- Email: [email protected]
Nosocomial Infection and Related Risk Factors in a General Surgery Service: A Prospective Study
Journal of Infection (2003) 46: 17±22 doi:10.1053/jinf.2002.1073
Nosocomial Infection and Related Risk Factors in a General Surgery Service: A Prospective Study P. Vazquez-Aragon*1, M. Lizan-Garcia2, P. Cascales-Sanchez1, M. T. Villar-Canovas3 and D. Garcia-Olmo1 1
General Surgery Service, 2Preventive Medicine Unit and 3Anesthesiology Service, Albacete General and University Hospital, Albacete, Spain
Objectives: The aim of this study was to quantify the frequency of nosocomial infection in the general surgery service of a tertiary-care hospital and to identify associated risk factors. Methods: A prospective, longitudinal, descriptive and analytical study was made from January 1995 to December 1998 of a clinical cohort of 2794 patients who underwent a surgical procedure with a post-surgery stay of more than 48 h. The criteria for infection were those defined by the Center for Disease Control and Prevention (CDC) of the USA. Results: The most frequent nosocomial infection was surgical infection (SI), with a global cumulative incidence (CI) of 7.7%, ranging from 3.4% for clean surgery to 23.7% for dirty surgery. The next most frequent were urinary tract infection (UTI) and bacteremia (1.5%) and nosocomial pneumonia (NP) (0.5%). The global CI of SI decreased from 11.7% in 1995 to 4% in 1998. An ASA classification higher than 2 multiplied the risk of SI by 1.76; with respect to UTI multiplied the risk by 2.13; the risk of NP by 5.93 and multiplied the risk of B by 4.72. Conclusions: The most frequent nosocomial infection was surgical infection. An ASA higher than 2, the stay prior to surgery; the number of days with a urinary catheter, with a central venous catheter and with mechanical ventilation; as well as the improvement in the use of antimicrobial prophylaxis, were all factors that influenced the frequency of nosocomial infection. # 2002 The British Infection Society. Published by Elsevier Science Ltd. All rights reserved.
Introduction Throughout their history, hospitals have coexisted with nosocomial infection (NI), which has been defined as ``that infection that was not present neither was incubating at the moment of patient admission at a hospital'' [1]. During the past 20 years it has been reported that between 6% and 14% of patients that enter general hospitals develop a NI. In order of frequency, urinary tract infection (UTI) accounts for 42% of all infections; followed by surgical infection (SI), previously defined as ``surgical wound infection'' which accounts for 24%; nosocomial pneumonia (NP), 15±20%; and bacteremia (B), 5% [2]. In Spain, according to the report of the EPINE project, excluding short-stay patients (less than two days), the
Please address all correspondence to: P. Vazquez Aragon, C/ CABA n 17, 5 Izda, 02001 Albacete, Spain. Tel.: 34 967 224203; Fax: 34 967 243952; E-mail address: [email protected] (P.Vazquez-Aragon). 0163±4453/03/$35.00
prevalence of nosocomial infection was 8% in 1998 [3]. The risk factors that depend on the patient are referred to as intrinsic risk factors. Certain procedures and the use of certain instrumentation are considered to be extrinsic risk factors. In the Study on the Efficacy of Nosocomial Infection Control (SENIC) project, Haley et al. [4], using multivariate analysis, developed an index for the risk of infection of a surgical wound. In 1990, Culver et al. [5], modified the SENIC index and proposed a new one, the ``National Nosocomial Infection Surveillance (NNIS)'' index. The aim of this study was to quantify the frequency of nosocomial infection in the general surgery service of a tertiary-care hospital in the Spanish National Health System and to identify associated risk factors.
Patients and Methods The Albacete Hospital Complex in Albacete, Spain, is a tertiary-care center of the Spanish National Health
# 2002 The British Infection Society. Published by Elsevier Science Ltd. All rights reserved.
18
P. Vazquez-Aragon et al.
Service, with 832 beds, of which 80 are assigned to the general surgery service. From 1/1/1995 to 31/12/1998, 14,089 patients underwent general surgical procedures. A prospective, longitudinal, descriptive and analytical, clinical cohort study was made. The clinical records of 2794 patients who underwent elective surgery or who underwent emergency surgery, with a post-surgery stay of more than 48 h (19.8%), using the clinical records, were examined. The data from the patients records were collected during the hospital stay of the patients by a trained nosocomial infection surveillance team from the Preventive Medicine Unit. We codified the surgical procedures according to ICD-9-CM with four digits and divided them subsequently into 10 groups for this study. We used the criteria for NI defined by the US Center for Disease Control (CDC) in 1998 [1]. We then examined demographic and administrative variables, NI variables, surgical operation variables, and variables that included information related to the use of antimicrobial agents. Methods of analysis Analysis of data We performed a descriptive and univariate study using EPI-INFO version 6.2 (CDC, Atlanta, GA, USA) and we used the STATISTIX 3.5 program (Analytical Software, St Paul, MN) for logistic regression (LR) analysis. Descriptive analysis of the studied cases We studied the characteristics of the patients, starting with age and sex distribution. We calculated the means (X), standard errors (SE) and dispersion measurements for the quantitative variables and determined the distributions of frequencies of qualitative variables. Subsequently, we determined the CI of SI, UTI, NP and B, in global form, in specific form and according to the NNIS risk index. The system for surveillance of hospital infections in our hospital has involved classification of patients in terms of the NNIS index since 1997, with a cut-off time of 120 min as ponderal measurement for all operations, since at the start of the surveillance system, the casuistry and the lack of information about the mean times for each operative procedure did not allow fitting the cut point for each operative procedure category. We compared the distributions of SI, UTI, NP and bacteremia among the collected variables and examined possible associations by the two-tailed Student's t-test or the Kruskall±Wallis test for quantitative variables and by the chi-squared test with the Yates correction or by Fisher's exact test for qualitative variables.
