Our intervention reduced the proportion of patients with prescriptions exceeding the recommended maximum daily dose, with a trend in reducing the proportion of patients with at least one prescription exceeding the recommended standard daily dose by more than 30%.
Most prescriptions exceeding the maximum or standard daily dose were ACE-inhibitors (ACEI) and angiotensin-receptor blockers (ARB).
A sensitivity analysis excluding these drugs failed to show a significant reduction of the recommended daily maximum dose at 6 months (data not shown). This suggests that the intervention effect was mostly related to ACEI and ARB.
Studies show that treatment with ACEI and ARB reduces the progression of CKD
], e.g. Benazepril was associated with a 52% reduction in the level of proteinuria and a 23% reduction in the rate of decline in renal function
]. While lower doses may be sufficient to treat hypertension
], dose escalation particularly in CKD may be beneficial due to a nephroprotective effect
]. Although ACEI and ARB have a wide therapeutic margin, high doses bear a risk of hyperkalemia, hypotension, and acute deterioration of renal function
]. Therefore, special attention is required when starting these drugs. In order to account for this complex situation, the DOSING programme provided information on ACEI/ARB dosage in a text format, including the specific recommendations for patients with CKD or hypertension. At the same time it allowed the calculation of an individual dose reduction for patients with hypertension.
Our results suggest that physicians using the DOSING programme reduced the dose below the maximum but kept it above the standard daily dose recommended for patients. A further sensitivity analysis showed that this was unrelated to the primary diagnosis of the patient (CKD or hypertension; data not shown). A possible explanation for this could be that the information provided by the programme was too complex and physicians were not certain which patients required a dose reduction.
In CKD patients, metformin may cause life-threatening lactic acidosis
], nitrofurantoin peripheral neuritis
], and a standard daily dose of 300 mg allopurinol may result in a hypersensitivity syndrome
]. No patient had been prescribed nitrofurantoin and fewer patients than in other studies
] were prescribed excessive metformin doses and, thus, differences between intervention and control group were not significant. However, the intervention group showed an 18% reduction in high dose allopurinol prescriptions.
These results suggest that DOSING may be effective in correcting dosing errors that are due to lack of knowledge or awareness (e.g. of serious adverse effects) and for which the evidence provided is unambiguous, but further studies with larger sample sizes are necessary to confirm these findings.
Most previously published evidence on effective interventions to improve correct dosage adjustment in CKD patients comes from inpatient settings
], even though dosage-related medication errors are equally relevant in outpatient primary care settings
]. Clinical decision support systems (CDSS) have been successful in reducing medication errors and adverse drug events in inpatient settings
]. In primary care, CDSS has also shown benefits, but variability among the types and methods of implementation and inconsistent use was noted
]. Rapid and easy access to the requested information
] and patient-specific advice
] are important prerequisites for the use of CDSS in busy practices. The strength of the DOSING programme is that it provides patient-specific prescribing advice at the point of care, which has the potential to improve attention and user response
In a concomitant qualitative study reported elsewhere
], we performed telephone interviews with family physicians in the intervention group in order to assess usability and the content of the programme. Participating physicians found the content of DOSING generally helpful and informative, and reported that its use improved their awareness of patients with impaired renal function.
As a CD version, DOSING proved feasible and effective in small, not fully computerized family practices; yet the intervention effect may be larger if DOSING is integrated into the EHR with a direct link to the medication and the patient information needed to calculate creatinine clearance.
Feasibility and effectiveness of the intervention may be improved if medication checks, implemented as a continuous background process, would create alerts when appropriate during the prescription of drugs. However, most available commercial EHR systems currently lack the functionalities to provide decision support
There are several possible limitations of the study.
First, selection bias, because the participating practices might have been especially motivated to improve prescribing in CKD patients, therefore introducing a study effect in control practices. In Hesse, physicians are required to participate in pharmacotherapy quality circles regularly. The infrequent prescription of potentially dangerous medications suggests that both groups were conscious of their prescribing behaviour. Therefore, the demonstrated effect may underestimate the real potential of the intervention in less well trained primary care practices.
Second, an intervention period of 6 months is relatively short. Further studies are needed to evaluate the long-term effects of the intervention.
Third, our study did not evaluate clinical outcomes for patients such as adverse effects of drugs requiring dose reduction, and we did not assess confounding factors such as the patients’ compliance to drug therapy. However, our study aimed to assess whether the use of a software tool for the estimation of renal function and the provision of information on corresponding dosage adjustment is feasible in busy small primary care practices, and whether it can effectively change prescription behaviour of family physicians for patients with impaired renal function. In order to measure these effects, we selected process rather than outcome parameters, such as the number of inappropriate prescriptions for these patients. Future studies should assess the effect of the intervention on patient-related clinical outcomes.
Finally, unlicensed use of metformin in patients with mild renal impairment is currently subject to debate
] making automated decision support difficult without considering the particularities of the individual case. In these situations our tool provided both, information on the labelled contraindication as well as guidance for dose adjustment to minimize the risk of accumulation.
Similarly, strict dose reductions of ACEI and ARB by an automated decision support system will only be appropriate if relevant patient characteristics such as the indication (hypertension vs. renal protection in CKD) are available and considered.