jhrnal
of Hospital
Prevalence
Infection
(1992) 20, 1-13
of hospital-acquired EPINE
Working
infections
in Spain
Group*
Accepted for publication
14 October 1991
Summary: In May 1990 a prevalence survey of hospital-acquired infections was conducted in 123 Spanish hospitals, in which 38 489 patients were studied. There was an 8.5% prevalence of infected patients and a 9.9% prevalence of infections. The most common infections were those of the urinary tract (27.7%), surgical wound (22.7%) and lower respiratory tract (15.4%) and bacteraemia (10.6%). Th ere was a 5.9% prevalence of patients with surgical wound infection and 3.5% after clean surgery. An aetiological diagnosis was made in 58% of the infections. Gram-negative bacteria were dominant, Escherichia coli (16.3%) and Pseudomonas aeruginosa (11.5 %) being the most prevalent; 33.8% of the patients were receiving antimicrobial agents. The following procledures were shown to be significantly associated with hospital-acquired infections: urinary catheterization, parenteral nutrition, mechanical ventilation and tracheostomy. The degree of contamination during surgery was also a significantly associated risk factor. The survey provided extensive information on the distribution of infections and the use of antibiotics in clinical services, as well as the differences between hospitals according to their size and .the presence of certain risk factors. Keywords: Spain.
Prevalence
survey;
hospital-acquired
infection;
antibiotic
use;
Introduction Hospital-acquired infections (HAI) pose a problem currently attracting considerable attention, with great human, social and economic importance, due to their associated morbidity and mortality. In many hospitals, prevalence studies are part of the routine HA1 surveillance, since they are easy to perform, provide general descriptive information on the problem and are helpful in evaluating the impact of control measures.lm3 *EPINE Writing and Steering Committee: Josep Vaqut, Vail d’Hebrdn Hospital, Universitat Auton&m, Barcelona; Vicente Mange, Hospital Ram& y Cajal, Madrid; Juan Garcia Caballero, Hospital La Paz, Madrid; Josk Rossell6, Vail d’Hebr& Hospital, Universitat Auto&ma, Barcelona; Jo& Barrio, Hospital de Sant Pau, Barcelona; Francisco Calbo, Hospital Carlos Haya, Mblaga; M. Carmen Sienz, Hospital Clz’nico Uniuersitario, Salamanca; A. Hernindez, Hospital La Fe, Vale&a; Plicido L6pez Encinar, Hospital Clinico Universitario, Valladolid; J. Luis Vaquero, Hospital de1 Rio Hortega, Valladolid; Jo& R. SLenz, Hospital N.S. Aranzazu, San Sebastidn; Vicente Dominguez, Hospital Juan Canalejo, La CoruKa; Rosa Coello, Hospital Chico Universitario, Madrid; Pedro Blasco, Hospital Virgin Macarena, Sevilla; J. Luis Arribas, Hospital Miguel Servet, Zaragoza; A. Gonzilez Torla, Hospital N.S. Covadonga, O&do; Maruja Santos, Hospital La Fe, Vale&a; Rafael Herruzo, Hospital La Paz, Madrid; Spain. Correspondence to: Dr J. VaquC, Barcelona, Spain. 0,9556701/92/010001
+I3
Servei
de Medicina
Preventiva,
tO3.00/0
Ciutat
Sanitlria
0
1
1992
Vall d’Hebr&,
The
Hospital
Infection
08035,
Society
2
EPINE Working
Group
Prevalence surveys performed in countries or large geographical areas define the overall situation and, when the sample is comprehensive or sufficiently representative at hospital level, may indicate differences between hospitals. The World Health Organisation promoted a prevalence study between 1983 and 1985, in which data from 47 hospitals in 14 countries4 were analysed. In Europe, multicentre prevalence surveys have been conducted in Sweden,’ Denmark,6 Norway,7 England and Wales,’ ” In Spain, a pilot study was carried Italy,’ Belgium” and Czechoslovakia. out in 1986, in which a group of hospital patients from virtually all regions was studied; l2 later, a second study was carried out in Catalonia.i3’i4 The European prevalence study of urinary tract infection and bacteraemia as well as the national studies performed in should also be mentioned,” Australiai and Thailand.i7 A prevalence survey was conducted in a large group of Spanish public acute-care hospitals, in order to study HAI and their associated factors, as well as community-acquired infections (CAI) and the use of antimicrobial agents.
Patients
and methods
All patients hospitalized in the centres participating in the investigation were included. Data were collected from the clinical records and information provided by the doctors and nurses in charge of the patients, in addition to bedside confirmation of extrinsic risk factors. This task was carried out at each centre by the HAI surveillance and control team, as constituted in the individual hospitals on the advice of the Infection Control Committee. The survey was conducted during a 2-week period in May 1990. At each hospital one person was charged with coordinating and reviewing the study; usually this person was a medical epidemiologist, less frequently a medical microbiologist, infectious diseases clinician or infection control nurse. The study protocol’8 had been widely tested in previous surveys,‘3,‘4 and all its variables were explicitly defined. The intrinsic risk factors studied were: coma, renal failure, diabetes mellitus, neoplasia, immu”iodeficiency, neutropenia, cirrhosis, hypoalbuminaemia, pressure sores, alcoholism, smoking and drug addiction. Extrinsic factors, including those applied during the stay in hospital included urinary catheterization, peripheral vascular catheterization, central vascular catheterization, central catheterization with peripheral insertion, parenteral nutrition, tracheostomy and mechanical ventilation. Information was also compiled about the surgical operations performed. The baseline risk of the patient was classified according to three categories of clinical prognosis: severe (survival of less than a year); moderate (life expectation between 1 and 5 years); and mild (the rest of the cases).
