- Email: [email protected]
The national nosocomial infections surveillance system: Plans for the 1990s and beyond
The National Nosocomial Infections Surveillance System: Plans for the 1990s and Beyond ROBERT P. GAYNES,M.D., DAVIDH. CULVER,Ph.D.,T. GRACEEMORI,R.N.,MS., TERESA C. HORAN,M.P.H., c.I.c., SHAILENN. BANERJEE, Ph.D.,JONATHAN R. EDWARDS, M.s., WILLIAMR. JARVIS,M.D., JAMESS. TOLSON,B.S.,TONYAS. HENDERSON, B.S.,JAMESM. HUGHES,M.D.,WILLIAMJ. MARTONE,M.D., ANDTHENATIONAL NOSOCOMIAL INFECTIONS SURVEILLANCE SYSTEM, Atlanta Georgia
The National Nosocomial Infections Surveillance (NNIS) System is an ongoing collaborative surveillance system among the Centers for Disease Control (CDC) and United States hospitals to obtain national data on nosocomial infections. This system provides comparative data for hospitals and can be used to identify changes in infection sites, risk factors, and pathogens, and develop efficient surveillance methods. Data are collected prospectively using four surveillance components: hospital-wide, intensive care unit, high-risk nursery, and surgical patient. The limitations of NNIS data include the variability in case-finding methods, infrequency or unavailability of culturing, and lack of consistent methods for postdischarge surveillance. Future plans include more routine feedback of data, studies on the validity of NNIS data, new components, a NNIS consultant group, and more rapid data exchange with NNIS hospitals. Increasing the number of NNIS hospitals and cooperating with other agencies to exchange data may allow NNIS data to be used better for generating benchmark nosocomial infection rates. The NNIS system will continue to evolve as it seeks to find more effective and efficient ways to measure the nosocomial infection experience and assess the influence of patient risk, changes in the delivery of hospital care, and changes in infection control practices on these measures.
T
he Centers for Disease Control (CDC) National Nosocomial Infections Surveillance (NNIS) System is the only national source of nosocomial infections data in the United States. Many individuals at the CDC have been involved in the team effort of the NNIS system. However, the NNIS system requires the voluntary effort of the many infection control practitioners and hospital epidemiologists at the participating NNIS hospitals.
OBJECTIVES OF THE NATIONAL NOSOCOMIAL INFECTIONS SURVEILLANCE SYSTEM In its broadest terms, the NNIS system attempts to estimate the magnitude and nature of nosocomial infections in the United States. NNIS data aid in the prevention of nosocomial infections by identifying changes in infection rates and sites, risk factors, outcomes, pathogens, and antimicrobial resistance. Data on observed changes are used to redirect infection control and prevention efforts to areas of new or increasing concern [ 11. In addition, the NNIS system can provide hospitals with meaningful comparative data on nosocomial infections to aid in the prevention effort. Data for interhospital comparison must adjust for differences in the distribution of confounding variables such as duration of patient stay or device utilization. Several recent analyses attempt to provide hospitals with these types of data [2-41. Another NNIS objective is to assist hospitals in developing efficient and effective data collection and analysis methods which allow infection control personnel to use their time more efficiently.
HISTORICAL BACKGROUND
Diseases, Centers for Disease Control, Atlanta, Georgia. Hospital Infections Program A-07, Building 3, Room B16A, Centers for Disease Control, 1600 Clifton Road N.E., Atlanta, Georgia 30333.
3B-116s
September
16, 1991
The American
Journal
of Medicine
The NNIS system began in 1970 when selected U.S. hospitals sent data on nosocomial infections to the CDC for aggregation into a national database. Initially, all participating hospitals were required to conduct hospital-wide surveillance, i.e., all patients in the hospital were monitored for infections at all body sites. A major change occurred in 1986, when several alternative protocols for performing surveillance, called surveillance Volume
91 (suppl3B)
CONFERENCE
components, were introduced. As a part of this NNIS revision, definitions of nosocomial infections, including the use of clinical and laboratory algorithms, were expanded. The NNIS system also introduced microcomputer software designed specifically for the NNIS hospitals to facilitate the entry and analysis of data collected under the surveillance component protocols.
