Community acquired , nursing home acquired and hospital acquired pneumonia : A five-year review of the clinical , bacteriological and radiological characteristics

Department of Medicine, The Wellesley Hospital and The Toronto Hospital, University of Toronto, Toronto, Ontario Correspondence and reprints: Dr Charles K Chan, The Toronto Hospital, 200 Elizabeth Street, 10 EN-220, Toronto, Ontario M5G 2C4. Telephone 416-340-3235, fax 416-971-6427 Received for publication March 20, 1995. Accepted August 2, 1995 CW CHOW, LR LEE-PACK, N SENATHI RA GAH, M RAWJI, M CHAN, CK CHAN. Com mu nity ac quired, nurs ing home ac quired and hos pi tal ac quired pneu mo nia: A fiveyear re view of the clini cal, bac te rio logi cal and ra dio logi cal char ac ter is tics. Can J In fect Dis 1995;6(6):317325.

T HE FIRST RE PORT OF A POPU LA TION SUR VEY OF PNEU MO NIA was by August Hirsch in 1881 (1), in which he de scribed pneu mo nia to oc cur in epi dem ics and to be more preva lent in the eld erly, the sick and those of low so cio eco nomic status.This re port had fol lowed the first suc cess ful cul ture of a bac terial or gan ism from a pa tient with pneu mo nia by Louis Pas teur, who aptly named the or gan ism Strep to coc cus pneu mo niae.Hirsch's re port ap pears to dif fer lit tle from sub se quent surveys of pneu mo nia.Mor bid ity and mor tal ity in pa tients with pneu mo nia in the re cent era re main high de spite the de vel opment of ef fec tive an ti mi cro bial agents and in creas ingly bet ter un der stand ing of the patho gene sis and eti ol ogy of pneu monia.
Ac cu rate as sess ment of the over all bur den of pneu mo nia on health care re mains in ac cu rate since pneu mo nia is not a re port able ill ness and the ma jor ity of pa tients are man aged in the am bu la tory set ting.How ever, sta tis ti cal fig ures from the United States sug gest that over 3.3 mil lion cases of com munity ac quired (CA) pneu mo nia oc cur an nu ally (2).It is the fifth lead ing cause of death in those over 65 years of age (3).Although low in the out-patient set ting, mor tal ity is re ported to be 25% in those re quir ing hos pi tali za tion.High est mor tal ity rates oc cur in the in ten sive care unit (ICU) set ting (4)(5)(6)(7)(8)(9)(10)(11), where it has been re ported to be as high as 47% (12).Higher mor bid ity and mor tal ity were ob served in the eld erly (6,8,13) and in those with con cur rent ill nesses such as chronic ob -struc tive pul mo nary dis ease (COPD), dia be tes mel li tus, chronic re nal fail ure, con ges tive heart fail ure and chronic alco hol abuse (6,(8)(9)(10).Para doxi cally, in creased lon gev ity permits op por tu nity for de vel op ment of medi cal ill nesses and is likely re spon si ble for the lack of sig nifi cant im prove ment in the over all prog no sis of pneu mo nia.In the past dec ade, there has been an in crease in the pool of sus cep ti ble hosts, par ticu larly pa tients with com pro mised im mune de fence sys tems.There has been a marked in crease in hu man im mu no de fi ciency virus (HIV) dis ease and in sur vi vors of bone mar row and or gan trans plan ta tion who are on chronic im mu no sup pres sive agents.At the same time, the de vel op ment of newer and more pow er ful an ti mi cro bial agents has met with a par al lel emer gence of re sis tant and new strains of in fect ing or ganisms.
Re cent pub li ca tion of guide lines for the em piric management of CA pneu mo nia (14)(15)(16)(17) un der scores the im por tance of con tinu ally up dat ing our knowl edge of this per sis tent but ever-changing dis ease.The pres ent study was un der taken to re view the con tem po rary epi de mi ol ogy of CA, nurs ing home (NA) and hos pi tal ac quired (HA) pneu mo nia in a community-based hospital.We sought to char ac ter ize the pa tient popu la tion at risk, the spec trum of clini cal and roentge no graphic pres en ta tion, the pat tern of of fend ing pathogens, the re sponse to man age ment and pa tient out come.
We re cently re ported the re sults of an in terim analysis of the ini tial 1300 pa tients (18).The com plete analy sis of the final data set at 1622 pa tients is re ported here.

