Computed Tomography in Established Adult Respiratory Distress Syndrome

Computed Tomography in Established Adult Respiratory Distress Syndrome* Correlation With Lung Injury Score Catherin e M . Owens, MB ; Timothy W . Evans, MD , PhD ; Brian F . Keogh, MD ; and David M . Han sell, MB Study objectives: To identify serial changes in the appearances of the lungs on computed tomography (CT) in patients with established adult re spiratory distress syndrome (ARDS). Second, to evaluate any relationship between the extent of morphologic abnormalities on CT anatomic and physiologic derangement using a numeric score of the severity of lung injury. Design: Retrospective, descriptive. Setting: Adult intensive care unit, Department of Radiology, and outpatient department of a national tertiary referral center. Patients: Eight patients meeting diagnostic criteria for ARDS able to tolerate CT scanning during the acute phase of their illness and who survived to be reevaluated during convalescence. Interv entions: Mechanical ven tilatory support. Conventional intensive care support of other failed syst ems as appropriate. Measurements and results: Thin-section CT scans of the lungs categorized as to ext ent (calculated percent volum e of abnormal lung), di stribution, and dominant disea se pattern. Concurrent lung injury score (LIS) was recorded at the time of the CT during the acute phase of illness (mean, 26 da ys; range, 3 to 48 da ys after precipitating event) and at follow-up (96; 17 to 187 da ys). On initial CT scans, disease patterns included ground-glass opacification (8/8), parenchymal distortion (8/8), multifocal areas of consolidation (6/8), reticular opacities (6/8), and linear opacities (5/8). On follow-up scans, there was clearing of consolidation in all patients, but

The adult respiratory d istress synd rome (ARDS) is cha racterized by refract or y hypoxemia secondar y to nonh ydrostatic pulmonar y edema and is associat ed with a wid e variety of pr ecipitating factors, man y not directly in volving the lun g. 1,2 Mortality has rem ain ed alm ost un chan ged since th e synd rome was first described in 1967.3 However , much is now kn own a bout th e pathophysiology of th e condition : histop athologic examina tion of tissue obta ined from pat ients with ARDS sugg ests that th e condition evo lves through exuda tive , inflammat ory, and fibr opr oliferative phases.v'' However , th e exte nt to which th e pathologic chan ges that cha rac te rize ARDS ar e *From the Departments of Radiology, (Drs. Owens and Hansell), Anesthetics and Intensive Care (Drs. Evans and Keogh), Royal Brompton National Heart and Lung Hospital, London, England. Manuscript received November 29, 1993; revision accepted March 8, 1994

ground-glass opacification persisted in four of eight patients. The reticular pattern persisted unchanged in five of eight patients, became more extensive in two of eight, and developed in one. A reticular pattern was most pronounced in areas that had been densely consolidated previously. Evidence of parenchymal distortion, present on the initial scan in all patients, persisted in six of eight patients. Computed tomographic features suggestive of emphysema developed in one patient. The LIS revealed moderate to severe ARDS in all patients initially; this decrea sed to a mild or zero LIS at follow-up . Overall, there was 76.9 % ± 5.3 % abnormal lung on the initial CT scan and 34.5 ± 9.3 % on th e follow-up CT scan. There was a significant correlation between the extent of abnormalities on CT and LIS (r=0.75, p
reflect ed in abnormaliti es of lung compliance and gas exchange rem ains uncl ear. Co m puted tomography (CT) has been used at sing le tim e point s to evaluate possible relationships bet ween CT den sity of lung par en ch yma and gas excha nge during th e application of positive end -expira tory pr essur e (PEEP)6 and following changes in bod y position ," but it has not been used to follow up surv ivors. Furthermore , to our knowled ge, no attempt has been mad e to correlate persiste nt morphologic abno rmalities on CT in sur vivors with indices of respirat or y fun ction . Th e aim s of this study were th er efore twofold: fir st, to exte nd pr eviou s work in thi s area by describing and quantifying th e extent of lun g injury in patients with ARDS with particular refer ence to those features th at wer e potentiall y reversible , both during the acute and convalescent stages of th e synd rome ; and second, CHEST 1106 161 DECEMBER. 1994

1815

Table I-Demographic Details of the Patients Studied

Patient No.*

Time to 1st CT, (d)

Time to 2nd CT, (d )

LIS (1)

LIS (2)