Nosocomial
infection
in Spain
3
The criteria of the Centers for Disease Control, Atlanta, USAl were used for defining infections, except that asymptomatic bacteriuria was not accepted as a criterion for urinary tract infection. A patient with more than one site of infection was considered to have multiple infections; the survey data forms provided room for four infections per person. The date of onset of the infection (the day on which the symptoms that led to microbiological investigation appeared, or antibiotic therapy was begun) was recorded. The results of microbiological and serological investigations were usually available within 7 days following the end of the clinical investigation; those not available were separately categorized. We used two measures of the frequency of HA1 or CAI: the prevalence of infections (the ratio between the number of clinically active infections and the total number of patients studied, expressed as a percentage) and the prevalence of infected patients, which is the percentage of patients with active infection in relation to the total number of patients studied. Confidence intervals were calculated and hypotheses were tested on the basis of infected patient prevalence. Since the stays of patients in obstetric services are usually short and their infection levels are low, the prevalence of infections and of infected patients were also calculated after exclusion of these patients. The prevalence of patients receiving antibiotics was the proportion of patients receiving these drugs at the time of the survey in relation to the total number of patients. Length of stay prior to infection was defined as the number of days between admission and the onset of the infection. The variable ‘total number of intrinsic risk factors’ was calculated by adding up the number of factors of this type for each patient. A similar accumulative variable was also defined for extrinsic factors. The crude association between a risk factor and the presence of infection was assessed by the Odds Ratio; its confidence interval was estimated by the Cornfield method.20 The combined study of the possible risk factors involved in the development of HA1 was conducted by means of multiple logistic regression models. In these models, the presence or absence of nosocomial infection, overall or for the main different infection sites, was the dependent dichotomous variable. Those patients with an infection included in the study were considered cases, and those without HA1 controls. In these models we used the variable ‘length of stay prior to infection’ for patients with HA1 and the interval between admission and date of entry to the study for those without. In this way, we avoided using the interval between the onset of infection and the day of study, which is influenced by the prolongation of stay due to infection. One hundred and twenty-six hospitals participated in the investigation, and in total 38 857 patients Twere studied. Of these, 123 hospitals met the quality requirements and 38 489 patients were eligible for overall assessment, and these are the basis of this report. Since 135 centres were invited to participate in the study, the proportion of participants can be
4
EPINE Working
Group
considered very high (93.3%). The hospitals were classified into three categories: small hospitals (fewer than 200 beds, 50 centres, 40.7%); medium-sized hospitals (between 200 and 500 beds, 46 centres, 37.4%); and large hospitals (over 500 beds, 27 centres, 21.9%). Hospital services were classified into six large areas: medicine, surgery, obstetrics and gynaecology, paediatrics, intensive care, and other services, which last included psychiatry, geriatrics etc. The intensive care area included the ICU, the Coronary Care Unit and the Transplantation Units for adults. Results
Prevalence of HAI
and CAI
In the initial assessment of the whole group of 38 489 patients 3800 HA1 and 5917 CA1 were detected (Table I). The overall prevalence of HA1 was 9.9%, and that of infected patients 8.5% (95% CI 8.2-8.7). The overall prevalence of CA1 was 15.4% and that of patients with infection 13.4% (95% CI 13-l-13.7). Th e median prevalence of HA1 in the 123 centres was 9.4% and that of CA1 16.1%. After exclusion of the patients of the obstetric services, the general HA1 prevalence rose to 10.3% and that of infected patients to 8.8% (a non-significant difference compared with the overall prevalence, x2 = 3.69, P= 0.054); with respect to CAI, the prevalence of infections rose to 16.1% and that of infected patients to 15% (a significant difference compared with the overall prevalence, x2 = 6.61, P= 0.01). The lowest prevalence of patients with HA1 was recorded in the small hospitals (7*9%), while large and medium-sized centres recorded a similar prevalence (no significant differences between the three types of hospital, x2= 2.67, P= 0.26; nor after excluding patients in obstetric services,
x2= 3.92, P=O.14). In 33 hospitals (26.8%) the HA1 prevalence was below 7%; in 27 centres (22%) it ranged between 7% and 9%) and in 63 centres it was over 9%. The HA1 prevalence in the ICU was high: 48.7% in adults and 26.2% in paediatrics. In the area of medicine, the prevalence was high in haematology (15.2%) and rehabilitation (17.3 %) services; neurology (13.9%) and nephrology (12.1 “A) services were also above the general Spanish prevalence rate. In the surgical area there were also some services with HA1 prevalences above the general percentage, but not very high: the adult burns unit (12.9%), neurosurgery (12.9%) and general surgery (11.7%). The average age (f SD) among patients with HA1 was 49.15 years (f 25.36); 55.1% of the cases were men and 44.9% women. The highest prevalence of patients with HA1 was noted in children aged 1 to 10 year old stratum had the lowest prevalence (4.5%). From this category onwards, there was a linear significant trend of increase for the prevalence (Mantel-Haenszel extended test, x2= 113.4, P75 Stay prior to infection: 8-i4 days 15-21 21-28 29-35 36-42 43-60 >60 Community-acquired infection Type of hospital: medium-sized large
Odds Ratio
infections (95%
CI)
1.8 2.9 7.1
(1.6-2.1) (2.6-3.4) (5.7-8.9)
1.1 I.4 1.6 1.3 1.8
(0.9-1.2) (1.2-1.6) (1.2-1.9) (1.1-1.5) (1.5-2.2)
1.6 I.8 2.2 1.8 1.1
(14-1.8) (1.6-2.1) (1.8-2.6) (1.6-2.0) (1.0-1.2)
I.0 1.0 1.1 0.9 1.1
(0.8-l .3) (0.8-1.3) (0.9-1.3) (0.8-1.2) (091’4)
I.5 1.9 2.6 2.6 2.5 2.9 2.9 0.5
(1.331.7) (1.7-2.3) (2.1-3.1) (2.1-3.2) (1.9-3.3) (2.4-3.8) (2.3-3.7) (0.4-0.6)
I.1 1.1
(0.9-1.3) (0.9-1.2)
Comparison categories: absence of extrinsic risk factors, absence of intrinsic risk factors, mild baseline risk, number of diagnoses = 1, absence of surgical operation, male sex, age = 15-30 years, hospital stay prior to the infection = l-7 days, patient without CAI, type of hospital = small.
.
Nosocomial
categories. The two categories presence of CA1 was protective. Microbiological
infection
in Spain
9
of hospital size were not significant. The None of the age categories was significant.
diagnosis
The microbiological diagnosis (culture or serological) was available in 58% of the HA1 and in 41% of the CAI. Among HAI, 57.8% had positive cultures and 14.1% had negative cultures, while in 24.7% cultures had not been carried out and in 3.1% the results were too late for inclusion (see above); in only 0.3% of cases was serological testing carried out. Among CAI, 30.1% had positive cultures and 20.8% had negative cultures, while in 30.6% cultures had not been carried out and in 7.5% the results were too late for inclusion. Serological testing was more important in CAI, accounting for 11% of cases. Overall and for each infection site the proportion of positive cultures in relation to the total number performed was lower in CA1 than in HAI. UT1 was the type of HA1 with the highest percentage of positive cultures (67% of the cases), followed by BACT (64.7%), SW1 (60%), LRI (49.1%) and the group composed of other sites (48.3%). Gram-negative organisms, were identified in 43.9% of the cases and Gram-positive in 29.4%. The aetiological agents were identified in 50.8% of HA1 and 49.2% of CAI. Table IV shows the classification of the isolated organisms. Among HA1 the most frequent Gram-negative pathogen was Escherichia coli (16.3% of total), followed by Pseudomonas aeruginosa (11.5%); the most frequent Gram-positive organism was Staphylococcus aweus (11.2%). Staphylococcus epidermidis and other coagulase-negative staphylococci also exceeded 5 %. Escherichia coli was also the most prevalent organism in CA1 (14.1%), with human immunodeficiency virus (HIV; antibody detection, 12.6% of positive diagnoses by culture or serology) in
Table Groups
IV. Microbiological
diagnoses
(culture
or serology)
and main organisms
and main pathogens
Type
of infection
HA1
CA1
Total Groups Gram-negative bacteria Gram-positive bacteria Anaerobes Other microorganisms Main pathogens E. coli P. aeruginosa Proteus mirabilis S. aweus Human immunodeficienc virus Enterococcus spp.