As in the case of crude measurements of the magnitude of association for qualitative variables in two categories, we used the relative risk (RR) in the univariate study and we used odds ratios (OR) and 95% confidence intervals (CI95) (as estimators of RR) in the LR analysis according to Cornfield's method. Statistical significance was accepted for P < 0.05, with a 95% confidence limit. In LR analysis, the dependent variables were SI, UTI, NP and B and the independent variables were age, sex, duration of operation (with a cut-off time of 120 min), duration of hospital stay prior to surgery, antibiotic prophylaxis (received or not), surgical class (contamination grade of the operation), operative procedure, ASA classification, emergency procedure, days with a urinary catheter (measured until infection), total days with mechanical ventilation during and post-surgery (measured until infection), and total days with a central venous catheter (measured until infection).
Results The members of our cohort ranged in age from 4 and to 98 years (58.8 17.3; median 63). There were 1129 males (40.4%) and 1665 females (59.6%). Among the admissions, 80.8% were programmed and the 19.2% were emergencies. All the patients were classified according to an index of intrinsic risk, assessed in terms of the ASA classification, as follows: ASA-I, 756 patients (27%); ASA-II, 1196 patients (42.8%); ASA-III, 701 patients (25.1%); ASA-IV, 139 patients (5%); and ASA-V, 2 patients (<1%). A total of 816 patients (29.2%) had a urinary catheter for a mean duration of 1.5 4.3 days (maximum, 60 days). Among the total cohort, 145 (5.2%) needed mechanical ventilation for a mean of 0.2 1.9 days (maximum, 55 days), and 339 patients (12.1%) of the cohort had a central venous catheter for a mean of 1.0 3.4 days (maximum, 41 days). During their stay in the hospital, 314 patients had some type of hospital infection (i.e., the global incidence was 11.2%). The most frequent nosocomial infection in the cohort was SI with a rate of global CI of 7.7%, ranging from 3.5% for clean surgery to 23.7% for dirty surgery, followed in frequency by B (1.5%), then UTI (1.5%), and finally NP with a global AI of 0.5%. The global CI of SI varied between 11.7% in 1995 and 4% in 1998 (Table I). The global CI of NI in terms of NNIS index is shown in Table II. In 775 cases (27.7%), the duration of surgery was 120 min or longer (89.7 62.8 min; median, 84.5; maximum, 420 min). Open cholecystectomy (16.7%),
Infection Risk Factors in a General Surgery Service colorectal surgery (16.4%) and herniorrhaphy (14.7%) were the most frequent operative procedures (Table III). In 75.5% of cases, the hospital stay prior to surgery was less than 5 days. The mean stay prior to surgery of the cohort was 4.8 10.4 days and the mean postsurgery stay was 8.1 7.8 days. The results of LR analysis with odds ratios and their confidence intervals for the different types of NI are shown in Table IV. Male gender multiplied the risk of SI by 1.54; each 3 days of stay prior to surgery multiplied the risk of SI by 1.06; each 10 min of operation time multiplied the risk by 1.03; emergency procedure multiplied the risk by Table I. Cumulative incidence of nosocomial infection. Year
SI UTI NP B
1995 (%)
1996 (%)
1997 (%)
1998 (%)
Total (%)
11.7 2.1 0.5 1.7
9.0 1.2 0.7 2.0
5.5 0.8 0.5 1.1
4.0 1.9 0.2 1.3
7.7 1.5 0.5 1.5
19
1.49; clean-contaminated surgery multiplied the risk by 0.80; contaminated surgery multiplied the risk by 1.95; dirty surgery multiplied the risk by 4.0; an ASA classification higher than 2 multiplied the risk by 1.76 and each day with a urinary catheter multiplied the risk by 1.04. The antibiotic prophylaxis appears as a protection factor against SI. With respect to UTI, clean-contaminated surgery multiplied the risk of UTI by 3.42, contaminated surgery multiplied the risk of UTI by 2.96 and dirty surgery multiplied the risk by 2.7; every day with a urinary catheter multiplied the risk of UTI by 1.05; and an ASA classification higher than 2 multiplied the risk of UTI by 2.13. With respect to NP, each day of hospital stay prior to surgery multiplied the risk of NP by 1.04; each day with mechanical ventilation multiplied it by 1.14 and an ASA classification higher than 2 multiplied it by 5.93. Each day of hospital stay prior to surgery multiplied the risk of B by 1.03; each minute of operation time multiplied the risk by 1.01; the emergency admission multiplied it by 4.24; an ASA classification higher than 2 multiplied it by 4.72; and each day with CVC multiplied it by 1.15.
Table II. Global cumulative incidence of nosocomial infection by NNIS index. NNIS
SI No SI
UTI SI
NP
B
Total
AI (%)
No UTI
UTI
Total
AI (%)
No NP
NP
Total
AI (%)
No B
B
Total
AI (%)
0 1 2 3
1256 811 362 150
29 60 67 59
1285 871 429 209
2.26 6.89 15.62 28.23
1274 857 422 199
11 14 7 10
1285 871 429 209
0.86 1.61 1.63 4.78
1285 866 424 205
0 5 5 4
1285 871 429 209
0.00 0.57 1.17 1.91
1285 866 410 190
0 5 19 19
1285 871 429 209
0.00 0.57 4.43 9.09
Total
2579
215
2794
7.70
2752
42
2794
1.50
2780
14
2794
0.50
2751
43
2794
1.54
Table III. Distribution of the cohort by NNIS index and operative procedure. Operative procedure
NNIS index 0
Mastectomy Thyroidectomy Esophagus-gastric surgery Colorectal Surgery Appendectomy Open Cholecystectomy Acute Cholecystitis Laparoscopic Cholecystectomy Herniorraphy Other procedures Total
1 186 242 10 0 82 227 4 230 276 28
(48%) (64%) (38.5%) (0%) (54.6%) (58.7%) (3.8%) (7.03%) (67%) (30%)
1285 (45.99%)
166 123 10 62 52 182 29 86 120 41
2 (44%) (32.5%) (38.5%) (13.5%) (34.6%) (39%) (27.8%) (26.3%) (29%) (44%)
871 (31.18%)
26 13 4 237 16 48 42 11 15 17
3 (6.8%) (3.5%) (5.4%) (51.6%) (10.8%) (10.3%) (40.6%) (3.4%) (3.6%) (18.5%)
429 (15.35%)
1 0 2 160 0 9 29 0 1 7
Total (0.2%) (0%) (7.6%) (34.9%) (0%) (2.1%) (27.8%) (0%) (0.3%) (0.3%)
209 (7.48%)
379 378 26 459 150 466 104 327 412 93
(13.5%) (13.5%) (0.9%) (16.4%) (5.4%) (16.7%) (3.7%) (11.7%) (14.7%) (3.3%)
2794 (100.00%)