1 (post-CABG) 2 (RTA) 3 (RTA) 4 (RTA) 5 (pneumonia) 6 (abdominal operation) 7 (RTA) 8 (RTA) Mean SEM

60 5 48 30 5 12 3 45 26 7.81

137 30 104 177 18 134 154 110 108 20.1

!.i5 1.75 2.25 1.25 2.0 1.0 1.75 2.0 1.72 0.14

1.0 0 0.5 1.0 0.5 0 0.5 1.33 0.6 0.17

sta te. An LIS was per for med immedia te ly be fore th e pati ent leavin g the IC U for C T scanning. Th e scan was repeat ed as soon as possib le after eac h pati ent was dischar ged f rom hospital an d becam e fully a m bu lato ry . A seco nd LI S was perf ormed on th e sam e d ay as th e CT sca n using a standard ches t rad iog ra ph and arterial gas ana lysis (Corn ing 170 blood gas ana lyzer , Co rn ing UK, Essex).

Analysis of CT ima ges

In our unit , patients with ARDS wh o are conside red fit enough undergo CT sca nn ing fo r th e eva lua tion of di sease exte nt and det ection of supe rve n ing com plica tions ar e selec ted . The study pop ulati on was confine d to those wh o survived to be review ed as outpatie nts. Th e records a nd sca ns of eight pati ents (three mal e; mean age, 30 years; range, 15 to 59 yea rs) wer e exam ined (Ta ble 1). Fiv e patients presented after trauma sustained in road traffi c accide n ts, on e a fte r an abdo mi na l opera tion, one a fter cardioth oracic sur ger y, and one pati ent afte r a se vere pn eumococcal pn eumonia . At the tim e of en try into th e study , eac h pati ent fulfilled the di agn ostic criteria for ARDS used in our unit: 8 an antece de nt histor y of a pr ecipitating cond itio n, cha nges on che st rad iog ra ph suggestive of pulmonar y ede ma, and a ratio of Pa02:FI02 of <1 50 mm Hg (20 kPa) in the presence of norm al plasma pr oteins and a pulmonar y artery occlusi on pressur e <15 mm Hg.

Each CT image was assessed by two obser ver s for th e pr esen ce and d istribution (the lun g was di vid ed in to six zones : upper , m iddle, lower ; ante rior or posteri or) of five patterns of d isease as follows: (1) ground-g lass opacification: ha zy areas of incr eased a tte nua tio n of the lun g par ench ym a, without obsc ura tio n of th e underl ying vasc ula r ma rkings and bron ch ial wal ls.l? (2) consolidat ion : characterized by marked ly in cr eased atten ua tio n of the lun g par en ch ym a with obscuratio n of th e underl yin g vasc ular markings usuall y acco m panied by an air b ronch ogr am; (3) reti cular opacities: mo st frequen tly representing int erlobular septa thi cken ed b y ede ma or fibrosis; (4) lin ear opacities: th icken ed int erl obular sep ta or fibrou s stra nds tr aver sing areas of destroyed lung; and (5) parenchymal distortion : sho wn as di stortion a nd dilatat ion of br on chi , so-ca lled " trac tion bro nch iec tasis." Th e total exten t of disease was deri ved by both observe rs visually estimating th e per centage of a bnorma l lun g to the near est 5 % at five pred efined CT levels: (1) origin of the great vessels; (2) m id-arch of th e aorta ; (3 ) ma in carina ; (4) pul monar y venous conflue nce ; an d (5 ) 1 ern above th e right d om e of d iaphragm . Allowa nce was mad e for d ifferences in lun g volu mes a t th ese levels: the volume adjustm ent was mad e using a we igh ting factor , ca lculate d fr om CT measur em ents of ten randoml y selected patients with d iffuse lun g disea se. The outlines of the lungs at th e five CT levels wer e traced onto paper ; th ese wer e cut out and weighed using an elec tro n ic balan ce. Th e rati os of th e weights (and thus th e lung volu mes ) for the five levels were (mean ± SD ) as follows: level 1=0.129±0.017; level 2 = 0.190 ± 0.01l ; level 3 = 0.222 ± 0.008; level 4=0. 228 ±0.009; an d le vel 5 = 0.230 ± 0.015 . The mean of the percentage of abnorma l lu ng estima ted by th e two observers at eac h level was multiplied by the cor respondi ng ratio. Ad d ing th e five adjusted figu res gave an estim ation of th e ove ra ll per centage volume of a bno rmal lun g .