No.
(%I
No.
W)
No.
(%)
2377 1591 138 1307
(43.9) (29.4)
1392 1030 81 245
(50.7) (37.5) ( 2.9) ( 8.9)
985 561 57 1062
(37.0) (21.0)
823 413 157 488
(15.2)
(16.3) (11.5) ( 3.7) (11.2)
376 97 56 181
(14.1) ( 3.6)
( 9.0)
447 316 101 307
; ;:;I
( 5-3)
212
( 7-7)
335 74
(12.6) ( 2.8)
;2:::j
; ;:;;
13;:;;
:Y 335 286
10
EPINE Working
second place. Mycobacterium more than 5% of the cases.
tuberculosis
Group
and S. auwus were also isolated
in
Antibiotics At the time of the study, 13 007 patients were receiving antibiotics, an overall prevalence of 33.8% (Table V). Forty-three percent of the antimicrobial agents were prescribed in surgery and 32.6% in medicine. The area with the greatest prevalence was intensive care (52.2%). Eighty-three percent of the patients with HA1 were being treated with antimicrobial agents, a proportion roughly similar to that in CA1 (80%). In the group of patients without infection, 21.1% were receiving antibiotics. The prevalence of antibiotic usage was higher in small (35.1%) than in medium-sized hospitals (33*8%), and higher in the latter than in large hospitals (33.1%) (significant differences between the three types of hospital, x2= 14.62, P= 0.0006). The more commonly prescribed antimicrobial agents were aminoglycosides (17.8%), followed by aminopenicillins (14.1%) and third generation cephalosporins (11.2%). The most widely used product was gentamicin: general use 8.8%; 8.2% in HAI, 7.5% in CA1 and 9.8% in patients without infection. Discussion
In the number of patients studied, this is the largest multicentre prevalence study yet published. +17 The overall HAI prevalence observed (9.9’/,) is an intermediate value, between the extremes observed in the studies shown in Table VI. The results of the present study show a considerable Table
V. Prevalence of antibiotic use according service and type of infection
Service and type of infection
Hospital service Medicine Surgery Intensive care Obstetrics & gynaecology Paediatrics Other services Total Types of infection Hospital-acquired Community-acquired Both types No infection * P%, Prevalence.
Prescribed antibiotics
to hospital
Patients with antibiotics
N
(%)
N
P%*
6121
(32.6)
y;:;
7::;;
3954 5976 558
32.7 37.1 52.2
:;2 212 18 780
&cl; ( 1.1) (100.0)
1048 1312 159 13 007
22.4 37.1 15.7 33.8
2698 4128 229 6410
82.9 SO.1 91.6 21.1
4365 6643
g:;;
Nosocomial Table Country
VI. Prevalence
and reference
Sweden5 Denmark6 Norway’ England & Wales* Europe” (8 countries)” Italy9 14 Countries, WHO4 Belgium’“,b Australia16 Spain12 Czechoslovakia” Catalonia’3,‘4 ThailandI Spain (this study)
infection
studies in several
11
in Spain countries
Year
Number of hospitals studied
1975 1978179 1980 1’980 1982 1983 1983-85 1984 1’984 1’986 1988 1988 1988 1990
5 2512.5 15 43 40 130 47 106 269 39 23 33 23 123
and WHO Number of patients studied 3657 1363/1X7 7833 18 163 3899 34 577 28 861 8723 28 643 10 470 12 260 7434 6805 38 489
study Prevalence of HA1 per 100 patients 10.5 10.5/12.1 9.0 9.2 7.3 6.8 9.9 10.3 8.1 11.2 6.1 12.1 11-7 9.9
aOnly urinary infections and bacteraemia were studied. b Only urinary and surgical infections and bacteraemia were studied.
improvement in comparison with those obtained in the general Spanish the two studies were conducted by very different study of 1986, l2 although methods and with different definitions. HAIs were greatly concentrated in the intensive care units and to a lesser degree in certain paediatric, surgical and medicine services. Small hospitals had a lower prevalence, as was to be expected, in view of the fact that more serious pathologies are seen, and more complex and aggressive therapies are given, in medium-sized and large centres. The CA1 prevalence was Ihigh, with considerable impact in the areas of medicine and paediatrics; these infections, however, were the main diagnosis in only 5% of the patients. From an epidemiological perspective (not a biological one), the presence of CA1 protects against the development of HAI. This could possibly be due to the CA1 patients receiving antibiotics after admission and therefore being protected against a nosocomial infection, or alternatively, a distortion could exist, due to under-reporting of patients with both types o’f infection. The distribution of HA1 by site was that commonly seen. There were, nevertheless, marked differences from the WHO study4 in which the percentages were: UT1 22.1%; SW1 25.1%; LRI 20*6% and BACT 3.7%. In our study, there was a high proportion of primary (7.6%) and secondary (3 %) bacteraemia, which was related to the significant proportion of severely ill and immunosuppressed patients (43% of the BACT were diagnosed in intensive care, neonatal and haematology units). In the logistic model of each infectious site, specific instrumentation procedures showed a high degree of significance. In surgical wound infections, the degree of contamination has a determinant effect. According to traditional concepts, the crude analysis showed that the older the patient,
12
EPINE Working
Group
the greater the risk of nosocomial infection. However, the logistic adjustment indicated that increased risk could be due to the concomitant circumstances of age, such as the baseline risk and the number of diagnoses. The idea of using the variable number of diagnoses was suggested by Haley et a1.21: the three categories were significant in the overall model and its behaviour is similar to that of the baseline risk. While the latter helps to define how seriously ill the patient is, the number of diagnoses can explain the complexity of the baseline situation and the multi-system extension of the process. Two categories of the number of intrinsic factors showed statistical significance in the overall model. This variable could explain part of the impact of multisystem involvement. With respect to the microbiological studies, our results showed that the percentage of culture requests and the percentage of positives of the United States22 were not attained; however, in relation to other European studies6-9”2 the comparison is more favourable. The pattern of isolated pathogens was roughly similar to that seen in other studies,4,“,‘4,23 with the exception of the outstanding current role of HIV and Ad. tuberculosis. Worthy of note is the preponderance of Gram-negative over Gram-positive