PA TIENTS AND METH ODS
Study popu la tion: Pa tients were iden ti fied from a com puter da ta base of all dis charges from the Well es ley Hos pi tal in Toronto, On tario dur ing a five-year pe riod from April 1987 to March 1992.All pa tients with a di ag no sis of pneu mo nia encoded in the dis charge di ag no sis were in cluded in the study.The Well es ley Hos pi tal is a university-affiliated, 400-bed commu nity hos pi tal that func tions mostly as a pri mary and sec ondary cen tre with an nual to tal hos pi tal dis charge of 15,000 to 16,000.A small por tion of dis charges origi nates from a ter ti ary re fer ral base due to prox im ity to The Prin cess Mar ga ret Hos pital/On tario Can cer In sti tute, the ma jor on col ogy re fer ral hos pital in south ern On tario and from the only burn unit in the city, lo cated in the Well es ley Hos pi tal.Pa tients with a di ag no sis of Pneu mo cys tis car inii pneu mo nia were ex cluded from the study.Those with HIV dis ease were not ex cluded.Data col lec tion: In for ma tion was manu ally ex tracted ret rospec tively from the hos pi tal charts and re corded on a da tabase sheet be fore fi nal en try into a com puter da ta base pro gram.Va lid ity of data cod ing and en try was en sured by dou ble en try.The fol low ing in for ma tion was gath ered: first, pa tient demo graph ics such as age and sex; sec ond, con current medi cal ill nesses (de fined as COPD, asthma, bron chiec tasis, dia be tes mel li tus, chronic re nal dis ease, con ges tive heart dis ease, un der ly ing ma lig nancy, im mune de fi ciency [both pri -mary and ia tro genic such as long term use of im mu no sup pressive agents] and chronic al co hol abuse); past and cur rent smok ing his tory was re corded sepa rately; third, labo ra tory find ings in clud ing com plete blood counts, se rum so dium, chest roent ge no gram (CXR) (in the ma jor ity of cases, the of ficial ra di ol ogy re port was used al though at times only the results of the clini cal in ter pre ta tion re corded on the chart were avail able; fo cal in fil trate was de fined as in fil trate lo cal ized to one lobe; dif fuse in fil trate was de fined as in volve ment of two or more lobes) and mi cro bi ol ogy re sults; fourth, clini cal pres enta tion in clud ing fe ver (tem pera ture over 38°C), chills, sweat, de creased level of con scious ness, cough, spu tum pro duc tion and dysp nea; fifth, du ra tion and type of an ti mi cro bial agents used and pres ence of ad verse drug re ac tions; sixth, de vel opment of com pli ca tions (de fined as lung ab scess, em pyema, pleu ral ef fu sion, con ges tive heart fail ure and pneu motho rax); sev enth, need for oxy gen ther apy and ICU ad mis sion; eighth, length of stay in hos pi tal and in ICU; ninth, fi nal out come (see sec tion on out come meas ure ments).
The di ag no sis of pneu mo nia was con sid ered to be CA if the di ag no sis was made within the first 72 h of ad mis sion, HA if made af ter 72 h and NA if the pa tient was a resi dent of a nursing home and the di ag no sis was made within the first 72 h.Mi cro bio logi cal data: All mi cro bio logi cal data were ex tracted from the of fi cial Mi cro bi ol ogy De part ment labo ra tory re port form.Since these pa tients were not part of a pro spec tive clinical study, in ves ti ga tions and work-ups for pneu mo nia were not stan dard ized.Nev er the less, the di ag nos tic work-up re flect Co-morbidity was de fined as chronic ob struc tive pul mo nary dis ease, asthma, bron chiec ta sis, dia be tes mel li tus, chronic re nal dis ease, con ges tive heart dis ease, un der ly ing ma lig nancy, im mune de fi ciency (both pri mary and ia tro genic such as long term use of im mu no sup pres sive agents) and chor nic al co hol abuse; com pli ca tions were de fined as lung ab scess, em pyema, pleu ral ef fu sion, con ges tive heart fail ure and pneu motho rax.ICU In ten sive care unit the stan dard clini cal prac tice at the Well es ley Hos pi tal for pneu mo nia and in cluded spu tum, if avail able, for Gram stain, cul ture and sen si tiv ity; and blood for aero bic and an aero bic cul tures.All speci mens were proc essed through the Mi cro biol ogy De part ment at the Well es ley Hos pi tal.Blood cul tures were grown in Bac tec (Becton-Dickinson, Mary land) bot tles and spu tum sam ples were han dled in the stan dard mi cro biologi cal labo ra tory fash ion.Tests for atypi cal agents such as Le gionella spe cies and my co bac te ria, se rum se rol ogy, tho racen te sis, per cu ta ne ous nee dle as pi ra tion and bron cho scopy with bron choal veo lar lavage (BAL) with or with out pro tected brush were done only on se lected pa tients, at the dis cre tion of the at tend ing and/or the con sult ing phy si cians.When or dered, test ing for Le gionella was by the di rect fluo res cent an ti body method fol lowed by cul ture.My co bac te ria was in ves ti gated ini tially with acid fast stain and fluo res cent an ti body stain followed by cul tures at the On tario Health Min is try's My co bac terial Labo ra to ries.If re quested, acute and con va les cent an ti body ti tres for Myco plasma pneu mo niae were done.All or gan isms iso lated from blood, spu tum and urine cultures within 72 h of di ag no sis of pneu mo nia were re corded.Mul ti ple iso lates from a sin gle pa tient were re corded as such in each pa tient file.How ever, these were treated as in di vid ual iso lates in the fi nal analy sis.Out come meas ure ments: Pa tients were con sid ered to have im proved or to be cured if the pa tient was alive and off an ti bi otics at the end of the hos pi tali za tion pe riod.Those who discharged them selves against medi cal ad vice were in cluded in the 'i mproved' group be cause it can be as sumed that they were alive at the time of dis charge.Death cer tifi cates of those who died dur ing the hos pi tal stay were re viewed.If the death cer tifi cate listed pneu mo nia as a cause of death, it was con sid -ered to be a pneu mo nia death.Need of oxy gen ther apy, ICU ad mis sion and de vel op ment of com pli ca tions of pneu mo nia as de fined above were also re corded.

