Construction and implementation: Using both methods, medians, and quartiles, as well as geometric means and confidence intervals (10), we elaborated ceftriaxone EC, and general calculations are presented in Table 1.
Table 1
DOT* of ceftriaxone (2017–2020) and calculated quartiles
Month | Days of therapy (DOT) Ceftriaxone | Quartile method** | Geometric mean (GM) and confidence intervals (CI) method ^ |
2017 | 2018 | 2019 | 2020 | Q1 | Q2 | Q3 | Superior CI | GM | Inferior CI |
January | 48 | 30 | 20 | 18 | 18 | 20 | 30 | 3,75 | 3,09 | 2,43 |
February | 18 | 19 | 20 | 16 | 16 | 19 | 20 | 3,20 | 2,89 | 2,58 |
March | 39 | 19 | 31 | 18 | 18 | 19 | 31 | 3,83 | 3,08 | 2,32 |
April | 28 | 21 | 17 | 31 | 17 | 21 | 31 | 3,89 | 3,10 | 2,32 |
May | 37 | 16 | 27 | 16 | 16 | 16 | 27 | 3,70 | 2,95 | 2,21 |
June | 28 | 14 | 28 | 8 | 8 | 14 | 28 | 4,23 | 2,69 | 1,15 |
July | 23 | 15 | 21 | 14 | 14 | 15 | 21 | 3,33 | 2,78 | 2,23 |
August | 24 | 27 | 18 | 13 | 13 | 18 | 27 | 3,82 | 2,90 | 1,98 |
September | 40 | 19 | 21 | 9 | 9 | 19 | 21 | 3,85 | 2,73 | 1,61 |
October | 26 | 15 | 16 | 3 | 3 | 15 | 16 | 4,64 | 2,18 | 0,00 |
November | 29 | 19 | 10 | 10 | 10 | 10 | 19 | 3,42 | 2,54 | 1,66 |
December | 37 | 25 | 14 | 18 | 14 | 18 | 25 | 3,64 | 2,92 | 2,21 |
* DOT x 1000 patient-days ** Q1: percentile 25, Q2: percentile 50, Q3: percentile 75 ^ DOT (days of therapy) used to construct endemic channel and depicted in GM and CI methods should be in natural logarithm |
As mentioned, different zones of antimicrobial use were defined according to historic DOTs. Figure 2 describes the variation in the historical use of ceftriaxone, with a slight increment in the last 3 months of the year and again from February to March, in concordance with the most intense respiratory infections and rainy season in Bogotá, which is associated with increased use of ceftriaxone because of more frequent cases of bacterial pneumonia and complicated pneumonia. In this historical use, the actual use was compared.
According to each zone, ASP analysis EC as follows: a value located in the success zone indicates that the use of antibiotics (DOT) for that period has a lower frequency than expected, one located in the security zone indicates that the antibiotic consumption shows a stable behavior, and a value located in the alert zone indicates that the number has a higher frequency than expected, and it is necessary to study the situation to determine the origin of the variation. A value located in the epidemic zone indicates that the DOTs reported for that period present a situation that requires immediate control actions.
Medians and quartiles Vs. geometric means and confidence interval method: When comparing both methods for the elaboration of endemic channels, we found that although the method of geometric means and confidence intervals requires much more data processing, making it more cumbersome, it is more sensitive than the median and quartiles method to detect changes in DOT with respect to the previous history of antimicrobial consumption, as can be observed in ampicillin sulbactam (AS) when a change in the guidelines of prophylaxis was made (Fig. 3).
Since January 2022, ampicillin-sulbactam has been the first indication as part of the primary surgical prophylaxis regimen for appendicitis. This change represents an increase in AS consumption. Using the median and quartile method (3 B), this change appears only as a slight increase, but in the same zone (safety zone), while with the geometric mean and CI method (3A), the same change was more easily seen (change reached the alert zone). The possibility of detecting small differences in antimicrobial consumption, such as ASP, makes this model an ideal method for us (Fig. 3). For this reason, since 2020, in our institution, we decided to implement EC with geometric measurement of rates and confidence intervals using the natural logarithm for each DOT per month per year, mean, standard deviation, and Student’s t, to obtain upper and lower confidence intervals, which, in addition to being more sensitive according to our findings, have also been described in the literature as a more robust statistical data analysis method (10).
One year after implementation: From 2021, every month, DOTs were analyzed in constructed endemic channels. In Fig. 4, we appreciate the behavior of three broad-spectrum antibiotics and a narrow-spectrum antibiotic (ampicillin-sulbactam) comparing ECs from all hospital services and the Pediatric Intensive Care Unit (PICU).
As shown in Fig. 4A-B, ceftriaxone presented an increase in consumption during the final trimester of 2021 and 2022, compared with previous years. While most of the time in 2021, its use was in the success zone (less than expected compared with previous years) in general hospital services, its use was more pronounced in the PICU, where it was in the alert zone most of the time.
In Fig. 4C-D Piperacillin–tazobactam (PT) showed an important increase in use with respect to historical consumption, especially in the PICU during the first 8 months of 2021, where its use was in the alert zone until August 2021. From September 2021, there is a decrease in PICU and general hospital use. During November and December 2021, its use was lower than that in the same period in previous years.
Vancomycin (Fig. 4E-D) has a relatively stable behavior; in general, it remains at the upper limit of the safety zone, but its use in the PICU has been greater than its historical use, especially during the first trimester of 2022. Ampicillin–sulbactam (Fig. 4G-H) has reported an increase in its use since January 2022, reaching the alert zone during this period. In the PICU, its use has been stable compared with historic consumption, remaining in the success zone during the first three months of the year in 2022, similar to what happened in the first trimester of 2021.