ARTICLES

NatZorral nosocomial infections surveillance system (NNIS): Description of surveillance methods T. Grace Emori, RN, MS David H. Culver, PhD Teresa C. Horan, MPH, CIC William Ft. Jarvis, MD John W. White, PhD David R. Olson, PhD Shailen Banerjee, PhD Jonathan R. Edwards, MS William J. Martone, MD Robert P. Gaynes, MD James M. Hughes, MD Atlanta, Georgia

The National Nosocomial Infections Surveillance System (NNIS) is an ongoing collaborative surveillance system sponsored by the Centers for Disease Control (CDC) to obtain national data on nosocomial infections. The CDC uses the data that are reported voluntarily by participating hospitals to estimate the magnitude of the nosocomial infection problem in the United States and to monitor trends in infections and risk factors. Hospitals collect data by prospectively monitoring specific groups of patients for infections with the use of protocols called surveillance components. The surveillance components used by the NNIS are hospitalwide, intensive care unit, high-risk nursery, and surgical patient. Detailed information including demographic characteristics, infections and related risk factors, pathogens and their antimicrobial susceptibilities, and outcome, is collected on each infected patient. Data on risk factors in the population of patients being monitored are also collected; these permit the calculation of risk-specific rates. An infection risk index, which includes the traditional wound class, is being evaluated as a predictor of the likelihood that an infection will develop after an operation. A major goal of the NNIS is to use surveillance data to develop and evaluate strategies to prevent and control nosocomial infections. The data collected with the use of the surveillance components permit the calculation of risk-specific infection rates, which can be used by individual hospitals as well as national health-care planners to set priorities for their infection control programs and to evaluate the effectiveness of their efforts. The NNIS will continue to evolve in finding more effective and efficient ways to assess the influence of patient risk and changes in the financing of health care on the infection rate. (AM J INFECT CONTROL 1991;19:19-35)

From the Hospital Infections Program, Center for Infectious Diseases, Centers for Disease Control. Reprint requests: T. Grace Emori, RN, MS, Hospital Infections Program, Center for Infectious Diseases, Centers for Disease Control United States Public Health Service, Department of Health and Human Services, Atlanta, GA 30333. 17/46/25024

The National Nosocomial Infections Study (NNIS) began in 1970, when the Centers for Disease Control (CDC) invited selected hospitals to routinely report their nosocomial infection surveillance data for aggregation into a national data base. Later renamed the National Nosocomial Infections Surveillance system, it 19

Amencan Journal of

20

Emori

et al.

is the only source of national data on the epidemiology of nosocomial infections in the United States.‘s4 The objectives of the NNIS are as follows: 1. To estimate the incidence of nosocomial infections in the United States. 2. To identify trends in infection rates, sites, risk factors, patient outcomes, nosocomial pathogens, and antimicrobial resistance. 3. To provide hospitals with comparative data on nosocomial infections that they can use to evaluate their prevention and control efforts. 4. To develop efficient and effective data collection and analysis methods for nosocomial infection control. 5. To conduct collaborative research studies to describe the epidemiology of emerging infections and pathogens, assess the importance of potential risk factors, and further characterize mechanisms of antimicrobial resistance in nosocomial pathogens. At the inception of the NNIS, participating hospitals used the surveillance method and infection definitions developed by the CDC in pilot studies.‘-’ All patients were continuously monitored for infections at all sites, and the denominator was the number of patients discharged during the surveillance period. This system remained essentially unchanged until October 1986, when the need for more precise measurements of nosocomial infection risks and outcomes in specific patient groups led to the introduction of the NNIS surveillance components. Hospitals participating in the NNIS now have flexibility in designing their own surveillance programs, since they are able to select the surveillance components to use for the period of time they desire. This article describes the current methods used in the NNIS to collect nosocomial infection data. CNANACl’EN1STlCS OF PARTICIPATlwQ llOdPItALS

Hospitals that are eligible to participate in the NNIS provide general medical-surgical inpatient services to adults or children requiring acute care. These hospitals may be of any size