20
P. Vazquez-Aragon et al.
Table IV. Coefficients, standard errors, odds ratios, 95% confidence intervals, and P values for variables included in the logistic regression analysis of nosocomial infection. Variable
Coefficient
SE
OR
CI95
P
ÿ4.0841 ÿ4.5226 ÿ2.5914 ÿ3.7478 ÿ3.7671 ÿ1.6997
2042.2 1.5509 7.1581 1.0290 2.0809 1.9926
1.54 1.06 1.03 1.49 0.84
1.13±2.09 1.01±1.08 1.00±1.07 0.99±2.24 0.57±1.25
0.0035 0.0003 0.0003 0.0703 0.3937
ÿ2.1759 ÿ8.9678 ÿ1.4943 ÿ6.5142 3.5214
2.5279 2.3781 2.5486 1.5631 1.5329
0.8 1.95 4 1.76 1.04
0.49±1.32 1.17±3.24 2.37±7.34 1.29±2.41 1.02±1.07
0.3894 0.0002 0.0000 0.0000 0.0123
ÿ5.4947 5.3234
4.2915 1.5402
1.05
1.02±1.09
0.0005
1.2287 1.0837 0.9947 7.5818
4.7893 5.3742 5.7178 3.3771
3.42 2.96 2.7 2.13
1.34±8.74 1.03±8.47 0.88±8.29 1.10±4.14
0.0103 0.0438 0.0819 0.0248
Pneumonia Constant Stay prior to surgery (at 1-day intervals) Mechanical ventilation (at 1-day intervals) ASA > II
ÿ6.7231 3.5619 1.3034 1.7798
6.1242 1.5723 3.4477 6.7783
1.04 1.14 5.93
1.00±1.07 1.06±1.22 1.57±22.38
0.0235 0.0002 0.0086
Bacteremia Constant Stay prior to surgery (at 1-day intervals) Duration of operation (at 1-min intervals) Emergency procedure ASA > II CVC (at 1-day intervals)
ÿ6.8251 3.3532 5.0045 1.4443 1.5516 1.4039
4.5578 1.1778 1.6995 3.5069 4.0221 2.0242
1.03 1.01 4.24 4.72 1.15
1.01±1.06 1.00±1.02 2.13±8.43 2.15±10.35 1.11±1.20
0.0044 0.0032 0.0000 0.0001 0.0000
Surgical infection Constant Gender Stay prior to surgery (at 3-day intervals) Duration of operation (at 10-min intervals) Emergency procedure Prophylaxis Surgical class Clean-contaminated Contaminated Dirty ASA > II Urinary catheter (at 1-day intervals) Urinary tract infection Constant Urinary catheter (at 1-day intervals) Surgical class Clean-contaminated Contaminated Dirty ASA > II
ASA > II: ASA III, IV or V.
A microbiological diagnosis was obtained in 67.2% of NI. Half of the cultured strains were Gram-positive cocci and 37% were Gram-negative bacilli. The most frequently isolated pathogens were Enterococcus faecalis (19.2%) and Escherichia coli (13.1%) for SI, Enterococcus faecalis (38.9%) and Candida albicans (22.2%) for UTI, Neumococcus (26%) and Staphylococcus aureus (17%) for NP and Staphylococcus epidermidis (21.6%) and other Staphylococcus for B. A total of 1422 patients (57.8%) received antimicrobial prophylaxis. The perioperative antibiotic prophylaxis was indicated in clean with implants, clean-contaminated and contaminated surgery patients following the antibiotic prophylaxis protocol from our hospital, with prophylaxis administration 1-h before surgery. Three hundred and twenty-four patients with clean surgery (with and without implants) received antibiotic prophylaxis. The mean duration of prophylaxis was 2.5 days, and cephoxitin was the most frequently
used antibiotic (18.7%). The antimicrobial agents used most frequently for prophylaxis were second-generation cephalosporins. The antibiotic association used most frequently was clyndamycin plus gentamicin. Sixty-five patients (2.3%) died during their stay in hospital. The ratio of the number of patients who died who had had a NI to the number of patients with a NI was 31/314 (9.9%). Of 65 patients who died, 16 had a SI, namely, approximately one-quarter of the total.
Discussion Among the 2794 patients in the cohort, 1162 (41.6%) were more than 65 years old and 223 (8%) were more than 80 years old. It has been estimated that, in the USA alone, 27 million surgical procedures are performed annually and almost a third of the patients subjected to these procedures are 65 years of age or older.