Investigations

Data Analysis

All scans wer e perf orm ed on an elec tro n beam ultraf ast CT scanne r (Ima tron, San Francisco). Three-millim et er sections wer e pe rforme d at 10-mm int er vals fro m th e lung apices to bases in th e supine posit ion in all pati ents. Scan ac q uisition tim e was 200 ms and images wer e reconstructed using a high spa tia l resolution reconstruction a lgorithm. On the init ial CT sca n, ven tilation was suspe nde d at full inspira tion . Th e CT im ages we re viewed at window levels an d widths appropriate fo r lung par ench ym a (leve l, 500 H U; width, 900 to 1,500 H U). Immedi at el y be fore each scan, a lung injury score (LIS ), as d escribed b y Murray et al,? was per form ed . Th e LIS is d esign ed to pr ovid e a quantitat ive m ean s of following the ph ysiologic va riables of patients with acute lun g injury. In sum ma ry , lun g com plia nce , the fra cti onal inspired oxygen concentration, th e d egr ee of PEEP used , a nd a chest rad iogra ph ar e score d numer icall y. A score greate r than 2.5 is required before ARD S is d iagnosed. A score of 0.1 to 2.5 is conside red indicati ve of mild to moderate lun g injury .

Data are pr esented in th e text as mean ± SEM exce p t wh er e otherwise sta ted . Th e per centage of persistently abnormal lun g at both initial sca n and follow- up was correlated with th e LIS using linear regr ession ana lysis. A p value less than 0.05 was conside red sig nifica nt.

*RTA=road tra ffic accident ; CABG=corona ry artery bypass graft surgery.

to relate an y abnormalities to a severity score of lung injury. METHODS ArW M ATERIALS

Patient Population

Protocol All pat ients we re tr ea ted in the IC U using modes of m ech anical ven tilation, an F IO z, and inotropic agents sufficien t to produce an oxygen deli ver y index of 300 mL/m in /m 2. Th e initial CT sca n was pe rformed as soon as th e pati ent could be tr an sfe rred safely to th e CT scanning unit in a m ech ani call y ventilat ed

1816

R ESULTS

The Royal Brompton National Heart and Lung Hospital is a tertiary referral center and all patients had been tr eated for vary ing per iods in other units prior to transfer to our own IC U. Furthermore, they wer e all consid er ed too unstable to be transferred immediately to the CT scanning unit on arrival. Conseq uently, the mean time int erv al from th e pr ecipitating eve nt leading to the development of ARDS to the initial scan was 26 da ys (range, 3 to 48 da ys). A number of pati ents had particularl y difficult and prolonged clinical cour ses in the IC U and the mean tim e int er val to th e follow-up scans from th e pr ecipitating eve nt was 108 days (ran ge, 18 to 177 da ys) from the pr ecipitating event (Table 1). All patients CT in Established ARDS: Correlation With Lung Injury Score (Owens et sO

Table 2-Pulmonary Abnormalities Recorded on Initial CT Scan* Case 1'\0.

Gro und- Glass Op acificat ion

2

U-P M- A+P L -A + P All zones

Areas of Consolida tion M- A+P L-A +P

4

All zones V-A + P

All zones, spari ng lun g pe rip hery None L-A+P

5

All zones

6 7

M-A+P L -A + P All zones

8

All zones

All zones

3

Reti cular Op aciti es

Lin ear Op aciti es

Par en ch ym al Distor tion

%

Abnormal Lung

V- P

None

M-A + P L- A+P

84

M- A+P L- A+ P None

None

M-A+P L- A+ P All zones M- A+P L- A+P All zones (emphyse ma) All zones (em physe ma) All zones

86

L-P L-P

None

All zones M-P L- P All zones

None

None

L-A

All zones

U- A+ P M- A+P U- A M-A L-A+P

None

L-A+P

U- A M-P L-A+P

All zones

93 66 60 53 86 87

*U= up per zone ; M=mid zone; L=lower zone ; A=anterior (nondepe ndent) lun g; P=poster ior (de pendent) lun g.

