Catheter-Associated Urinary Tract Infections, Bacteremia, and Infection Control Interventions in a Hospital: A Six-Year Time-Series Study

Catheter-associated urinary tract infections (CAUTIs) are among the most common healthcare-associated infections. Urine catheters are often reservoirs of multidrug-resistant (MDR) bacteria and sources of pathogens transmission to other patients. The current study was conducted to investigate the correlation between CAUTIs, MDR bacteremia, and infection control interventions, in a tertiary-care hospital in Athens, from 2013 to 2018. The following data were analyzed per month: 1. CAUTI incidence; 2. consumption of hand hygiene disinfectants; 3. incidence of isolation of MDR carrier patients, and 4.incidence of bacteremia/1000 patient-days [total resistant a.Gram-negative: carbapenem-resistant Pseudomonas aeruginosa, Acinetobacter baumannii, and Klebsiella pneumoniae; b.Gram-positive: vancomycin-resistant Enterococci and methicillin-resistant Staphylococcus aureus]. The use of scrub disinfectant solutions was associated with decreased CAUTI rate in Total Hospital Clinics (OR: 0.97, 95% CI: 0.96–0.98, p-value: <0.001) and in Adults ICU (OR: 0.79, 95% CI: 0.65–0.96, p-value:0.018) while no correlation was found with isolation rate of MDR-carrier pathogens. Interestingly, an increase in total bacteremia (OR: 0.81, 95% CI: 0.75–0.87, p-value:<0.001) or carbapenem-resistant bacteremia correlated with decreased incidence of CAUTIs (OR: 0.96, 95% CI: 0.94–0.99, p-value: 0.008). Hand hygiene measures had a robust and constant effect on infection control, reducing the incidence of CAUTIs.


Introduction
Urinary tract infection (UTI) is one of the most common healthcare-associated infections (HAIs), and approximately two-thirds of these infections are attributed to an indwelling urethral catheter [1]. The first pan-European point prevalence survey of HAIs

Setting
The study was conducted prospectively from January 2013 to December 2018 in a 300-bed private tertiary-care hospital in Athens. The hospital consists of: 1. one Adults Clinic with Internal Medicine, Oncology, Hematology, Surgery Departments and one ICU; 2. one Obstetrics and Gynecology Clinic with one neonatal ICU, and 3. one Pediatrics Clinic with one pediatric ICU.

Infection Control Measures
As we already described [17,18], the following infection control measures were applied: 1. active surveillance of carbapenem-resistant A. baumannii, P. aeruginosa, and K. pneumoniae, and methicillin-resistant Staphylococcus aureus (MRSA) and vancomycin-resistant Enterococci (VRE); 2. implementation of a CAUTI bundle, which consisted of aseptic insertion and maintenance techniques, catheter change guidelines, and discontinuation indications; 3. promotion of hand hygiene before and after providing healthcare to patients; 4. carriage screening (pharyngeal, axillary-rectal, nasal) cultures followed by isolation of MDR carrier patients; and 5. audit of CAUTI bundle and ASP on a monthly basis.

Data Collection
Data were collected prospectively on a monthly basis: 1. number of CAUTIs was collected from Clinical Infectious Diseases Department; 2. catheterization was detected and recorded manually by clinical visits from Nurse Department; 3. hand disinfectant solutions use was collected from Pharmacy Department; 4. number of bacteremia was collected from Microbiology Department. The medical/nurse hospital quality procedures and laboratory diagnostic procedures were supervised by Quality Assurance Department and did not change throughout the six year-study period.

Detection of Bacteremia
Bacteremia was detected through Gram stains and blood cultures. The automated VITEK 2 system (Biomerieux) was used for isolation, identification, and antibiotic susceptibility testing. The CLSI breakpoints were applied. The susceptibility of bacteria was determined by Kirby-Bauer test, MIC semi-automated testing, and/or E-test.