bacteria in both HA1 and CAI. The overall prevalence of antibiotic use was higher than in other multicentre studies.4,6,‘4,‘5,24 Prevalence studies are affected by an over-representation bias due to patients with long-term hospitalization and they are not adequate for establishing causal inferences. They are, nevertheless, easy to perform, very versatile and the results that they provide are of undoubted descriptive interest for understanding the general dimension and features of the problem of nosocomial infections. This survey was sponsored by the Spanish Association of Hospital Hygiene and Preventive Medicine, with the technical collaboration of Laboratorios Beecham S.A. Statistical analysis was performed by J. Vaqut, J. Rossello and L. Armadans. The coordinators in the individual hospitals were as follows. Anduhci’u: M. Torres, I. Fernindez, F. Rodriguez, J. M. Bautista, S. Oiia, D. Roman, R. Gilvez, R. Rodriguez-Contreras, S. Oliver, J. L. Martin Ruiz, F. Lopez, J. Lorente, P. Alamillos, M. de Cueto, R. Fernandez, S. Duefias, R. Creagh, D. Sureda. Arugdn: P. Muniesa, P. Prieto. Asturias: E. Tui%n, T. Fernandez, P. Prendes, A. Torreblanca. Baleares: 0. Hidalgo. Canarias: G. Lopez, J. Alamo. Custillu-La Muncha: J. Martinez, A. Biurrun, E. Ruiz, J. Medrano. Cast&z-Ledn: J. de la Lama, N. Coladas, J. Moro, F. Benito, M. Mendo, A. Sacristan, G. Fernindez, A. Hita, P. Vicente, E. Robles, J. M. Olona, B. Bermejo, J. M. Bayas, M. Gurgui, L. Blanco. Cutulunyu: M. Campins, Drobnic, J. M. Garces, E. Plass, N. Sanjuan, N. Miserachs, M. DurQn, M. Esteve, C. M. T. Martinez, L. R/loner, R. Priu, A. Brotons, F. Rossell, I. Garcia, A. Casagran, Verdaguer, L. Gurgui, J. Alijotas, R. Pujol, M. Oliveras, J. M. SBnchez, X. Puig, J. Brau, F. Ribas, A. Anglada, A. Vilamala, J. Baucells, C. Clemente, T. Puig, J. J. Guardiola, F. Ballester, A. Manonelles, V. Romeu, J. Galvany, J. Luna. Extremaduru: J. Jimenez, A. Cerrillo. Guliciu: J. J. Gestal, M. L. Cal, V. Souto, J. L. Lopez, B. Uriel, A. Blanco, M. Cueto. Madrid: A. Asensio, N. Valverde, J. M. Garcia Paez, P. Gomez, A. Blanco, B. Perez, S. Sanchez, V. Valls. Murciu: F. Botia, J. Piqueras, A. Mas. Comunidud Vulenciunu: L. Ferreres, E. Monte, F. Asensi, S. Cervellb, V. C. Zanon, C. Gonzalez, C. Oltra, J. F. Navarro. P&s Vusco: J. L. Novales, J. M. Arevalo, M. E. Elosegui.
Nosocomial
infection
in Spain
13
References 1. Casewell MW. Surveillance of infection in hospitals. J Hasp Infect 1980; 1: 293-297. 2. Crede W, Hierholzer WJ Jr. Surveillance for quality assessment: I. Surveillance in infection control success reviewed. Infect Control Hosp Epidemiol 1989; 10: 470-474. 3. French GL, Cheng AFB, Wong SL, Donnan S. Repeated prevalence survey for monitoring effectiveness of hospital infection control. Lancet 1989; 2: 1021-1023. 4. Mayon-White RT, Ducel G, Kereselidze T, Tikhomirov E. An international survey of the prevalence of hospital-acquired infection. J Hasp Infect 1988; 11 (Suppl. A): 43-48. 5. Bernarder S, Hambreaus A, Myrback, Nystrijm B, Sundelof B. Prevalence of hospital-associated infection in Five Swedish Hospitals in November 1975. &and J Infect Dis 1978; 10: 66-70. 6. Jepsen OP, Mortensen N. Prevalence of nosocomial infections and infection control in Denmark. J Hasp Infect 1980; 1: 237-244. 7. Hovig B, Lystad A, Opsjon H. A prevalence survey of infections among hospitalized patients in Norway. NIPH Annals 1981; 4: 49-60. 8. Meers PD, Ayliffe GAJ, Emmerson AM, Leigh DA, Mayon-White RT, Mackintosh CA, Stronge JL. Report on thse national survey of infections in hospitals, 1980. r Hasp Infect 1981; 2 (Suppl.): 141. 9. Moro ML, Stazi MA, Marasca D, Greco D, Zampieri A. National prevalence survey of hospital-acquired infections in Italy, 1983. J Hasp Infect 1986; 8: 72-85. 10. Mertens R, Kegels G, Stroobant A. et al. The national prevalence survey of nosocomial infections in Belgium, 1984. J Hasp Infect 1987; 9: 219-229. 11. Srimovi H, Bartonovi A, Bolek S, Krecmerovi M, Subertova V. National prevalence survey of hospital-acquired infections in Czechoslovakia. J Hasp Infect 1988; 11: 328-334. 12. Ministerio de Sanidad y Consumo. Encuesta epidemiologica de prevalencia en un punto de la infection hospitalaria y uso clinico de antimicrobianos en Espafia. Anilisis descriptive de resultados. Madrid, Direction General de Planificacibn Sanitaria, 1986. 13. Grupo de Trabajo Epincat. Prevalencia de las infecciones nosocomiales en Catalufia. (I). Infecciones y factores de riesgo. Med Clin (Bore) 1990; 95: 41-52. 14. Grupo de Trabajo Epincat. Prevalencia de las infecciones nosocomiales en Catalufia. (II). Germenes y antimicrobianos. Med C’lin (Bare) 1990; 95: 161-168. 15. Jepsen OB, Larsen SO, Dankert J, et al. Urinary-tract infection and bacteremia in hospitalized medical patients. A European multicentre prevalence survey on nosocomial infection. J Hosp Infect 1982; 3: 241-252. 16. McLaws ML, Gold J, King K, Irwig LM, Berry G. The prevalence of nosocomial and community-acquired infections in Australian hospitals. MedJ Aust 1988; 149: 582-590. 17. Danchaivijitr S, Chokloikaew S. A national prevalence study on nosocomial infections 1988. J Med Assoc Thai 1989; 72 (Suppl.): l-6. 18. Proyecto Epine. Protocolo de1 estudio Epine1990. Madrid, Sociedad Espaiiola de Higiene y Medicina Preventiva Hospitalarias, 1990. 19. Garner JS, Jarvis WR, Emori TG, Horan TC, Hughes JM. CDC definitions for nosocomial infections, 1988. Am J Infect Control 1988; 16: 128-l 40. 20. Cornfield J. A statistical problem arising from retrospective studies. In: Neyman J, Ed. Proceedings of the Third Berkeley Symposium, Volume IV. Berkeley: University of California Press 1956; 135-148. 21. Haley RW, Culver DH, Morgan WM, White JW, Emori TG, Hooton TM. Identifying patients at high risk of surgical wound infection. A simple index of patients susceptibility and wound contamination. Am r Epidemiol 1985; 121: 206-215. 22. Jarvis WR, White JW, Munn VP, et al. Nosocomial infection surveillance, 1983. CDC Surveillance Summaries. MMWR 1984; 33 (Suppl. 2): 9-21. 23. Marples RR, Mackintosh CA, Meers PD. Microbiological aspects of the 1980 national prevalence survey of infections in hospitals. J Hosp Infect 1984; 5: 172-180. 24. Leigh DA. Antimicrobial usage in forty-three hospitals in England. J Antimicrob Chemother 1982; 9: 75-84.