were mechanically ventilated during the first scan and wer e br eathing spontaneously at th e second . Th e mean duration of ventilatory support was 35 da ys (ra nge, 13 to 54 da ys). Computed tomographic abn ormalities identified on the initial and follow-up scans ar e sum marized in Tables 2 and 3. The pr esen ce or absence of the five categories of disease was recorded for the initial scan . Th ese wer e ground-glass opacification (8/ 8), parenchy mal distortion (8/ 8), multifocal ar eas of consolidation (6/8), re ticular opacities (6/ 8), and linear opacities (5/ 8). Bilateral pleural effusions were pr esent in two patients and unila teral effusions wer e pr esent in th ree . Small , shallow, bilateral anterior

pn eumothoraces were pr esent in two patients and unilateral pneumothoraces were pr esent in three. On the subseq uent CT scan s, the mu ltifocal ar eas of consolidation had resolved in all cases. Groundglass opacification persisted in four of eight (F ig 1), although the extent and severity was less marked than in the acute phase. Th e reti cular pattern resolved completely in one patient, partially resolved in one, persisted unchanged in four of eig ht, and became mor e extensive in two of eight patients. The persistent reticular pattern , reflecting int erstitial fibrosis, was most pronounced in ar eas that had previously been densel y consolidated (F ig 2). Linear opacities persisted unchanged in four of eight patients, dev el-

Ta ble 3- Pulmo nary Abnormalities Recorded on Follow-up CT Scan* Pati ent No.

2

3 4

Ground-Glass Opacifi cation

Areas of Co nsolida tion

Par ench ym al Distorti on

%

Abnorm al Lun g

V- P

Nonc

V-A

None

lvi- A L-A All zones more extensive than ac ute ph ase All zones

U- P M- A+P L- A+ P M- A L- A

66

M-A+P L- A+P None

L-P

All zones

43

M- A+P L - A+P L- A+P

All zones (em physema) All zones (em physem a) All zones All zones

20

None

None

None

6 7

IvI -A + P L-A + P None None

None None

8

L- P

None

5

Linear Opacities

All zones

U- P M-A+P

U-A, M-A, L-A+P Intralobul ar fibrosis None

Reti cular Opacities

None

All zones d ue to em physe ma None U-A +P M- A L- A All zones

L-A U-A+ P M- A U-A M-P L-A+P

All zones

8

58 0 18

63

*U=upper zone; M=mid zone; L=lower zone; A= anterior (nonde pende nt) lun g; P =posterior (de pende nt) lun g. CHEST 1 106/ 61 DECEMBER , 1994

1817

FIGURE 1. Pati ent 5. Top , Dense consolidation in the left lower lobe with patchy areas of ground-glass opacificati on bilaterall y. Mode rate-sized left pleural effusion. Bottom, Follow-up CT 13 days later ; persistent areas of ground-glass opacification. Complete resolution of the consolida tion and pleur al effusion.

oped in three of eight, and resolved completely in one pati ent. Lung distortion, shown by bronchial dilation and distortion , was pr esent to some extent in all patients on th e initial scan and persist ed unchanged in six of eight patients (Fig 3). Th ere was partial resolution of dilated subsegmental bronchi in one patient and features suggestive of progression of em physema in one nonsmoking patient. Two patients ha ve subsequently developed clinical bronchiectasis, and two oth er patients had unequivocal evide nce of em phy sema on follow-up CT scans . All patients had an LIS in excess of 2.5 on admission to our ICU , but this had fallen to 1.72 ± 0.14 at th e time of th e initial scan (Table I ). Mean LIS at follow-up was 0.6 ± 0.17. There was a significant corr elation between the extent of CT abnormalities and LIS for all tim e points (r=0.75, p
.....

F IGURE 2. Pati ent 7. Top, Multifoca l areas of consolida tion in the upper lobes. A rig ht pleur al effusion is part ially responsible for the opacification in the posterior part of the right hem ithorax. Bott om, Follow-up CT scan 5 months later ; a reticular and linear pattern is seen in areas where the consolida tion has resolved.

patient population seem ed to correlate closely. The CT appea ra nces are known to reflect histologic abnormaliti es revealed on open lun g biops y spe cime n

DISC USSION

We hav e shown that the CT changes observed in patients with ARDS who survive are variable in extent and character and change considerably over tim e as th e disease resolves. Moreover , the morphologic and functional abnormalities observed in this

FIGURE 3. Pati ent 1. CT scan 4 months afte r the pr ecipitatin g insult showing distortion and dilation of segme ntal and subsegment al br onchi in the lower lobes in associati on with a widespread reticular patt ern consistent with established int erstitial fibrosis.