METHODS Hospitals currently participating in the NNIS system collect information on patients requiring acute care. Under Section 308(d) of the Public Health Service Act (42 U.S.C. 242b, k, and m(d)), the identity of NNIS hospitals is confidential. Data are released outside of the NNIS system only in aggregate form. Hospitals apply for membership in the NNIS system and must have adequate personnel support for infection control, a microcomputer compatible with the requirements of the NNIS software, and approval from hospital administration to participate in the NNIS system. Currently, 115 United States hospitals participate in the NNIS system. The current NNIS methods include four standardized surveillance protocols or surveillance components 151: Hospital-wide Adult and pediatric intensive care units High-risk nursery Surgical patient The surveillance components may be used singly or in any combination. These components are designed to form the foundation of a hospital’s overall surveillance program and allow flexibility for incorporating other aspects into their surveillance efforts. Once selected, a component must be followed for at least 1 month. NNIS hospitals use standard CDC definitions of nosocomial infections, published in 1988 161. The data collected on patients with nosocomial infections are essentially the same regardless of component. The differences between the components lie in the group of patients monitored for nosocomial infections and in the summary or denominator data that are collected in that group. The denominator data dictate the type of infection rates that may be calculated for a NNIS component. In the hospital-wide surveillance component, all patients are monitored for nosocomial infection at all sites. The denominator data collected each month consist of the total number of discharges from each NNIS-defined service and, optionally, the total number of patient-days spent on each service. These data allow calculation of overall September
16, 1991
ON NOSOCOMIAL
INFECTIONS
/ GAYNES
ET AL
infection rates and rates for each site of infection by service. In addition, because hospital-wide surveillance has been used since the beginning of the NNIS system, trend analyses covering 10 or more years can be performed 171. Because of difficulties in controlling for differences in case mix (e.g., severity of illness, medical interventions, or average duration of patient stay), caution is needed when any data from the hospitalwide component are used for interhospital comparison. In addition, many hospitals, particularly large hospitals, have insufficient resources for surveillance, causing significant underreporting of infections. Because of these problems, other components were developed so that surveillance efforts could be targeted and more specific denominators could be acquired. In the adult and pediatric intensive care units (ICUs) surveillance component, infection control practitioners monitor all ICU patients in one or more critical care units for nosocomial infections at all sites. The denominator data collected include the type of unit being monitored (e.g., medical, surgical, coronary) and the total number of admissions, patient-days, and device (central intravenous catheter, urinary catheter, or ventilator) days. With these data, we can calculate overall infection rates and device-associated infection rates for bloodstream, urinary tract, and respiratory infection sites. For example, central intravenous catheter-associated bloodstream infection rates are defined as: Number of catheter-associated bloodstream infections Number of central intravenous catheter days
x 1,000
By attempting to control for the exposure to the principal risk factor, such a device-associated rate can help to identify outliers, i.e., hospital units in which the rate is significantly above that in similar units (Figure 1). With the high-risk nursery (HRN) surveillance component, all HRN patients are monitored for nosocomial infections at all sites. The denominator data collected include the number of admissions, patient-days, and device-days within each of three birthweight categories ( < 1,500 g, 1500-2500 g, and >2,500 g). This allows calculation of overall infection rates and device-associated infection rates for bloodstream and respiratory infection sites, much as in the ICU component. The surgical patient (SP) surveillance component has two options: limited and detailed. In the SP surveillance component limited option, all pa-
The American
Journal
of Medicine
Volume
91 (suppl3B)
3B-117s
CONFERENCE
m
$ ‘ti :: 1 %
2
31:
0
ON NOSOCOMIAL
/ GAYNES
ET AL
4 indicates the Median rate
8 6 4
Rate IO
m -a .E 2
INFECTIONS
12
14
16
18
2
*
0
22
24
26
Surgical (n = 2.8, median = 5.1)
8 6
r” ‘i;
20
4 Rate 8
2
10
12
14
16
18
20
22
24
26
Pediatric (n = 15, median = 8.9)
‘Number
of Bloodstream
Infections per 1000 Central IV Catheter Days
tients undergoing all NNIS-defined operative procedures are monitored for infections. The summary data collected include a) the total number of procedures performed in each NNIS-defined operative procedure category, b) the total number of procedures performed in the four traditional wound classes (i.e., clean, clean-contaminated, contaminated, and dirty), and c) the total number of procedures performed by each surgeon (CDC does not receive data on individual surgeons). This allows calculation of infection rates by each of these factors. Thus, much like the hospital-wide component, detailed information on infections is collected with a general overall denominator in each category. An attempt to examine risk factors that are known to be associated with surgical wound infections was incorporated into the surgical patient surveillance component detailed option. All patients undergoing selected NNIS-defined operative procedures are monitored. This component differs from any other in that detailed information (e.g., age, wound class, duration of surgery, urgency of procedure, the surgeon(s), and whether the procedure was the result of trauma to the patient) is collected on every such patient. If a patient develops a nosocomial infection, further information on the infection is collected. Infection rates within categories of each potential risk factor (e.g., wound class) can be calculated. Alternatively, infection rates within the categories of a composite risk index can also be calculated. OVERVIEW OF RESULTS Device-associated rates from the adult and pediatric ICU component are shown in Figure 1. The hospital whose medical ICU had a rate of 41.4 catheter-associated bloodstream infections (BSI) per 1,000 catheter days is readily identified as an 3B-118s
