Hospital-acquired morbidity on a neurology service.

HOSPITAL ACQUIRED MORBIDITYON A NEUROLOGY SERVICEStephen Q. Shafer, MD, MPH, John C.M. Brust, MD, Edward B. Healton, MD, andJoan B. Mayo, MS, MBANew York, New York

Clinical services must monitor hospital-acquired morbidity, but what rates are ex-pected specifically for neurology inpatients isnot evident from published studies. We studiedprospectively 1317 consecutive admissions toa neurology service in a university-affiliatedcity hospital from 1987 to 1990 and recorded allnosocomial infections, nosocomial pneumo-nia, and decubitus ulcers of stage Ill or IV. Overthe 3-year period, 6.8% of patients had :1nosocomial infection (and almost half of thesehad a nosocomial bloodstream infection); 3.1%had ¢1 case of nosocomial pneumonia; 1.2%developed severe decubitus ulcers, and 8.4%had one or more of the three complications.The incidence of nosocomial infection exceedsthat expected from multihospital studies. Howmuch of the excess is peculiar to neurologypatients and how much can be attributed tofactors in our community and at our hospitalcannot be determined from this study. Further-more, our statistics are not meant as norms,but as initial estimates for quality assurance. (JNati Med Assoc. 1993;85:31-35.)

Key words * nosocomial infection * nosocomialpneumonia decubitus ulcers * quality assurance

* morbidity

Most figures on hospital-acquired morbidity are fromintensive care units, entire hospitals, or multihospitalsurveys. There are few studies that have recorded the

From the Department of Neurology, Harlem Hospital Center,the Columbia University College of Physicians and Surgeons,and the Infection Control Service, Harlem Hospital Center, NewYork, New York. Requests for reprints should be addressed toDr Stephen Q. Shafer, Dept of Neurology, Harlem HospitalCenter, Room 16-103, 506 Lenox Ave, New York, NY 10037.

occurrence of nosocomial pneumonia, nosocomialinfection, or severe decubitus ulcers on a neurologyservice. This article reports the 3-year incidence ofthese occurrences on a city hospital neurology service.

METHODSHarlem Hospital Center is a 678-bed university-

affiliated municipal hospital in an inner-city area notedfor poor health conditions. In 1979 to 1981, death ratesin Harlem were twice those for US whites and 50%higher than those for US blacks.' On the 32-bedneurology service, stroke is the most common reasonfor admission, with one third of discharge diagnosesfalling into an ICD-9 stroke rubric category and 19%into an epilepsy or seizure category. Table 1 lists the 12diagnosis related groups (DRGs) that each accountedfor more than 1% of discharges. In sum, these 12 DRGsaccounted for two thirds of the discharges.From October 1, 1987 through September 30, 1990,

the Harlem Hospital Center Department of Neurologyquality assurance committee prospectively monitoredthe incidence of three cardinal hospital-acquired ill-nesses: nosocomial infections, nosocomial pneumonia,and decubitus ulcers. One full-time practitioner did allcase finding in 1317 consecutive admissions. Thefindings were validated in monthly meetings with wardstaff and with the hospital's infection control service.

Nosocomial infection surveillance was only forbacteria and fungi. Evidence of infection (eg, fever,leukocytosis, and pyuria) had to be manifested noearlier than 72 hours after admission. A pathogenicorganism had to be cultured and judged by the treatingclinician to represent infection. We attributed blood-stream infection to urinary tract infection if the sa'mepathogen was recovered from both sites on the sameday. Bloodstream infection with concurrent urinarytract infection was declared if a different organism wascultured from each site on the same day. In this article,

JOURNAL OF THE NATIONAL MEDICAL ASSOCIATION, VOL. 85, NO. 1 31

HOSPITAL-ACQUIRED MORBIDITY

TABLE 1. DIAGNOSIS RELATED GROUPS(DRGs) ACCOUNTING FOR 1% OR MORE OFDISCHARGES FROM NEUROLOGY SERVICE,AS PERCENTAGE OF TOTAL ADMISSIONS