of Hospital
Prevalence
Infection
(1992) 20, 1-13
of hospital-acquired EPINE
Working
infections
in Spain
Group*
Accepted for publication
14 October 1991
Summary: In May 1990 a prevalence survey of hospital-acquired infections was conducted in 123 Spanish hospitals, in which 38 489 patients were studied. There was an 8.5% prevalence of infected patients and a 9.9% prevalence of infections. The most common infections were those of the urinary tract (27.7%), surgical wound (22.7%) and lower respiratory tract (15.4%) and bacteraemia (10.6%). Th ere was a 5.9% prevalence of patients with surgical wound infection and 3.5% after clean surgery. An aetiological diagnosis was made in 58% of the infections. Gram-negative bacteria were dominant, Escherichia coli (16.3%) and Pseudomonas aeruginosa (11.5 %) being the most prevalent; 33.8% of the patients were receiving antimicrobial agents. The following procledures were shown to be significantly associated with hospital-acquired infections: urinary catheterization, parenteral nutrition, mechanical ventilation and tracheostomy. The degree of contamination during surgery was also a significantly associated risk factor. The survey provided extensive information on the distribution of infections and the use of antibiotics in clinical services, as well as the differences between hospitals according to their size and .the presence of certain risk factors. Keywords: Spain.
Prevalence
survey;
hospital-acquired
infection;
antibiotic
use;
Introduction Hospital-acquired infections (HAI) pose a problem currently attracting considerable attention, with great human, social and economic importance, due to their associated morbidity and mortality. In many hospitals, prevalence studies are part of the routine HA1 surveillance, since they are easy to perform, provide general descriptive information on the problem and are helpful in evaluating the impact of control measures.lm3 *EPINE Writing and Steering Committee: Josep Vaqut, Vail d’Hebrdn Hospital, Universitat Auton&m, Barcelona; Vicente Mange, Hospital Ram& y Cajal, Madrid; Juan Garcia Caballero, Hospital La Paz, Madrid; Josk Rossell6, Vail d’Hebr& Hospital, Universitat Auto&ma, Barcelona; Jo& Barrio, Hospital de Sant Pau, Barcelona; Francisco Calbo, Hospital Carlos Haya, Mblaga; M. Carmen Sienz, Hospital Clz’nico Uniuersitario, Salamanca; A. Hernindez, Hospital La Fe, Vale&a; Plicido L6pez Encinar, Hospital Clinico Universitario, Valladolid; J. Luis Vaquero, Hospital de1 Rio Hortega, Valladolid; Jo& R. SLenz, Hospital N.S. Aranzazu, San Sebastidn; Vicente Dominguez, Hospital Juan Canalejo, La CoruKa; Rosa Coello, Hospital Chico Universitario, Madrid; Pedro Blasco, Hospital Virgin Macarena, Sevilla; J. Luis Arribas, Hospital Miguel Servet, Zaragoza; A. Gonzilez Torla, Hospital N.S. Covadonga, O&do; Maruja Santos, Hospital La Fe, Vale&a; Rafael Herruzo, Hospital La Paz, Madrid; Spain. Correspondence to: Dr J. VaquC, Barcelona, Spain. 0,9556701/92/010001
+I3
Servei
de Medicina
Preventiva,
tO3.00/0
Ciutat
Sanitlria
0
1
1992
Vall d’Hebr&,
The
Hospital
Infection
08035,
Society
2
EPINE Working
Group
Prevalence surveys performed in countries or large geographical areas define the overall situation and, when the sample is comprehensive or sufficiently representative at hospital level, may indicate differences between hospitals. The World Health Organisation promoted a prevalence study between 1983 and 1985, in which data from 47 hospitals in 14 countries4 were analysed. In Europe, multicentre prevalence surveys have been conducted in Sweden,’ Denmark,6 Norway,7 England and Wales,’ ” In Spain, a pilot study was carried Italy,’ Belgium” and Czechoslovakia. out in 1986, in which a group of hospital patients from virtually all regions was studied; l2 later, a second study was carried out in Catalonia.i3’i4 The European prevalence study of urinary tract infection and bacteraemia as well as the national studies performed in should also be mentioned,” Australiai and Thailand.i7 A prevalence survey was conducted in a large group of Spanish public acute-care hospitals, in order to study HAI and their associated factors, as well as community-acquired infections (CAI) and the use of antimicrobial agents.
Patients
and methods
All patients hospitalized in the centres participating in the investigation were included. Data were collected from the clinical records and information provided by the doctors and nurses in charge of the patients, in addition to bedside confirmation of extrinsic risk factors. This task was carried out at each centre by the HAI surveillance and control team, as constituted in the individual hospitals on the advice of the Infection Control Committee. The survey was conducted during a 2-week period in May 1990. At each hospital one person was charged with coordinating and reviewing the study; usually this person was a medical epidemiologist, less frequently a medical microbiologist, infectious diseases clinician or infection control nurse. The study protocol’8 had been widely tested in previous surveys,‘3,‘4 and all its variables were explicitly defined. The intrinsic risk factors studied were: coma, renal failure, diabetes mellitus, neoplasia, immu”iodeficiency, neutropenia, cirrhosis, hypoalbuminaemia, pressure sores, alcoholism, smoking and drug addiction. Extrinsic factors, including those applied during the stay in hospital included urinary catheterization, peripheral vascular catheterization, central vascular catheterization, central catheterization with peripheral insertion, parenteral nutrition, tracheostomy and mechanical ventilation. Information was also compiled about the surgical operations performed. The baseline risk of the patient was classified according to three categories of clinical prognosis: severe (survival of less than a year); moderate (life expectation between 1 and 5 years); and mild (the rest of the cases).