1818

CT in Established ARDS : Correlation With Lung Injury Score (Owens et aQ

in certain diffuse lung diseases, particularl y tho se invol ving fibrosis .'! The patients in our population were considered too unwell to tolerate op en lung biopsy to allow pathologic correlation with th e CT findings. Furthermore, the lung injury in ARDS is known to be patchy and representative sampling is difficult. However, the range of histopathologic changes seen in ARDS have been well-documented in samples taken from patients with disease of var ying severity and in diff er ent stages of ev olution up to and including post mortem . The natural history of th e syndrome has also been characteri zed using plain chest radlography.lvln most cases, th e disease pr ocess is nonuniform , some parts of th e lun g showing no changes at all , but overall three distinct stages can be recognized. Stage I can be regarded as radiologically latent and occurs during th e first few hours following any precipitating insult , oft en antedating th e need for mechanical ventilatory support. At a pathologic level , pulmonary capillary congestion, endothe lial ce ll swe lling, and exte nsive microatelectasis ar e described . Pri or to ventilation , lung volumes may be decr eased reflecting th e onset of diffuse mi cro atelect asis. At this sta ge, th er e is onl y a small increa se in extrav ascular water conte nt of lun gs, which ma y be reflected in subtle peribronchovascular obscuration or diffuse ground-glass opacification .P'P This phenomenon was obser ved at least in part on th e first CT scan tak en in all patients in th e current study, although to a ver y variable exte nt. It seems rea sonabl e to expect such variation in view of th e con siderable tim e period th at ela psed between th e onset of th e disea se and th e first CT, although no formal relationship between thi s delay and th e degree of ground-glass opacification could be established, probably because the radiologic features of ARDS ar e frequ ently modified according to the nature of pr ecipitating e vent and th e complications of supportive th erapy.V This seems unlik ely to be th e whole explanation, how ever , as several of our study population still had ev ide nce of such appearances in convalesce nc e (Table 3). The second stage of ARDS is a direct consequen ce of th e incr eased pulmonary vascula r permeability that charact erizes th e condition. Ultrastr uctural changes ar e e vident in both the alveolar epithelium and the capillary endothelium .l" The resultant pulmonary ede ma and th e outpouring of hemorrhagic proteinaceous fluid and hyaline membranes into th e air spaces cla ssically results in radiographic consolidation with associated air bronchograms. The latter were visible in five of eight of th e stud y group. The degree of consolidation reflects th e type , severity, and duration of lung injury. In th e current study, CT revealed th e consolida tion to be variable in distribution,

even wh en th e plain chest radiograph suggested it to be uniform . Second, although previous studies have shown gr eater consolidation in dependent lung regions , with more normal aeration of th e nondependent lung ," in the current study, consolidation tended to be concentrated at th e periphery. This difference may reflect the timing of our initial scans, which were not alw ays performed immediately following the onset of clinical lung injury, or might be consequence of th e methods of ventilatory th erapy used. In particul ar , th e trend toward pressur e-c ontrolled , inve rse-ra tio places th e em phasis on act ive alveolar recruitment , and we speculate th at thi s might obvia te th e well-recognized tendency toward dependent con solid ation .f A chronic, third phase of ARDS occurs in survivors. The early histopathologic features of this stage may be det ectabl e as early as 3 to 7 da ys postinjury; th ese are characte rized by hyperplasia of type 2 pn eurnocyt es, fibroblastic infiltration , a nd deposition of conne cti ve tissue.l" At thi s sta ge, th e high permeability pulmon ar y ede ma begins to regr ess. Continuing ph ysiologic impairment is probably th e conseq ue nce of decr eased tissue co m plia nce, V / Q imbalance, diffusion impairment , a nd d estruction of the micr ovascular bed. Radiogr a phicall y, consolidation becomes less confluent and ar eas of lucency and int erstitial or groundglass opacification 17 supe rvene as was seen on CT in our own patient group. The pathologic pro cess respon sible for th e pattern of gro und -glass opacification is ofte n complex and may be due to inte rstitial thi ckening (by edema , inflammator y cells or fibrosis), partial filling of th e air spa ces, or a combination of th e two. lO Areas of lucen cy in th e lung peripher y can develop at the sites of previous lung consolidation and th e transformation of lung into contiguous cy stic air spaces referred to as "adult bronchopulmonary dysplasia" has been described, but it is un usual.' 8 In th e current study, th e residuum of int er stitial fibro sis was seen as a reticular pattern and linear band-like opacifications, causing so-call ed tracti on br onchiectasis of the subsegmental bronchi. Data conc erning seq uential pathologic changes in th e lun gs of survivor s is limited, but increased collagen conte nt has been observed as ea rly as 2 week s after th e initial injury.l '' The proliferative phase of ARDS is dynamic and reports have shown histologic resolution of initial abnormalities suggesting reparativ e pro cesses and acti ve remodeling . As far as we ar e aware, th e CT data pr esent ed in th e current study represent th e first attempt to describe follow-up morphologic ch an ges in survivors with ARDS . We chose th e LIS of Murray et al 9 to add functional data at each tim e point to quantify as accurately as possible the degr ee of ph ysiologic abnormality conseCHEST I 106 I 6 I DECEMBER , 1994