% ofDRG Discharges

014 Specific cerebrovasculardisorders except for TIA 29

024 Seizure or headache age>69 and/or cc 13

025 Seizure or headache age 19to 69 without cc 9

015 TIA and precerebralocclusion 4

750 Alcohol dependency 3012 Degenerative nervous

system disorders 2429 Organic disturbances and

mental retardation 2294 Diabetes age >35 1010 Nervous system neoplasms

age >69 and/or cc 1011 Nervous system neoplasms

age <70 without cc 1013 Multiple sclerosis and

cerebellar ataxia 1134 Hypertension 1

TOTAL 67

Abbreviations: TIA = transient ischemic attack andcc = complication or comorbidity.

only the first nosocomial infection of the admission isconsidered,, although about 20% of patients had morethan one.

Nosocomial pneumonia was declared when two ormore of the following signs or symptoms developedmore than 72 hours after admission: fever, tachypnea,hypoxemia, adventitious sounds, or tracheobronchialsecretions from which a pathogen was cultured. Aconfirmatory radiograph was not required. Noso-comial pneumonia was listed separately from noso-comial infection because in hospital-acquired pneu-monia a culture is not as reliable as in othernosocomial infection.2'3 Only the first nosocomialpneumonia of the admission is considered in thisarticle.

Decubitus ulcers were categorized by stage.4 Lesssevere decubitus ulcers (stages I and II) were notmonitored for the entire 3 years. Stage III ulcers weredefined by loss of dermis with exposure of subcutane-ous tissue or muscle; stage IV ulcers were defined bypenetration of fascial planes or exposed bone.

TABLE 2. NUMBER OF PATIENTS WITH:1 NOSOCOMIAL INFECTION OR

¢1 NOSOCOMIAL PNEUMONIA IN 1317CONSECUTIVE NEUROLOGY ADMISSIONS, BY

SITE OF NOSOCOMIAL INFECTION

Total Rate per 1000Count Admission

BSI without UTI 26 19.7BSI due to UTI 7 5.3BSI, concurrent UTI 6 4.6

BSI subtotal 39 29.6

UTI no BSI 49 37.2

Other NI 1 0.8

Total with _1 NI,excluding NP 89 67.6

¢1 NP with no otherNI 20

Total with :1 NI or¢'1 NP 109 82.8

Abbreviations: BSI = bloodstream infection,UTI=urinary tract infection, Nl=nosocomial in-fection, and NP = nosocomial pneumonia.

RESULTSNosocomial Infection

Eighty-nine individuals (68/1000 admissions) had atleast one nosocomial infection. The distribution of firstnosocomial infection is listed by site in Table 2.Thirty-nine patients (30/1000) had a bloodstreaminfection as their first nosocomial infection. Thebloodstream infection was caused by urinary tractinfection in seven cases (5/1000 admissions). Gram-negative organisms were isolated in all seven, butEscherichia coli was isolated from just one. In six othercases of bloodstream infection, there was a concurrenturinary tract infection; in only one of these was E colithe bacteremic organism. Gram-negative organismswere recovered in 24/39 episodes of bloodstreaminfection (eleven Enterobacter, five Pseudomonas, fourKlebsiella, two Acinetobacter, and two others). Gram-positive organisms caused 13 of the first nosocomialinfections that presented as bloodstream infection (fourStaphylococcus aureus, five other Staphylococcusspecies, and four Streptococcus species). In twopatients, the first nosocomial infection was a Candidabloodstream infection.

The first nosocomial infection was a urinary tractinfection without bloodstream infection in 49 patients(37/1000). Escherichia coli accounted for 30 of theseinfections and Klebsiella for six. A total of 62 patients

32 JOURNAL OF THE NATIONAL MEDICAL ASSOCIATION, VOL. 85, NO. 1

HOSPITAL-ACQUIRED MORBIDITY

(47/1000) had a urinary tract infection as the firstnosocomial infection, of whom 13 also had a blood-stream infection in the same episode.Only one case of nosocomial meningitis due to

bacteria or fungus occurred in the 1317 admissions.This was in a cerebral infarction patient who hadStreptococcus pneumoniae bacteremia and meningitiswith the same organism. A single case of nosocomialerysipelas, confirmed by a dermatologist, representedthe only nosocomial infection in this series that was nota bloodstream infection, a urinary tract infection, orpneumonia.

