Bongard Frederic , Darryl Sue. Diagnosis and Treatment Critical Care

Подождите немного. Документ загружается.



PULMONARY DISEASE

547

of patients with pulmonary embolism but may not reflect

findings in patients already in the ICU. Nevertheless, deep

venous thrombosis of the proximal leg veins remains the most

frequent source of pulmonary thromboemboli.

The clinical manifestations of pulmonary thromboem-

bolism reflect two pathologic processes: obstruction of the

pulmonary circulation resulting in hemodynamic compro-

mise and gas-exchange abnormalities. The degree of circula-

tory compromise depends on the size and number of

thromboemboli and the preembolic state of the right side

of the heart and pulmonary circulation. Some authors have

used the term massive pulmonary embolism to describe the

angiographic occlusion of two or more lobar pulmonary

arteries or greater than 50% of the pulmonary circulation,

while others reserve this label for patients who have hemody-

namic instability directly related to the embolic event. The use

of this radiographic description of a “massive pulmonary

embolism” has started to fall out of favor as these large or

multiple emboli may or may not be associated with circula-

tory collapse and shock. Patients with a previously normal

pulmonary circulation and right ventricular function gener-

ally can tolerate occlusion of even a large pulmonary artery

with maintenance of sufficient cardiac output to avoid shock.

However, acute pulmonary thromboembolism in a patient

with preexisting pulmonary hypertension or heart failure

with minimal reserve may lead to acute right-sided heart fail-

ure and subsequent circulatory collapse. The same may hap-

pen to a previously normal patient in whom a very large

embolus lodges in the main pulmonary artery or who has

multiple moderately sized emboli in several major branches.

Occlusion of pulmonary arteries results in decreased

regional perfusion of the lungs. If ventilation to these areas is

maintained, then high

Q areas contribute to increased dead

space ventilation. Minute ventilation requirements increase to

maintain Pa

within normal range. Arterial hypoxemia is

much more common. Although the mechanism of hypoxemia

is not completely understood, it probably results from a com-

bination of ventilation-perfusion mismatching from atelecta-

sis, redistribution of pulmonary blood flow, and increased

blood transit time. Occasionally, acute pulmonary hyperten-

sion leads to the opening of a patent foramen ovale with intraa-

trial right-to-left shunting and severe refractory hypoxemia.

The manifestations of pulmonary thromboembolism

often appear to be greater than can be explained by the degree

of vascular occlusion by thrombi. Although this is often due

to a lack of cardiopulmonary reserve in patients with chronic

illness, it is probable that vasoactive and bronchoactive sub-

stances play a role as well as normal compensatory processes

within the lung circulation and parenchyma. Among poten-

tial candidates for participation in this response are products

released by platelets and endothelial cells. In addition, occlu-

sion of a pulmonary artery is associated with a decreased

amount and effectiveness of surfactant in the region of lung

supplied by that vessel contributing to atelectasis.

Pulmonary infarction is another potential manifestation

of pulmonary thromboembolism. However, this diagnosis

does not seem to alter outcome or management other than

causing different abnormalities on chest x-ray and somewhat

different clinical manifestations. Pulmonary infarction is

uncommon in pulmonary embolism because of the dual sys-

temic and pulmonary artery blood supply to the lung.

Patients who present with pulmonary infarction are more

frequently those with congestive heart failure, in whom both

pulmonary venous congestion and systemic perfusion may

be compromised at baseline.

Because the lungs receive the total cardiac output, a

number of other emboli can make their way into the pul-

monary arterial circulation. These include pieces of tumor,

including adenocarcinomas that have eroded into the sys-

temic veins; foreign bodies such as broken intravenous

catheters or particulate matter accidentally or deliberately

injected into veins; fat and tissue emboli from orthopedic

injury, operative procedures, or bone marrow infarction as

seen in acute chest syndrome in patients with sickle cell dis-

ease; air introduced through intravenous lines, lung rupture,

or decompression during ascent from underwater diving;

and amniotic fluid introduced into the systemic circulation

during a tumultuous obstetric delivery. The pathophysio-

logic consequences of these emboli depend somewhat on

the clinical situation, the size and number of emboli, and

concomitant medical problems. Of note, fat emboli may

result in a distinct syndrome with systemic manifestations as

a result of the breakdown of free fatty acids within the

microcirculation.