Definitions
CAUTI in a patient with an indwelling urethral catheter is defined as the onset of signs and/or symptoms compatible with UTI and no other source of infection along with ≥10 3 colony-forming units (cfu)/mL of ≥1 bacterial species in a single urine specimen or in a midstream voided urine specimen from a patient whose catheter was removed the past 48 h [19]. Signs and symptoms compatible with CAUTI include new onset or worsening of fever, rigors, altered mental status or lethargy with no other identified cause; costovertebral angle tenderness; acute hematuria; pelvic discomfort; and in those whose catheters have been removed, dysuria, urgent or frequent urination, or suprapubic pain or tenderness [19]. Isolation rate of MDR-carrier patients was expressed as a percentage of isolated patients per 100 admissions. Bacteremia was defined as a laboratory-confirmed bloodstream infection, either primary (not related to an infection at another body site) or secondary (thought to be seeded from a site-specific infection at another body site) [20,21]. New episode of bacteremia within a month period was defined as a new episode of bacteremia due to a different pathogen strain or due to the same pathogen strain but with different phenotype of resistance. The incidence of total bacteremia is the sum of total Gram-positive and Gram-negative bacteremia. The incidence of total carbapenem-resistant Gram-negative bacteremia refers to carbapenem-resistant A. baumannii, P. aeruginosa and K. pneumoniae bacteremia. The incidence of total resistant Gram-positive bacteremia refers to the incidence of MRSA and VRE bacteremia [21]. Hand hygiene concerns scrub disinfectant solutions with chlorhexidine, alcohol 70% disinfectant solutions with chlorhexidine, and/or simple soap [18].

Statistical Analysis
As already described [17,18], an analysis of time trends in the intervention and outcome variables during the study period was initially performed. The variable under investigation was the dependent variable. Time since the beginning of the study (in months) was the independent variable in the regression models and was entered through appropriate restricted cubic splines. In order to capture potential seasonality effects, Fourier series terms of time (1st and 2nd order) were also used in the models. Standard errors (SE) and corresponding 95% confidence intervals (CI) were estimated using the robust (sandwich) variance estimator to adjust for potential violations of models' assumptions. Estimated values for start and end of the study period and corresponding 95% CIs were estimated through a simplification of the models. Spline time terms were replaced by a single linear time trend or two piecewise linear terms to capture average long-term trend. A linear regression model was applied for consumption of hand disinfectants. When the outcome of interest was CAUTI or bacteremia rates, Poisson regression models were applied with number of cases as dependent variable and the appropriate number of catheter-days or patient-days, respectively, used as an offset after logarithmic transformation [17]. When the percentage over total number of hospitalizations was the outcome of interest (e.g., isolations), binomial regression models were applied with the number of cases as the dependent variable and the appropriate number of hospitalizations as the binomial denominator [17]. Associations between outcomes and interventions were studied by introducing appropriate independent variables into the models [22]. The effects of the independent variables were initially tested separately for current ("month 0") and lagged values (months −1, −2 and −3). In case of statistical significance (p-value < 0.05) or indicative significance (0.05 < p-value < 0.10) for more than one case (e.g., in month 0 and in month −1) and association direction was the same (e.g., positive for both), average value was used as independent variable. In cases where the direction of the association was different (e.g., positive for "month 0" and negative for "month −1"), the results of the respective models are presented separately [22]. All p-values reported throughout the manuscript have not been adjusted for multiple testing. Analyses were conducted using Stata version 14.2 (Stata Corporation, College Station, TX, USA).

Results
A total of 95,228 admissions occurred during the entire study period. Overall, 12.84% of hospitalized patients underwent catheterization. There were 379 CAUTI episodes among 12,228 catheterized patients; therefore, the CAUTI rate during the six-year study period was 5.28 episodes of CAUTI/1000 catheter days. The monthly CAUTI incidence in all Hospital Clinics and in Adults ICU are depicted in Figure 1 ues (months −1, −2 and −3). In case of statistical significance (p-value <0.05) or indicat significance (0.05 < p-value <0.10) for more than one case (e.g., in month 0 and in mon −1) and association direction was the same (e.g., positive for both), average value was us as independent variable. In cases where the direction of the association was different (e positive for "month 0" and negative for "month −1"), the results of the respective mod are presented separately [22]. All p-values reported throughout the manuscript have been adjusted for multiple testing. Analyses were conducted using Stata version 1 (Stata Corporation, College Station, TX, USA).