RE SULTS
Dur ing the five-year pe riod of the study (April 1987 through March 1992) 1782 of the 74,435 pa tients dis charged (2.4%) from the Well es ley Hos pi tal had a di ag no sis of pneu mo nia.Charts of 160 pa tients (9%) were ir re trieva bly lost.The remain ing 1622 dis charges avail able for re view com prised the study popu la tion.
There were more men than women (964 cases or 59.4% ver sus 658 cases or 40.6%, re spec tively).Mean age was 64.4 years (range 16 to 103 years).One thou sand and twenty-two cases (63%) were CA, 138 (8.5%) were NA and 462 (28.5%) were HA.These pa tients rep re sented an ill popula tion.One thou sand five hun dred and forty-two (95%) had at least one con comi tant medi cal con di tion.The ma jor ity (73%) had two or more con cur rent medi cal ill nesses.The three groups were simi lar with re spect to pres ence of co-morbidity.The NA group was older and had a lower in ci dence of a smoking his tory (Ta ble 1).Com mon con comi tant medi cal con ditions were car diac dis ease, chronic al co hol use, COPD and un der ly ing ma lig nancy (Ta ble 2).
Ini tial CXR was re ported on 1568 pa tients (97%) and was ab nor mal in 97% of these, al though a sur pris ing 3% of pa - tients were re ported to have a nor mal CXR (Ta ble 3).In the 54 pa tients for whom re ports of the ad mis sion CXR could not be found, the data were treated as in com plete and were not included in the sub se quent analy sis.Fo cal ra dio graphic ab normal ity was the most com mon find ing in all three groups: 69% in CA, 77% in NA and 68% in HA (Ta ble 3).There was no sig nificant dif fer ence in the roent ge no graphic pres en ta tion in the three groups.Ap proxi mately three-quarters of the pa tients were docu mented to have cough and/or spu tum pro duc tion at the time of pres en ta tion.Fe ver and chills were re ported in about half of the pa tients in each group (Ta ble 4).
Few pa tients were started on an ti mi cro bial ther apy be fore hos pi tal ad mis sion.Forty of 1022 pa tients in the CA group were docu mented to have been on an ti bi ot ics at the time of ad mis sion, eight in the NA group (6%) and eight in the HA group (2%).
In 85% of all pa tients, at tempts were made to iden tify an infect ing or gan ism by means of cul tures of spu tum, blood, urine, BAL and/or se rol ogy (Ta ble 5).Cul tures of both blood and spu tum were ob tained in 79% of pa tients.In ves ti ga tions were per formed more of ten in the CA group.Sum mary of all Over all, the pre domi nant or gan isms found from all in ves ti gations were Hae mo phi lus in flu en zae (iso lated from 204 patients), Staphy lo coc cus au reus (from 152 pa tients) and S pneu mo niae (from 143 pa tients).The pat tern of or gan isms de tected dif fered in the three groups.In the CA group, H in fluen zae, S pneu mo niae and S au reus were the most com mon or gan isms.In the HA group, S au reus and Gram-negative organ isms pre domi nated.In the NA group, 22 of the 32 Gramnegative or gan isms iso lated were from ei ther blood (two) or spu tum (20) (Ta bles 6,7).The main or gan isms docu mented from blood cul tures were S pneu mo niae and S au reus (Ta ble 7).Yield of in ves ti ga tion of bron cho scopy speci mens and serum and/or urine se rol ogy in the iden ti fi ca tion of the in fect ing or gan ism was low.Two hun dred and forty-seven pa tients (15%) were ad mitted to ICU.The high est ICU ad mis sion rate, not sur pris ingly, was in the HA group (28%) and the low est in the NA group (3%).The av er age length of ICU stay was 7.1, 5 and 9.5 days in the CA, NA and HA groups, re spec tively.Those ad mit ted to ICU were simi lar to the over all group in terms of pres ence of co-morbid ill nesses (Ta ble 1).