Clinical Features

The clinical features of deep venous thrombosis and pul-

monary thromboembolism are intertwined, and both can

present diagnostic difficulties. Deep venous thrombosis

causes nonspecific clinical findings and is sometimes found

in patients with pulmonary embolism in whom thrombosis

was previously unsuspected. The diagnosis of pulmonary

thromboembolism can be difficult because it too presents

with nonspecific symptoms, signs, and laboratory tests that

suggest other acute lung and heart diseases. In the critically

ill patient with preexisting cardiac or respiratory failure,

pneumonia, atelectasis, pleural effusion, or infection, diag-

nosing superimposed pulmonary thromboembolism may be

even more difficult. It is often suspected when a critically ill

patient undergoes acute deterioration from previous baseline

findings manifested by tachypnea, tachycardia, and impaired

gas exchange.

A. Deep Venous Thrombosis—

1. Symptoms and signs—Obstruction of the deep venous

system of the leg may result in edema of the lower part or the

entire leg, pain, tenderness, redness, and other nonspecific

features. Findings are notoriously unreliable and insensitive,

with as many as 50% of patients with deep venous thrombo-

sis being asymptomatic or lacking abnormal physical find-

ings. Clinically significant venous thrombi may not

completely occlude the vascular lumen, or collateral veins

and lymphatics may prevent swelling. Most blood clots do



CHAPTER 24

548

not elicit an inflammatory response unless there is additional

vascular injury, so redness and warmth are most often not

present in uncomplicated deep venous thrombosis.

Importantly, patients with proximal deep venous thrombosis

may have a somewhat greater tendency to have silent disease

compared with those who have calf vein involvement.

However, swelling above or below the knee, recent immobil-

ity, cancer, and fever were found to have diagnostic value

in proximal acute deep venous thrombosis. Only 5% of

95 patients had none of these five findings, whereas 42% had

between two and five of these features.

The differential diagnosis of deep venous thrombosis in

symptomatic patients includes cellulitis and other soft tissue

infections, popliteal cysts with below-the-knee swelling,

septic or inflammatory arthritis, lymphatic obstruction or

inflammation, external compression of deep veins, trauma,

and hematomas. The finding of unilateral leg swelling does

not rule out primary disease in pelvic iliac veins.

2. Diagnostic workup—Imaging studies for deep venous

thrombosis include contrast venography, duplex compres-

sion ultrasonography, impedance plethysmography, radio-

fibrinogen uptake scans, CT venography, and MRI. Contrast

venography uses injection of radiocontrast material into

peripheral veins of the legs to identify obstructing thrombi

in the deep venous system. Between 20% and 50% of proxi-

mal venous thrombi usually can be identified as intralumi-

nal filling defects. Costs and complications make this test less

desirable.

Duplex ultrasound uses a combination of real-time ultra-

sound imaging to demonstrate normal venous collapse with

direct compression and Doppler venous flow assessment.

Because collapse with compression is the most important

feature, a more accurate name for this technique is compres-

sion ultrasonography. Failure to collapse or demonstration of

intraluminal echoes is indirect evidence of deep venous

thrombosis. Compression ultrasonography is practical for

the popliteal vein, the common femoral vein, and often the

superficial femoral vein. It is generally regarded as having as

high as 89–98% sensitivity for proximal deep venous throm-

bosis and comparable specificity. It is less reliable in imaging

the venous plexus in the calf, and it cannot detect thrombi

limited to iliac or pelvic veins. As discussed below, compres-

sion ultrasonography has been incorporated into numerous

diagnostic algorithms for the diagnosis of pulmonary

thromboembolism.

Impedance plethysmography uses the change in electrical

impedance of the lower extremity when blood flows out of

the leg venous system after release of a pressure cuff. Failure

to change impedance is presumptive evidence of proximal

deep venous thrombosis. This test has been reported to be

highly accurate for diagnosing proximal deep venous throm-

bosis, although other forms of venous obstruction and con-

gestive heart failure can give false-positive results. In

addition, the accuracy of the test is likely to be hospital-

dependent or operator-dependent. It has been recommended

that each facility establish the accuracy of impedance

plethysmography compared with contrast venography.

