A pharmacy medication alert system based on renal function in older patients

Setting and study population

The study was conducted in Arnhem, a city in the East of the Netherlands with nearly 148 000 inhabitants. Seven GPs, belonging to the same pharmacotherapy audit meeting group, participated in the study. Five pharmacists who worked in close collaboration with this group selected the patients in their pharmacy computer system. Patients aged ≥70 years in the care of the participating GPs were eligible if they were on GP-prescribed maintenance therapy of blood-glucose-lowering or cardiovascular drugs (for example, digoxin, diuretics, or inhibitors of the renin–angiotensin system [RAS], including angiotensin-converting enzyme [ACE] inhibitors and angiotensin II receptor blockers [ARBs]). Patients also used at least one ‘target drug’ on the inclusion date of 4 January 2010. ‘Target drugs’ were defined as drugs requiring therapeutic advice in patients with decreased renal function considering the Dutch dosing guideline for impaired renal function.17 Patients with an estimated glomerular filtration rate (eGFR) <10 ml/min/1.73 m² were excluded (Figure 1).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Selection of the study population.

The GPs already used a computerised medication monitoring system. This system generated an alert when the GP prescribed a target drug in patients labelled as having CKD, but it could not consider the eGFR level.

The use of drug dispensing data and laboratory test results in this study complied with Dutch privacy regulations.

Renal function monitoring

Actual eGFR was defined as an eGFR value measured within the last 12 months. If an actual eGFR was unknown, the GP requested the patient to undergo a blood test for renal function. The laboratory provided serum creatinine and an eGFR (ml/min/1.73m²) calculated by the normalised four-variable Modification of Diet in Renal Disease (MDRD).18 Serum creatinine was measured enzymatically (Modular, Roche diagnostics) and was IDMS (isotope dilution mass spectrometry) calibrated. The actual values of eGFR data were provided to determine drug-specific risk.

Assessment of renal function alerts

In this study, the pharmacists used a pharmacy medication alert system (PMAS) built by one of the authors in a Microsoft® Access® database. The system, which was an addition to the current pharmacy computer system, assessed the medication in relation to the reported eGFR and provided an alert for target drugs according to the Dutch guidelines for drug administration in reduced renal function.17 These guidelines include drug-specific cut-off values for eGFR, accompanied by a therapy-adjustment advice.

After receiving a data file from the GP with the eGFR of the included patients, the pharmacists linked the eGFR in the PMAS. Simultaneously, the patient’s actual medication was electronically imported from the usual pharmacy computer system into the PMAS. An alert was generated to stipulate action if the eGFR was lower than the cut-off value of the target drug (Table 1). The software could not correct for invalid dose or dose interval, so the pharmacist assessed the alerts for these aspects based on the guideline recommendations presented in a text box. The pharmacist formulated therapeutic advice for either dosage adjustment, to stop the drug, or to substitute it by a non-contraindicated drug. Once a week, the pharmacist communicated the therapeutic advice to the GP by a list. The GP evaluated the therapeutic advice in relation to the clinical context of each individual patient, and responded with agreement or disagreement. Predefined reasons for disagreement could be checked on the list and the GP was asked to give supplementary comments in a free-text box. The list was returned to the pharmacist.

Outcome

The outcome of the study was the frequency of therapeutic advice formulated by the pharmacist (expressed as a proportion of the total number of target drugs). The management of the therapeutic advice by the GP was also studied.

Statistical analysis

All relevant patient data were entered into a Microsoft Access 2003 database and further analysed with SPSS Statistics (version 17.0) for descriptive statistics (mean, frequency, range).

Assessment of renal function alerts

The computer software generated 212 alerts (15.9%) in a total of 1333 target drugs, because the eGFR was lower than the predefined cut-off value of the target drug. After the pharmacist assessed the actual medication for correct dose and dose interval, 93 alerts (7.0%) appeared to be correct and seven alerts (0.5%) were missing. Therefore, action to adjust therapy was considered necessary in 112 prescriptions in 74 patients (8.4% of the target drugs, 11.4% of the patients). Additionally, pharmacists gave advice in 31 prescriptions of target drugs, even though the eGFR was just above the cut-off value. Eventually, 143 therapeutic recommendations (10.7% of the target drugs) concerning 89 patients (13.7% of analysed study population) were included for analysis of the GP responses. The drugs most frequently involved were diuretics (41.3% of therapeutic advice), blood-glucose-lowering drugs (14.0%), digoxin (11.2%), and RAS inhibitors (10.5%). Almost all prescriptions that received an alert were chronic prescriptions taken by the patient for a longer period of time.

