Impact of an Infection Control Program in an Intensive Care Unit in France
Objective. To evaluate the impact of an infection control program in an intensive care unit (ICU).
Design. Prospective before‐after study. Two 6‐month study periods were compared; between these periods, an infection control program based on isolation was implemented.
Setting. Polyvalent ICU of Montpellier Teaching Hospital.
Patients. Any patient who was hospitalized in the ICU for >48 hours and was discharged during 1 of the 2 periods.
Main Outcome Measures. The main patient‐related variables were sex, age at admission, type of patient (surgical, medical, or trauma), Simplified Acute Physiology Score II, length of ICU stay, need for intubation, duration of exposure to invasive devices, onset of nosocomial infection and pathogens responsible, and death. We compared the 2 study periods with respect to the incidence of 4 nosocomial infections (pneumonia, urinary tract infection, bacteremia, and catheter‐associated infection), the frequency of infection with the main multidrug‐resistant pathogens, and patient survival.
Results. Patients in periods 1 and 2 were similar with regard to sex, age, physiology score, and exposure to invasive devices. The rates of infection with multidrug‐resistant pathogens were significantly lower during period 2 than during period 1 (infection rate: 28.1% of patients in period 1 and 9.6% of patients in period 2 [
]; pneumonia rate: 32.6% of patients in period 1 and 4.2% of patients in period 2 [
]). The mortality rate among patients with nosocomial pneumonia was 38.2% in period 1 and 4.3% in period 2 (
).
Conclusions. After implementation of an infection control program, the rate of infection with multidrug‐resistant pathogens decreased, as did the mortality rate among patients with nosocomial pneumonia.
Received January 9, 2004; accepted April 8, 2005; electronically published January 6, 2006.
The development of infection control programs is recent. Most US hospitals established their infection control programs in the 1970s or 1980s.1 In France, such programs began in 1988 with the creation of a Nosocomial Infections Committee in every hospital. This growing interest in infection control programs can be explained by the socioeconomic and human impact of nosocomial infections. In French intensive care units (ICUs), nosocomial infections are frequent,1,2 increasing the mortality rate and the cost of the hospital stay.3‐6 Moreover, nosocomial infections due to multidrug‐resistant (MDR) pathogens (hereafter, “MDR infections”) result in high mortality rates3,7,8 and high costs3,9 compared with nosocomial infections due to antibiotic‐susceptible pathogens.
The French Nosocomial Infection National Technical Committee (NINTC) has published standard recommendations for fighting nosocomial infections and the spread of MDR pathogens.10 The main preventive measures recommended are aimed at preventing cross‐transmission through use of hygiene measures, patient isolation, and control of antibiotic use.10 Weist and colleagues11 found that 37.5% of nosocomial infections in a surgical ICU are the result of cross‐transmission. However, studies assessing the impact of infection control programs in ICUs have heterogeneous findings, and the debate about the use of isolation still exists.
In July 1999, an outbreak of infection due to a multidrug‐resistant strain of Acinetobacter occurred in a medical‐surgical ICU at the University Hospital of Montpellier. The source was a patient transferred from a Spanish ICU; 6 cases occurred within 8 weeks after his admission. The ICU was closed for 2 weeks in an attempt to keep the epidemic under control. After the outbreak, an infection control program was implemented in the ICU. The aim of this study was to assess the effectiveness of this infection control program, which followed NINTC recommendations.
Methods
This study was performed in the polyvalent ICU of the University Hospital of Montpellier, which contains 20 private rooms (5 units of 4 beds each).
Study Design
A prospective before‐after study, comparing two 6‐month periods, was chosen for evaluating the benefit of the infection control program: the period 1 ran from January 1, 1999, through June 30, 1999; period 2 ran from October 1, 1999, through March 31, 2000 (after the outbreak occurred). Every patient who was hospitalized in the ICU for >48 hours and whose discharge date fell within 1 of the 2 periods of investigation was included in the study.