September
16, 1991
The American
Journal
of Medicine
Figure 1. Distribution of central intravenous catheter-associated bloodstream infection rates by type of intensive care unit, Intensive Care Unit Component, National Nosocomial Infections Surveillance System, 1987-1989. Catheterassociated bloodstream infection rate is defined as the number of central intravenous catheter bloodstream infections per 1,000 central intravenous catheter
outlier. However, the identification of an outlier does not necessarily define a problem in that ICU. Rather, the outlier status only suggests an area for investigation by hospital personnel. In August 1988 a report was sent to NNIS hospitals with a distribution of data similar to that presented in Figure 1. One outlier hospital investigated the possible reasons for their elevated catheter-associated BSI rate and found several practicerelated problems. After taking corrective action, the hospital reduced its infection rate substantially. This use of NNIS data as a prevention tool was recently reported by Selva et al. @I. Other analyses of NNIS data, found elsewhere in these Proceedings, suggest that more meaningful intrahospital and interhospital comparisons can be made using these device-associated rates [2,31. Analysis of nosocomial infections among the neonates from the HRN component confirmed the striking differences from infections among adult patients reported in other studies (Figure 2) [9,10]. The site distribution of nosocomial infections shows a high percentage of BSIs for all birthweight groups. More than 85% of these occurred more than 2 days after birth and were usually catheter-associated. This suggests that surveillance and prevention efforts in HRNs should emphasize this site of infection. Using data taken from the SP componentlimited option, the pooled mean surgical wound infection @WI) rates for operative procedures can be calculated. For example, the rate for herniorrhaphy was 1.1 SW1 per 100 operations; for colon surgery it was 7.3. These rates give an estimate of the SW1 rates by operative procedure for the United States. However, for purposes of interhospital comparison, they ignore known risk factors, e.g., wound class or a patient’s intrinsic susceptibility to infection, which could be responsible for an Volume
91 (suppl3B)
CONFERENCE
individual hospital’s or surgeon’s variation from this pooled mean. The NNIS risk index attempts to control for some of this variation and is a substantial improvement over the stratification of operations done by the traditional method using wound class [41. LIMITATIONS OF THE NATIONAL NOSOCOMIAL INFECTIONS SURVEILLANCE SYSTEM Certain limitations are inherent in NNIS data, largely because it is a multi-institutional system with hundreds of data collectors. Diagnostic testing may vary from hospital to hospital, either because physicians make use of tests (e.g., blood culturing) differently or certain tests, such as viral cultures, may not be available at all hospitals. A previous study has shown that differences in diagnostic testing can result in differences in nosocomial infection rates [ill. However, unless there are marked differences in testing at an institution, intrahospital comparisons should still be valid. Surveillance intensity may vary from hospital to hospital either because case-finding methods vary or the efforts devoted to case finding vary even when the hospitals use the same method. This may be due partly to variation in the personnel resources available for surveillance. Alternatively, because hospitals voluntarily participate in the NNIS system, personnel at these hospitals may be more motivated to perform surveillance and institute control measures. This has possible implications when extrapolating results from NNIS hospitals to the general United States hospital population. Although approximately one third of NNIS hospitals report using some method of postdischarge surveillance for wound infections, no standard method exists. Developing a standard method that hospitals can conduct efficiently will remain a major methodological challenge for the 1990s. Other biases may limit the applicability of NNIS results to all hospitals, particularly because the NNIS system does not contain any hospital with fewer than 100 beds. Notably, one half of all United States hospitals have fewer 100 beds [121. In addition, the number of hospitals in the NNIS system is still small, which presents problems when performing certain stratified analyses. Finally, establishing a multihospital surveillance system has certain inherent problems. For efficiency and practicality, the amount of data collected must be limited, but it is often difficult to know in advance what data will be needed for an analysis. Thus, important information may be missing in NNIS analyses, such as information on September
16, 1991
ON NOSOCOMIAL
INFECTIONS
Other
Skin-Soft
/ GAYNES
ET AL
Tissue
~Bloodstream 32.3% Figure 2. Distribution ot nosocomial lntections by site In neonates, high-risk nursery component, National Nosocomial Infections Surveillance System, 1986-1990. EENT = Eye, Ear, Nose, and Throat; GI = Gastrointestinal; SWI = Surgical Wound Infection.
antibiotic use for examination tance trends.