GP response to pharmacist advice

The GP immediately agreed with 71 recommendations (49.7% of the therapeutic advice) concerning 52 patients (8.0% of the study population). The GP most frequently disagreed with the advice on diuretics, blood-glucose-lowering drugs, digoxin, and RAS-inhibitors. Within each of these therapeutic groups, the GP immediately disagreed with one-third of the advice. The responses of the GPs are shown in Table 3.

Summary

The use of a PMAS based on renal function resulted in therapeutic advice for a substantial number of drugs in older patients with diabetes or cardiovascular disease. The GP immediately agreed with half of the advice. Overall, in 5% of the prescriptions, the GP agreed to rectify the prescription.

The GPs used a medication monitoring system based on the Dutch G-standard,17 the national drug database, which is used by all professional parties in Dutch health care. Despite this monitoring system, pharmacists still formulated additive therapeutic advice in 11% of the target drugs. What could be the reasons for this? First, it is known that a high number of medication alerts may cause ‘alert fatigue’ in the prescriber.10 In the case of repeat prescriptions in particular, alerts were ignored. The extra effort to seek a renal function and to weigh the choice and dosage of the drug may cost too much time. Second, this observation could be explained because at the time of prescription, an actual eGFR was not available in more than 20% of the patients. Finally, it is important to consider the prescribing context. The alerts concerned chronic medication that the patient may have been using for a longer period of time, with an established clinical effect and with the patient accustomed to take them. Change of drug choice under these circumstances may disrupt the flow of treatment.

The use of a PMAS reduced the number of alerts compared to the current pharmacy computer system. A more sophisticated clinical decision support system could further reduce the number of irrelevant alerts by incorporating invalid dose or dose-interval algorithms that can weigh comorbidity and other patient-related risk factors that may affect the reliability of the eGFR,19 and by linking laboratory to pharmacy data. Currently, some of these principles are already incorporated in new versions of medication monitoring systems.

Strengths and limitations

This study revealed the benefit of therapeutic advice automatically generated by a PMAS based on renal function. The clinical relevance is substantial: prescribing of target drugs to older patients with diabetes or cardiovascular disease is a daily activity in primary care, and the risk of complications related to renal function is high.4 Primary care studies on compliance to dosing guidelines in patients with CKD are rare.6 Recently, Bhardwaja et al demonstrated in a large US study of 32 917 patients with an eGFR below 50 ml/min/1.73 m², that an alert system in the pharmacy can result in a reduction of medication errors from 49% to 33%.14

The data may not be generalisable to other settings because of the small number of participating practices, but the underlying problem of medication safety in relation to renal function and the intervention of a PMAS is of general interest.

The advice given was based on a single eGFR value obtained not more 1 year previously. This was for pragmatic reasons: the renal function of patients who are in a diabetes or hypertension control system should be monitored yearly. However, variability in serum creatinine measurements necessitates at least two creatinine measurements,20,21 and even more frequent monitoring of renal function is needed in patients who are not stable.22

Comparison with existing literature

GPs immediately agreed with half of the therapeutic advice. This is in accordance with the acceptance rate in a study in which clinical pharmacists gave therapeutic recommendations to GPs based on the medical records of 200 patients with diabetes or hypertension.23 In a hospital setting, the acceptance rate was the same: 55% of the pharmacist advice was accepted by the clinician.15

Besides the predefined reasons for disagreement, the GPs were not very explicit with their comments in the free-text box. Disagreement could be explained by a difference between the dosage guidelines and clinical practice. An example is the advice to start with low doses of RAS inhibitors to prevent adverse drug reactions, whereas current clinical practice guidelines do advise to prescribe RAS inhibitors in high doses in order to protect kidney function (with monitoring of renal function and serum potassium).24–26 Meanwhile, the advice in the Dutch dosage guidelines has been adjusted to clinical practice.

Implications for research and practice

To optimise drug prescribing in patients with decreased renal function, many steps need to be taken: systematic renal function monitoring in patients on target drugs, linking the laboratory to the pharmacy, assessment of the alerts by both pharmacist and GP, and communication with the patient on the proposed prescription change. When implementing a PMAS, all above-mentioned steps deserve attention.

A PMAS based on renal function resulted in therapeutic advice in one of every nine target drugs in older patients on blood-glucose-lowering or cardiovascular drugs. After weighing the clinical context, the GP agreed with half of the advice. Collaboration between the GP and pharmacist, using their clinical and pharmacological expertise respectively, can contribute to patient safety.