Impedance plethysmography must be performed with the leg

held immobile and cannot be used if there is a plaster cast on

the suspected leg. In a comparison trial of compression ultra-

sonography and impedance plethysmography, the predictive

value of compression ultrasonography was significantly

higher in symptomatic outpatients.

Radiofibrinogen leg scans demonstrate the inclusion of

radiolabeled fibrinogen into actively forming thrombi. This

test is useful primarily for identifying calf and lower-thigh

deep venous thrombosis. Only clots that are actively forming

can be located with this method. It is poor in detecting prox-

imal deep venous thrombosis. This test is used rarely today

except in research protocols.

MRI and CT pulmonary angiography accompanied by

venography are currently being investigated for their role in

the workup of thromboembolic disease. The advantage of

these studies would be evaluation of the pulmonary vascular

system for emboli combined with evaluation of the pelvis

and lower extremity venous system for their source in a sin-

gle study.

Owing to its invasive nature, contrast venography is obvi-

ously not appropriate for screening patients at risk for deep

venous thrombosis. Both impedance plethysmography and

compression ultrasonography can be used effectively for this

purpose. In patients at high risk for deep venous thrombosis,

such as those with pelvic or hip trauma and those with criti-

cal medical illness, compression ultrasonography is probably

most sensitive and specific for proximal vein thrombosis,

even though there has been greater experience with imped-

ance plethysmography. In patients with deep venous throm-

bosis limited to the calf, serial compression ultrasonography

of patients is needed to identify the 15–20% of patients who

extend their thrombi into the proximal veins. Most clots have

been found to extend proximally within the first 7 days. Thus

studies have recommended follow-up examinations within

2–3 days and again in 7–10 days if clinical suspicion for deep

venous thrombosis remains high.

B. Pulmonary Embolism

1. Symptoms and signs—The clinical features of pul-

monary embolism have been accurately described, and

review of these findings demonstrates their nonspecificity.

Table 24–3 summarizes data from a series of 500 patients

comparing the clinical signs and symptoms in patients with

proven pulmonary emboli by pulmonary angiography (202

patients) versus those without pulmonary embolism.

Dyspnea and chest pain were the most common complaints;

tachycardia, tachypnea, rales, and an increased intensity of

the pulmonic component of the second heart sound were the

most frequent findings on physical examination. The classic

findings of hemoptysis, chest pain, and dyspnea were

uncommon as a triad. Fewer than one-third of the total

group had symptoms or signs suggesting deep venous

thrombosis. While it is correct that these clinical findings do



PULMONARY DISEASE

549

not distinguish patients with pulmonary embolism from

those with other severe heart and lung diseases, patients sus-

pected of pulmonary embolism who have cyanosis, hypoten-

sion, shock, or syncope deserve particular consideration.

Symptoms and signs can establish meaningful probabili-

ties for pulmonary embolism. Several clinical models for esti-

mating probability have been developed and validated.

Patients suspected of pulmonary embolism were stratified

into low, moderate, and high probability, and these classifica-

tions were confirmed by subsequent testing (Table 24–4).

This clinical estimate of pulmonary embolism likelihood is

the first step in diagnosis, followed by laboratory and imaging

studies. Objectively derived clinical prediction rules should be

used for judging “pretest” probability of pulmonary

embolism. These have been derived for nonhospitalized

patients suspected of pulmonary embolism and may not be

valid in hospitalized or ICU patients.

2. Laboratory findings—Arterial blood gases most often

show mild to moderate hypoxemia, increased P(

–a)

, and

mildly reduced Pa

. Almost all patients with pulmonary

embolism have a Pa

of less than 80 mm Hg, but no

absolute level of Pa

can be used to exclude the diagnosis.

Diagnostic accuracy may be improved somewhat by using

the P(

–a)

difference rather than the Pa

,but again,a

clear distinction between those with and those without pul-

monary embolism cannot always be made. In the Prospective

Investigation of Pulmonary Embolism Diagnosis (PIOPED)

study, 7% of patients with angiographically documented

pulmonary emboli had completely normal arterial blood gas

measurements. Conversely, another study found that

patients suspected of having a venous thromboembolic event

who had a normal P(

–a)

had only a 1.8% chance of hav-

ing a pulmonary embolism. Severe hypoxemia refractory to

oxygen administration may indicate the opening of a patent

foramen ovale. Hypercapnia is unusual and suggests the

presence of underlying lung disease.