Infection Risk Factors in a General Surgery Service In our series, the large number of patients (816; 29%) with a urinary catheter seems very high in relation to data published (between 5% and 17%) in other reports [6,7]. This difference might be due to the large number of patients of more than 65 years of age who needed a urinary catheter and, also because insertion of a catheter is included in the preoperative protocol for specified surgical procedures. The NNIS index appears to be the best indicator of intrinsic risk of SI, being apparently better than the traditional classification in terms of the contamination grade of the surgical wound [8]. Although it was initially developed to predict SI risk, the index also predicts reasonably well the risks of postoperative infections in other sites. The success of the NNIS index in predicting the risk of developing postoperative pneumonia, bloodstream or urinary tract infections suggests that the index is a general expression of the propensity to develop infection and probably reflects the general nature of the risk factors that comprise the index [5]. The CI of NI was 11.2%, resembling the published results for other series (1.4±17.3%) [9,10]. Our rates of SI were higher but, with respect to other NI such as UTI, NP and B, our data are similar and, in some cases, incidences are even lower than those reported by Emori et al. [10]. The system for surveillance of hospital infections in our hospital has involved classification of patients in terms of the NNIS index since 1997, with a cut-off time of 120 min for all operations. However, if we had considered times equal to or above the 75th percentile of the duration of surgery, cases included in groups 1, 2 and 3 would have been classified as 0, 1 and 2, respectively. This error in classification leads to an overestimate of the rates of infection in the lowest groups. In the total cohort, only 775 (27.7%) operations lasted 120 min or longer. A remarkable decrease in the incidence of SI is shown apparent over the course of the study, probably as a result of the establishment of a system for surveillance of NI in our hospital and of an improved understanding by surgeons of the infection rates associated with different types of surgery, as well as the more frequent and more appropriate use of antimicrobial prophylaxis protocols including the change in administration time 1-h before surgery. In this way, the antibiotic prophylaxis appears as a protection factor against SI. The hospital stay prior to surgery is another factor to take in account in the decrease of the SI. It is widely accepted that the hospital stay prior to surgery is a risk factor for NI [11]. The mean stay prior to surgery of our cohort was 4.8 days, which seems long since 19.2% of admissions were emergencies, 70% of patients were
21
classified as ASA II and, since 1997, a day-surgery unit has been in use, with almost all admissions occurring on the same day as surgery. Thus, the lengthy stay prior to surgery might be explained by the fact that, some years ago, preoperative studies and diagnostic tests were performed after the patient had been admitted to hospital. The duration of a surgical intervention of more than two hours is accepted as a risk factor for SI. In our cohort, as in other series, the mean duration of surgery for patients with SI was of 31.7 min longer than for noninfected patients. Patients with a central venous catheter and those with a urinary catheter had higher rates of SI. In our series, the incidence of SI in diabetic patients was not significantly different from that in non-diabetic patients. The ASA score is itself an index, designed to asses preoperatively the overall physical status of the patient and constitutes a critical component of the NNIS index, included in an attempt to measure intrinsic patient susceptibility and has the advantage of being readily available at the time of surgery [5]. Culver et al. [5] showed that ASA score as a single predictor of SI risk was at least as good as the traditional wound classification. In our study, an ASA classification higher than 2 appears as a risk factor in every studied infection, being this increase of the risk more important in NP and B. This fact clearly confirms this score as a predictor of nosocomial infections other than SI in surgical patients. A noticeable decrease was also observed for the other types of NI over the course of the study, and all these factors appear to explain this decrease in NI between 1995 and 1998. The use of prophylaxis in clean surgery is controversial. Almost one quarter of the Spanish surgeons use antibiotic prophylaxis with any type of clean surgery, according to the National Survey of Surgical Infection carried out in 1995 by the Spanish Association of Surgeons [12]. In our series, 28% of the patients with clean surgery received prophylactic treatment. The cited survey indicated that, as in our series, the antibiotics used most frequently for prophylaxis were first- and second-generation cephalosporins, which are used by 72% of the Spanish surgeons. Multivariate analysis revealed three variables that affected the risk of UTI: the contamination grade of the surgery; the number of days with an urinary catheter and the primary health status of the patient. The principle ``if possible, do not catheterize'', as well as established protocols of asepsia and careful assessment daily of the necessity for the urinary catheter, with its removal as soon as possible, constitute the basis for prevention of UTI.
22
P. Vazquez-Aragon et al.
With respect to NP, mechanical ventilation (which facilitates the aspiration of polluted droplets), together with poorer health status, might predispose a patient to NP. A direct program for prevention of NP should focuse on decreasing, as far as possible, the hospital stay prior to surgery of the patient and on decreasing as far as possible, the number of days with mechanical ventilation. Increases in the number of days with CVC increase the possibility of colonization, both extrinsically and intrinsically, of the system and depending on the patient's status, can even facilitate the development of bacteremia. A program for prevention of bacteremia should focuse on decreasing, as far as possible, the hospital stay prior to surgery, the establishment of preoperative preparation protocols, and a decrease, as far as possible, in the number of days with CVC. We hope that implementation of the procedures that we suggest will help to reduce the frequency of NI on surgical services.
References 1 Garner JS, Jarvis WR, Emori TG, Horan TC, Hughes JM. CDC definitions for nosocomial infections, 1998. Am J Infect Control 1988; 16: 128±140. 2 Haley RW, Culver DR, White W, Emon TG. The nationwide nosocomial infection rate. A new need for vital statistics. Am J Epidemiol 1985; 121: 159±167.
3 Grupo de Trabajo EPINE 98. Estudio de prevalencia de las infecciones nosocomiales en los hospitales espanÄoles. 9 estudio. Madrid: Sociedad EspanÄola de Medicina Preventiva, Salud PuÂblica e Higiene. 1998. 4 Haley RW, Culver DH, Morgan WM, Emori TG, Munn VP, Hooton TM. Identifying patients at high risk of surgical wound infection: a simple multivariate index of patient susceptibility and wound contamination. Am J Epidemiol 1985; 121: 206±215. 5 Culver DH, Horan TC, Gaynes RP, Martone WJ, Jarvis WR, Emori TG, Banerjee SN, Edwards JR, Tolson JS, Henderson TS, Hughes JM. Surgical wound infection rates by wound class, operative procedure, and patient risk index. Am J Med 1991; 91: 152S±157S. 6 Shapiro M, Simehen E, Izvaeli S. Multivariate analysis of risk factors for acquiring bacteriuria in patients with indwelling urinary catheters for longer than 24 hours. Infect Control 1984; 32: 525±532. 7 Vasile I, Mogos D, Paun I, Paun M, Florescu M. Urinary infections with hospital germs in general surgery. Chirurgia 1998; 93: 317±321. 8 Horan TC, Culver DH, Gaynes RP, Jarvis WR, Edwards JR, Reid CR. Nosocomial infections in surgical patients in the United States, January 1986±June 1992. Infect Control Hosp Epidemiol 1993; 14: 73±80. 9 Cruse P, Foord R. The epidemiology of wound infection. A 10-year prospective study of 62,939 wounds. Surg Clin North Am 1980; 60: 27±40. 10 Emori TG, Culver DH, Horan TC. National Nosocomial Infections Surveillance System (NNIS): description of surveillance methods. Am J Infect Control 1991; 19: 19±35. 11 Bueno Cavanillas A, RodrõÂguez-Contreras R, Delgado RodrõÂguez M, Moreno Abril O, LoÂpez Gigosos R, Guillen Solvas J, GaÂlvez Vargas R. Preoperative stay as a risk for nosocomial infection. Eur J Epidemiol 1991; 7: 670±676. 12 Cainzos M, Lozano F, Alcaraz P et al. Profilaxis antibioÂtica en cirugõÂa limpia. Cir Esp 1996; 59: 7±10.
