Clinician ordering practices for voriconazole therapeutic drug monitoring: experiences of a referral laboratory.

Monitoring of serum voriconazole concentrations has been proposed to optimize therapeutic effect and minimize toxicity. However, little is known about the clinical use of voriconazole therapeutic drug monitoring (TDM) by treating physicians. Four hundred seventy-eight episodes corresponding with 161 adult patients (mean three TDM episodes per patient; range 1-31, at a mean interval 43.6 days [range 1-266] between repetitions) performed at a state reference laboratory in Australia during a 30-month period were reviewed. Information about voriconazole dose was provided on only nine (1.9%) request forms. Timing of voriconazole TDM in relation to the previous dose was stated in 189 (39%) episodes and corresponded with peak measurements in 16; interval measurements (taken on average 3.3 [range 3-10.5] hours from the preceding dose) in 15; and trough measurements in 158 episodes. Of the 158 trough concentration measurements, only 66 (42%) were between 1 and 5.5 mg/L, the suggested therapeutic range. Similarly, only 33% (98 of 298) of all the random TDM episodes achieved voriconazole concentrations greater than 2.05 mg/L, previously associated with favorable outcomes. Compared with trough TDM, random episodes were significantly more likely to result in undetectable (less than 0.1 mg/L) concentrations (45 of 298 [15.1%] versus 12 of 158 [7.6%]; P = 0.021, odds ratio 2.16, 95% confidence interval: 1.11-4.22). Among patients with multiple TDM episodes, there was no correlation between the initial and final trough or between the initial and final random concentrations. Only 44% (eight of 18) of patients with multiple trough TDM had final concentration within 1 to 5.5 mg/L; and only 26% (15 of 58) of patients with multiple random TDM had final concentration greater than 2.05 mg/L. Adoption of consistent and clear guidelines on voriconazole TDM use and education of physicians ordering the test is required because the majority of testing performed was inappropriate and prone to suboptimal interpretation.