INFECTION CONTROL

and ownership, affiliated with a medical school, and located anywhere in the United States. Although participation is voluntary, hospitals must apply for membership in the NNTS and must have adequate personnel support for infection control, a microcomputer that is compatible with the requirements of the NNIS surveillance software, and approval from hospital administration to participate in the NNIS. One hundred sixteen hospitals currently participate in the NNIS. When compared with all U.S. hospitals, these institutions overrepresent teaching hospitals and underrepresent small nonteaching hospitals.8 NNtS SUNVW.tAMOB C-S All NNIS data are collected with the use of four standardized protocols, called surveillance components: hospital-wide, adult-and pediatric intensive care unit, high-risk nursery, and surgical patient. The components may be used singly or simultaneously, but, once selected, they must be used for a minimum of one calendar month. The decision about which surveillance components to use and for how long is made by each hospital. Each hospital is urged to develop a long-term surveillance plan and to set objectives for their surveillance. All infections reported to the NNIS must have occurred in patients who were admitted to the hospital and who stayed at least overnight; patients on the psychiatry, physical medicine and rehabilitation, or extended-care services are excluded. All infections are categorized into major and specific infection sites by the use of standard CDC definitions that include clinical and laboratory criteria.’ The data collected on infected patients are the same, regardless of the surveillance component used (Fig. 1). Hospitals may also collect additional data of special interest on infected patients for their own use. The summary or denominator data, which include the number of patients exposed or the number of patient-days of exposure to selected risk factors, are different for each component. The data reported to the CDC must conform to the protocol of the selected surveillance component before they are entered into the national data base.

Volume 19 Number 1

National nosocomial infections surveillance system 21

February 1991

Hospitabwide

surveillance

component

The hospital-wide component is similar to the surveillance method used before October 1986 to collect NNIS data in that all NNIS patient admissions are monitored for nosocomial infections at all body sites. The required denominator data are the number of patients discharged or admitted during the month from each of 17 NNIS-defined hospital services (Table 1) and the total number of vaginal and cesarean-section deliveries performed. Optionally, hospitals may collect the following data: (1) the number of patient-days spent on each NNIS-defined service; (2) the number of discharges (or admissions) from each ward; and (3) the number of discharges in each diagnosis-related group (DRG).” The distribution of infections and risk factors among infected patients, as well as site-specific infection rates for each of the hospital services and for wards and DRGs, may be calculated with the use of hospital-wide surveillance data. Adult and pediatric intensive surveillance component

care unit (ICU)

The ICU surveillance component focuses on patients in intensive care areas designated for critically ill patients. An ICU-associated infection is one that was not present or incubating at the time of the patient’s admission to the ICU but became apparent during the ICU stay or within 48 hours after transfer from the ICU. Hospitals with more than one ICU may do surveillance in any or all ICUs, but in the selected units all of the patients are monitored for infections at all sites. Data are collected for 10 types of ICU: burn, coronary care, medical, combined medical and surgical, neurosurgical, pediatric, respiratory, surgical, trauma, and other. The denominator data collected are the total number of patients in the ICU during the month, total number of patient-days, urinary catheter-days, central line-days, and ventilatordays (Fig. 2). Calculation of site-specific infection rates can be based on the appropriate denominator (e.g., catheter-associated urinary tract infections divided by the total number of indwelling urinary catheter-days). A severity-of-illness index, which is an ad-