Nosocomial
infection
in Spain
3
The criteria of the Centers for Disease Control, Atlanta, USAl were used for defining infections, except that asymptomatic bacteriuria was not accepted as a criterion for urinary tract infection. A patient with more than one site of infection was considered to have multiple infections; the survey data forms provided room for four infections per person. The date of onset of the infection (the day on which the symptoms that led to microbiological investigation appeared, or antibiotic therapy was begun) was recorded. The results of microbiological and serological investigations were usually available within 7 days following the end of the clinical investigation; those not available were separately categorized. We used two measures of the frequency of HA1 or CAI: the prevalence of infections (the ratio between the number of clinically active infections and the total number of patients studied, expressed as a percentage) and the prevalence of infected patients, which is the percentage of patients with active infection in relation to the total number of patients studied. Confidence intervals were calculated and hypotheses were tested on the basis of infected patient prevalence. Since the stays of patients in obstetric services are usually short and their infection levels are low, the prevalence of infections and of infected patients were also calculated after exclusion of these patients. The prevalence of patients receiving antibiotics was the proportion of patients receiving these drugs at the time of the survey in relation to the total number of patients. Length of stay prior to infection was defined as the number of days between admission and the onset of the infection. The variable ‘total number of intrinsic risk factors’ was calculated by adding up the number of factors of this type for each patient. A similar accumulative variable was also defined for extrinsic factors. The crude association between a risk factor and the presence of infection was assessed by the Odds Ratio; its confidence interval was estimated by the Cornfield method.20 The combined study of the possible risk factors involved in the development of HA1 was conducted by means of multiple logistic regression models. In these models, the presence or absence of nosocomial infection, overall or for the main different infection sites, was the dependent dichotomous variable. Those patients with an infection included in the study were considered cases, and those without HA1 controls. In these models we used the variable ‘length of stay prior to infection’ for patients with HA1 and the interval between admission and date of entry to the study for those without. In this way, we avoided using the interval between the onset of infection and the day of study, which is influenced by the prolongation of stay due to infection. One hundred and twenty-six hospitals participated in the investigation, and in total 38 857 patients Twere studied. Of these, 123 hospitals met the quality requirements and 38 489 patients were eligible for overall assessment, and these are the basis of this report. Since 135 centres were invited to participate in the study, the proportion of participants can be
4
EPINE Working
Group
considered very high (93.3%). The hospitals were classified into three categories: small hospitals (fewer than 200 beds, 50 centres, 40.7%); medium-sized hospitals (between 200 and 500 beds, 46 centres, 37.4%); and large hospitals (over 500 beds, 27 centres, 21.9%). Hospital services were classified into six large areas: medicine, surgery, obstetrics and gynaecology, paediatrics, intensive care, and other services, which last included psychiatry, geriatrics etc. The intensive care area included the ICU, the Coronary Care Unit and the Transplantation Units for adults. Results
Prevalence of HAI
and CAI
In the initial assessment of the whole group of 38 489 patients 3800 HA1 and 5917 CA1 were detected (Table I). The overall prevalence of HA1 was 9.9%, and that of infected patients 8.5% (95% CI 8.2-8.7). The overall prevalence of CA1 was 15.4% and that of patients with infection 13.4% (95% CI 13-l-13.7). Th e median prevalence of HA1 in the 123 centres was 9.4% and that of CA1 16.1%. After exclusion of the patients of the obstetric services, the general HA1 prevalence rose to 10.3% and that of infected patients to 8.8% (a non-significant difference compared with the overall prevalence, x2 = 3.69, P= 0.054); with respect to CAI, the prevalence of infections rose to 16.1% and that of infected patients to 15% (a significant difference compared with the overall prevalence, x2 = 6.61, P= 0.01). The lowest prevalence of patients with HA1 was recorded in the small hospitals (7*9%), while large and medium-sized centres recorded a similar prevalence (no significant differences between the three types of hospital, x2= 2.67, P= 0.26; nor after excluding patients in obstetric services,
x2= 3.92, P=O.14). In 33 hospitals (26.8%) the HA1 prevalence was below 7%; in 27 centres (22%) it ranged between 7% and 9%) and in 63 centres it was over 9%. The HA1 prevalence in the ICU was high: 48.7% in adults and 26.2% in paediatrics. In the area of medicine, the prevalence was high in haematology (15.2%) and rehabilitation (17.3 %) services; neurology (13.9%) and nephrology (12.1 “A) services were also above the general Spanish prevalence rate. In the surgical area there were also some services with HA1 prevalences above the general percentage, but not very high: the adult burns unit (12.9%), neurosurgery (12.9%) and general surgery (11.7%). The average age (f SD) among patients with HA1 was 49.15 years (f 25.36); 55.1% of the cases were men and 44.9% women. The highest prevalence of patients with HA1 was noted in children aged 1 to 10 year old stratum had the lowest prevalence (4.5%). From this category onwards, there was a linear significant trend of increase for the prevalence (Mantel-Haenszel extended test, x2= 113.4, P75 Stay prior to infection: 8-i4 days 15-21 21-28 29-35 36-42 43-60 >60 Community-acquired infection Type of hospital: medium-sized large
Odds Ratio
infections (95%
CI)
1.8 2.9 7.1
(1.6-2.1) (2.6-3.4) (5.7-8.9)
1.1 I.4 1.6 1.3 1.8
(0.9-1.2) (1.2-1.6) (1.2-1.9) (1.1-1.5) (1.5-2.2)
1.6 I.8 2.2 1.8 1.1
(14-1.8) (1.6-2.1) (1.8-2.6) (1.6-2.0) (1.0-1.2)
I.0 1.0 1.1 0.9 1.1
(0.8-l .3) (0.8-1.3) (0.9-1.3) (0.8-1.2) (091’4)
I.5 1.9 2.6 2.6 2.5 2.9 2.9 0.5
(1.331.7) (1.7-2.3) (2.1-3.1) (2.1-3.2) (1.9-3.3) (2.4-3.8) (2.3-3.7) (0.4-0.6)
I.1 1.1
(0.9-1.3) (0.9-1.2)
Comparison categories: absence of extrinsic risk factors, absence of intrinsic risk factors, mild baseline risk, number of diagnoses = 1, absence of surgical operation, male sex, age = 15-30 years, hospital stay prior to the infection = l-7 days, patient without CAI, type of hospital = small.
.
Nosocomial
categories. The two categories presence of CA1 was protective. Microbiological
infection
in Spain
9
of hospital size were not significant. The None of the age categories was significant.
diagnosis
The microbiological diagnosis (culture or serological) was available in 58% of the HA1 and in 41% of the CAI. Among HAI, 57.8% had positive cultures and 14.1% had negative cultures, while in 24.7% cultures had not been carried out and in 3.1% the results were too late for inclusion (see above); in only 0.3% of cases was serological testing carried out. Among CAI, 30.1% had positive cultures and 20.8% had negative cultures, while in 30.6% cultures had not been carried out and in 7.5% the results were too late for inclusion. Serological testing was more important in CAI, accounting for 11% of cases. Overall and for each infection site the proportion of positive cultures in relation to the total number performed was lower in CA1 than in HAI. UT1 was the type of HA1 with the highest percentage of positive cultures (67% of the cases), followed by BACT (64.7%), SW1 (60%), LRI (49.1%) and the group composed of other sites (48.3%). Gram-negative organisms, were identified in 43.9% of the cases and Gram-positive in 29.4%. The aetiological agents were identified in 50.8% of HA1 and 49.2% of CAI. Table IV shows the classification of the isolated organisms. Among HA1 the most frequent Gram-negative pathogen was Escherichia coli (16.3% of total), followed by Pseudomonas aeruginosa (11.5%); the most frequent Gram-positive organism was Staphylococcus aweus (11.2%). Staphylococcus epidermidis and other coagulase-negative staphylococci also exceeded 5 %. Escherichia coli was also the most prevalent organism in CA1 (14.1%), with human immunodeficiency virus (HIV; antibody detection, 12.6% of positive diagnoses by culture or serology) in
Table Groups
IV. Microbiological
diagnoses
(culture
or serology)
and main organisms
and main pathogens
Type
of infection
HA1
CA1
Total Groups Gram-negative bacteria Gram-positive bacteria Anaerobes Other microorganisms Main pathogens E. coli P. aeruginosa Proteus mirabilis S. aweus Human immunodeficienc virus Enterococcus spp.