The infection control program consisted of the following steps. First, preexisting standard preventive measures were strengthened. The staff's awareness of standard hygiene measures was increased; use of nonsterile, single‐use gloves and gowns was required after each patient contact. An antibiotic control policy was put in place, which specified that only senior physicians could prescribe antibiotics. Second, new measures were introduced. Hand disinfection with an alcohol‐based rub was required after each patient contact. Patient were cohorted by dividing the ICU into 4 areas (1 area consisted of 1 or 2 units according to the number of patients), and each nurse was specifically allocated to work in only 1 of the areas. The areas were as follows: (1) an emergency reception area (especially for patients with multiple trauma); (2) a contamination area for patients colonized or infected with MDR pathogens; (3) a protected area for noncolonized patients or patients infected with antibiotic‐susceptible pathogens; and (4) a quarantine area for patients whose infection status was unknown; as soon as their infection status was determined, these patients were directed to either the protected area or the contamination area.
During period 1, patients were screened for MDR pathogens during the first 48 hours of their ICU stay and then once each week; this screening schedule was maintained during period 2. Patients were screened for methicillin‐resistant Staphylococcus aureus (MRSA) from nasal swab specimens, for all pathogens from tracheal aspiration specimens, for MRSA and gram‐negative bacilli from armpit swab specimens, and for gram‐negative bacilli from rectal swab specimens. Neither the ICU structure nor the staff (4 physicians, 3 residents, and 1 nurse for 4 patients) nor the diagnostic criteria for nosocomial infection were modified between periods 1 and 2.
Collection of Data
Information on the following variables was recorded for each patient at admission to the ICU and/or daily by ICU staff members: sex, age, type of patient (surgical, medical, or trauma), location before ICU admission, presence of infection, carriage of an MDR pathogen at the time of admission, Simplified Acute Physiology Score II (SAPS II) score,12 length of ICU stay, death, intubation, duration of exposure to invasive devices (ie, mechanical ventilation, use of intravascular devices, and urinary catheterization), onset of nosocomial infection and pathogen(s) responsible (2 pathogens for a single infection could be entered into the database), and duration of antibiotic therapy among infected patients (administration of 2 antibiotics for 10 days was counted as 20 days of antibiotic therapy).
For the diagnosis of nosocomial infections we used the criteria of the NINTC.10. The diagnosis of nosocomial pneumonia in patients undergoing mechanical ventilation was considered if patients developed persistent lung infiltrates and purulent tracheal secretions. These patients underwent bronchoalveolar lavage. The diagnosis of nosocomial pneumonia was confirmed if culture of a lavage fluid sample yielded >104 colony‐forming units per milliliter of fluid. The onset of nosocomial urinary tract infection was defined as detection of >105 organisms per milliliter in a urine culture. The onset of nosocomial bacteremia was defined as detection of pathogen(s) in 1 blood culture (or 2 blood cultures, in the case of coagulase‐negative Staphylococcus). The onset of catheter‐associated infection was defined as the presence of >103 colony‐forming units per milliliter in a quantitative culture.
Evaluation Criteria
The following factors were considered in the evaluation of the effectiveness of the infection control program: (1) the incidence of the 4 types of nosocomial infection (pneumonia, urinary tract infection, bacteremia, and catheter‐associated infection); (2) the frequency of occurrence of the main MDR pathogens defined by NINTC10 (MRSA, and MDR strains of Enterobacteriaceae, Acinetobacter, and Pseudomonas species); and (3) the patient’s survival.
Statistical Analysis
Comparisons between 2 categorical variables were done using the χ2 test or Fisher’s exact test. Because the continuous variables were not normally distributed, comparisons of means were performed with the Mann‐Whitney nonparametric test. The incidence of nosocomial infections was assessed as follows: the number of nosocomial infections was divided by the number of person‐days of exposure to invasive devices. Two incidence rates were compared by calculating 95% confidence intervals (CIs). A step‐wise logistic regression model was used to estimate the effect of the infection control program on mortality, taking into account already known risk factors for death. A P value <.05 was considered statistically significant. The statistical analysis was done with SAS software, version 8.02 (SAS Institute).
Results
One hundred seventy one patients were included in period 1, and 165 patients in period 2. The patients included in the 2 periods were similar with regard to their characteristics at the time of admission (sex, age, type of patient, location before ICU admission, presence of infection, carriage of an MDR pathogen, and SAPS II score) and other characteristics (median length of ICU stay, and median duration of intubation, mechanical ventilation, and venous, arterial, or urinary catheterization) (Table 1).