Results
A total of 95,228 admissions occurred during the entire study period. Overall, 12.8 of hospitalized patients underwent catheterization. There were 379 CAUTI episod among 12,228 catheterized patients; therefore, the CAUTI rate during the six-year stu period was 5.28 episodes of CAUTI/1000 catheter days. The monthly CAUTI incidence all Hospital Clinics and in Adults ICU are depicted in Figure 1 with robust standard errors, seasonality terms and linear or piecewise linear long term trend: log = β0 +β1t-+β2t + +β3 × sin(2πt/12) + β4 × cos(2πt/12) + β5 × sin(4πt/12) + β6 × cos(4πt/12) + log(ventila days) with N being the number of cases and t being time since study start in months (t-and t+ pie wise linear time terms). Trends shown with grey area derived similarly but spline terms of ti were used for long term trend instead of piecewise linear terms.
The time trends of CAUTI rate during the six-year study period in the entire hosp and divisions are shown in Table 1. The incidence of CAUTI decreased significantly in Hospital Clinics and Departments and also in the Adults ICU (p-value: <0.001 for all co parisons). models with robust standard errors, seasonality terms and linear or piecewise linear long term trend: log(N) = β 0 +β 1 t − +β 2 t + +β 3 × sin(2πt/12) + β 4 × cos(2πt/12) + β 5 × sin(4πt/12) + β 6 × cos(4πt/12) + log(ventilator-days) with N being the number of cases and t being time since study start in months (t − and t + piecewise linear time terms). Trends shown with grey area derived similarly but spline terms of time were used for long term trend instead of piecewise linear terms.
The time trends of CAUTI rate during the six-year study period in the entire hospital and divisions are shown in Table 1. The incidence of CAUTI decreased significantly in all Hospital Clinics and Departments and also in the Adults ICU (p-value: <0.001 for all comparisons). Table 2 shows time trends of incidence of isolations per 100 admissions. A significant increase in the rate of isolation of patients with MDR pathogens was observed in Total Hospital Departments and in Adults ICU (p-value ≤ 0.001), while in Adults Clinic the increase was significant up to April 2015 only (p-value < 0.001) and in Adults Departments up to February 2015 only (p-value < 0.001). CAUTI: catheter-associated urinary tract infection; ICU: intensive care unit; EVSP: estimated value start period; EVEP: estimated value end period; CI: confidence interval. All estimates derived from Poisson regression models with robust standard errors, seasonality terms and linear or piecewise linear long-term trend: log(N) = β 0 +β 1 t − +β 2 t + +β 3 × sin(2πt/12) + β 4 × cos(2πt/12) + β 5 × sin(4πt/12) + β 6 × cos(4πt/12) + log(catheter-days) with N being the number of cases and t being time since study start in months (t − and t + piecewise linear time terms; when piecewise linear long term trend was not required a single time term was used). % Relative changes/year derived as [exp(12 × β 1 , 2 )−1] × 100%. ICU: intensive care unit; EVSP: estimated value start period; EVEP: estimated value end period; CI: confidence interval1; All estimates derived from binomial logistic regression models with robust standard errors, seasonality terms and piecewise linear long term trend: logit(π) = β 0 +β 1 t − +β 2 t + +β 3 × sin(2πt/12) + β 4 × cos(2πt/12) + β 5 × sin(4πt/12) + β 6 × cos(4πt/12) with π being theprobability of isolation and t being time since study start in months (t − and t + piecewise linear time terms). % Relative changes/year derived as [exp(12 × β 1 , 2 )−1] × 100%.          or averaged values over more than one month. Incidence Rate Ratio (IRR) refers to increases incidence of bacteremia, denoted in column labeled "per (n) unit". All estimates derived from Poisson regression models with robust standard errors, seasonality effects and spline terms of time: log(N) = β 0 + β 1 V + β 2 S 1 (t) +β 3 S 2 (t) +β 4 S 3 (t) + β 5 × sin(2πt/12) + β 6 × cos(2πt/12) + β 7 × sin(4πt/12) + β 8 × cos(4πt/12) +log(catheter-days)with N being the number of cases, t being time since study start in months, S(t) being spline terms of t and V referring to the current month covariate (month 0) value, lagged values (months −1, −2, −3) or averaged values over more than one month. Incidence Rate Ratios derived as exp(n × β 1 ) with n given in column labeled "per (n)". Table 4 shows time trends in the consumption of hand disinfectant solutions per category of disinfectant. There was a statistically significant increase in the consumption of alcohol disinfectant solutions and all hand disinfectant solutions in all Departments and Clinics (p-value ≤ 0.001). In Adults Clinic Departments separately the increase was