Antibiotics prescription and guidelines adherence in elderly: impact of the comorbidities

Antibiotics are among the most prescribed drugs in the world [1]. Since the 2000s, there has been an increase of more than 20% of antibiotics prescriptions in elderly patients [2]. Antibiotics have undeniable benefit effects, but have also side effects that could represent a serious threat to public health [3]. Good quality of antibiotics prescription, defined by an adapted use of an antibacterial agent during an infection (molecule, dose and duration) and by the adherence of prescription guidelines, is associated with less side-effect [4, 5]. Previous studies found that between 25 and 75% of antibiotics prescriptions did not meet the guidelines [6, 7]. To our knowledge, no studies have examined the relationship between the number of comorbidities and the adherence of antibiotics prescriptions guidelines in elderly patients [8], and considering the others confounding factors associated with antibiotic prescription [9‐13].

The morbidity burden is defined by Valderas et al. as the total burden of types of illness having an impact on an individual’s physiological reserve [14]. The presence of a high burden of comorbidity increases early mortality in bacterial infection, the number of hospitalizations and has a high public health impact [15]. A recent study in primary care found that a high burden of comorbidities is associated with an increased in antibiotic initiation [16]. However, this study did not specially examine whether treatments were initiated for bacterial infection and whether they were prescribed according to guidelines.

The objective of this study was to examine the relationship between a high comorbidity index score and the adherence to antibiotics prescriptions guidelines, in elderly patients.

In this study, the following variables were negatively associated with antibiotic prescription according to the guidelines: Charlson’s comorbidity index score, prescription of 2 antibiotics, patients aged 85 years and older.

The prevalence of antibiotics prescriptions in adhrence with guidelines (46%) is consistent with other studies but is higher than another study carried out in a French nursing home [6, 26]. Indeed, Boivin et al [6] have found a prevalence of 17%. We make the assumption that this lower prevalence could be explained by three main arguments. First, quinolones prescriptions were considered as inappropriate by Boivin et al, and appropriate in this study, according to the guidelines [18, 19]. Second, we did not consider if antibiotics prescriptions were reevaluated and switched at 48-72 h because of a high number of missing data. Third we assume that the MacGeer criteria could improve the accuracy of the diagnosis and could increase the number of appropriate prescriptions.

In this study, a high Charlson’s comorbidity index score was a risk factor of antibiotics prescriptions not in adherence with guidelines. To our knowledge, this is the first description of this association. Previous studies have examined the association between antibiotics initiation and presence of comorbidities [27, 28], but without considering if the prescription was in adherence with guidelines. We assume that, together, two antibiotics bad practices could explain this result: inappropriate antibiotic initiation and absence of antibiotics prescription reevaluation. First, a physician could prescribe antibiotics earlier, and inappropriately in order to anticipate decompensation of comorbidities related to sepsis [28]. Second, Boivin et al. have found that a higher burden of comorbidities is associated with an absence of a systematic antibiotic prescription reevaluation [6]. Moreover, the study was conducted in a clinical ward where patients are treated by different senior doctors. Thus, prescribers were not always the one who reevaluated and stopped the prescription. We assume that this organization could also explain the difficulty to reevaluate and/or to stop the antibiotic.

In this study, 85 years of age and more has been negatively associated with antibiotic prescription in accordance with guidelines. To our knowledge, this is the first description of this association, that was previously described in another study that analyzed factors associated with inappropriate prescriptions in the elderly [11]. We assume that this result could be explained by two main arguments. First, age could have an impact on an earlier initiation of the antibiotics prescription. Indeed, Brooke et al. [29], in a study about factors influencing antibiotics prescription, proposed that physician may have fear of a rapid clinical aggravation in elderly patients, even in absence of clear evidence of bacterial infection. Second, as it was proposed for the Charlson comorbidity index score, the physician could have fear to stop antibiotics prescriptions in elderly patients even in absence of a high burden of comorbidities.

In this study, prescription of two antibiotics has been negatively associated with antibiotics prescription according to the guidelines. Kawanami et al. have already found this association in a recent study [13]. Our study confirms this result in a more specific population, vulnerable to side effects associated with antibiotics prescription [5]. This result can be explained by two arguments. First, physicians could initiate broad spectrum antibiotics prescriptions considering that the diagnosis of site of infection is not certain because of atypical clinical presentations [4]. Then, for the same reason, deprescribing to a narrower spectrum antibiotic may not be done [30]. In our study, no patient who initially received ceftriaxone or imidazole had further deprescription to a narrower-spectrum antibiotic. Second, we assume that a switch for another antibiotic can lead to inappropriate prescriptions. In this study, only three prescriptions were in accordance to the guideline when the antibiotic was switched for a narrower spectrum molecule, with a longer treatment duration.

Although it was previously described [9, 13], we did not find an association between guidelines adherence and the site of infection or the administration of a parenteral antimicrobial treatment. We assume that two methodological biases could explain these results. First our study may have been underpowered to detect these associations. Second, the population characteristics could be different. Indeed, Kawanami et al. included hospitalized patient, without age limit and described more gastro-intestinal infections, peritoneum infections or bone infections, associated with an inappropriate antibiotic prescription [9, 13]. Moreover, in our study, no patient has been treated with cephalosporin 1st generation or aminoglycoside, which are antibiotics associated with inappropriate prescription [9, 13].

This study has limitations. First, this study had an observational design and we report the result of a retrospective analysis. Therefore, hidden bias could exist and may have underestimated the relationship between some factors and guidelines adherence. Second, because of the retrospective design, it was difficult to collect social or economic or morbimortality parameters that could be associated with guidelines adherence. Indeed, this study was not designed to analyze whether guidelines adherence is cost effective or reduced morbidity and mortality. Moreover, guidelines sometimes ignore vulnerable population such as frail elderly patients. Third, it is a French monocentric and study and the results may not be fully applicable to others settings with others guidelines. However, we assume that this population is representative of a population particularly vulnerable to infectious disease and with a higher risk of side-effects associated with antibiotics prescriptions [5]. Indeed, patients were predominantly female, with a median age of 87 years old, required assistance for activities of daily living, and the mean Charlson’s index score was 2.48. Fourth, the Mac Geer criteria may not be fully adapted for the accuracy of the diagnosis of infection. We used the Mac Geer criteria because they take into account atypical semiological presentation in elderly patients [17], and we assume that they can be useful to asses antibiotics prescription in this population.

This study highlights the importance of co-morbidities in the prescription of antibiotics and more particularly on compliance with the guidelines. Further research is needed to identify other parameters involved in adherence with guidelines. Considering the impact on public health, these factors should be integrated into research protocols to improve and optimize antibiotic prescriptions.