PE Present

(

= 202)

PE Absent

(

= 298)

Symptoms

Dyspnea

Sudden onset

Gradual onset

78%

29%

20%

Chest pain

Pleuritic

Nonpleuritic

44%

16%

30%

10%

Orthopnea 1% 9%

Fainting 26% 13%

Hemoptysis 9% 5%

Cough 11% 15%

Palpitations 18% 15%

Signs

Tachycardia

HR >100 beats/min

24% 23%

Cyanosis 16% 15%

Hypotension (SBP <90 mm Hg) 3% 2%

Neck vein distention 12% 9%

Unilateral leg swelling 17% 9%

Fever >38°C 7% 21%

Crackles 18% 26%

Wheezes 4% 13%

Pleural friction rub 4% 4%

Data from Miniati M et al: Accuracy of clinical assessment in the

diagnosis of pulmonary embolism. Am J Respir Crit Care Med

1999;159:864–71.

Table 24–3. Symptoms and signs in 500 patients with

suspected pulmonary embolism (PE).

Table 24-4. Estimating clinical probability of pulmonary

embolism (PE).

Points

Risk factors

Age >65 years 1

Previous DVT or PE 3

Surgery under general anesthesia or fracture 2

of the lower limb within 1 month

Active malignant condition (solid or hematologic 2

malignant condition, currently active or

considered cured <1 year)

Symptoms

Unilateral lower limb pain 3

Hemoptysis 2

Clinical signs

Heart rate 75–94 beats/min 3

Heart rate ≥95 beats/min 5

Pain on lower limb deep venous palpation and 4

unilateral edema

Clinical probability of pulmonary embolism

Low (8%) Total = 0–3

Intermediate (28%) Total = 4–10

High (74%) Total ≥11

Modified from Le Gal G et al: Prediction of pulmonary embolism in

the emergency department: The revised Geneva score. Ann Intern

Med 2006;144:165–71.



CHAPTER 24

550

Sinus tachycardia is a frequent and nonspecific finding in

acute pulmonary embolism. Supraventricular tachycardia

and atrial fibrillation are sometimes present. Features

suggesting acute right-sided heart strain on the ECG occur

relatively infrequently; these include acute right-axis devia-

tion, P pulmonale, right bundle branch block, and

inverted T waves and ST-segment changes in right-sided

leads. Electrocardiographic patterns such as an S wave in lead

I, a Q wave in lead III, and an inverted T wave in lead III

(“S1Q3T3”) and S waves in leads I, II, and III (“S1S2S3”)

were considered highly predictive of pulmonary embolism.

These observations were found in fewer than 12% of patients

with pulmonary emboli. In the differential diagnosis of pul-

monary embolism, the ECG is particularly useful to assess

the presence of myocardial ischemia and infarction.

D-dimer, a fibrin degradation product, is found in the

plasma of patients with deep venous thrombosis and pul-

monary embolism. D-dimer is the result of plasmin action

(thrombolysis) on fibrin monomers that have undergone

cross-linking by factor XIII to form fibrin polymers. Various

methodologies are available to measure D-dimer levels. ELISA

D-dimer assays have a higher sensitivity and negative predictive

value (91–100%) when compared with latex agglutination

techniques. However, the older ELISAs were more labor-

intensive, required skilled personnel, and took hours to com-

plete, making them less useful clinically in an emergent

situation, whereas the semiquantitative latex agglutination

studies can be performed at the bedside. A D-dimer level of less

than 500 μg/L by ELISA is considered the cutoff for excluding

venous thromboemboli. Newer latex whole blood agglutina-

tion techniques have demonstrated consistently high negative

predictive values in patients with low pretest probability of dis-

ease. Current recommendations are to combine this laboratory

finding with the pretest clinical probability as well as some

other noninvasive evaluation to guide decision making for

diagnosis and management. The one exception would be in the

face of a low clinical pretest probability, when the finding of a

low D-dimer may be enough to exclude venous thromboem-

bolic disease. Lastly, D-dimer is of limited use in a number of

clinical scenarios, which are associated with elevated D-dimer

levels as part of the disease state including surgery or trauma in

the past 3 months, underlying malignancy, sepsis with or with-

out disseminated intravascular coagulation (DIC), inflamma-

tory states, pregnancy, or abnormal liver function.