Nosocomial Infection and Related Risk Factors in a General Surgery Service: A Prospective Study P. Vazquez-Aragon*1, M. Lizan-Garcia2, P. Cascales-Sanchez1, M. T. Villar-Canovas3 and D. Garcia-Olmo1 1
General Surgery Service, 2Preventive Medicine Unit and 3Anesthesiology Service, Albacete General and University Hospital, Albacete, Spain
Objectives: The aim of this study was to quantify the frequency of nosocomial infection in the general surgery service of a tertiary-care hospital and to identify associated risk factors. Methods: A prospective, longitudinal, descriptive and analytical study was made from January 1995 to December 1998 of a clinical cohort of 2794 patients who underwent a surgical procedure with a post-surgery stay of more than 48 h. The criteria for infection were those defined by the Center for Disease Control and Prevention (CDC) of the USA. Results: The most frequent nosocomial infection was surgical infection (SI), with a global cumulative incidence (CI) of 7.7%, ranging from 3.4% for clean surgery to 23.7% for dirty surgery. The next most frequent were urinary tract infection (UTI) and bacteremia (1.5%) and nosocomial pneumonia (NP) (0.5%). The global CI of SI decreased from 11.7% in 1995 to 4% in 1998. An ASA classification higher than 2 multiplied the risk of SI by 1.76; with respect to UTI multiplied the risk by 2.13; the risk of NP by 5.93 and multiplied the risk of B by 4.72. Conclusions: The most frequent nosocomial infection was surgical infection. An ASA higher than 2, the stay prior to surgery; the number of days with a urinary catheter, with a central venous catheter and with mechanical ventilation; as well as the improvement in the use of antimicrobial prophylaxis, were all factors that influenced the frequency of nosocomial infection. # 2002 The British Infection Society. Published by Elsevier Science Ltd. All rights reserved.
Introduction Throughout their history, hospitals have coexisted with nosocomial infection (NI), which has been defined as ``that infection that was not present neither was incubating at the moment of patient admission at a hospital'' [1]. During the past 20 years it has been reported that between 6% and 14% of patients that enter general hospitals develop a NI. In order of frequency, urinary tract infection (UTI) accounts for 42% of all infections; followed by surgical infection (SI), previously defined as ``surgical wound infection'' which accounts for 24%; nosocomial pneumonia (NP), 15±20%; and bacteremia (B), 5% [2]. In Spain, according to the report of the EPINE project, excluding short-stay patients (less than two days), the
Please address all correspondence to: P. Vazquez Aragon, C/ CABA n 17, 5 Izda, 02001 Albacete, Spain. Tel.: 34 967 224203; Fax: 34 967 243952; E-mail address: [email protected] (P.Vazquez-Aragon). 0163±4453/03/$35.00
prevalence of nosocomial infection was 8% in 1998 [3]. The risk factors that depend on the patient are referred to as intrinsic risk factors. Certain procedures and the use of certain instrumentation are considered to be extrinsic risk factors. In the Study on the Efficacy of Nosocomial Infection Control (SENIC) project, Haley et al. [4], using multivariate analysis, developed an index for the risk of infection of a surgical wound. In 1990, Culver et al. [5], modified the SENIC index and proposed a new one, the ``National Nosocomial Infection Surveillance (NNIS)'' index. The aim of this study was to quantify the frequency of nosocomial infection in the general surgery service of a tertiary-care hospital in the Spanish National Health System and to identify associated risk factors.
Patients and Methods The Albacete Hospital Complex in Albacete, Spain, is a tertiary-care center of the Spanish National Health
# 2002 The British Infection Society. Published by Elsevier Science Ltd. All rights reserved.
18
P. Vazquez-Aragon et al.
Service, with 832 beds, of which 80 are assigned to the general surgery service. From 1/1/1995 to 31/12/1998, 14,089 patients underwent general surgical procedures. A prospective, longitudinal, descriptive and analytical, clinical cohort study was made. The clinical records of 2794 patients who underwent elective surgery or who underwent emergency surgery, with a post-surgery stay of more than 48 h (19.8%), using the clinical records, were examined. The data from the patients records were collected during the hospital stay of the patients by a trained nosocomial infection surveillance team from the Preventive Medicine Unit. We codified the surgical procedures according to ICD-9-CM with four digits and divided them subsequently into 10 groups for this study. We used the criteria for NI defined by the US Center for Disease Control (CDC) in 1998 [1]. We then examined demographic and administrative variables, NI variables, surgical operation variables, and variables that included information related to the use of antimicrobial agents. Methods of analysis Analysis of data We performed a descriptive and univariate study using EPI-INFO version 6.2 (CDC, Atlanta, GA, USA) and we used the STATISTIX 3.5 program (Analytical Software, St Paul, MN) for logistic regression (LR) analysis. Descriptive analysis of the studied cases We studied the characteristics of the patients, starting with age and sex distribution. We calculated the means (X), standard errors (SE) and dispersion measurements for the quantitative variables and determined the distributions of frequencies of qualitative variables. Subsequently, we determined the CI of SI, UTI, NP and B, in global form, in specific form and according to the NNIS risk index. The system for surveillance of hospital infections in our hospital has involved classification of patients in terms of the NNIS index since 1997, with a cut-off time of 120 min as ponderal measurement for all operations, since at the start of the surveillance system, the casuistry and the lack of information about the mean times for each operative procedure did not allow fitting the cut point for each operative procedure category. We compared the distributions of SI, UTI, NP and bacteremia among the collected variables and examined possible associations by the two-tailed Student's t-test or the Kruskall±Wallis test for quantitative variables and by the chi-squared test with the Yates correction or by Fisher's exact test for qualitative variables.
As in the case of crude measurements of the magnitude of association for qualitative variables in two categories, we used the relative risk (RR) in the univariate study and we used odds ratios (OR) and 95% confidence intervals (CI95) (as estimators of RR) in the LR analysis according to Cornfield's method. Statistical significance was accepted for P < 0.05, with a 95% confidence limit. In LR analysis, the dependent variables were SI, UTI, NP and B and the independent variables were age, sex, duration of operation (with a cut-off time of 120 min), duration of hospital stay prior to surgery, antibiotic prophylaxis (received or not), surgical class (contamination grade of the operation), operative procedure, ASA classification, emergency procedure, days with a urinary catheter (measured until infection), total days with mechanical ventilation during and post-surgery (measured until infection), and total days with a central venous catheter (measured until infection).