TA BLE 5 Rou tine in ves ti ga tions in pa tients ad mit ted to hos pi tal with pneu mo nia
Only 39 pa tients (2.4%) re quired home oxy gen ther apy at the time of dis charge al though, not sur pris ingly, more pa tients re quired oxy gen ther apy dur ing their course in hos pi tal.One hun dred and four pa tients (6.4%) de vel oped com pli ca tions of pneu mo nia, most com monly in the HA group.The most common com pli ca tion was the de vel op ment of pleu ral ef fu sion.
The av er age length of stay for the en tire group was 24 days.Shorter hos pi tali za tion courses of 16.3 and 17.4 days were seen in the CA and NA groups, re spec tively.The av er age hos pi tali za tion in the HA group was 42.9 days (Ta ble 1).
In all, 1261 pa tients (78%) were deemed to have im proved or to be cured at the time of dis charge.Three hun dred and sixty-one pa tients (22%) died dur ing ad mis sion to hos pi tal.The high est mor tal ity rate was in the HA group (34%) and the low est in the CA group (16%) (Ta ble 1).Mor tal ity in the NA group was 25%.Death cer tifi cates were re viewed in 320 cases (89%).Pneu mo nia was listed as a con trib ut ing cause of death in 165 of the 320 deaths (52%).Seventy-five of 149 deaths in the CA group, 10 of 30 in the NA group and 70 of 141 in the HA group were at trib uted to pneu mo nia (Ta ble 9).