1819

quent on the structur al changes. Although the LIS incorporates estimations of com pliance and the degree to which PEEP is used , the hypoxemia and chest radiography scores enable its application to patients who are self-ventilating, as in convalescence. Th e correlation between the overall percentage abnormality of the CT scans and LIS was statistically significant. This might be thought to be due in part to mathem atical coupling, as a radiographic score is incorporat ed into the LIS. Furthermore, only the radiographic and hypoxemia scores wer e calculated at the LIS at the second study point , as all the patients were self-ventilating . This influence on the statistical relationship, however , is unlikely to be important as the cross-sectional nature of CT pro vides a different and mor e precise depiction of th e extent of pulmonary abnormalities compared with two-dimensional chest radiographs. As ARDS evolves in sur vivors, functi onal recover y is most rapid in the first 6 months and reaches a plateau 1 year after the onset. 20 A red uced carbon monoxide diffusing capacity is the most common abnorma lity of lun g fun ction , bein g reported in 7 of 16 nonsm okers at 1 year. Reductions of carbon monoxid e diffusing capacity are often still appar ent even wh en corrected for alveolar volume, suggesting that the most important defect is a loss of pulmonary capillary surface area.21

warranted . We have documented the changes in CT appearanc es observ ed in survivors followin g severe ARDS and correla ted them with a severity score incorporating ph ysiologic variables. Th e implications of these obser vations for management and how modification s in therapy affect outcome in patients with ARDS remain to be determined, but with th e mor e aggressive treatment and imaging of critically ill patients, it seems appropriate to becom e acquainted with the natural history at clinical , ima ging, and path ologic levels. Th e increasing use of CT in the evaluation of pati ent s with ARDS25 will allow further insights into the natural histor y of th e disease and its long-term effects on the lung . R EFERENCES

Sever e exercise-induced desaturation is uncommon 6 or mor e months after ARDS, in contrast with marked decr eases in oxygenation observed on exertion in pati ents with established idiopathic pulmonar y fibrosis. In our own study, CT evidence of interstitial fibrosis was gene rally limited and unlik ely to account for th e entire functional deficit. Few studi es hav e addressed th e question of whether poor post-ARDS lung function can be pr edicted by features of the patient or the patient's illness.22 Th e patient's age and smoking histor y hav e not proven to be accurate predictors of subsequent lung function .F' but ample experime nta l evide nce suggests that even brief exposure to high concentrations of oxygen causes lun g injur y and can contribute to the development of fibrosis.23 Some studies ha ve correlated prolong ed exposure to an FI0 2 above 0.6 with irr eversible lung injury, but oth ers ha ve failed to find an association.r! We wer e unabl e to find an y obvious association between the degree of residual abnormality on CT I LlS in con valescence and th e extent to which a high FI02 was required to support the patient in the acute phase of the syndrome. Th e high percentage of abnormalities detected at follow-up in our study group ma y reflect th e relatively short tim e interval from diagnosis to the follow-up scan (mean = 96 da ys; range, 17 to 187 da ys) and the CT study of ARDS sur vivors over a longer period is