No.
(%I
No.
W)
No.
(%)
2377 1591 138 1307
(43.9) (29.4)
1392 1030 81 245
(50.7) (37.5) ( 2.9) ( 8.9)
985 561 57 1062
(37.0) (21.0)
823 413 157 488
(15.2)
(16.3) (11.5) ( 3.7) (11.2)
376 97 56 181
(14.1) ( 3.6)
( 9.0)
447 316 101 307
; ;:;I
( 5-3)
212
( 7-7)
335 74
(12.6) ( 2.8)
;2:::j
; ;:;;
13;:;;
:Y 335 286
10
EPINE Working
second place. Mycobacterium more than 5% of the cases.
tuberculosis
Group
and S. auwus were also isolated
in
Antibiotics At the time of the study, 13 007 patients were receiving antibiotics, an overall prevalence of 33.8% (Table V). Forty-three percent of the antimicrobial agents were prescribed in surgery and 32.6% in medicine. The area with the greatest prevalence was intensive care (52.2%). Eighty-three percent of the patients with HA1 were being treated with antimicrobial agents, a proportion roughly similar to that in CA1 (80%). In the group of patients without infection, 21.1% were receiving antibiotics. The prevalence of antibiotic usage was higher in small (35.1%) than in medium-sized hospitals (33*8%), and higher in the latter than in large hospitals (33.1%) (significant differences between the three types of hospital, x2= 14.62, P= 0.0006). The more commonly prescribed antimicrobial agents were aminoglycosides (17.8%), followed by aminopenicillins (14.1%) and third generation cephalosporins (11.2%). The most widely used product was gentamicin: general use 8.8%; 8.2% in HAI, 7.5% in CA1 and 9.8% in patients without infection. Discussion
In the number of patients studied, this is the largest multicentre prevalence study yet published. +17 The overall HAI prevalence observed (9.9’/,) is an intermediate value, between the extremes observed in the studies shown in Table VI. The results of the present study show a considerable Table
V. Prevalence of antibiotic use according service and type of infection
Service and type of infection
Hospital service Medicine Surgery Intensive care Obstetrics & gynaecology Paediatrics Other services Total Types of infection Hospital-acquired Community-acquired Both types No infection * P%, Prevalence.
Prescribed antibiotics
to hospital
Patients with antibiotics
N
(%)
N
P%*
6121
(32.6)
y;:;
7::;;
3954 5976 558
32.7 37.1 52.2
:;2 212 18 780
&cl; ( 1.1) (100.0)
1048 1312 159 13 007
22.4 37.1 15.7 33.8
2698 4128 229 6410
82.9 SO.1 91.6 21.1
4365 6643
g:;;
Nosocomial Table Country
VI. Prevalence
and reference
Sweden5 Denmark6 Norway’ England & Wales* Europe” (8 countries)” Italy9 14 Countries, WHO4 Belgium’“,b Australia16 Spain12 Czechoslovakia” Catalonia’3,‘4 ThailandI Spain (this study)
infection
studies in several
11
in Spain countries
Year
Number of hospitals studied
1975 1978179 1980 1’980 1982 1983 1983-85 1984 1’984 1’986 1988 1988 1988 1990
5 2512.5 15 43 40 130 47 106 269 39 23 33 23 123
and WHO Number of patients studied 3657 1363/1X7 7833 18 163 3899 34 577 28 861 8723 28 643 10 470 12 260 7434 6805 38 489
study Prevalence of HA1 per 100 patients 10.5 10.5/12.1 9.0 9.2 7.3 6.8 9.9 10.3 8.1 11.2 6.1 12.1 11-7 9.9
aOnly urinary infections and bacteraemia were studied. b Only urinary and surgical infections and bacteraemia were studied.
improvement in comparison with those obtained in the general Spanish the two studies were conducted by very different study of 1986, l2 although methods and with different definitions. HAIs were greatly concentrated in the intensive care units and to a lesser degree in certain paediatric, surgical and medicine services. Small hospitals had a lower prevalence, as was to be expected, in view of the fact that more serious pathologies are seen, and more complex and aggressive therapies are given, in medium-sized and large centres. The CA1 prevalence was Ihigh, with considerable impact in the areas of medicine and paediatrics; these infections, however, were the main diagnosis in only 5% of the patients. From an epidemiological perspective (not a biological one), the presence of CA1 protects against the development of HAI. This could possibly be due to the CA1 patients receiving antibiotics after admission and therefore being protected against a nosocomial infection, or alternatively, a distortion could exist, due to under-reporting of patients with both types o’f infection. The distribution of HA1 by site was that commonly seen. There were, nevertheless, marked differences from the WHO study4 in which the percentages were: UT1 22.1%; SW1 25.1%; LRI 20*6% and BACT 3.7%. In our study, there was a high proportion of primary (7.6%) and secondary (3 %) bacteraemia, which was related to the significant proportion of severely ill and immunosuppressed patients (43% of the BACT were diagnosed in intensive care, neonatal and haematology units). In the logistic model of each infectious site, specific instrumentation procedures showed a high degree of significance. In surgical wound infections, the degree of contamination has a determinant effect. According to traditional concepts, the crude analysis showed that the older the patient,
12
EPINE Working
Group
the greater the risk of nosocomial infection. However, the logistic adjustment indicated that increased risk could be due to the concomitant circumstances of age, such as the baseline risk and the number of diagnoses. The idea of using the variable number of diagnoses was suggested by Haley et a1.21: the three categories were significant in the overall model and its behaviour is similar to that of the baseline risk. While the latter helps to define how seriously ill the patient is, the number of diagnoses can explain the complexity of the baseline situation and the multi-system extension of the process. Two categories of the number of intrinsic factors showed statistical significance in the overall model. This variable could explain part of the impact of multisystem involvement. With respect to the microbiological studies, our results showed that the percentage of culture requests and the percentage of positives of the United States22 were not attained; however, in relation to other European studies6-9”2 the comparison is more favourable. The pattern of isolated pathogens was roughly similar to that seen in other studies,4,“,‘4,23 with the exception of the outstanding current role of HIV and Ad. tuberculosis. Worthy of note is the preponderance of Gram-negative over Gram-positive