of antibiotic
resis-
NEW DIRECTIONS AND USES FOR NATIONAL NOSOCOMIAL INFECTIONS SURVEILLANCE DATA The dissemination of results both to NNIS hospitals and to all hospitals was sporadic in the 1980s. Data from the ICU, HRN, and SP components required some time for accumulation, since their introduction in the fall of 1986. Publication of NNIS data is a part of these Proceedings [2-4,13,141. In addition, regular publications of NNIS analyses will appear in peer-reviewed journals in the coming months. The dissemination of NNIS data is a top priority. The NNIS system has led in the development of quality indicators, i.e., infection rates that better lend themselves to intrahospital and interhospital comparison. The NNIS risk index for SWI rates and device-associated rates for critical care units appear to lend themselves to these comparisons [2-41. These indicators will provide a basis for improved prevention and control efforts [15]. Revision of current surveillance component protocols and Interactive Data Entry and Analysis System (IDEAS), the microcomputer software, are planned. New components will be needed for the NNIS system. For example, a component for immunosuppressed hospitalized patients may be useful for improved prevention efforts for these high-risk patients. Validation studies in the NNIS system are planned to attempt to compare case-finding meth-
The American
Journal
of Medicine
Volume
91 (suppl3B)
38-l
19s
CONFERENCE
ON NOSOCOMIAL
INFECTIONS
/ GAYNES
ET AL
ods, measure intensity of surveillance, and determine the impact of variations in diagnostic testing. Technology advancements will enable the NNIS system to be more timely in its analysis. For example, a telecommunications computer could allow much more extensive and rapid data capture on patients, devices, procedures, and other parameters, and should be available soon at CDC. Adding more hospitals to the NNIS system would allow performance of additional stratified analyses and better estimation of the distribution of various types of infection rates. One major difficulty at present is in ensuring that appropriate resources are available for supporting greater numbers of hospitals. This includes ways to train new infection control personnel so that the quality of the data in the NNIS system continues to improve. Data exchange with other agencies such as the Joint Commission on Accreditation of Health-Care Organizations, the Health Care Financing Administration, the Food and Drug Administration, and other agencies will prevent duplication of efforts in infection control and allow data from the NNIS system to be used to its greatest potential. Finally, input from the NNIS hospitals that make up this surveillance system has been useful in the past. A NNIS consultant group made up of infection control practitioners and hospital epidemiologists would keep the NNIS system on the cutting edge of its field. To that end, a recent initiative for a Hospital Infections Control Practices Advisory Committee is under consideration by the Secretary of Health and Human Services. The NNIS system has many of the limitations of any multi-institutional surveillance system. Although the system has served an important function in the field of hospital epidemiology, its potential has not been fully realized, not only for infection control but in surveillance of noninfectious complications of hospitalization. With the continued support and the much appreciated involvement of its
3B-120s
September
16,
1991
The American
Journal
of Medicine
participating hospitals, the NNIS system will continue to evolve as it seeks to find more effective and efficient ways to measure the nosocomial infection experience and assess the influence of patient risk, changes in the delivery of hospital care, and changes in infection control practices on these measures. REFERENCES 1. Horan TC, White JW, Jarvis WR, et a/. Nosocomial infection surveillance, 1984. MMWR 1986; 35: 17SS-29. 2. Gaynes RP, Martone WJ, Culver DH, et a/. Comparison of rates of nosocomial infections in neonatal intensivecare units in the United States. Am J Med 1991; 91tSuppl3B): 192-6. 3. Jarvis WR, Edwards JR, Culver DH, et al. Nosocomial infection rates in adult and pediatric intensive care units in the United States. Am J Med 1991; 91(Suppl3!3): 185-91. 4. Culver DH, Horan TC, Gaynes RP, et al. Surgical wound infection rates by wound class, operative procedure, and patient risk index. Am J Med 1991; 91(Suppl 38): 152-7. 5. Emori TG, Culver DH, Horan TC, et al. National Nosocomial Infections Surveillance (NNIS) System: description of surveillance methodology. Am J Infect Control 1991; 19: 19-35. 6. Garner JS, Jarvis WR, Emori TG, Horan TC, Hughes JM. CDC definitions for nosocomial infections, 1988. Am J Infect Control 1988; 16: 128-40. 7. Banerjee SN, Emori TG, Culver DH, et al. Secular trends in nosocomial bloodstream infections in the United States, 1980-1989. Am J Med 1991; 91fSuppl3B): 86-9. 8. Selva J, Toledo A, Maroney A, Forlenza S. The value of participation in the CDC-National Nosocomial Infection Surveillance (NNIS) System in a large teaching hospital. Proceedings of Association for Practitioners of Infection Control, 1989: Sixteenth Annual Educational Conference, Reno, Nevada, May 21-26,1989. 9. Welliver RC, McLaughlin S. Unique epidemiology of nosocomial infections in a children’s hospital. Am J Dis Child 1984; 138: 131-5. 10. Jarvis WR. The epidemiology of nosocomial infections in pediatric patients. Pediatr Infect Dis 1987; 6: 344-51. 11. Haley RW, Culver DH, Morgan WM, White JW, Emori TG, Hooton TM. Increased recognition of infectious diseases in US hospitals through increased useof diagnostic tests, 1970-1976. Am J Epidemiol 1985; 121: 168-81. 12. American Hospital Association. Hospital Statistics, 1989-1990. Chicago, American Hospital Association, 1989. 13. Jarvis WR, and the Epidemiology Branch, Hospital Infections Program. Nosocomial outbreaks: The Centers for Disease Control’s Hospital Infections Program Experience, 1980-1990. Am J Med 1991; 91(Suppl3B): 101-6. 14. Emori G, Banerjee SN, Culver DH, et a/. Nosocomial infections in elderly patients in the United States, 1986-1990. Am J Med 1991; 91fSuppl): xx-xx. 15. Fuchs PC. Will the real infection rate please stand up? Infect Control 1987; 8: 235-6.
