A cluster randomized trial evaluating electronic prescribing in an ambulatory care setting

Merrick F Zwarenstein, Katie N Dainty, Sherman Quan, Alex Kiss and Neill KJ Adhikari

Trials. 2007 Oct 3;8:28

Question

Desktop-based Computerized Physician Order Entry (CPOE) has been shown to decrease medication errors. Does electronic medication prescribing using a handheld Personal Digital Assistant (PDA) with decision support result in fewer medication errors?

Background and Rationale

Medication errors, adverse drug events (ADEs), and potential adverse drug events are prevalent, with the largest study suggesting rates of 6.5 ADEs and 5.5 potential ADEs per 100 hospital admissions. Errors resulting in preventable ADE’s occurred most often at the stages of ordering (56%) and administration (34%). There is less data on errors, but one study found 605 (19%) of 3216 medication doses were given in error.

There is reliable evidence than CPOE reduces ADEs, but the systems used have all been desktop-based. A systematic review of physician order entry systems found an 81% decrease in total medication errors, 55% decrease in non-intercepted serious medication errors, 13% decrease in cases of inappropriate dosing, and 24% decrease in inappropriate frequency of nephrotoxic drugs.

There is widespread use and acceptance among physicians of PDAs. This is the first study to rigorously evaluate PDA-based prescribing systems in which drug interaction, dosing alerts, and adverse event reporting are fully integrated.

Methods

A field-tested Canadian electronic prescribing software package was used for this study. The study was a 65 week, cluster randomized, parallel study with time as the element of randomization. The PDA software was available or not available for a week at a time, with the weeks randomized. Physicians used the PDA or paper prescription pad to write all prescriptions based on randomization. The principle outcome measure was total prescribing error, measured by pharmacy callbacks. Rates will be compared between intervention (PDA) and control weeks, with the physicians acting as their own controls. Types of errors recorded will include interactions, dosing and route of administration, and legibility.

Discussion

This study will be the first randomized controlled trial to evaluate the effectiveness of electronic prescribing in reducing prescription errors in an ambulatory setting using PDAs.

Commentary

This paper describes the protocol for a study looking at electronic prescribing using PDAs. CPOE has been shown to effectively reduce medication errors significantly at the point of prescribing, but most studies have been done using desktop-based systems. Most studies have focused on ADEs, while information about errors, which did not cause ADEs, has been far less available. This study looks specifically at PDAs and focuses on prescribing errors, which occur at the point of ordering. They use a novel method to randomize and control with randomization done by ‘week’ with the unit of analysis the prescription. The servers the PDAs communicate with were switched on and off by week, allowing the physicians to be their own controls.

No outcomes were given in this paper, which only describes the study design. There are some potential sources of error in the study. The first is if physicians continue to use paper prescription pads during the “on” time when they should be using PDAs, which will affect the intention to treat analysis. Another source of error is relying on pharmacy callbacks, since pharmacists could also miss interaction or dosing errors affecting the numerator.

Many physicians currently use PDAs and feel comfortable with them in their current workflow. It would potentially be a large benefit if they were able to incorporate computerized decision support into their current workflow easily and without disruption to their workflow.

The authors point out “It is imperative to evaluate these applications in pragmatic, real-world fashion so that we can determine if they are truly effective or more of a hindrance than help.”

Charles Laudenbach