bacteria in both HA1 and CAI. The overall prevalence of antibiotic use was higher than in other multicentre studies.4,6,‘4,‘5,24 Prevalence studies are affected by an over-representation bias due to patients with long-term hospitalization and they are not adequate for establishing causal inferences. They are, nevertheless, easy to perform, very versatile and the results that they provide are of undoubted descriptive interest for understanding the general dimension and features of the problem of nosocomial infections. This survey was sponsored by the Spanish Association of Hospital Hygiene and Preventive Medicine, with the technical collaboration of Laboratorios Beecham S.A. Statistical analysis was performed by J. Vaqut, J. Rossello and L. Armadans. The coordinators in the individual hospitals were as follows. Anduhci’u: M. Torres, I. Fernindez, F. Rodriguez, J. M. Bautista, S. Oiia, D. Roman, R. Gilvez, R. Rodriguez-Contreras, S. Oliver, J. L. Martin Ruiz, F. Lopez, J. Lorente, P. Alamillos, M. de Cueto, R. Fernandez, S. Duefias, R. Creagh, D. Sureda. Arugdn: P. Muniesa, P. Prieto. Asturias: E. Tui%n, T. Fernandez, P. Prendes, A. Torreblanca. Baleares: 0. Hidalgo. Canarias: G. Lopez, J. Alamo. Custillu-La Muncha: J. Martinez, A. Biurrun, E. Ruiz, J. Medrano. Cast&z-Ledn: J. de la Lama, N. Coladas, J. Moro, F. Benito, M. Mendo, A. Sacristan, G. Fernindez, A. Hita, P. Vicente, E. Robles, J. M. Olona, B. Bermejo, J. M. Bayas, M. Gurgui, L. Blanco. Cutulunyu: M. Campins, Drobnic, J. M. Garces, E. Plass, N. Sanjuan, N. Miserachs, M. DurQn, M. Esteve, C. M. T. Martinez, L. R/loner, R. Priu, A. Brotons, F. Rossell, I. Garcia, A. Casagran, Verdaguer, L. Gurgui, J. Alijotas, R. Pujol, M. Oliveras, J. M. SBnchez, X. Puig, J. Brau, F. Ribas, A. Anglada, A. Vilamala, J. Baucells, C. Clemente, T. Puig, J. J. Guardiola, F. Ballester, A. Manonelles, V. Romeu, J. Galvany, J. Luna. Extremaduru: J. Jimenez, A. Cerrillo. Guliciu: J. J. Gestal, M. L. Cal, V. Souto, J. L. Lopez, B. Uriel, A. Blanco, M. Cueto. Madrid: A. Asensio, N. Valverde, J. M. Garcia Paez, P. Gomez, A. Blanco, B. Perez, S. Sanchez, V. Valls. Murciu: F. Botia, J. Piqueras, A. Mas. Comunidud Vulenciunu: L. Ferreres, E. Monte, F. Asensi, S. Cervellb, V. C. Zanon, C. Gonzalez, C. Oltra, J. F. Navarro. P&s Vusco: J. L. Novales, J. M. Arevalo, M. E. Elosegui.
Nosocomial
infection
in Spain
13
References 1. Casewell MW. Surveillance of infection in hospitals. J Hasp Infect 1980; 1: 293-297. 2. Crede W, Hierholzer WJ Jr. Surveillance for quality assessment: I. Surveillance in infection control success reviewed. Infect Control Hosp Epidemiol 1989; 10: 470-474. 3. French GL, Cheng AFB, Wong SL, Donnan S. Repeated prevalence survey for monitoring effectiveness of hospital infection control. Lancet 1989; 2: 1021-1023. 4. Mayon-White RT, Ducel G, Kereselidze T, Tikhomirov E. An international survey of the prevalence of hospital-acquired infection. J Hasp Infect 1988; 11 (Suppl. A): 43-48. 5. Bernarder S, Hambreaus A, Myrback, Nystrijm B, Sundelof B. Prevalence of hospital-associated infection in Five Swedish Hospitals in November 1975. &and J Infect Dis 1978; 10: 66-70. 6. Jepsen OP, Mortensen N. Prevalence of nosocomial infections and infection control in Denmark. J Hasp Infect 1980; 1: 237-244. 7. Hovig B, Lystad A, Opsjon H. A prevalence survey of infections among hospitalized patients in Norway. NIPH Annals 1981; 4: 49-60. 8. Meers PD, Ayliffe GAJ, Emmerson AM, Leigh DA, Mayon-White RT, Mackintosh CA, Stronge JL. Report on thse national survey of infections in hospitals, 1980. r Hasp Infect 1981; 2 (Suppl.): 141. 9. Moro ML, Stazi MA, Marasca D, Greco D, Zampieri A. National prevalence survey of hospital-acquired infections in Italy, 1983. J Hasp Infect 1986; 8: 72-85. 10. Mertens R, Kegels G, Stroobant A. et al. The national prevalence survey of nosocomial infections in Belgium, 1984. J Hasp Infect 1987; 9: 219-229. 11. Srimovi H, Bartonovi A, Bolek S, Krecmerovi M, Subertova V. National prevalence survey of hospital-acquired infections in Czechoslovakia. J Hasp Infect 1988; 11: 328-334. 12. Ministerio de Sanidad y Consumo. Encuesta epidemiologica de prevalencia en un punto de la infection hospitalaria y uso clinico de antimicrobianos en Espafia. Anilisis descriptive de resultados. Madrid, Direction General de Planificacibn Sanitaria, 1986. 13. Grupo de Trabajo Epincat. Prevalencia de las infecciones nosocomiales en Catalufia. (I). Infecciones y factores de riesgo. Med Clin (Bore) 1990; 95: 41-52. 14. Grupo de Trabajo Epincat. Prevalencia de las infecciones nosocomiales en Catalufia. (II). Germenes y antimicrobianos. Med C’lin (Bare) 1990; 95: 161-168. 15. Jepsen OB, Larsen SO, Dankert J, et al. Urinary-tract infection and bacteremia in hospitalized medical patients. A European multicentre prevalence survey on nosocomial infection. J Hosp Infect 1982; 3: 241-252. 16. McLaws ML, Gold J, King K, Irwig LM, Berry G. The prevalence of nosocomial and community-acquired infections in Australian hospitals. MedJ Aust 1988; 149: 582-590. 17. Danchaivijitr S, Chokloikaew S. A national prevalence study on nosocomial infections 1988. J Med Assoc Thai 1989; 72 (Suppl.): l-6. 18. Proyecto Epine. Protocolo de1 estudio Epine1990. Madrid, Sociedad Espaiiola de Higiene y Medicina Preventiva Hospitalarias, 1990. 19. Garner JS, Jarvis WR, Emori TG, Horan TC, Hughes JM. CDC definitions for nosocomial infections, 1988. Am J Infect Control 1988; 16: 128-l 40. 20. Cornfield J. A statistical problem arising from retrospective studies. In: Neyman J, Ed. Proceedings of the Third Berkeley Symposium, Volume IV. Berkeley: University of California Press 1956; 135-148. 21. Haley RW, Culver DH, Morgan WM, White JW, Emori TG, Hooton TM. Identifying patients at high risk of surgical wound infection. A simple index of patients susceptibility and wound contamination. Am r Epidemiol 1985; 121: 206-215. 22. Jarvis WR, White JW, Munn VP, et al. Nosocomial infection surveillance, 1983. CDC Surveillance Summaries. MMWR 1984; 33 (Suppl. 2): 9-21. 23. Marples RR, Mackintosh CA, Meers PD. Microbiological aspects of the 1980 national prevalence survey of infections in hospitals. J Hosp Infect 1984; 5: 172-180. 24. Leigh DA. Antimicrobial usage in forty-three hospitals in England. J Antimicrob Chemother 1982; 9: 75-84.