Elevation in cardiac troponins in the setting of an acute

pulmonary embolism has been described. In patients with

either a moderate or large pulmonary embolism, troponin T

(TnT) levels greater than 0.1 ng/mL were seen in 32% of

patients in one trial. None of the patients with small emboli

had an elevation of this cardiac marker. In another study,

tropoinin I (TnI) levels greater than 0.4 ng/mL were seen in

21% of patients with pulmonary embolism, with levels

exceeding 2.3 ng/mL in 4% of patients. It is thought that the

strain on the right ventricle from the increased pulmonary

arterial resistance in the face of an acute embolism leads to

the right ventricular myocardial ischemia in these patients

and is associated with a poorer prognosis.

3. Imaging studies—Radiographic studies include nonspe-

cific tests such as chest x-rays, examinations of the pul-

monary circulation such as perfusion lung scans, CT

pulmonary angiography (spiral or helical CT scans), MRI,

and pulmonary angiograms, as well as studies directed at

finding deep venous thrombosis.

a. Chest x-ray—The chest x-ray is most useful in identi-

fying coexisting problems such as pneumonia, lung mass,

lymphadenopathy, pulmonary edema, atelectasis, or pleural

effusion. The most common findings in pulmonary

embolism without coexisting disease are nonspecific, includ-

ing no visible abnormality, enlarged cardiac silhouette, ele-

vated hemidiaphragm, atelectasis, or small pleural effusion.

A normal chest x-ray in a patient with shortness of breath

and hypoxemia should prompt a further evaluation for pul-

monary embolism. Findings suggestive of pulmonary vascu-

lar occlusion, such as an apparent cutoff of a segmental or

lobar pulmonary artery, regional hyperlucency of the lung

parenchyma or oligemia (Westermark’s sign), or a wedge-

shaped density consistent with pulmonary infarction

(Hampton’s hump if located in the periphery), may suggest

pulmonary embolism but are insensitive and lack specificity.

b. Radionuclide ventilation-perfusion scan—The venti-

lation-perfusion lung scan was previously the test used most

frequently to diagnose pulmonary embolism. Newer imaging

studies have replaced this nuclear medicine study in many

centers. In addition, scan results must be considered carefully

because these tests do not have perfect diagnostic accuracy,

and over 70% of patients undergoing this radiographic eval-

uation have indeterminate results. To perform the perfusion

scan, radionuclide-labeled macroaggregated albumin is

injected into a peripheral vein, after which the labeled parti-

cles become trapped in the pulmonary capillary bed.

Uniform distribution of the radionuclide throughout the

lung fields implies the absence of significant localized pul-

monary arterial obstruction, whereas a pulmonary

embolism occluding a pulmonary artery will result in a per-

fusion defect. Unfortunately, perfusion defects commonly

result from other causes, including focal vasoconstriction

accompanying atelectasis, pneumonia, or bronchospasm.

To improve diagnostic value, uniformity of ventilation is

assessed using a ventilation scan, performed by inhalation of

either radioactive Xenon or an aerosol containing a radiola-

beled solute. The perfusion and ventilation scans are then

compared. A sufficiently large perfusion defect without a cor-

responding ventilation defect in the same area (ie, mis-

matched defect) generally is considered supportive of the

diagnosis of pulmonary embolism. On the other hand, a

matched ventilation-perfusion defect generally is considered

indeterminate and not helpful in making the diagnosis of

pulmonary embolism or other kinds of heart or lung disease

such as pneumonia or bronchospasm.

By convention, ventilation-perfusion lung scans are inter-

preted as normal (no perfusion defects), low or high proba-

bility for pulmonary embolism, or intermediate probability

(sometimes called indeterminate) for pulmonary embolism.

Prospective studies have led to accepted diagnostic strategies



PULMONARY DISEASE

551

for pulmonary embolism. The PIOPED multicenter study

compared lung scan results with pulmonary angiography in

patients with suspected pulmonary embolism. Table 24–5 is

a modified summary of lung scan categories and criteria

used in 931 patients included in this study.