Results The members of our cohort ranged in age from 4 and to 98 years (58.8 17.3; median 63). There were 1129 males (40.4%) and 1665 females (59.6%). Among the admissions, 80.8% were programmed and the 19.2% were emergencies. All the patients were classified according to an index of intrinsic risk, assessed in terms of the ASA classification, as follows: ASA-I, 756 patients (27%); ASA-II, 1196 patients (42.8%); ASA-III, 701 patients (25.1%); ASA-IV, 139 patients (5%); and ASA-V, 2 patients (<1%). A total of 816 patients (29.2%) had a urinary catheter for a mean duration of 1.5 4.3 days (maximum, 60 days). Among the total cohort, 145 (5.2%) needed mechanical ventilation for a mean of 0.2 1.9 days (maximum, 55 days), and 339 patients (12.1%) of the cohort had a central venous catheter for a mean of 1.0 3.4 days (maximum, 41 days). During their stay in the hospital, 314 patients had some type of hospital infection (i.e., the global incidence was 11.2%). The most frequent nosocomial infection in the cohort was SI with a rate of global CI of 7.7%, ranging from 3.5% for clean surgery to 23.7% for dirty surgery, followed in frequency by B (1.5%), then UTI (1.5%), and finally NP with a global AI of 0.5%. The global CI of SI varied between 11.7% in 1995 and 4% in 1998 (Table I). The global CI of NI in terms of NNIS index is shown in Table II. In 775 cases (27.7%), the duration of surgery was 120 min or longer (89.7 62.8 min; median, 84.5; maximum, 420 min). Open cholecystectomy (16.7%),
Infection Risk Factors in a General Surgery Service colorectal surgery (16.4%) and herniorrhaphy (14.7%) were the most frequent operative procedures (Table III). In 75.5% of cases, the hospital stay prior to surgery was less than 5 days. The mean stay prior to surgery of the cohort was 4.8 10.4 days and the mean postsurgery stay was 8.1 7.8 days. The results of LR analysis with odds ratios and their confidence intervals for the different types of NI are shown in Table IV. Male gender multiplied the risk of SI by 1.54; each 3 days of stay prior to surgery multiplied the risk of SI by 1.06; each 10 min of operation time multiplied the risk by 1.03; emergency procedure multiplied the risk by Table I. Cumulative incidence of nosocomial infection. Year
SI UTI NP B
1995 (%)
1996 (%)
1997 (%)
1998 (%)
Total (%)
11.7 2.1 0.5 1.7
9.0 1.2 0.7 2.0
5.5 0.8 0.5 1.1
4.0 1.9 0.2 1.3
7.7 1.5 0.5 1.5
19
1.49; clean-contaminated surgery multiplied the risk by 0.80; contaminated surgery multiplied the risk by 1.95; dirty surgery multiplied the risk by 4.0; an ASA classification higher than 2 multiplied the risk by 1.76 and each day with a urinary catheter multiplied the risk by 1.04. The antibiotic prophylaxis appears as a protection factor against SI. With respect to UTI, clean-contaminated surgery multiplied the risk of UTI by 3.42, contaminated surgery multiplied the risk of UTI by 2.96 and dirty surgery multiplied the risk by 2.7; every day with a urinary catheter multiplied the risk of UTI by 1.05; and an ASA classification higher than 2 multiplied the risk of UTI by 2.13. With respect to NP, each day of hospital stay prior to surgery multiplied the risk of NP by 1.04; each day with mechanical ventilation multiplied it by 1.14 and an ASA classification higher than 2 multiplied it by 5.93. Each day of hospital stay prior to surgery multiplied the risk of B by 1.03; each minute of operation time multiplied the risk by 1.01; the emergency admission multiplied it by 4.24; an ASA classification higher than 2 multiplied it by 4.72; and each day with CVC multiplied it by 1.15.
Table II. Global cumulative incidence of nosocomial infection by NNIS index. NNIS
SI No SI
UTI SI
NP
B
Total
AI (%)
No UTI
UTI
Total
AI (%)
No NP
NP
Total
AI (%)
No B
B
Total
AI (%)
0 1 2 3
1256 811 362 150
29 60 67 59
1285 871 429 209
2.26 6.89 15.62 28.23
1274 857 422 199
11 14 7 10
1285 871 429 209
0.86 1.61 1.63 4.78
1285 866 424 205
0 5 5 4
1285 871 429 209
0.00 0.57 1.17 1.91
1285 866 410 190
0 5 19 19
1285 871 429 209
0.00 0.57 4.43 9.09
Total
2579
215
2794
7.70
2752
42
2794
1.50
2780
14
2794
0.50
2751
43
2794
1.54
Table III. Distribution of the cohort by NNIS index and operative procedure. Operative procedure
NNIS index 0
Mastectomy Thyroidectomy Esophagus-gastric surgery Colorectal Surgery Appendectomy Open Cholecystectomy Acute Cholecystitis Laparoscopic Cholecystectomy Herniorraphy Other procedures Total
1 186 242 10 0 82 227 4 230 276 28
(48%) (64%) (38.5%) (0%) (54.6%) (58.7%) (3.8%) (7.03%) (67%) (30%)
1285 (45.99%)
166 123 10 62 52 182 29 86 120 41
2 (44%) (32.5%) (38.5%) (13.5%) (34.6%) (39%) (27.8%) (26.3%) (29%) (44%)
871 (31.18%)
26 13 4 237 16 48 42 11 15 17
3 (6.8%) (3.5%) (5.4%) (51.6%) (10.8%) (10.3%) (40.6%) (3.4%) (3.6%) (18.5%)
429 (15.35%)
1 0 2 160 0 9 29 0 1 7
Total (0.2%) (0%) (7.6%) (34.9%) (0%) (2.1%) (27.8%) (0%) (0.3%) (0.3%)
209 (7.48%)
379 378 26 459 150 466 104 327 412 93
(13.5%) (13.5%) (0.9%) (16.4%) (5.4%) (16.7%) (3.7%) (11.7%) (14.7%) (3.3%)
2794 (100.00%)
