Our study shows that HA-CDI incidence has not increased during the COVID19 pandemic in two tertiary centers in Spain. We have also observed that the evolution of HA-CDI has been heterogeneous between the two participant centers. Thus, a significant decrease of HA-CDI incidence in every wave of the pandemic was observed in the HUV whereas incidence remained stable through the pandemic in HGUA. Interestingly, these figures were observed in the presence of a slightly higher antibiotic consumption during the COVID19 period in both hospitals, suggesting that, in the presence of strict infection control measures, hospital antibiotic consumption might have a lower impact than expected on HA-CDI.
At the beginning of the pandemic, some experts expressed concern that COVID19 could impact CDI rates, especially in the elderly [5]. However, initial studies have reported no impact or a trend for a lower incidence that anticipated by historical data. In the first study assessing this issue, Ponce-Alonso et al. reported a marked decrease in HO-HCFA incidence during the first wave in Madrid when compared to a historical control period in spite of a mild increase of antibiotics consumption [9]. A similar effect has been observed during the first wave in Dublin [10], Rome [11] and Mexico [12]. On the other hand, studies conducted during the first wave in New York [6] and Singapore [7] have found no effect of the pandemic on the incidence of CDI. Recently, Baker and coworkers did not found variations in HA-CDI incidence with respect to monthly COVID19 hospitalization rates during 2020 among 148 HCA Healthcare-affiliated hospitals in USA [8].
In the line with these previous reports, our work demonstrates that COVID19 has not promoted an increase in HA-CDI incidence during the first year of the pandemic. Of note, a marked decrease of HA-CDI during the first three waves of the pandemic was seen in the HUV in a similar manner to what has been seen in the first wave in other centers [9-12]. This behavior was even more pronounced when only HO-HCFA cases were considered. These results suggest that, at least in some centers, the implementation of infection control measures and the reorganization of health-care activities during the pandemic have led to collateral beneficial effects in terms of Clostridioides difficile nosocomial transmission.
Antimicrobial exposure is one of most relevant individual factors associated with CDI development [3,18,19]. Besides direct patient exposure to antibiotics, the influence of antibiotic use on the risk of CDI may also operate at aggregate levels (i.e. ward or hospital use of antibiotics). In Scotland, limiting hospital use of high-risk antibiotics was associated with a substantial decline in CDI, especially those caused by epidemic ribotypes [20]. Recent studies conducted in the US have shown that facilities achieving significant reductions in hospital antibiotic use experienced reductions of HO-HCFA CDI, which was more evident for reductions in quinolones and cephalosporins [21,22]. By contrast, a lack of effect on CDI has been seen after 3 years of starting the PIRASOA program, an antimicrobial stewardship program implemented in hospitals of the Public Health System of Andalusia, Spain [23]. While PIRASOA has led to a significant decrease in total antimicrobial consumption and a reduction of the incidence of infections due MDROs in the Andalusian public health-care system, the incidence of CDI increased during the same period [23]. Likewise, other studies have not found a correlation between monthly cephalosporin hospital use and the risk of CDI [24].
In our study, a stable or declining incidence of HA-CDI was seen in the presence of a mild increase of total antibiotic consumption during the pandemic, including significant increases in certain high-risk antibiotics such as cephalosporins. Similarly, some of the previous studies reporting a decrease of CDI incidence during the first wave of the pandemic also found a concomitant increase of antibiotic consumption [6,9]. These data support that the impact of hospital antibiotic use on CDI rates is not uniform and depends on several factors, such as the degree of implementation of infection control measures, the prevalence of high-risk ribotypes or the level of use of specific antibiotic classes such as quinolones. In this sense, a “saturation effect” has been described for quinolones, as CDI risk rises until reaching a certain threshold of use beyond which risk plateaus [21]. In the line of this, we speculate that the heterogeneous effect of COVID19 between the two participant hospitals observed herein might be explained by different baseline levels of hospital antibiotic pressure. Thus, the lower baseline antibiotic pressure might have facilitated the observed decrease in CDI incidence during COVID19 in the HUV, whereas this was not seen in the HGUA as a consequence of a higher baseline antibiotic consumption and a marked increase of cephalosporin use during COVID19. While previous studies have suggested that reductions in CDI incidence can be expected in institutions with high-level of antibiotic consumption [21,22], it is possible that mild changes in antibiotic use in settings where low to intermediate level of consumption has already been achieved might no further impact on CDI.
Our study has some limitations that must be acknowledged. First, as our study was observational and retrospective, definitive causal relationships cannot be inferred from our observations. Second, variations in CDI rates could be explained by under diagnosis due to the exceptional situation within hospitals during the COVID19 pandemic. We found this unlikely, as the standardized number of stool samples sent to the laboratory during the pandemic was comparable to the previous year. As a slight decrease in CDI diagnostic test during the COVID19 period was observed in the HUV, we cannot completely exclude that this could partially affect HA-CDI incidence rates. However, it would not justify the marked reductions of HO-HCFA rates that we have observed. Third, the direct effect of SARS-CoV2 infection on CDI risk in an individual basis could only be assessed in the HGUA, as in the HUV we could only assess monthly incidence of HA-CDI but not the incidence of CDI among hospitalized COVID19 patients. Finally, we compared antibiotic consumption between COVID19 and non COVID19 periods by means of DDD per 1000 OBD calculation but we could not evaluate if durations of antibiotic regimens differ between both periods, which would have been of interest, as the duration of antibiotic exposure has been linked to CDI risk [18]. On the other hand, previous studies have focused only in HO-HCFA CDI, which could underestimate the impact of COVID19 hospitalization in health-care associated CDI, as many of hospital acquired CDI cases become apparent after hospital discharge. In our study, we have included incident CDI cases up to 4 weeks after previous admission, which gives a better picture of the impact of the pandemic on HA-CDI. Besides, previous studies have evaluated the overall consumption of antibiotics during the pandemic without distinguishing between different classes. We have compared overall consumption during COVID19 but also by antibiotic families, as CDI risk is not uniform across them. Finally, changes in HA-CDI incidence is highly influenced by clinical suspicion and microbiological detection. To solve this, our analysis has also included standardized data of stool samples sent to microbiology during COVID19 and non-COVID19 periods. These are strengths of our study.