Of patients with suspected pulmonary embolism, lung

scans indicated a high probability of pulmonary embolism in

13% and were normal or nearly normal in 14%. Intermediate

or low probability was the conclusion in 73%. The sensitiv-

ity, specificity, and likelihood ratios of lung scan interpreta-

tions for angiographically diagnosed pulmonary emboli are

shown in Table 24–6. The likelihood of pulmonary embolism

parallels the interpretation of the lung scans, especially when

used in conjunction with the pretest clinical likelihood of

pulmonary embolism. A reading of any probability (low,

intermediate, or high) of pulmonary embolism on lung scan

resulted in 98% sensitivity but low specificity. Unfortunately,

only 41% of cases of pulmonary embolism had high-

probability lung scans, whereas 42% had intermediate-

probability scans and 17% had low-probability scans. It is

emphasized that patients with low-probability lung scans are

found to have pulmonary emboli about 15–30% of the time.

These and other data show that ventilation-perfusion lung

scans are of greatest utility when they show high probability for

pulmonary embolism (87% likelihood of pulmonary embolism

in all patients suspected of pulmonary embolism; 96% if clinical

suspicion is high) or when they are normal (nil to 4% likelihood

of pulmonary embolism during long-term follow-up even if clin-

ical suspicion was high). Thus high-probability lung scans effec-

tively predict pulmonary embolism, whereas a normal scan (no

perfusion defects) effectively excludes pulmonary embolism.

Treatment or withholding therapy can be guided by this rule with

considerable confidence. Unfortunately, the majority of patients

suspected of having pulmonary embolism fall into intermediate

or low probability, and substantial numbers within each

Table 24–5. PIOPED lung scan interpretation criteria

(modified).

High probability

≥ 2 large (>75% of the segment), OR

≥ 2 moderate (25–75% of the segment) plus one large, OR

≥ 4 moderate mismatched perfusion lung scan defects

Low probability

Nonsegmental perfusion defects only, OR

One moderate mismatched segmental perfusion defect with normal

chest x-ray, OR

Any perfusion defect with a larger chest x-ray abnormality, OR

Limited number of large or moderate perfusion defects with

matching ventilation defects (with normal or mildly abnormal

chest x-ray)

Intermediate probability

Not falling into normal, low-, or high-probability categories

Borderline high or borderline low

Difficult to categorize as high or low

Normal

No perfusion defects, OR

Perfusion outlines exactly the shape of lungs seen on chest x-ray

(chest x-ray or ventilation lung scan may be abnormal)

Table 24-6. Sensitivity, specificity, likelihood ratios, and posttest probabilities of ventilation-perfusion radionuclide

lung scans and CT pulmonary angiograms for pulmonary embolism.

For a patient with suspected pulmonary embolism, estimate pretest probability (clinical information or clinical information plus results of prior tests).

Look up posttest probability at intersection of pretest probability (column) and test result (row).

Pretest Probability

∗

8% 28% 50% 74%

Ventilation-Perfusion Radionuclide Scan

Scan Result Sensitivity Specificity Likelihood Ratio Posttest Probability

High probability 41% 97% 17.1 60% 87% 94% 98%

Indeterminate 41% 62% 1.1 9% 30% 52% 76%

Low probability 16% 60% 0.4 3% 13% 29% 53%

Near-normal or normal 2% 81% 0–0.1 0–1% 0–4% 0–9% 0–22%

CT Pulmonary Angiogram

Positive 78–97% 53–100% 3.5–32 23–74% 58–93% 78–97% 91–99%

Negative 0.05–0.48 0–4% 2–16% 5–32% 12–58%

∗

Using the revised Geneva Score (see Table 24–4).



CHAPTER 24

552

category will or will not actually have the disease. It should be

emphasized that these studies have characterized mostly patients

who were not critically ill, in whom the appropriate studies could

be performed and compared. The predictive value of lung scans

may not be comparable in patients in the ICU.