aptation of a previously published clinical classification system, is being evaluated for standardizing ICU infection rate.” The clinical classification codes are shown in Table 2. Every patient in the unit is assessed once a week by direct observation by surveillance or ward personnel, who assign a score depending on the severity of illness. Individual patient scores are combined and averaged into a weekly and then a monthly severity-of-illness score for the unit monitored. High-risk nursery (HRN) surveillance component The HRN surveillance component focuses on infants in the level III nursery, which is defined by the American Academy of Pediatrics as one that provides multisystem support or critical care for unstable neonates and is staffed by a full-time board-certified pediatrician with special qualifications in neonatal medicine and by nurses specially trained in perinatal care.** In the HRN component, all neonates in the level III nursery are monitored for infections at all body sites. An HRN-associated infection is one that was not present or incubating at the time of the neonate’s admission into the HRN but which became apparent during the HRN stay or within 48 hours after transfer from the HRN. Neonatal infections that occur as the result of passage through the birth canal are considered nosocomial, whereas those that are acquired through transplacental transmission are not.g Denominator data are collected separately for each of three birthweight categories51500 gm, 1501 to 2500 gm, and >2500 gmand include the total number of patients in the HRN during the month, total number of patient-days, umbilical catheter/central linedays, and ventilator-days (Fig. 3). Site-specific infection rates can be calculated on the basis of the appropriate denominator (e.g., ventilatorassociated pneumonia divided by the total number of ventilator-days). Surgical

patient

surveillance

component

In this component, patients who have undergone NNIS operative procedures (Table 3) are

22

American Journal of iNtECl!ON CONTROL

Emori et al.

NNIS INFECTION Idedkn

NNID #: Discharge

ID %-Type: N

DRG:

Admission

Sex: M

Patient name:

Patient ID #:

Servke:

wa No. W%W12 EXT.02/29/93

WORKSHEET

dd

years/montix/dey

YY

OPH ORT PED PLS

GU GVN MED NS ONC

BT cs DEN ENT GS

Age:

Ward:

date: -----. mm

F

HRN -

Maternally aquired: Birtliwt in grams:

Y N (A) s 1500

OB -

Vaginal delivery:

Y

N

WBN -

Maternally

Y

N

acquired:

(8) 1501-2500

(C) > 2500

opt1 :

m

Y

N

Type of ICU:

B

C

M

MS

N

NS

P

R

S

T

0: tw=-w

=‘N re:

AMP HN 006

APPY HVST ORES

BILI CABG CARD CHOL MAST NEPH OBL OCVS OSKN PROS PRST SB -

Date of of.leratfon: mm i&bud&as Gem&-Y Emergency:

C

Infection date: m

CC Y

-

hrs.

min

YY

D

U

Surgeon: ASA clasdficetion: Trauma: Y N

1

2 3 Implant:

4 Y

5 N

-

AS8 mmSUTfdd OUTI lndwelffng urfnary catheter: Other Madder instrumentation:

1 PNEU 1 CXR: Del Ventifator:

Poss Neg Y N

Y Y

El

Incisional

@El

g$

Not Done

Spacifk

infect&n:

Y

Died:

Y

line(s):

Y

N N

TPN: Umtilical

Y

oplfer~ef

Y Y

N N

N

N

Refatknahii

to death:

opt3: Additfenat

catheter:

site:-

device/prccedure:

N

specifksite:

um(s).yy

Perfpheral

Invasive

bloodstream

Deep If deep,

N N

Majorsite:

Secondary

FX GAST CRAN CSEC FUS HER OES OEVE OGIT OGU OMS ONS SPLE THOR TP VHVS VS VBHN XtAP

Duration of operation:

dd CO N N

COLO OENT SKGR

Ffefde

a3 -

ud------d3 ~

Fig. 1. NNIS infection worksheet.

a5 b5

CA

CO

NR

U

Volume 19 Number 1

National nosocomial infections surveillance system 23

February 1991

Laboratory NNID#: Laboratory diagnosis: Culturespecimen: B Opt4:

Data Infection ID #:--

C BX

A V N CSF DD

ID

NSD

R

S

ST

U

1. AMIKACIN 2. AMOXICILLINI CLAVULANlC ACID 3. AMPICILLIN 4. AMPICILLIN/ SULBACTAM 5. AZLOCILLIN 6. AZTREONAM 7. CARBENJCILLIN or TICARCILUN 8. CEFACLOR 9. CEFAMANDOLE 10. CEFAZOUN 11. CEFIXIME 12. CEFONICID 13. CEFOPERAZONE 14. CEFOTAXIME 15. CEFOTETAN 16. CEFOXITIN 17. CEFTAZlDlME 16. CEmi!OXIME 19. CEFTRIAXONE 20. CEFUROXIME 21. CEFUROXIME AXETIL 22. CEPHALOTHIN 23. CHLORAMPHENICOL 24. CINOXACIN 25. CIPROFLOXACIN 26. CLINDAMYCIN 27. ENOXACIN 28. ERYTHROMYCIN 29. GENTAMICIN 30. IMIPENEM 31. KANAMYCIN 32. METHICILLIN 33. METRONIDAZOLE 34. MEZLOCILLIN 36. MOXALACTAM 36. NAFCILLIN 37. NALIDIXIC ACID 38. NETILMICIN 39. NfTROFURANTOlN 40. NORFLOXACIN 41. OXAClLLlN 42. PENICILLIN 43. PIPERACILLIN 44. RIFAMPIN 45. SULFISOXAZOLE 46. TETRACYCLINE 47. TICARCILLIfV CLAVULANIC ACID 48. TOBRAMYCIN 49. TRIMETHOPRIM 60. TRIMETHOPRIMI SULFAMETHOXAZOLE 51. VANCOMYCIN

Fig. 1. continued

VC OTH

24 Table

Emori et al. 1. NNIS-defined

services

Medicine Medicine (MED) Medical oncology (ONC) Surgery Burn/trauma (BT) Cardiac surgery (CS) Dental surgery (DEN) General surgery (GS) Neurosurgery (NS) Plastic surgery (PLS) Ophthalmology (OPH) Orthopedic surgery (ORT) Otolaryngology (ENT) Urology (GU) Gynecology (GYN) Obstetric (OB) Pediatric (PED) Newborn Well-baby nursery (WBN) High-risk nursery (HRN)

monitored for infections at the surgical wound site only or at all body sites, depending on the choice of the hospital. The cohort of patients includes those who have operations during the same month. There are two options for collecting denominator data, detailed or limited. Detailed option. In the detailed option, infection control personnel select the procedures they wish to follow from the NNIS operative procedures (Table 3). All patients undergoing the selected operative procedures are observed for infections. Risk factor data are collected on each patient; these include age, sex, date of operation, duration of surgery, wound class, use of general anesthesia, the ASA class,13 whether the operation was an emergency or was necessitated by trauma, and, optionally, the surgeon’s name or code (Fig. 4). Infection rates can be calculated for each type of denominator data collected (e.g., operative procedure and wound class), as well as for combinations of denominator data (e.g., operative procedure by wound class). In addition, the hospitals are able to calculate surgeon-specific infection rates if the surgeon’s name or code is collected. Data on surgeons are not reported to the CDC. An infectionrisk index that combines measures of patient risk (i.e., duration of surgery), ASA class, and wound class (i.e., clean, clean-contaminated, contaminated, and dirty) is currently being

Table

2. NNIS ICU severity-of-illness

classification

cllnical

codes

Postoperative patient requiring routine postoperatlve observation but not requiring intensive nursing or physician care. These patients are usually discharged from the ICU within 48 hours: some of these patiei!ts could have been handled in a recovery roar Physiologically stable patients requiring p:ophylactic overnight observation without intensive nursing or medical care (the medical equivalent of the routine postoperative patient). Examples of such patients would be those admitted to rule out a myocardlal infarction and patients who are stable but have had drug Ingestion Physiologically stable patlents requiring intensive nursing and monitoring (e g., a stable pattent in coma or with chronic renal failure) Physiologically unstable patinets requlnng intensive nursing and medical care with the need for frequent reassessment and adjustment of therapy (e.g patients with cardiac arrhythmias, diabetic ketoacidosls without coma, septic shock, disseminated intravascular coagulation) Physiologically unstable patients who are In coma or shock (e.g BP