Comparison of Nosocomial Infection Incidence
Incidences were calculated on the basis of the total number of nosocomial infections, the number of infections due to MDR pathogens, and the number of infections due to antibiotic‐susceptible pathogens. The incidence of infection associated with arterial and venous catheters was not calculated because of the low number of infections (there were 5 arterial catheter–associated infections during period 1, and 3 during period 2; there were 8 venous catheter–associated infections during period 1, and 6 during period 2).
Table 2 shows incidences and rates of infection for pneumonia, urinary tract infection, and bacteremia during periods 1 and 2. There was no statistically significant difference in the rates of pneumonia and urinary tract infection during periods 1 and 2. The incidence of bacteremia was stable during periods 1 and 2.
Comparison of the Rates of MDR Infection and the Durations of Antibiotic Therapy Among Infected Patients
Table 3 summarizes data on the number of infected patients and the types of infection, according to the resistance profile of the pathogen responsible. (Because a patient could contract more than 1 infection, the number of infections was greater than the number of infected patients, and the number of pathogens was greater than the number of infections.) To evaluate the impact of preventive measures on the pathogens responsible for infection, we calculated rates of MDR infection among infected patients as follows (Table 4): the number of infections of a specific type was divided by the total number of infections (ie, the total number of infections due to MDR pathogens plus infections due to antibiotic‐susceptible pathogens).
The overall rate of MDR infection (defined as the percentage of patients with at least 1 infection due to an MDR pathogen, whatever the site of infection), decreased significantly from period 1 to period 2 (28.1% vs 9.6% of patients; 
) (Table 4). The rate of MDR pneumonia also decreased significantly from period 1 to period 2 (32.6% vs 4.2% of patients; 
). The rate of MDR urinary tract infection was 12.5% during period 1 and 2.4% during period 2. The rate of MDR bacteremia was 6.7% (1 of 15 patients) during period 1 and 30.0% (3 of 10 patients) during period 2. Among the pathogens monitored, MDR strains of Enterobacteriaceae, Acinetobacter, and Pseudomonas species were less frequently detected during period 1. However, MRSA, which was not the most frequently identified pathogen during period 1, became the most frequently detected bacterium during period 2 (Table 3). The median duration of antibiotic therapy among infected patients tended to decrease from period 1 to period 2 (34.5 vs 20.0 days; 
).
Comparison of Mortality Rates
After determining the incidence of nosocomial infections and the proportion that were MDR infections, we looked for an effect of preventive measures on survival. The overall mortality rate among all patients was stable: 18.1% (31 of 171 patients) during period 1, and 16.2% (26 of 165 patients) during period 2 (
, odds ratio (OR), 1.17; 95% CI, 0.63‐2.15). To take into account potential confounding factors, we used a univariate analysis to determine the association between the factors studied and death. Factors significantly or nearly significantly (
) associated with death were included in the logistic regression model; those factors were the following: age (
), SAPS II score (
); nosocomial bacteremia (
; OR, 3.47 [95% CI, 1.3‐9.15]); nosocomial pneumonia (
; OR, 1.76 [95% CI, 0.84‐3.67]); type of patient (
; OR, 2.88 [95% CI, 1.44‐5.8]) (recoded from 3 into 2 categories, medical‐surgical and trauma, because the mortality rates among medical patients [23.3%] and surgical patients [23.5%] were similar); and location before admission to the ICU (
; OR, 1.66 [95% CI, 0.86‐3.19]) (recoded from 3 into 2 categories: hospital and outside hospital). To look for an effect of preventive measures, we included in the model the variable “period” (2 categories: period 1 and period 2).
The logistic regression model identified the following factors as risk factors for death: SAPS II score (increase of 1 unit; 
; OR, 1.055 [95% CI, 1.031‐1.079]); type of patient (medical‐surgical; 
; OR, 2.7 [95% CI, 1.3‐5.5]); and nosocomial bacteremia (
; OR, 3.2 [95% CI, 1.2‐6.7]). Pneumonia approached statistical significance as a risk factor for death (
; OR, 2.1 [95% CI, 0.9‐4.6]). Age, a factor classically associated with death, was not found to be a statistically significant risk factor, but age was strongly associated with type of patient (median age of trauma patients, 37.0 years; median age of medical‐surgical patients, 54.1 years; 
) and with SAPS II score (
; 
).