DIS CUS SION
Our five-year re view of all pa tients dis charged from hos pital with a di ag no sis of pneu mo nia dur ing the pe riod from April 1987 to March 1992 de scribes the most con tem po rary popula tion of in-patients with pneu mo nia, and is two to three times larger than the most re cently pub lished re ports (5,6).Data from these 1622 cases of pneu mo nia con tinue to sup port previ ous ob ser va tions that pneu mo nia in the 1990s re mains a se ri ous ill ness with sig nifi cant mor bid ity and mor tal ity.Pneumo nia ac counted for 2.6% of all dis charges from our hos pi tal dur ing this pe riod.
Al though it ap pears that few pa tients were started on outpatient an ti mi cro bial ther apy, the high in ci dence of co-morbid ill ness, with 95% hav ing at least one con comi tant ill ness and 73% hav ing two or more, sug gests that, as a group, these elderly pa tients are in deed ill.These con tem po rary hos pi tal ized patients are, in fact, a sicker popu la tion than pre vi ous co horts.Pre vi ous stud ies (4,5,7, 9-11) re ported a lower in ci dence of comorbid ill ness in pa tients hos pi tal ized with pneu mo nia.In the Fang et al (5) study popu la tion of 359 pa tients with hos pi tal ized CA pneu mo nia 30.6% had no un der ly ing medi cal dis ease, and in the 719 pa tients stud ied by Mar rie et al (6) 13% had no concur rent medi cal ill ness.The age, sex dis tri bu tion and length of hos pi tali za tion of pre vi ous co horts were com pa ra ble with ours (Ta ble 10).
The in ci dence of HA pneu mo nia in our pa tient popu la tion was 6 in 1000.This is com pa ra ble with pre vi ous re ports from Can ada and the United States (19,20).
Pre vi ous stud ies have re ported S pneu mo niae to be the pre domi nant or gan isms in all groups with pneu mo nia (4)(5)(6)(7)(8)(9)(10)(11)13,21,22).In the com mu nity set ting, H in flu en zae and S au reus have been found to be im por tant patho gens as well.In our study, H in flu en zae was the most com mon or gan ism isolated over all.How ever, patho gen ic ity can not be veri fied because the ma jor ity of the iso lates were from spu tum rather than blood cul tures.Among or gan isms re cov ered from blood cul tures, S pneu mo niae re mains the pre domi nant patho gen.
Le gionella pneu mo niae is an un com mon cause of pneu mo nia in our popu la tion in con trast to pre vi ous re ports (4)(5)(6)9,10,23).This is likely a re flec tion of re gional dif fer ences in the prevalence of Le gionella in fec tions.Al though a higher in ci dence of Gram-negative or gan isms was found in NA and HA pneu mo nia, S au reus re mained a signifi cant patho gen, par ticu larly in the HA group.These find ings are con sis tent with other re ports of no so comial pneu mo nia (24)(25)(26)(27)(28)(29)(30)(31)(32).
The pat tern of oc cur rence of or gan isms for the three types of pneu mo nia (Ta ble 6) con cur with those out lined in the recent guide lines on the ini tial man age ment of pneu mo nia, and fur ther sup port their rec om men da tions of em piric an ti mi crobial ther apy (14,15,17).
In 26% of the NA group, there was no at tempt to docu ment a patho gen (Ta ble 5).This may rep re sent a more prag matic and less in va sive thera peu tic ap proach to this sub popu la tion.It is, how ever, sur pris ing that 12% in the CA group and 16% of HA group did not have any cul tures or se rol ogy done.The reasons for this are not clear.
As with pre vi ous stud ies (5), CXR ap pear ance and the pres ence of spe cific pre sent ing signs and symp toms such as cough, spu tum and de creased level of con scious ness were non spe cific and were not as so ci ated with any par ticu lar etiologic agent.
The com pli ca tion rate in our study was low.How ever, we con fined our defi ni tion of com pli ca tions (see Pa tients and Meth ods Sec tion) to those that could be con fi dently at trib utable to pneu mo nia.Other com pli ca tions such as re nal fail ure or pul mo nary com pli ca tions can not be at trib uted solely to pneu mo nia in a popu la tion with high co-morbid ill ness, particu larly in a ret ro spec tive con text.
Over all mor tal ity in the pres ent study was 22%, with the high est mor tal ity rate seen in the HA group and the low est in the CA group, where it was 16%.Over all mor tal ity in CA pneumo nia in the 719 pa tients de scribed by Mar rie et al (6) and in the 359 pa tients de scribed by Fang et al ( 5) was 21% and 13.7%, re spec tively (Ta ble 9).There fore, de spite be ing a sicker popu la tion with more con comi tant ill nesses, the over all mor tal ity is com pa ra ble with that of pre vi ous groups (4)(5)(6)(7)(8)(9)(10)(11)22,23).In re view of the death cer tifi cates of 320 of the 361 deaths in our se ries, pneu mo nia was found to be a contrib ut ing cause of death in 165 cases.Spe cific cause of death was not re ported in ei ther of the two groups de scribed by Marrie et al (6) and Fang et al (5).We could not con fi dently re port any deaths as a di rect re sult from pneu mo nia since guide lines for com plet ing death cer tifi cates are not strict and open to inter pre ta tion.Moreo ver, the death cer tifi cates in our pa tients were com pleted by vari ous phy si cians.Thus, pneu mo nia listed on the death cer tifi cate as the cause or as a con trib ut ing cause of death was treated in the same man ner for the purpose of this study.
Stud ies have shown that pa tients with pneu mo nia and con cur rent ill ness have more se vere dis ease (9,10).The high in ci dence of co-morbidity in our pa tient popu la tion is likely a re flec tion of the cur rent ad mis sion cri te ria for pneu mo nia that ad mit pri mar ily sick, eld erly pa tients with se vere dis ease.The cur rent trend to man age more pa tients in the out-patient setting for eco nomic rea sons may have in fact ad versely bi ased the mor tal ity out come of a study based on an in-patient popula tion.
We rec og nize limi ta tions to the in ter pre ta tion of the data due to the ret ro spec tive na ture of this study.The defi ni tion of pneu mo nia was based on avail able clini cal and labo ra tory data and the di ag no sis was made by the at tend ing or con sulting phy si cians.How ever, in stud ies com par ing the ac cepted di ag nos tic cri te ria for pneu mo nia with the gold stan dard (a patho logi cal as sess ment of pul mo nary tis sue), these cri te ria were found to be nei ther spe cific nor in fal li ble (26,33).The strength of the study is the gen er aliz abil ity of the re sults to other community-based hos pi tals in North Amer ica.Al though com plete and de fini tive data are not avail able in the work-up and man age ment of pneu mo nia, the in for ma tion avail able mir rors what cli ni cians have to rely on in gen eral prac tice.
Al though our yield from blood and spu tum cul tures is compa ra ble with pre vi ous re ports, patho gen ic ity of or gan isms recov ered from spu tum or BAL fluid can not be as cer tained.The mi cro bio logi cal re sults of this study may be over-implicating some or gan isms while under-representing oth ers, par ticu larly those that re quired spe cial tests for de tec tion (eg, vi ral agents that are de tected on acute and con va les cent ti tres).This is par ticu larly true in cases where the pu ta tive or gan ism(s) were iso lated from spu tum cul tures.The ret ro spec tive na ture of the study did not al low us to ac cept or re fute the re sults based on the clini cal con text or on the qual ity of the spu tum ex am ined.There fore, we chose to rec ord all iso lates ob tained.How ever, dif fer en ti at ing a colo nizer from a patho gen in cul tures of res pira tory se cre tions is a com mon clini cal prob lem and one that is not read ily cor rect able with out in va sive pro ce dures and even in the most strin gent pro spec tive stud ies.

