The use of antimicrobials is frequently associated with the appearance of resistant strains of microorganisms.1,2 Therefore, it is important to use antibiotics appropriately to avoid increases in antimicrobial resistance and improve patient outcomes.3–6 In order to promote rational use, antimicrobial consumption must to be measured and trends in consumption must also be monitored.7 Hospitals need reliable methods to determine their consumption of antibiotics and to compare it with consumption at other centers.8–12 Knowledge of the consumption of antibiotics in hospitals and in the ICU is one of the main pillars in the guidelines to enhance antimicrobial stewardship programs.13,14

The best method to quantify the use of antibiotics has yet to be defined. Antibiotic consumption is usually measured by calculating aggregated ratios on their use with the aim of enabling comparative evaluations. These ratios mainly include one of two reference units: the defined daily dose (DDD) or the days of therapy (DOT). Currently there is also other method to measure the consumption of antibiotics that is the prescribed daily dose (PDD) but that has the difficult to measure without a computerized system at the patient level.7

The DDD of a determinate antimicrobial is determined annually by the World Organization of Health (WHO); this unit represents the assumed average maintenance dose per day for a drug used for its main indication in adults.15 On the other hand, the DOT represents the days of antibiotic therapy administered to a patient, regardless of the number of doses administered or dosage strength.16

Using the DDD to measure the consumption of antimicrobials in hospitals is problematic. For many, but not all, parenteral drugs, the WHO sets the DDD at relatively low doses that often do not reflect dosages usually prescribed for hospitalized adults. Different studies found that the DDD format overestimated true prescription practice by 40–53%.17–19 Moreover, the DDD may be especially inappropriate in intensive care units (ICU). In critically ill patients, drastic derangements in physiological parameters can have large effects on the pharmacokinetics of antimicrobials, often requiring modify the doses administered.20 Therefore, using the DDD will lead to errors in the measurement of the consumption of antibiotics in ICUs.

The other recommended method to measure the consumption of antimicrobials, the DOT is easy to apply in settings where prescriptions are recorded at the individual level.17,21–23 Nevertheless, DOTs also have limitations, mainly derived from the calculations in patients undergoing therapy with more than one antibiotic.

Our hypothesis is that, given the differences between DDD and the doses of antimicrobials used in critical patients, the DDD method overestimates the consumption of antibiotics, especially in these patients, and for this reason we designed this study with the objective of measuring aggregate consumption of antibiotics in an ICU of a university hospital with DDD and DOT methods, to contrast these measures and confirm the differences between both methods.

During the study period, 2393 adult patients (mean age, 62±15 years; mean APACHE II score at admission, 14±2) were admitted to the ICU, representing 18,126 patients-days. The diagnosis of admission was acute respiratory failure (22.5%), neurological diseases (21.7%), sepsis/septic shock (18.6%), cardiovascular diseases (16.5%), polytrauma (12.8%), and others (7.9%).

Total median antimicrobial drug use measured by DDDs was 535.3 (IQR 319.8–845.5) vs. 344.0 (IQR 117.2–544.5) when measured by DOTs, p<0.001. Global antimicrobial use was 36.7% higher when measured by DDD than by DOT. Median consumption normalized to 100 patient-days was higher when calculated by DDD [2.98 (IQR 1.76–5.25)/100 patient-days] vs. [1.89 (IQR 0.64–3.0)/100 patient-days] when calculated by DOT, p<0.001. The correlation analysis of antimicrobial drug use per 100 patient-days found these two measures were significantly different (p<0.001) (Fig. 1).

Figure 1.

Correlation of total median antimicrobial drug use measured by defined daily dose (DDDs) per 100 patient-days and days of therapy (DOTs) per 100 patient-days (p<0.001).

(0.07MB).

In the 15 most commonly used antibacterial drugs, the difference between DDDs and DOTs was major in 9 (60%), moderate in 4 (26.6%), and minor in 2 (13.3%) (Table 1). The mean administered daily doses for 46.6% of these drugs were higher than the WHO's DDD recommendations. The greatest differences between DDDs and DOTs were found for cloxacillin, amikacin, cefepime, and daptomycin. The smallest differences between DDDs and DOTs were found for antibacterial drugs usually administered once or twice per day that have limited ranges of daily dosage, such as azithromycin and vancomycin.

In the six antifungal drugs analyzed, the difference between DDDs and DOTs was major in 3 (50%) (Table 2), and the mean administered daily doses were 34.4% higher than the WHO's DDD recommendations. The greatest differences between DDDs and DOTs were found for amphotericin B (91.6%) and fluconazole (57.5%).

To our knowledge, this is the first analysis of different methods to measure antimicrobial drug use in critically ill adult patients. As in studies done in other hospital populations,17–19 the DDD system of measuring antimicrobial use overestimated the actual consumption of most antimicrobial drugs.