The logistic regression did not retain the variable “period.” However, the mortality rate among patients with nosocomial pneumonia was higher during period 1 than during period 2 (38.2% [13 of 34 patients] vs 4.3% [1 of 23 patients]; 
; [OR, 13.0; 95% CI, 2.2‐77.4]). To ascertain whether this reduction was related to a different distribution of risk factors for death, as determined by the logistic regression model, we compared their distribution between periods 1 and 2. SAPS II score and age did not differ significantly between periods 1 and 2 (
and 
, respectively, for SAPS II and age). However, there was a smaller percentage of medical‐surgical patients during period 2 than during period 1 (21.7% vs 38.2%; 
). Even though this result was not statistically significant, we performed a stratified analysis to take into account the variable “type of patient.” The Mantel‐Haenszel adjusted odds ratio for the factor “period” was 14.45 (95% CI, 1.9‐107.9; 
). Therefore, we conclude that patients with nosocomial pneumonia were less likely to die after preventive measures were initiated.
We did not compare the mortality rate among patients with nosocomial bacteremia between the 2 periods because of the low numbers of such patients.
Discussion
The infection control program had a significant impact on the incidence of nosocomial MDR infections (Table 4) and on the mortality rate among patients with nosocomial pneumonia. Similar results have been obtained by the rational use of antibiotics.13,14 Moreover, the incidence of pneumonia and the percentage of patients with pneumonia tended to decrease.
A reduction in the rate of nosocomial MDR infection was found for nosocomial MDR infections overall (ie, considered together), and for MDR pneumonia. A similar trend was observed for MDR urinary tract infection (Table 4). This reduction in the rates of these MDR infections was probably due to the isolation of infected patients, a practice that prevented cross‐transmission.15 The small number of patients with bacteremia precluded an assessment of the evolution of the rate of MDR bacteremia. We performed checks to ensure that this positive effect was not the result of the following 4 confounding factors.
The positive effect was not the result of greater exposure to antibiotics during period 2, because the duration of antibiotic therapy was similar between period 2 (1787 days of antibiotic therapy in 3204 days of hospitalization [percentage of days during which patients received antibiotic therapy, 55.8%) and period 1 (2082 days of antibiotic therapy in 3452 days of hospitalization [percentage of days during which patients received antibiotic therapy, 60.3%]). The positive effect was not the result of a global reduction in the rate of nosocomial MDR infections (relative to the total number of infections due to MDR or antibiotic‐susceptible pathogens) in the hospital as a whole during period 2, because this rate was stable from 1997 to 1999 (MDR Enterobacteriaceae infection rate of 16.1% in 1997, 19.2% in 1998, and 15.1% in 1999; MDR Pseudomonas infection rate of 20% from 1997 through 1999; MDR Acinetobacter infection rate of 75% in 1997, 80% in 1998, and 63% in 1999; MRSA infection rate of 34% in 1997, 35% in 1998, and 33% in 1999).
The positive effect was not the result of higher rates of carriage of MDR pathogens at admission during period 1, because in the ICU this rate was stable between period 1 and period 2 (Table 1). And finally, the positive effect was not the result of undetected outbreaks during period 1. If undetected outbreaks of the same magnitude as that of July 1999 had occurred, they would have been detected. Moreover, the strain of Acinetobacter isolated during the outbreak in July 1999 had never before been detected in this department. However, the lack of preventive measures during period 1 was probably associated with a higher rate of cross‐transmission at the origin of clusters of infection.