Volume
91 (suppl3B)
The National Nosocomial Infections Surveillance (NNIS) System is an ongoing collaborative surveillance system among the Centers for Disease Control (CDC) and United States hospitals to obtain national data on nosocomial infections. This system provides comparative data for hospitals and can be used to identify changes in infection sites, risk factors, and pathogens, and develop efficient surveillance methods. Data are collected prospectively using four surveillance components: hospital-wide, intensive care unit, high-risk nursery, and surgical patient. The limitations of NNIS data include the variability in case-finding methods, infrequency or unavailability of culturing, and lack of consistent methods for postdischarge surveillance. Future plans include more routine feedback of data, studies on the validity of NNIS data, new components, a NNIS consultant group, and more rapid data exchange with NNIS hospitals. Increasing the number of NNIS hospitals and cooperating with other agencies to exchange data may allow NNIS data to be used better for generating benchmark nosocomial infection rates. The NNIS system will continue to evolve as it seeks to find more effective and efficient ways to measure the nosocomial infection experience and assess the influence of patient risk, changes in the delivery of hospital care, and changes in infection control practices on these measures.
T
he Centers for Disease Control (CDC) National Nosocomial Infections Surveillance (NNIS) System is the only national source of nosocomial infections data in the United States. Many individuals at the CDC have been involved in the team effort of the NNIS system. However, the NNIS system requires the voluntary effort of the many infection control practitioners and hospital epidemiologists at the participating NNIS hospitals.
OBJECTIVES OF THE NATIONAL NOSOCOMIAL INFECTIONS SURVEILLANCE SYSTEM In its broadest terms, the NNIS system attempts to estimate the magnitude and nature of nosocomial infections in the United States. NNIS data aid in the prevention of nosocomial infections by identifying changes in infection rates and sites, risk factors, outcomes, pathogens, and antimicrobial resistance. Data on observed changes are used to redirect infection control and prevention efforts to areas of new or increasing concern [ 11. In addition, the NNIS system can provide hospitals with meaningful comparative data on nosocomial infections to aid in the prevention effort. Data for interhospital comparison must adjust for differences in the distribution of confounding variables such as duration of patient stay or device utilization. Several recent analyses attempt to provide hospitals with these types of data [2-41. Another NNIS objective is to assist hospitals in developing efficient and effective data collection and analysis methods which allow infection control personnel to use their time more efficiently.
HISTORICAL BACKGROUND
Diseases, Centers for Disease Control, Atlanta, Georgia. Hospital Infections Program A-07, Building 3, Room B16A, Centers for Disease Control, 1600 Clifton Road N.E., Atlanta, Georgia 30333.
3B-116s
September
16, 1991
The American
Journal
of Medicine
The NNIS system began in 1970 when selected U.S. hospitals sent data on nosocomial infections to the CDC for aggregation into a national database. Initially, all participating hospitals were required to conduct hospital-wide surveillance, i.e., all patients in the hospital were monitored for infections at all body sites. A major change occurred in 1986, when several alternative protocols for performing surveillance, called surveillance Volume
91 (suppl3B)
CONFERENCE
components, were introduced. As a part of this NNIS revision, definitions of nosocomial infections, including the use of clinical and laboratory algorithms, were expanded. The NNIS system also introduced microcomputer software designed specifically for the NNIS hospitals to facilitate the entry and analysis of data collected under the surveillance component protocols.