1 Pett y TL, Ashbaugh DG. The adult respira tory distress syndrome: clinical features, factors influencing progn osis and principles of managemen t. Chest 1971; 60:233-39 2 Fowler AA, Ham man RF, Good JT, Benson KN, Baird M, Eberle OJ, et al. Adult respiratory distress syndrome: risk with common predis positions. Ann Intern Med 1983; 98:593-97 3 Ashbaug h DG, Bigelow DB, Pett y T L, Levine BE. Acute respiratory distress in adults. Lancet 1967; 2:319-23 4 Alberti ne KH. Ultrastructura l abnorma lities in increased-permeabilit y pulmonary ede ma. Clin Chest Med 1985; 6:345-66 5 Hill JD, Ratliff JL , Parr ott JCW , et al. Pulm onary pathology in acute respirator y insufficiency: lung biopsy as a diagnosti c tool. J Thorac Cardiovasc Surg 1976; 71:64-71 6 Gattin oni L, Pesenti A, Bombin o M, Baglioni S, Rivolta M, Rossi F, et al. Relationships between lung comp uted tomograph ic density, gas exchange and PEEP in acu te respiratory failure. Anesthesiology 1988; 69:824-32 7 Gattinoni L, Pelosi P, Vitale G, Psenti A, D'Andrea L, Mascheroni D. Body position changes redistribute lung comput ed-tomographic density in patient s with acute respiratory failur e. Anesthesiology 1991; 74:15-23 8 Macnaughton PO, Evans TW . Management of the adult respiratory distress syndrome. Lancet 1992; 339:469-72 9 Murr ay JF , Matthay MA, Luce JM, Flick MR. An expanded definition of the adult respiratory distress synd rome. Am Rev Respir Dis 1988; 138:720-23 10 Rem y-Jard in M, Rem y J, Giraud P, Watt ine L, Gosselin B. CT assessment of grou nd glass opacity: seminology and significance. J Th orac Imaging 1993; 8:249-64 11 Muller NL. Clinical value of high resolution CT in chro nic diffuse lung disease. AJR 1991; 157:1163-70 12 Aberle DR, Brown K. Rad iologic conside rations in the adult respiratory distresssyndrome. Clin Chest Med 1990; 11:737-54 13 Goodman LR, Putm an CE o Diagnostic imaging in ac ute cardiopulmonary disease. Clin Chest Med 1984; 5:247-64 14 Greene R. Adult respirat ory distress syndrome: acute alveolar damage. Radiology 1987; 163:57-66 15 Greene R, Zapol WM, Snider MT. Early bedside detection of pulmon ary vascular occlusion during ac ute respirator y failur e. Am Rev Respir Dis 1981; 124:593-601 16 Bachofen H, Bachofen M, Weibel ER. Ultrastructural aspects of pulmonary ede ma. J Th orac Imaging 1988; 3:1-7 17 Greene R, Jantsch H, Boggis C, et al. Respiratory distress syndrome with new considerations. Radi ol Clin North Am 1983; 21:699-708 18 Churg A, Golden J, Fligial S. Bronchopulm onary d ysplasia in

1820

CTin Established ARDS: Correlation With Lung Injury Score (Owens et all

the adult. Am Rev Respir Dis 1983; 127:117-20 19 Fukuda Y, Ishizaki M, Masuda Y. Th e role of intra-alveolar fibrosis in the process of pulmonar y struct ural remodelling in patients with diffuse alveolar dam age. Am J Path ol 1987; 126:171-82 20 Ghio AJ, Elliot CG, Crapo RO, et aI. Respiratory imp airm ent after the adu lt respiratory distress syndrome. Am Rev Respir Dis 1989; 139:1158-62 21 Elliot CG, Rasmusson BY, Cra po RO, et aI. Prediction of pulmonary function abnorma lities afte r ad ult respirator y distress

syndro me [abstrac t]. Am Rev Respir Dis 1987; 135:634-38 22 Elliot CG, Morris AH , Ceng iz M. Pulmonary fu nction and exercise gas exchange in survivors of adult respira tory distress syndrome. Am Rev Respir Dis 1981; 123:492-95 23 Elliot CG. Pulm onar y sequelae in survivors of adult respiratory distress syndrome. Clin Chest Med 1990; 1l :789-800 24 T yler DC, Pavlin J, Boatm an ES, et aI. Oxygen and resolution of lung injury. Pediatr Pulmonol 1986; 2:75-81 25 Bone RC. Th e ARDS lung: new insights from computed tomograph y. JAMA 1993; 269:2134-35

CHEST / 106 / 6 / DECEMBER, 1994

1821