This reduction of the rate of nosocomial MDR infections should have 2 consequences. First, it should slow the spread of nosocomial MDR infections. Indeed, ICUs have a greater percentage of patients with nosocomial infections than do other wards,16 and patients with nosocomial infections are often referred, after their ICU stay, to other wards or institutions. Second, it should provide an economic benefit, because nosocomial infections due to MDR pathogens are more expensive to treat than nosocomial infections due to antibiotic‐susceptible pathogens.9
The logistic regression did not show an effect of preventive measures on the overall mortality rate for all patients, despite the reduction in the mortality rate among patients with pneumonia. However, because patients with pneumonia comprise only ∼20% of all patients, a reduction in the mortality rate among these patients is not sufficient to influence the overall mortality rate. The decrease in the mortality rate among patients with pneumonia could be explained by the fact that the incidence of MDR pneumonia was lower during period 2 than during period 1. This finding suggests that some of the mortality among patients with nosocomial pneumonia may be attributable to MDR pneumonia. A prospective study by Rello et al7 in a medical‐surgical ICU (which compared 38 patients with methicillin‐sensitive S. aureus–associated pneumonia and 11 with MRSA‐associated pneumonia) found a relationship between mortality and MRSA–associated pneumonia. However, this result was not confirmed by a multivariate analysis, and the patients with methicillin‐sensitive S. aureus–associated pneumonia were older than those with methicillin‐sensitive S. aureus–associated pneumonia. The mortality rate decreased only among patients with pneumonia; those with bacteremia could not be compared because of the low number of cases.
Both the incidence of pneumonia and its rate of infection tended to decrease after preventive actions were taken (Table 2). This nonsignificant result must be interpreted cautiously, given the low power of the test (
). This downward trend was not found for urinary tract infection and bacteremia; instead, our results suggest an increase in the rate of urinary tract infection. This increase could be explained by a change in the kinds of antibiotics prescribed during period 2 (broad‐spectrum antibiotics were prescribed less frequently), a consequence of a better antibiotic‐control policy. As shown in Table 2, the incidence of bacteremia and the rate of infection were similar during both periods; at any rate, it would have been difficult to evaluate the evolution of the incidence of bacteremia, given the low number of events.
For 2 reasons, the variations in the rates of pneumonia and urinary tract infection were not the result of a different distribution of risk factors for nosocomial infections between periods 1 and 2. First, the duration of exposure to invasive devices17‐21 (a major risk factor) was taken into account by calculating the incidence per 1000 days of exposure. Second, the SAPS II scores, which reflect the severity of illness at the time of admission, were similar during periods 1 and 2. However, many studies have not found this the SAPS II score to be a major risk factor.5,19
The main weakness of this study was the absence of a measure of the degree of compliance with preventive measures. The measure of the degree of compliance with cohorting should not be necessary. With regard to the policy for control of antibiotics, the sole indirect indicator of its effectiveness is the median duration of antibiotic therapy among infected patients; this variable tended to decrease from period 1 to period 2 (34.5 vs 20.0 days). However, no data were available on the kinds of antibiotics prescribed (especially broad‐spectrum antibiotics), the increased use of nonsterile single‐use gloves and gowns, and the practice of systematic hand disinfection after each contact with patients. However, because the effect of potential confounding factors has been discussed and ruled out, we can consider that the preventive measures had a positive impact, even though they may not have been implemented completely. Another weakness of the study design was that it was impossible to distinguish the effect of technical isolation from the effect of cohorting patients or the effect of the antibiotic‐control policy.
In conclusion, the infection control program was effective. The isolation of infected patients contributed to a decrease in the rate of nosocomial MDR infections and a decrease in the mortality rate among patients with nosocomial pneumonia. Moreover, the incidence of pneumonia and the rate of infection tended to decrease. However, the impact of the preventive measures depends on the staff’s motivation. The 6‐month follow‐up period is not sufficient to assess the long‐term benefit of this infection control program. Therefore the question is: will the staff's motivation and benefits of the program continue over the long term? Answering that question would require additional evaluation of the program’s impact, both medical and economic.
Acknowledgment
We gratefully thank Marie‐Ange Grosbois for the English review of the manuscript.
References
- 1. Haley RW, White JW, Culver DH, Hughes JM. The financial incentive for hospitals to prevent nosocomial infections under the prospective payment system: an empirical determination from a nationally representative sample. JAMA 1987; 257:1611‐1614.
- 2. Nosocomial Infection National Technical Committee. Infections nosocomiales: constats et objectifs. Hygiène 1997; 621:15‐16.
- 3. Bueno‐Cavanillas A, Delgado‐Rodriguez M, Lopez‐Luque A, Schaffino‐Cano S, Galvez‐Vargas R. Influence of nosocomial infection on mortality rate in an intensive care unit. Crit Care Med 1994; 22:55‐60.