CON CLU SIONS
Our large scale study on a con tem po rary se ries of pa tients hos pi tal ized for pneu mo nia sug gests that the spec trum of micro or gan isms im pli cated in pneu mo nia in the 1990s is simi lar to that of the pre vi ous dec ades.How ever, the types of patients be ing hos pi tal ized for pneu mo nia have changed to include pri mar ily eld erly pa tients with mul ti ple co-morbidity.Hence, de spite sig nifi cant ad vances in the de vel op ment of diag nos tic tests and an ti mi cro bi als, the over all mor tal ity rate for the cur rent popu la tion re mains high.

AC KNOW LEDGE MENTS:
The authors thank the staff at the Health Re cords De part ment of the Well es ley Hos pi tal for their co op era tion and as sis tance in lo cat ing over 1600 charts.This work was sup ported in part by a grant from Roche Can ada Ltd.

TA BLE 2 Pres ence of co-morbid con di tions in pa tients with pneu - mo nia
*Ex cludes leu ke mia and lym phoma.CA Com mu nity ac quired; COPD Chronic ob struc tive pul mo nary dis ease; HA Hos pi tal ac quired; HIV Hu man im mu no de ficiency vi rus; NA Nurs ing home ac quired TA

BLE 3 Roent ge no graphic pres en ta tion of pa tients with pneu - mo nia
Re port of ini tial chest roent ge no grams was avail able in 1568 pa tients.Fo cal infil trate was de fined as in fil trate lo cal ized to one lobe; dif fuse in fil trate was defined as in volve ment of two or more lobes

pa tients with com mu nity ac quired (CA), nurs ing home ac quired (NA) and hos pi tal ac quired (HA) pneu mo nia in whom com mon patho gens were iden ti fied
Or gan isms iden ti fied on cul tures of blood, spu tum, bron choal veo lar lavage fluid, urine plus urine and se rum se rol ogy.Pa tients who had the same or gan ism iso lated from mul ti ple sites were counted only once TA BLE

7 Num bers of or gan isms re cov ered from dif fer ent cul ture sites in 1622 pa tients with pneu mo nia
BAL fluid, urine) plus urine and se rum se rol ogy found the high est yield from spu tum cul tures.Yield of a posi tive re sult from spu tum and/or blood cul ture in the over all group was 53%.Only 6% of pa tients had posi tive blood cul tures.A sum mary of all in ves ti ga tions done to identify the of fend ing patho gen is pre sented in Ta bles 6, 7 and 8.
CA Com mu nity ac quired; HA Hos pi tal ac quired; NA Nurs ing home ac quired CAN J IN FECT DIS VOL 6 NO 6 NO VEM BER/DE CEM BER 1995 cul tures (blood, spu tum,

TA BLE 10 Re cent stud ies of pneu mo nia Pre sent study 1995 Mar rie et al 1989 (6)
The re sults are for the en tire study group.Num bers in pa ren the ses are the results in the com mu nity ac quired group.Cases stud ied by Fang et al and Mar rie et al were com mu nity ac quired pneu mo nia.Pro Pro spec tive; Retro Ret ro spective *