observed only in alcohol disinfectant solutions (p-value ≤ 0.001). In the ICU it is noticeable that there was a statistically significant increase in scrub and all hand disinfectant solutions (p-value = 0.001). Combining the results with Figure 1b   or averaged values over more than one month. Incidence Rate Ratios (IRR) refers to increases denoted in column labeled "per (n) units". All estimates derived from Poisson regression models with robust standard errors, seasonality effects and spline terms of time: log(N) = β 0 + β 1 V + β 2 S 1 (t) +β 3 S 2 (t) +β 4 S 3 (t) + β 5 × sin(2πt/12) + β 6 × cos(2πt/12) + β 7 × sin(4πt/12) + β 8 × cos(4πt/12) +log(catheter-days) with N being the number of cases, t being time since study start in months, S(t) being spline terms of t and V referring to the current month covariate (month 0) value, lagged values (months −1, −2, −3) or averaged values over more than one month. Incidence Rate Ratios derived as exp(n × β 1 ) with n given in column labeled "per (n)". Table 4 shows the time trends in the incidence of different bacteremia/1000 patientdays. The incidence of total bacteremia increased significantly in the entire hospital and divisions (p-value ≤ 0.001), which is attributed to the increase in the number of blood cultures and admissions. However, there was no statistically significant difference in the incidence of total bacteremia from resistant pathogens. In total Hospital Clinics and Departments, the trend in the incidence of total carbapenem-resistant Gram-negative pathogens decreased not significantly, while in Adults ICU increased not significantly. The analysis per pathogen showed a significant decrease only for carbapenem-resistant P. aeruginosa in total Hospital Clinics (p-value = 0.027) and Departments (p-value = 0.042), and in Adults Clinic (p-value = 0.031) and Departments (p-value = 0.051). For carbapenem-resistant K. pneumoniae and A. baumannii the incidence did not change significantly. Interestingly the latter had zero incidence in the hospital departments. The incidence of resistant Gram-positive pathogens remained very low and stable throughout the entireperiod; for this reason, we did not apply a linear model. Finally, in the Adults ICU the analysis per pathogen did not show any statistical change for any of the carbapenem resistant Gram-negative bacteria.
The correlation of CAUTI with the incidence of different bacteremia is shown in Table 5. Every increase in the incidence of total bacteremia the current and the previous month correlated with a decreased CAUTI rate, in total Hospital Clinics and Departments and in Adults Clinics and Departments (p-value ≤ 0.001). There was a negative correlation for total carbapenem-resistant Gram-negative bacteremia in total Hospital Clinics (p-value = 0.008) and Adults Clinics (p-value = 0.042) three months before, but in all Hospital Departments (p-value = 0.048) and Adults Departments (p-value = 0.060), we noticed a positive correlation for current month and negative three months before (p-value = 0.009 and 0.027, respectively). For each category of carbapenem-resistant Gram-negative pathogen, the negative correlation was constant in all Hospital Clinics/Departments and in Adults Clinics/Departments, and statistically significant especially for carbapenem-resistant K. pneumoniae bacteremia (p-value < 0.001). Every increase in their incidence two and three months earlier resulted in a decreased CAUTI rate. Every increase in the incidence of carbapenem-resistant P. aeruginosa bacteremia two and three months earlier significantly correlated with increased CAUTI rate (p-value < 0.001). This phenomenon was not repeated for the other Gram-negative resistant pathogens. However, there was a significant correlation between the incidence of carbapenem-resistant K. pneumoniae and a decrease in CAUTI incidence three months earlier (p-value = 0.018). Table 6 shows the correlation between CAUTI and infection control measures. Every increase in the consumption of either scrub or all disinfectant solutions the previous months significantly correlated with decreased CAUTI rate in all Hospital Clinics (p-value ≤ 0.001 and 0.004), Adults Clinic (p-value ≤ 0.001 and 0.006), and Adults Departments (p-value ≤ 0.001 and 0.005). In the Adults ICU, every increase in the consumption of alcohol and all disinfectant solutions current month correlated significantly with increased CAUTI (p-value = 0.012 and 0.016), while every increase the previous month in the consumption of scrub disinfectant solutions correlated significantly with decreased CAUTI rate (p-value = 0.018). Finally, the intervention of isolation of patients did not show a direct correlation with the CAUTI rate.