20
P. Vazquez-Aragon et al.
Table IV. Coefficients, standard errors, odds ratios, 95% confidence intervals, and P values for variables included in the logistic regression analysis of nosocomial infection. Variable
Coefficient
SE
OR
CI95
P
ÿ4.0841 ÿ4.5226 ÿ2.5914 ÿ3.7478 ÿ3.7671 ÿ1.6997
2042.2 1.5509 7.1581 1.0290 2.0809 1.9926
1.54 1.06 1.03 1.49 0.84
1.13±2.09 1.01±1.08 1.00±1.07 0.99±2.24 0.57±1.25
0.0035 0.0003 0.0003 0.0703 0.3937
ÿ2.1759 ÿ8.9678 ÿ1.4943 ÿ6.5142 3.5214
2.5279 2.3781 2.5486 1.5631 1.5329
0.8 1.95 4 1.76 1.04
0.49±1.32 1.17±3.24 2.37±7.34 1.29±2.41 1.02±1.07
0.3894 0.0002 0.0000 0.0000 0.0123
ÿ5.4947 5.3234
4.2915 1.5402
1.05
1.02±1.09
0.0005
1.2287 1.0837 0.9947 7.5818
4.7893 5.3742 5.7178 3.3771
3.42 2.96 2.7 2.13
1.34±8.74 1.03±8.47 0.88±8.29 1.10±4.14
0.0103 0.0438 0.0819 0.0248
Pneumonia Constant Stay prior to surgery (at 1-day intervals) Mechanical ventilation (at 1-day intervals) ASA > II
ÿ6.7231 3.5619 1.3034 1.7798
6.1242 1.5723 3.4477 6.7783
1.04 1.14 5.93
1.00±1.07 1.06±1.22 1.57±22.38
0.0235 0.0002 0.0086
Bacteremia Constant Stay prior to surgery (at 1-day intervals) Duration of operation (at 1-min intervals) Emergency procedure ASA > II CVC (at 1-day intervals)
ÿ6.8251 3.3532 5.0045 1.4443 1.5516 1.4039
4.5578 1.1778 1.6995 3.5069 4.0221 2.0242
1.03 1.01 4.24 4.72 1.15
1.01±1.06 1.00±1.02 2.13±8.43 2.15±10.35 1.11±1.20
0.0044 0.0032 0.0000 0.0001 0.0000
Surgical infection Constant Gender Stay prior to surgery (at 3-day intervals) Duration of operation (at 10-min intervals) Emergency procedure Prophylaxis Surgical class Clean-contaminated Contaminated Dirty ASA > II Urinary catheter (at 1-day intervals) Urinary tract infection Constant Urinary catheter (at 1-day intervals) Surgical class Clean-contaminated Contaminated Dirty ASA > II
ASA > II: ASA III, IV or V.
A microbiological diagnosis was obtained in 67.2% of NI. Half of the cultured strains were Gram-positive cocci and 37% were Gram-negative bacilli. The most frequently isolated pathogens were Enterococcus faecalis (19.2%) and Escherichia coli (13.1%) for SI, Enterococcus faecalis (38.9%) and Candida albicans (22.2%) for UTI, Neumococcus (26%) and Staphylococcus aureus (17%) for NP and Staphylococcus epidermidis (21.6%) and other Staphylococcus for B. A total of 1422 patients (57.8%) received antimicrobial prophylaxis. The perioperative antibiotic prophylaxis was indicated in clean with implants, clean-contaminated and contaminated surgery patients following the antibiotic prophylaxis protocol from our hospital, with prophylaxis administration 1-h before surgery. Three hundred and twenty-four patients with clean surgery (with and without implants) received antibiotic prophylaxis. The mean duration of prophylaxis was 2.5 days, and cephoxitin was the most frequently
used antibiotic (18.7%). The antimicrobial agents used most frequently for prophylaxis were second-generation cephalosporins. The antibiotic association used most frequently was clyndamycin plus gentamicin. Sixty-five patients (2.3%) died during their stay in hospital. The ratio of the number of patients who died who had had a NI to the number of patients with a NI was 31/314 (9.9%). Of 65 patients who died, 16 had a SI, namely, approximately one-quarter of the total.
Discussion Among the 2794 patients in the cohort, 1162 (41.6%) were more than 65 years old and 223 (8%) were more than 80 years old. It has been estimated that, in the USA alone, 27 million surgical procedures are performed annually and almost a third of the patients subjected to these procedures are 65 years of age or older.