Clusters of infection pointing to person-to-persontransmission did not stand out. There were, however,two pairs that may have represented this phenomenon.In October 1987, two very sick patients in adjoiningbeds both had Pseudomonas sepsis, and in November1987, two patients in different rooms each hadEnterobacter cloacae sepsis. The predominance of Ecoli urinary tract infections suggests that patients onthis floor were at exceptional risk for this type ofinfection.

Nosocomial PneumoniaForty-one patients (31/1000 admissions) had one or

more episode of nosocomial pneumonia. Twenty-six ofthese 41 were put on mechanically assisted ventilationbefore or during the pneumonia. At some time duringthe admission, 21 of the 41 patients also had : 1nosocomial infection that was not pneumonia. Nine ofthe 21 first nosocomial infections were bloodstreaminfections and 12 were urinary tract infections.

Decubitus UlcersDecubitus ulcers of stage III or IV occurred in 16

patients (12/1000 admissions). Six of these 16 hadarrived with less advanced bedsores. Five other pa-tients, admitted with stage III or IV decubitus ulcers,were not counted here. Thus, in 1306 patients arrivingwithout skin breakdown, only 10 (8/1000) developed denovo decubitus ulcers of stage III or IV. More than halfof the stage III or IV cases involved feet, usually theheel or lateral malleolus of a paralyzed limb. Decubitusulcers of stage I or II, tracked for only 30 of the 36months, appeared in patients admitted with intact skinabout four times as often as stage III or IV lesions.Thus, the probability of incurring a stage I or IIdecubitus ulcer might be estimated as 32/1000, and theprobability of developing a decubitus ulcer of any stagecan be estimated at about 40/1000.One hundred eleven (8.4%) patients of the 1317 had

TABLE 3. JOINT DISTRIBUTION OF DECUBITUSULCERS OF STAGE III and IV AND NOSOCOMIALINFECTION AND NOSOCOMIAL PNEUMONIA IN1317 CONSECUTIVE NEUROLOGY ADMISSIONS

No. of No. ofPatients With Patients

DecU Without DecU

NlwithNP 6 15NI without NP 6 62NP without NI 3 17Neither NP nor NI 2 1206Total 17 1300

Abbreviations: DecU = decubitus ulcer, NI = noso-comial infection, and NP=nosocomial pneumo-nia.

one or more of the three complications (nosocomialinfection, nosocomial pneumonia, or a severe decubitusulcer). The overlap among these three cardinal compli-cations is shown in Table 3. More than one of thecomplications occurred in 30 patients (2.3%), and 6(0.5%) had all three. The three complications cluster-patients were they independent, not even one personwith all three would have been expected.

DISCUSSIONThe incidence of bacteriologically documented first

nosocomial infections on this service (68/1000 or 6.8%)is higher than in several larger studies. Haley et alestimated a nationwide rate of 5.7% based on a randomsample of US hospitals.5 At the University of Virginia,the whole-hospital figure was 6%.6

Adjustments are needed, however, to compare aneurology service to another level of observation suchas an entire hospital. The national figure included13.7/1000 nosocomial infections due to surgical woundinfections, which are rare in neurology.5 These re-moved, the multihospital sample expectation falls to4.3%. Haley et al also counted nosocomial pneumonias(5.7/1000), which are not included in our nosocomialinfection incidence. If nosocomial pneumonia is ex-cluded from the multihospital figure, the expectationdrops to about 3.8%. The nosocomial infection rate onour service is almost twice that. If nosocomialpneumonia were counted as nosocomial infection, thenin our series 20 patients would be added to the 89 whosuffered nosocomial infection. This new figure (83/1000) is about double the comparable figure (43/1000)from the multihospital survey.The rate on our service is closer to the 12% reported

