.
The use of home-based oral cancer treatment has empowered the patient to be in charge of his/her health. After completing the initial encounter at the clinic, the patient becomes accountable for proper medication compliance at home. For example, taking an extra dose or failure to inform the physician about any emerging toxicity are significant safety issues.
The adherence to cancer treatment is considered high and fourth in rank among 17 diseases with a mean adherence rate of 79.1%. However, the compliance rate in oral antineoplastic medication is not well established and ranges between 46%–100%. This variation possibly arose because of the difference in patient population, treatment category, follow-up period, assessment procedures, and methods of tracking and reporting adherence. Even in clinical trials whereas multiple assessment measures implemented to assess the compliance, the low rate of reporting adherence data, which consist of 20% of clinical trials with oral antineoplastic agents is an obstacle toward obtaining reliable adherence rate.
Strategic planning by directing potential solutions toward the initial encounter and follow-up phone calls and visits could have a significant impact in reducing medication errors.
On the basis of our observation in our clinic, we have been working on multiple interventions summarized in Table 1. These possible improvement strategies include the development of best practice assessment tools and in-service training for all research staff during the initial patient encounter. This training consists of delivering good patient education about risks and benefits of treatment and providing clear instruction on how to take the medication and what to do if toxicity occurs.
Another key measure for improvement is streamlining the workflow for clinical trials with oral medications and implementing consistent and efficient process as a department standard operating procedures. Such standardization will allow conducting regular follow-up phone calls in the early stage of treatment, which serve as reinforcement of the treatment schedule and a way for early errors detection.
On the education level, a double checklist for patient and research staff will ensure comprehensive and standardized education by the staff and adequate information delivery to the patient. An example of this checklist of the initial encounter is shown in Table 2.
One of the main limitations in addressing medication errors is the lack of a standardized and accurate method to measure the adherence. Tracking treatment compliance in phase 1 relies on the patient's self-report, pill diaries log, and leftover pills. However, those methods are inadequate to represent the real adherence rate and give the patient a high share of responsibility in reporting.
One observational study at three pediatric oncology clinics used home visits as an assessment measure to report medical errors. This study has found that the rate of medication errors among all study sites was 72 during 92 visits. Among those medical errors, 40 had the probability to seriously harm the patient, and four did harm the patients. The rate of errors per 100 patients differed between sites and ranged between 40 and 121.
Other major issue arises when the patient reports false information, especially if the patient is aware of the reconciliation process. This could prevent the research team from proper detection of those errors. Similar fabrication has been reported earlier when some patients reported false information during the screening process to be eligible for the clinical trials.
Understanding the patient population is also important, many patients who are coming for phase 1 clinical trials failed standard treatment and are seeking experimental clinical trial as a last resort. The fear of being removed from the study for noncompliance might lead to falsely reported data, which represent a significant concern.
Another important aspect that needs to be considered is the substantial burden on the research staff performing the reconciliation process when matching the pill diaries, patient's report, protocol requirements, and unused medications. Therefore, driving toward automating the workflow in areas such as documentation and pills reconciliation could reduce medication errors for two main reasons; first, by avoiding human errors in the multistep reconciliation process and second, by driving the effort and time of the research staff to educate and interact with the patient. Examples of such measures that are easy to use are smart text in documentation and automatic pill counter.
The electronic pill diaries have also been used to measure the adherence. However, it will reflect the dispensation of the pill and not the actual administration. Also, the high cost of using this technology prevents widespread application.
Future directions should also focus on improving drug packaging for investigational drugs and avoiding having multiple pills with different strength. In addition, writing more cohesive protocols that have clear instruction and an easy schedule to follow by the patient.
Furthermore, using modern technology to help both the patient and research staff to track the oral medication administration will minimize errors and increase the detection of preventable errors. An example of such application is building a reminder function for treatment schedule through a patient's portal and even expand the alert capacity to generate electronic pill diary records.
Both patient safety and data integrity are crucial for early drug development. This requires complete adherence to the protocol to minimize any potential dosing errors. In this project, our goal was to shed light on the possible causes of oral medication errors and nonadherence in clinical trials. We proposed feasible measures that we have started implementing, including an educational component, training, and adherence monitoring tools. We will continue to monitor and evaluate our data to see any positive or negative impact from our interventions.
Financial support and sponsorship
The author disclosed no funding related to this article.
Conflicts of interest
The author disclosed no conflicts of interest related to this article.
For reprints contact:firstname.lastname@example.org