Infection Risk Factors in a General Surgery Service In our series, the large number of patients (816; 29%) with a urinary catheter seems very high in relation to data published (between 5% and 17%) in other reports [6,7]. This difference might be due to the large number of patients of more than 65 years of age who needed a urinary catheter and, also because insertion of a catheter is included in the preoperative protocol for specified surgical procedures. The NNIS index appears to be the best indicator of intrinsic risk of SI, being apparently better than the traditional classification in terms of the contamination grade of the surgical wound [8]. Although it was initially developed to predict SI risk, the index also predicts reasonably well the risks of postoperative infections in other sites. The success of the NNIS index in predicting the risk of developing postoperative pneumonia, bloodstream or urinary tract infections suggests that the index is a general expression of the propensity to develop infection and probably reflects the general nature of the risk factors that comprise the index [5]. The CI of NI was 11.2%, resembling the published results for other series (1.4±17.3%) [9,10]. Our rates of SI were higher but, with respect to other NI such as UTI, NP and B, our data are similar and, in some cases, incidences are even lower than those reported by Emori et al. [10]. The system for surveillance of hospital infections in our hospital has involved classification of patients in terms of the NNIS index since 1997, with a cut-off time of 120 min for all operations. However, if we had considered times equal to or above the 75th percentile of the duration of surgery, cases included in groups 1, 2 and 3 would have been classified as 0, 1 and 2, respectively. This error in classification leads to an overestimate of the rates of infection in the lowest groups. In the total cohort, only 775 (27.7%) operations lasted 120 min or longer. A remarkable decrease in the incidence of SI is shown apparent over the course of the study, probably as a result of the establishment of a system for surveillance of NI in our hospital and of an improved understanding by surgeons of the infection rates associated with different types of surgery, as well as the more frequent and more appropriate use of antimicrobial prophylaxis protocols including the change in administration time 1-h before surgery. In this way, the antibiotic prophylaxis appears as a protection factor against SI. The hospital stay prior to surgery is another factor to take in account in the decrease of the SI. It is widely accepted that the hospital stay prior to surgery is a risk factor for NI [11]. The mean stay prior to surgery of our cohort was 4.8 days, which seems long since 19.2% of admissions were emergencies, 70% of patients were
21
classified as ASA II and, since 1997, a day-surgery unit has been in use, with almost all admissions occurring on the same day as surgery. Thus, the lengthy stay prior to surgery might be explained by the fact that, some years ago, preoperative studies and diagnostic tests were performed after the patient had been admitted to hospital. The duration of a surgical intervention of more than two hours is accepted as a risk factor for SI. In our cohort, as in other series, the mean duration of surgery for patients with SI was of 31.7 min longer than for noninfected patients. Patients with a central venous catheter and those with a urinary catheter had higher rates of SI. In our series, the incidence of SI in diabetic patients was not significantly different from that in non-diabetic patients. The ASA score is itself an index, designed to asses preoperatively the overall physical status of the patient and constitutes a critical component of the NNIS index, included in an attempt to measure intrinsic patient susceptibility and has the advantage of being readily available at the time of surgery [5]. Culver et al. [5] showed that ASA score as a single predictor of SI risk was at least as good as the traditional wound classification. In our study, an ASA classification higher than 2 appears as a risk factor in every studied infection, being this increase of the risk more important in NP and B. This fact clearly confirms this score as a predictor of nosocomial infections other than SI in surgical patients. A noticeable decrease was also observed for the other types of NI over the course of the study, and all these factors appear to explain this decrease in NI between 1995 and 1998. The use of prophylaxis in clean surgery is controversial. Almost one quarter of the Spanish surgeons use antibiotic prophylaxis with any type of clean surgery, according to the National Survey of Surgical Infection carried out in 1995 by the Spanish Association of Surgeons [12]. In our series, 28% of the patients with clean surgery received prophylactic treatment. The cited survey indicated that, as in our series, the antibiotics used most frequently for prophylaxis were first- and second-generation cephalosporins, which are used by 72% of the Spanish surgeons. Multivariate analysis revealed three variables that affected the risk of UTI: the contamination grade of the surgery; the number of days with an urinary catheter and the primary health status of the patient. The principle ``if possible, do not catheterize'', as well as established protocols of asepsia and careful assessment daily of the necessity for the urinary catheter, with its removal as soon as possible, constitute the basis for prevention of UTI.
22
P. Vazquez-Aragon et al.
With respect to NP, mechanical ventilation (which facilitates the aspiration of polluted droplets), together with poorer health status, might predispose a patient to NP. A direct program for prevention of NP should focuse on decreasing, as far as possible, the hospital stay prior to surgery of the patient and on decreasing as far as possible, the number of days with mechanical ventilation. Increases in the number of days with CVC increase the possibility of colonization, both extrinsically and intrinsically, of the system and depending on the patient's status, can even facilitate the development of bacteremia. A program for prevention of bacteremia should focuse on decreasing, as far as possible, the hospital stay prior to surgery, the establishment of preoperative preparation protocols, and a decrease, as far as possible, in the number of days with CVC. We hope that implementation of the procedures that we suggest will help to reduce the frequency of NI on surgical services.
References 1 Garner JS, Jarvis WR, Emori TG, Horan TC, Hughes JM. CDC definitions for nosocomial infections, 1998. Am J Infect Control 1988; 16: 128±140. 2 Haley RW, Culver DR, White W, Emon TG. The nationwide nosocomial infection rate. A new need for vital statistics. Am J Epidemiol 1985; 121: 159±167.
3 Grupo de Trabajo EPINE 98. Estudio de prevalencia de las infecciones nosocomiales en los hospitales espanÄoles. 9 estudio. Madrid: Sociedad EspanÄola de Medicina Preventiva, Salud PuÂblica e Higiene. 1998. 4 Haley RW, Culver DH, Morgan WM, Emori TG, Munn VP, Hooton TM. Identifying patients at high risk of surgical wound infection: a simple multivariate index of patient susceptibility and wound contamination. Am J Epidemiol 1985; 121: 206±215. 5 Culver DH, Horan TC, Gaynes RP, Martone WJ, Jarvis WR, Emori TG, Banerjee SN, Edwards JR, Tolson JS, Henderson TS, Hughes JM. Surgical wound infection rates by wound class, operative procedure, and patient risk index. Am J Med 1991; 91: 152S±157S. 6 Shapiro M, Simehen E, Izvaeli S. Multivariate analysis of risk factors for acquiring bacteriuria in patients with indwelling urinary catheters for longer than 24 hours. Infect Control 1984; 32: 525±532. 7 Vasile I, Mogos D, Paun I, Paun M, Florescu M. Urinary infections with hospital germs in general surgery. Chirurgia 1998; 93: 317±321. 8 Horan TC, Culver DH, Gaynes RP, Jarvis WR, Edwards JR, Reid CR. Nosocomial infections in surgical patients in the United States, January 1986±June 1992. Infect Control Hosp Epidemiol 1993; 14: 73±80. 9 Cruse P, Foord R. The epidemiology of wound infection. A 10-year prospective study of 62,939 wounds. Surg Clin North Am 1980; 60: 27±40. 10 Emori TG, Culver DH, Horan TC. National Nosocomial Infections Surveillance System (NNIS): description of surveillance methods. Am J Infect Control 1991; 19: 19±35. 11 Bueno Cavanillas A, RodrõÂguez-Contreras R, Delgado RodrõÂguez M, Moreno Abril O, LoÂpez Gigosos R, Guillen Solvas J, GaÂlvez Vargas R. Preoperative stay as a risk for nosocomial infection. Eur J Epidemiol 1991; 7: 670±676. 12 Cainzos M, Lozano F, Alcaraz P et al. Profilaxis antibioÂtica en cirugõÂa limpia. Cir Esp 1996; 59: 7±10.