from an oncology service7 than to the 4% reported from

JOURNAL OF THE NATIONAL MEDICAL ASSOCIATION, VOL. 85, NO. 1 33

HOSPITAL-ACQUIRED MORBIDITY

another neurology service.6 This 4% figure, from theUniversity of Virginia in the early 1970s, does notsupport our hypothesis that neurology patients, oftenimmobile and beset by other systemic illnesses, are atmuch higher risk than hospital patients in general. Tworeports cannot decide this question, especially whenthey are separated by 20 years and perhaps other factors.A striking feature of nosocomial infections on this

service is the high rate of bloodstream infections, whichat 30 individuals/1000 admissions, is 10 times morelikely than in one multihospital sample (2.9/1000),5 andsix times greater than an average (5/1000) seen inselected municipal hospitals.8 Also remarkable is thehigh proportion (7/62) of urinary tract infection patientswho had a bloodstream infection, which was brought onby the urinary tract infection. Krieger et a19 observedthat only 2.7% of nosocomial urinary tract infectionsresulted in a bloodstream infection. Our figure forbloodstream infections due to nosocomial urinary tractinfections was 5.3/1000, far above the average hospital-wide rate of 0.73/1000 reported from Columbia, SouthCarolina 10 or the 1/1000 from Charlottesville, Vir-ginia.9A nosocomial pneumonia figure to compare with

ours (31/1000) is not readily available. One reviewgives the wide range of 4-50/1000 admissions ingeneral units and up to 220/1000 in intensive careunits."1 In some reports, nosocomial pneumonia isconsidered a nosocomial infection. The multihospitalsurvey of Haley et al indicates an incidence of5.7/1000.5 Wenzel et al recorded 6.7/1000 on theneurology service at the University of Virginia, Char-lottesville, Virginia.6 Our figure is over four timeshigher. As was the case with nosocomial infection, theUniversity of Virginia incidence contradicts our hy-pothesis that neurology patients run extra risk. Thosewith stroke are certainly more likely to have impaireddeglutition and airway defenses than most otherpatients.'2"13 Such a characteristic of one major patientgroup, it seems, may not always lead to high noso-comial pneumonia incidence on a neurology service.

Published statistics give only a broad range ofincidence figures for decubitus ulcers, and those aregiven without specifying stage. Versluysen refers totwo studies in Scotland that found incidences of 8.8%and 9.2%; at St Bartholomew's hospital (London), 32%(90/283) of patients admitted for elective or emergencyhip surgery developed pressure sores.14 Gerson found a2.7% incidence in three hospitals in Newfoundland;300 patients with "diseases of the nervous system andsense organs" had a 2.3% incidence.'5

Allman et al noted an incidence of 7.7% at the JohnsHopkins hospitals over 3 weeks in patients "atrisk ... expected to be confined to bed or chair for atleast one week."16 In that study, however, only one insix patients was "at risk." The hospital-wide pressuresore incidence would then be 7.7%/6, or about 1.3%. Atleast four in six of our patients, however, were "at risk"by these criteria. Considering the high proportion ofpatients on our neurology service who were "at risk"by the criteria of Allman and colleagues, the 0.8%incidence of severe decubitus ulcer and the 4%extrapolated estimate for decubitus ulcer of all stagesmay not be excessive.

CONCLUSIONThat incidences of two of the three complications

were higher than would be expected from publishedfigures is hardly encouraging. We note, however, thatpublished figures are seldom drawn from a comparablelevel of observation (ie, service, as opposed toinstitution or group of institutions), and almost neverfrom the same specialty. Our study cannot sort out theproportionate contributions to the apparent excessincidences from the following likely sources of varia-tion: hospital locale, timeliness of admission, case mixof diagnoses and comorbid conditions, hospital qualityof care, neurology service quality of care, and risks suchas severe immobility or combined bowel and bladderincontinence often associated with neurological dis-eases.Our record is offered not as a norm, but to help other

nonintensive care unit neurology services and generalmedicine services to identify problems and to measureany impact of quality assurance activities. Variations inmix of risk factors or in data-collection methodsthreaten the validity of comparisons between services.Within a service, however, data collection can becontrolled. We noted a decrease in nosocomial infec-tion, in nosocomial pneumonia, and in severe decubitusulcers in the latter half of the 3 years. The decline wasnot statistically significant, but it was heartening.