Discussion
In this six-year study we studied the relationship among infection control measures and outcomes with CAUTI in a tertiary-care hospital located in Athens. In our institution from 2013 to 2018 a CAUTI-bundle was implemented to promote the rational use of indwelling urinary catheters, resulting overall in a catheterization percentage of 12.84%, which stands within international references [23,24]. According to the 2011-2012 pan-European point prevalence survey of HAIs, the mean catheterization rate of hospitalized patients in acute care hospitals in European countries was 17.2%, while in Greece it was 30% [2]. In our study, the 6-year CAUTI rate was 5.27 infections per 1000 catheter days, which is within the range of pooled mean CAUTI rates (3.1-7.5 infections per 1000 catheterdays) found in acute care hospitals during 2015-2017, as reported by National Healthcare Safety Network [9].
During the last 15 years many guidelines for the prevention of CAUTI have been published [19,[25][26][27], and many programs with evidence-based educational and interventional bundles to reduce the incidence of CAUTI have been evaluated [28][29][30][31][32][33][34][35][36]. In our study a CAUTI-bundle was implemented along with other infection control interventions, resulting in a significant reduction in the incidence of CAUTI in all hospital Clinics and Departments, and in Adults ICU.
In our hospital, the most significant infection control interventions were the increased isolation of MDR carrier patients and the increased consumption of hand disinfectant solutions indicating adherence to hand hygiene. Furthermore, the most significant outcome was the decrease in the CAUTI rate. The decrease in the incidence of carbapenem-resistant Gram-negative bacteremia, either in total or per studied pathogen, although it was indicative, reflects the effect of isolation of patients with MDR pathogens, while very low or zero incidence of resistant Gram-positive pathogens during the six-year study period, reflects the effect of implementation of hand hygiene [17,21]. Our findings, in different clinics and departments of the hospital, depict the continuous need of infection control interventions with tailored frequency, evaluation, and implementation.
Until recently, it was more feasible to design a study regarding bundled horizontal infection prevention strategies for the prevention of HAI [28][29][30][31][32][33][34]. In our study, for the first time in the literature, we have results regarding the association of CAUTI with different bacteremia and with consumption of hand disinfectant solutions for the entire hospital as well as its Clinics and Departments separately. For CAUTI and total bacteremia, the correlation was always negative and significant, indicating no cross-infection between blood and UTIs. In case of total carbapenem-resistant Gram-negative bacteremia, the correlation was positive only for the current month, in Total Hospital Departments, showing the severity of such infections. For each category of carbapenem-resistant Gram-negative pathogens, the correlation with CAUTI was negative, in the entire hospital and its divisions, two and three months earlier, showing a preserve time effect of the implemented interventions. For Adults ICU the results diverged from one carbapenem-resistant bacterium to another. Especially for carbapenem-resistant P. aeruginosa bacteremia the correlation with increased CAUTI rate, two and three months earlier, showed a disability of the implemented interventions to have prolonged effect for this type of pathogen, and the need for further tailored infection control interventions regarding this hydrophilic Gram-negative bacterium. Interestingly, the incidence of carbapenem-resistant A. baumannii and K. pneumoniae correlated with the decreased CAUTI rate, indicating different infection control behavior of these pathogens.
From the results of CAUTI and hand disinfectant solutions, the correlation was negative in the entire hospital and its divisions giving a 2 to 3 month-lasting effect of implementation of hand hygiene. While no actual changes in catheter indwelling rates and indwelling duration was noted (data not shown), CAUTI-bundle education regarding hand sanitizer consumption resulted in decreased CAUTI rate in the hospital setting. Especially in the ICU both trough values of CAUTI rate coincide with the increase consumption of scrub and all hand disinfectant solutions implying the importance of hand washing along with disinfection in controlling nosocomial infections such as CAUTI.
Moreover, for the first time in the literature, the correlation of CAUTI with the isolation of MDR-carrier patients was investigated and non-significance was found. Worthy to mention though that both decrease in CAUTI and increase in isolation of patient directly correlated with decrease in incidence of MDR bacteremia.
Our study has several strengths. For the first a time series analysis was used to study the potential association among CAUTI rate, infection control measures, and MDR bacteremia. A clear strength is the prospective study design and the prolonged study period. The analysis of findings per clinic and department gave us the opportunity to end up with more accurate conclusions. Catheter-associated asymptomatic bacteriuria was not included in the definition of CAUTI, which is a potential limitation. Lastly, since the analysis of data included the study of associations between various outcomes and potential predictors in several clinics, several hypotheses were investigated. Some inflation of the Type I error beyond the typical 0.05 level may be considered, since we selected to present unadjusted p-values [37].

Conclusions
We prospectively studied the association of infection control measures and CAUTI incidence. The consumption of all hand disinfectant solutions and scrub correlated with decreased CAUTI rate in total Hospital and its divisions, while no correlation was found with the intervention of isolation of MDR carrier patients. Moreover, the correlation of CAUTI with MDR bacteremia was investigated and was found always negative, which indicates a constant and robust effect of the infection control program. Time series analysis can be used to inform evidence-based interventions and infection control policies. Since this study or data collection was based entirely on data abstraction from existing medical or laboratory record, with no interaction with the human subject concerned and with no collection of identifiable private information, an informed consent is not required from the patient.

Data Availability Statement:
The data that support the findings of this study are available on request from the corresponding author (H.C.M.). The data are not publicly available as data disclosure requires permission and ethical approval from Medical Ethical Committee of Athens Medical Center, Athens, Greece.