The main advantages of the DDD method are that it enables antimicrobial use to be compared using standardized methods across a wide number of healthcare settings and countries, because it can be used with relative ease where administrative records are less developed and where it is more feasible to count packages and vials that have been purchased or dispensed than to measure the number of days of antimicrobial therapy. The most important limitation of DDD methods is discrepancies between administered daily doses and the WHO-assigned DDD.16

The main advantages of the DOT method are that it is not influenced by changes in the DDD or by discrepancies between the administered daily doses and DDD. The most important limitation of DOT methods results is the difficulty of measurement when computerized pharmacy records are not available.16

Our results agree with those of a study in 130 American hospitals,16 where the measurements of aggregate hospital antibiotic use by DDDs per 1000 patient-days and DOTs per 1000 patient-days were discordant for many frequently used antimicrobial drugs due to differences between the administered dose and the WHO-assigned DDD. The authors concluded that DDD methods are useful for benchmarking purposes, but for many antimicrobial drugs cannot be used to make inferences about the number of DOTs or relative use.

Following this article, an editorial25 suggested that the DOT method was a good alternative to the DDD method and that the WHO International Working Group for Drug Statistics Methodology needs to revise the DDDs for many antimicrobials. It is also obvious that, DOT methods would be useful for other purposes, such as measuring antimicrobial use in other populations (e.g., pediatric or critical patients) in which the doses used are very different from the currently recommended DDDs. In this sense, the latest IDSA/SHEA guidelines for implementing an antibiotic stewardship program already suggest using the DOT method for monitoring antibiotic in preference to the DDD method.26

Our study in critical care patients found that the global antimicrobial use was 36.7% higher when measured by DDDs than when measured by DOTs, and the mean administered daily doses were greater than the DDDs in more than 40% of the antibiotics and more than 30% of the antifungals.

Although the DOT method is becoming the standard measure of antimicrobial use in the USA, there are some limitations regarding its use as a representation of drug exposure. For example, for an antimicrobial administered every other day as adjusted for renal function, the DOTs are not counted for the days between the days when the drug is administered. Thus, the DOT can underrepresent true drug exposure. Kubin et al.27 assessed the impact of accounting for drug exposure on days in- between administered renal-adjusted doses for five antibacterial drugs (ceftriaxone, piperacillin/tazobactam, vancomycin, tobramycin, and levofloxacin) for all patients, including ICU and non-ICU patients, comparing three different DOT methods: (1) the standard DOT method, counting only the actual days when antibiotics were administered; (2) counting DOT in between administered doses if the antibacterial was stopped and restored within 2 days; and (3) a custom-exposure method they developed to take into account variability in the exposure time between administered doses, depending on the specific drug-dosing guideline. When all patients were considered, the results obtained with the three approaches were similar for ceftriaxone, piperacillin/tazobactam, and levofloxacin; however, for vancomycin and tobramycin, the approaches that took in-between dosing days into account yielded higher DOTs than the standard method. When only ICU patients were considered, the differences were larger. Nevertheless, the differences between the three methods were not statistically significant.

In our study, most of the antimicrobial drugs were administered at a dosage that exceeded the WHO-assigned DDD. Our data concord with those reported by Polk et al.,16 who found a downward deviation of the correlation curve for the antibiotics that had higher DDD than DOT measures, as in our study; for example, they found that the mean dose of cefepime used was 50% greater that WHO-assigned DDD, similar to our study, where the cefepime dose measured by DOT was 53% higher than the WHO-assigned DDD.

The main limitations of our study are its retrospective design and the fact that it was done at a single center. Thus, our results should not be extrapolated to other populations admitted to hospitals with different characteristics. Nevertheless, although recommended daily doses may differ among hospitals, our local practice guidelines are similar to those applied in most ICUs and in many other areas of acute care hospitals. It should also be considered as a limitation of our study, the fact that we have only analyzed the global consumption of antimicrobials during the study period, without considering the variations in the doses due to different organic dysfunctions or the use of renal replacement techniques.

We conclude that, for most antibacterial and antifungal drugs used in critically ill patients, the measurement of aggregate antimicrobial use by DDDs per 100 patient-days and DOTs per 100 patient-days is discordant because the administered dose is dissimilar from the WHO-assigned DDD. Although not all ICUs have the pharmacy unit-dose records necessary for DOTs, DOT methods should be recommended to ensure more realistic measurement of antibacterial drug consumption and avoid the overestimation that occurs with the DDD method in adult critically ill patients.

Jordi Vallès was responsible for the design of the study, review of patient data and preparation of the article.

Samuel Fernández and Edgar Cortés contributed to the acquisition of patient data included in the study.

Anisi Morón was responsible for the pharmacy data of the patients included in the study.

Esther Fondevilla and Emili Diaz were involved in review and elaboration of the document.

Joan Carles Oliva contributed to the review and statistical analysis of the study.

This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.

The authors declare no conflict of interests.