AcknowledgmentsThe authors thank the staff of 1 6N, Harlem Hospital Center,

and B. Fox, RN, Hospital Infection Control Service.

Literature Cited1. McCord C, Freeman HP. Excess mortality in Harlem. N

Engl J Med. 1990;322:173-177.2. Stein D. Managing pneumonia acquired in nursing

homes: special concerns. Geriatrics. 1990;45:39-47.3. Tobin MJ, Grenvik A. Nosocomial lung infection and its

diagnosis. Crit Care Med. 1984;12:191-199.

34 JOURNAL OF THE NATIONAL MEDICAL ASSOCIATION, VOL. 85, NO. 1

HOSPITAL-ACQUIRED MORBIDITY

4. Shea JD. Pressure sores-classification and manage-ment. Clin Orthop. 1975;112:89-100.

5. Haley RW, Culver DH, White JW, Morgan WM, EmoriTG. The nationwide nosocomial infection rate: a new need forvital statistics. Am J Epidemiol. 1985;1 21:159-167.

6. Wenzel RP, Osterman CA, Hunting KJ. Hospital-acquired infections, II: infection rates by site, service andcommon procedures in a university hospital. Am J Epidemiol.1976; 1 04:645-651.

7. Rotstein C, Cummings KM, Nicolas AL, Lucey J,Fitzpatrick J. Nosocomial infection rates in an oncology center.Infect Control Hosp Epidemiol. 1988;9:13-19.

8. Maki DG. Nosocomial bacteremia. An epidemiologicaloverview. Am J Med. 1981 ;70:719-732.

9. Krieger JN, Kaiser DL, Wenzel RP. Urinary tract infectionetiology of bloodstream infections in hospitalized patients. JInfect Dis. 1983;1 48:57-62.

10. Bryan CS, Reynolds KL. Hospital-acquired bacteremicurinary tract infection. Epidemiology and outcome. J Urol.1984; 1 32:494-498.

11. Celis R, Torres A, Gatell JM, Almela M, Rodriguez-Roisin R, Augusti-Vidal A. Nosocomial pneumonia. A multivari-ate analysis of risk and prognosis. Chest. 1988;93:318-324.

12. Horner J, Massey EW. Silent aspiration following stroke.Neurology. 1988;38:317-319.

13. Veis SL, Logemann JA. Swallowing disorders in personswith cerebrovascular accidents. Arch Phys Med Rehabil.1 985;66:372-375.

14. Versluysen M. Pressure sores in elderly patients. Theepidemiology related to hip operations. J Bone Joint Surg.1985;67B:10-13.

15. Gerson LW. The incidence of pressure sores in activetreatment hospitals. Int J Nurs Stud. 1975;1 2:201-204.

16. Allman RM, Laprade CA, Noel LB, et al. Pressure soresamong hospitalized patients. Ann Intern Med. 1986;105:337-342.

35

vorrs I@~~~~~~~~~nrvuYOHIMBINE HOI

:~~

;g1~~~~~~~~~~~~~~~Ii 1 ao to.. a I pi n N z | l i

SS1i_E | | | een adeuatel st

-4. -4 _ ~hi d e though it is.g n ificn e nevousc

iaebyB a heic (eceponrit)s de cesaIsd ~~~toh ow-(h@er9it; hoeve S e it

= u 1 1 _ tobinet in male dxuaI

Wsiac cin n*'64 iseaseen andequatelyt' studie vet dto dd=~~~~ie at rmnawude °asat

nbf a iuA S Vp.tha mic centers an..C

--~ ry notrP*ied. or

:mustd ud iancy Neither isthi it

-4sto1dbe toI rit; honeei* dgtstd

imbecin-efmW~ied~

Ecanpbrofaestnse f r

_5~~~ '

~iistory it ble at ipharmacisntio nwide

PHRMCUThAS asCa

4,~ ~eafy New Jese 07744~ ~ ~ (0)46-80

(80 237-9083 ~ 4