METHODS Hospitals currently participating in the NNIS system collect information on patients requiring acute care. Under Section 308(d) of the Public Health Service Act (42 U.S.C. 242b, k, and m(d)), the identity of NNIS hospitals is confidential. Data are released outside of the NNIS system only in aggregate form. Hospitals apply for membership in the NNIS system and must have adequate personnel support for infection control, a microcomputer compatible with the requirements of the NNIS software, and approval from hospital administration to participate in the NNIS system. Currently, 115 United States hospitals participate in the NNIS system. The current NNIS methods include four standardized surveillance protocols or surveillance components 151: Hospital-wide Adult and pediatric intensive care units High-risk nursery Surgical patient The surveillance components may be used singly or in any combination. These components are designed to form the foundation of a hospital’s overall surveillance program and allow flexibility for incorporating other aspects into their surveillance efforts. Once selected, a component must be followed for at least 1 month. NNIS hospitals use standard CDC definitions of nosocomial infections, published in 1988 161. The data collected on patients with nosocomial infections are essentially the same regardless of component. The differences between the components lie in the group of patients monitored for nosocomial infections and in the summary or denominator data that are collected in that group. The denominator data dictate the type of infection rates that may be calculated for a NNIS component. In the hospital-wide surveillance component, all patients are monitored for nosocomial infection at all sites. The denominator data collected each month consist of the total number of discharges from each NNIS-defined service and, optionally, the total number of patient-days spent on each service. These data allow calculation of overall September
16, 1991
ON NOSOCOMIAL
INFECTIONS
/ GAYNES
ET AL
infection rates and rates for each site of infection by service. In addition, because hospital-wide surveillance has been used since the beginning of the NNIS system, trend analyses covering 10 or more years can be performed 171. Because of difficulties in controlling for differences in case mix (e.g., severity of illness, medical interventions, or average duration of patient stay), caution is needed when any data from the hospitalwide component are used for interhospital comparison. In addition, many hospitals, particularly large hospitals, have insufficient resources for surveillance, causing significant underreporting of infections. Because of these problems, other components were developed so that surveillance efforts could be targeted and more specific denominators could be acquired. In the adult and pediatric intensive care units (ICUs) surveillance component, infection control practitioners monitor all ICU patients in one or more critical care units for nosocomial infections at all sites. The denominator data collected include the type of unit being monitored (e.g., medical, surgical, coronary) and the total number of admissions, patient-days, and device (central intravenous catheter, urinary catheter, or ventilator) days. With these data, we can calculate overall infection rates and device-associated infection rates for bloodstream, urinary tract, and respiratory infection sites. For example, central intravenous catheter-associated bloodstream infection rates are defined as: Number of catheter-associated bloodstream infections Number of central intravenous catheter days
x 1,000
By attempting to control for the exposure to the principal risk factor, such a device-associated rate can help to identify outliers, i.e., hospital units in which the rate is significantly above that in similar units (Figure 1). With the high-risk nursery (HRN) surveillance component, all HRN patients are monitored for nosocomial infections at all sites. The denominator data collected include the number of admissions, patient-days, and device-days within each of three birthweight categories ( < 1,500 g, 1500-2500 g, and >2,500 g). This allows calculation of overall infection rates and device-associated infection rates for bloodstream and respiratory infection sites, much as in the ICU component. The surgical patient (SP) surveillance component has two options: limited and detailed. In the SP surveillance component limited option, all pa-
The American
Journal
of Medicine
Volume
91 (suppl3B)
3B-117s
CONFERENCE
m
$ ‘ti :: 1 %
2
31:
0
ON NOSOCOMIAL
/ GAYNES
ET AL
4 indicates the Median rate
8 6 4
Rate IO
m -a .E 2
INFECTIONS
12
14
16
18
2
*
0
22
24
26
Surgical (n = 2.8, median = 5.1)
8 6
r” ‘i;
20
4 Rate 8
2
10
12
14
16
18
20
22
24
26
Pediatric (n = 15, median = 8.9)
‘Number
of Bloodstream
Infections per 1000 Central IV Catheter Days
tients undergoing all NNIS-defined operative procedures are monitored for infections. The summary data collected include a) the total number of procedures performed in each NNIS-defined operative procedure category, b) the total number of procedures performed in the four traditional wound classes (i.e., clean, clean-contaminated, contaminated, and dirty), and c) the total number of procedures performed by each surgeon (CDC does not receive data on individual surgeons). This allows calculation of infection rates by each of these factors. Thus, much like the hospital-wide component, detailed information on infections is collected with a general overall denominator in each category. An attempt to examine risk factors that are known to be associated with surgical wound infections was incorporated into the surgical patient surveillance component detailed option. All patients undergoing selected NNIS-defined operative procedures are monitored. This component differs from any other in that detailed information (e.g., age, wound class, duration of surgery, urgency of procedure, the surgeon(s), and whether the procedure was the result of trauma to the patient) is collected on every such patient. If a patient develops a nosocomial infection, further information on the infection is collected. Infection rates within categories of each potential risk factor (e.g., wound class) can be calculated. Alternatively, infection rates within the categories of a composite risk index can also be calculated. OVERVIEW OF RESULTS Device-associated rates from the adult and pediatric ICU component are shown in Figure 1. The hospital whose medical ICU had a rate of 41.4 catheter-associated bloodstream infections (BSI) per 1,000 catheter days is readily identified as an 3B-118s