- (4. Fagon JY, Chastre J, Vuagnat A, Trouillet JL, Novara A, Gibert C. Nosocomial pneumonia and mortality among patients in intensive care units. JAMA 1996; 275:866‐869.
- 5. Girou E, Stephan F, Novara A, Safar M, Fagon JY. Risk factors and outcome of nosocomial infections: results of a matched case‐control study of ICU patients. Am J Respir Crit Care Med 1998; 157:1151‐1158.
- 6. Jarvis WR. Selected aspects of the socioeconomic impact of nosocomial infections: morbidity, mortality, cost, and prevention. Infect Control Hosp Epidemiol 1996; 17:552‐557.
- 7. Rello J, Torres A, Ricart M, et al. Ventilator‐associated pneumonia by Staphylococcus aureus: comparison of methicillin‐resistant and methicillin‐sensitive episodes. Am J Respir Crit Care Med 1994; 150:1545‐1549.
- 8. Romero‐Vivas J, Rubio M, Fernandez C, Picazo JJ. Mortality associated with nosocomial bacteremia due to methicillin‐resistant Staphylococcus aureus. Clin Infect Dis 1995; 21:1417‐1423.
- 9. Abramson MA, Sexton DJ. Nosocomial methicillin‐resistant and methicillin‐susceptible Staphylococcus aureus primary bacteremia: at what costs? Infect Control Hosp Epidemiol 1999; 20:408‐411.
- 10. French Noscomial Infection National Technical Committee Working Group. 100 Recommandations pour la surveillance et la prévention des infections nosocomiales. 2nd ed. Paris: Ministère de l'Emploi et de la Solidarité–Secrétariat d'Etat à la Santé et à l'action sociale; 1999.
- 11. Weist K, Pollege K, Schulz I, Ruden H, Gastmeier P. How many nosocomial infections are associated with cross‐transmission? A prospective cohort study in a surgical intensive care unit. Infect Control Hosp Epidemiol 2002; 23:127‐132.
- 12. LeGall JR. Evaluation en réanimation—risque de décès hospitalier. L’évaluation en réanimation. Paris: Masson; 1990.
- 13. Dupont H, Mentec H, Sollet JP, Bleichner G. Impact of appropriateness of initial antibiotic therapy on the outcome of ventilator‐associated pneumonia. Intensive Care Med 2001; 27:355‐362.
- 14. Geissler A, Gerbeaux P, Granier I, Blanc P, Facon K, Durand‐Gasselin J. Rational use of antibiotics in the intensive care unit: impact on microbial resistance and costs. Intensive Care Med 2003; 29:49‐54.
- 15. Safdar N, Maki DG. The commonality of risk factors for nosocomial colonization and infection with antimicrobial‐resistant Staphylococcus aureus, enterococcus, gram‐negative bacillii, Clostridium difficile, and Candida. Ann Intern Med 2002; 136:834‐844.
- 16. Eggiman P, Pittet D. Infection control in the ICU. Chest 2001; 120:2059‐2093.
- 17. Appelgren P, Hellstrom I, Weitzberg E, Soderlund V, Bindslev L, Ransjo U. Risk factors for nosocomial intensive care infection: a long‐term prospective analysis. Acta Anaesthesiol Scand 2001; 45:710‐719.
- 18. Farinas‐Alvarez C, Carmen Farinas M, Llorca J, Delgado Rodriguez M. Risk factors for nosocomial sepsis: a case‐control study. Med Clin (Barc) 2001; 116:765‐769.
- 19. McCusker ME, Perisse A, Roghmann MC. Severity‐of‐illness markers as predictors of nosocomial infection in adult intensive care unit patients. Am J Infect Control 2002; 30:139‐144.
- 20 Tejada Artigas A, Bello Dronda S, Chacon Valles E, et al. Risk factors for nosocomial pneumonia in critically ill trauma patients. Crit Care Med 2001; 29:304‐309.
- 21. Vanhems P, Lepape A, Savey A, Jambou P, Fabry J. Nosocomial pulmonary infection by antimicrobial‐resistant bacteria of patients hospitalized in intensive care units: risk factors and survival. J Hosp Infect 2000;45:98‐106.