September
16, 1991
The American
Journal
of Medicine
Figure 1. Distribution of central intravenous catheter-associated bloodstream infection rates by type of intensive care unit, Intensive Care Unit Component, National Nosocomial Infections Surveillance System, 1987-1989. Catheterassociated bloodstream infection rate is defined as the number of central intravenous catheter bloodstream infections per 1,000 central intravenous catheter
outlier. However, the identification of an outlier does not necessarily define a problem in that ICU. Rather, the outlier status only suggests an area for investigation by hospital personnel. In August 1988 a report was sent to NNIS hospitals with a distribution of data similar to that presented in Figure 1. One outlier hospital investigated the possible reasons for their elevated catheter-associated BSI rate and found several practicerelated problems. After taking corrective action, the hospital reduced its infection rate substantially. This use of NNIS data as a prevention tool was recently reported by Selva et al. @I. Other analyses of NNIS data, found elsewhere in these Proceedings, suggest that more meaningful intrahospital and interhospital comparisons can be made using these device-associated rates [2,31. Analysis of nosocomial infections among the neonates from the HRN component confirmed the striking differences from infections among adult patients reported in other studies (Figure 2) [9,10]. The site distribution of nosocomial infections shows a high percentage of BSIs for all birthweight groups. More than 85% of these occurred more than 2 days after birth and were usually catheter-associated. This suggests that surveillance and prevention efforts in HRNs should emphasize this site of infection. Using data taken from the SP componentlimited option, the pooled mean surgical wound infection @WI) rates for operative procedures can be calculated. For example, the rate for herniorrhaphy was 1.1 SW1 per 100 operations; for colon surgery it was 7.3. These rates give an estimate of the SW1 rates by operative procedure for the United States. However, for purposes of interhospital comparison, they ignore known risk factors, e.g., wound class or a patient’s intrinsic susceptibility to infection, which could be responsible for an Volume
91 (suppl3B)
CONFERENCE
individual hospital’s or surgeon’s variation from this pooled mean. The NNIS risk index attempts to control for some of this variation and is a substantial improvement over the stratification of operations done by the traditional method using wound class [41. LIMITATIONS OF THE NATIONAL NOSOCOMIAL INFECTIONS SURVEILLANCE SYSTEM Certain limitations are inherent in NNIS data, largely because it is a multi-institutional system with hundreds of data collectors. Diagnostic testing may vary from hospital to hospital, either because physicians make use of tests (e.g., blood culturing) differently or certain tests, such as viral cultures, may not be available at all hospitals. A previous study has shown that differences in diagnostic testing can result in differences in nosocomial infection rates [ill. However, unless there are marked differences in testing at an institution, intrahospital comparisons should still be valid. Surveillance intensity may vary from hospital to hospital either because case-finding methods vary or the efforts devoted to case finding vary even when the hospitals use the same method. This may be due partly to variation in the personnel resources available for surveillance. Alternatively, because hospitals voluntarily participate in the NNIS system, personnel at these hospitals may be more motivated to perform surveillance and institute control measures. This has possible implications when extrapolating results from NNIS hospitals to the general United States hospital population. Although approximately one third of NNIS hospitals report using some method of postdischarge surveillance for wound infections, no standard method exists. Developing a standard method that hospitals can conduct efficiently will remain a major methodological challenge for the 1990s. Other biases may limit the applicability of NNIS results to all hospitals, particularly because the NNIS system does not contain any hospital with fewer than 100 beds. Notably, one half of all United States hospitals have fewer 100 beds [121. In addition, the number of hospitals in the NNIS system is still small, which presents problems when performing certain stratified analyses. Finally, establishing a multihospital surveillance system has certain inherent problems. For efficiency and practicality, the amount of data collected must be limited, but it is often difficult to know in advance what data will be needed for an analysis. Thus, important information may be missing in NNIS analyses, such as information on September
16, 1991
ON NOSOCOMIAL
INFECTIONS
Other
Skin-Soft
/ GAYNES
ET AL
Tissue
~Bloodstream 32.3% Figure 2. Distribution ot nosocomial lntections by site In neonates, high-risk nursery component, National Nosocomial Infections Surveillance System, 1986-1990. EENT = Eye, Ear, Nose, and Throat; GI = Gastrointestinal; SWI = Surgical Wound Infection.
antibiotic use for examination tance trends.
of antibiotic
resis-
NEW DIRECTIONS AND USES FOR NATIONAL NOSOCOMIAL INFECTIONS SURVEILLANCE DATA The dissemination of results both to NNIS hospitals and to all hospitals was sporadic in the 1980s. Data from the ICU, HRN, and SP components required some time for accumulation, since their introduction in the fall of 1986. Publication of NNIS data is a part of these Proceedings [2-4,13,141. In addition, regular publications of NNIS analyses will appear in peer-reviewed journals in the coming months. The dissemination of NNIS data is a top priority. The NNIS system has led in the development of quality indicators, i.e., infection rates that better lend themselves to intrahospital and interhospital comparison. The NNIS risk index for SWI rates and device-associated rates for critical care units appear to lend themselves to these comparisons [2-41. These indicators will provide a basis for improved prevention and control efforts [15]. Revision of current surveillance component protocols and Interactive Data Entry and Analysis System (IDEAS), the microcomputer software, are planned. New components will be needed for the NNIS system. For example, a component for immunosuppressed hospitalized patients may be useful for improved prevention efforts for these high-risk patients. Validation studies in the NNIS system are planned to attempt to compare case-finding meth-
The American
Journal
of Medicine
Volume
91 (suppl3B)
38-l
19s
CONFERENCE
ON NOSOCOMIAL
INFECTIONS
/ GAYNES
ET AL
ods, measure intensity of surveillance, and determine the impact of variations in diagnostic testing. Technology advancements will enable the NNIS system to be more timely in its analysis. For example, a telecommunications computer could allow much more extensive and rapid data capture on patients, devices, procedures, and other parameters, and should be available soon at CDC. Adding more hospitals to the NNIS system would allow performance of additional stratified analyses and better estimation of the distribution of various types of infection rates. One major difficulty at present is in ensuring that appropriate resources are available for supporting greater numbers of hospitals. This includes ways to train new infection control personnel so that the quality of the data in the NNIS system continues to improve. Data exchange with other agencies such as the Joint Commission on Accreditation of Health-Care Organizations, the Health Care Financing Administration, the Food and Drug Administration, and other agencies will prevent duplication of efforts in infection control and allow data from the NNIS system to be used to its greatest potential. Finally, input from the NNIS hospitals that make up this surveillance system has been useful in the past. A NNIS consultant group made up of infection control practitioners and hospital epidemiologists would keep the NNIS system on the cutting edge of its field. To that end, a recent initiative for a Hospital Infections Control Practices Advisory Committee is under consideration by the Secretary of Health and Human Services. The NNIS system has many of the limitations of any multi-institutional surveillance system. Although the system has served an important function in the field of hospital epidemiology, its potential has not been fully realized, not only for infection control but in surveillance of noninfectious complications of hospitalization. With the continued support and the much appreciated involvement of its
3B-120s
September
16,
1991
The American
Journal
of Medicine
participating hospitals, the NNIS system will continue to evolve as it seeks to find more effective and efficient ways to measure the nosocomial infection experience and assess the influence of patient risk, changes in the delivery of hospital care, and changes in infection control practices on these measures. REFERENCES 1. Horan TC, White JW, Jarvis WR, et a/. Nosocomial infection surveillance, 1984. MMWR 1986; 35: 17SS-29. 2. Gaynes RP, Martone WJ, Culver DH, et a/. Comparison of rates of nosocomial infections in neonatal intensivecare units in the United States. Am J Med 1991; 91tSuppl3B): 192-6. 3. Jarvis WR, Edwards JR, Culver DH, et al. Nosocomial infection rates in adult and pediatric intensive care units in the United States. Am J Med 1991; 91(Suppl3!3): 185-91. 4. Culver DH, Horan TC, Gaynes RP, et al. Surgical wound infection rates by wound class, operative procedure, and patient risk index. Am J Med 1991; 91(Suppl 38): 152-7. 5. Emori TG, Culver DH, Horan TC, et al. National Nosocomial Infections Surveillance (NNIS) System: description of surveillance methodology. Am J Infect Control 1991; 19: 19-35. 6. Garner JS, Jarvis WR, Emori TG, Horan TC, Hughes JM. CDC definitions for nosocomial infections, 1988. Am J Infect Control 1988; 16: 128-40. 7. Banerjee SN, Emori TG, Culver DH, et al. Secular trends in nosocomial bloodstream infections in the United States, 1980-1989. Am J Med 1991; 91fSuppl3B): 86-9. 8. Selva J, Toledo A, Maroney A, Forlenza S. The value of participation in the CDC-National Nosocomial Infection Surveillance (NNIS) System in a large teaching hospital. Proceedings of Association for Practitioners of Infection Control, 1989: Sixteenth Annual Educational Conference, Reno, Nevada, May 21-26,1989. 9. Welliver RC, McLaughlin S. Unique epidemiology of nosocomial infections in a children’s hospital. Am J Dis Child 1984; 138: 131-5. 10. Jarvis WR. The epidemiology of nosocomial infections in pediatric patients. Pediatr Infect Dis 1987; 6: 344-51. 11. Haley RW, Culver DH, Morgan WM, White JW, Emori TG, Hooton TM. Increased recognition of infectious diseases in US hospitals through increased useof diagnostic tests, 1970-1976. Am J Epidemiol 1985; 121: 168-81. 12. American Hospital Association. Hospital Statistics, 1989-1990. Chicago, American Hospital Association, 1989. 13. Jarvis WR, and the Epidemiology Branch, Hospital Infections Program. Nosocomial outbreaks: The Centers for Disease Control’s Hospital Infections Program Experience, 1980-1990. Am J Med 1991; 91(Suppl3B): 101-6. 14. Emori G, Banerjee SN, Culver DH, et a/. Nosocomial infections in elderly patients in the United States, 1986-1990. Am J Med 1991; 91fSuppl): xx-xx. 15. Fuchs PC. Will the real infection rate please stand up? Infect Control 1987; 8: 235-6.
Volume
91 (suppl3B)