c. Combination imaging—Multiple approaches for

improving diagnostic accuracy when lung ventilation-

perfusion scanning is nondiagnostic have been described

(Figure 24–1). The traditionally used strategy is to perform

pulmonary angiography for all suspected patients in whom

intermediate probability lung scans are found (pulmonary

embolism present by angiography in 16–66% depending on

clinical suspicion) or in whom a low-probability lung scan is

associated with high or uncertain clinical suspicion (pul-

monary embolism found in 16–40%). A normal pulmonary

angiogram is quite accurate in excluding pulmonary

embolism, with fewer than 1% of patients with negative pul-

monary angiograms subsequently proving to have pulmonary

embolism without treatment during long-term follow-up or

at autopsy. This approach would be ideal if it were not for

problems encountered with obtaining pulmonary angiogra-

phy, including limited availability, uncertain reliability in all

hospitals, and risks of vascular catheterization and radi-

ographic contrast agents. Small amounts of contrast material,

selective pulmonary angiography, nonionic contrast material,

and careful preinjection measurement of pulmonary artery

pressure reduce the frequency of complications from pul-

monary angiography. Pulmonary hypertension is a relative

contraindication because of reported morbidity and mortal-

ity from injecting additional volume into a system already

under high pressure. Other complications, including death,

respiratory distress leading to intubation, renal failure, and

hematoma requiring transfusion, are reported to occur in

up to 4% of critically ill patients undergoing pulmonary

angiography. Pulmonary angiography used to be per-

formed in all patients being considered for thrombolytic



Figure 24–1. One suggested approach to a patient with suspected pulmonary embolism. After estimation of clinical

probability (see Table 24–4), most patients should have CT pulmonary angiogram performed unless a contraindication exists.

If the cumulative probability of clinical estimation plus CT angiogram results is sufficient to begin or withhold treatment,

diagnostic studies are completed. If not, further studies are required (eg, compression ultrasonography or pulmonary

angiography) until a decision can be reached. Any decision to begin or withhold treatment must take into account the risk

of treatment compared with the potential benefits of treatment.

Suspected pulmonary embolism

Estimate clinical probability

Low (8%) Moderate (28%) High (78%)

D-dimer (ELISA)

No treatment

DVT studies Treat

CT angiogram

High

Positive

Low Negative

Negative

Pulmonary

angiogram

If CT is contraindicated due to iodine allergy or creatinine clearance is prohibitive, consider

starting with V/Q scan or compression ultrasonography



PULMONARY DISEASE

553

therapy, but this recommendation is being questioned

because of the subsequent risk of bleeding (as high as 20%).

The major disadvantage of this approach is that a large num-

ber of patients would require pulmonary angiography

because of abnormal but nondiagnostic lung scans. In some

studies, patients making up this group comprise 40–60% of

the total enrolled.

A second approach recognizes the close relationship

between pulmonary thromboembolism and deep venous

thrombosis. Proximal vein deep venous thrombosis is found

on initial testing in about 50% of patients with pulmonary

embolism. In patients with low- or intermediate-probability

ventilation-perfusion lung scans but with high or uncertain

clinical probability of pulmonary embolism, compression

ultrasonography or impedance plethysmography may be

performed. A combination of high or uncertain clinical sus-

picion, abnormal (but not high-probability) lung scan, and

positive noninvasive test for deep venous thrombosis

strongly supports the diagnosis of pulmonary embolism.

Treatment should be initiated on the basis of this result. The

absence of evidence of deep venous thrombosis, however,

should not rule out pulmonary embolism because the false-

negative rate of a single noninvasive test for deep venous

thrombosis ranges from as low as 3% to as high as 30%.

Therefore, these patients must undergo pulmonary angiog-

raphy or serial duplex ultrasonography if the initial ultra-

sound is negative (see below). The combination of lung scan

and noninvasive deep venous thrombosis studies decreases

the number of pulmonary angiograms needed in these

patients from about 72% to 33%. Completely noninvasive

strategies for selected patients also have been proposed. In

patients suspected of pulmonary embolism who have abnor-

mal but nondiagnostic lung scans (ie, low or intermediate

probability) and adequate cardiopulmonary reserve (eg, lack

of respiratory failure, hypotension, severe underlying lung

disease, and severe tachycardia), serial noninvasive tests for

proximal lower extremity deep venous thrombosis are per-

formed (eg, impedance plethysmography or compression

ultrasonography). If evidence of deep venous thrombosis is

found initially or subsequently, treatment is started.

However, if no evidence of deep venous thrombosis is found,

anticoagulation is withheld while noninvasive deep venous

thrombosis studies are repeated at least twice over the next

10–14 days. In a study of 627 untreated patients with sus-

pected pulmonary embolism with nondiagnostic lung scans

and negative serial impedance plethysmographic studies over

2 weeks, pulmonary thromboembolism occurred in only

1.9% over the next 3 months. Treatment therefore can be

withheld in this group of patients with acceptable results. In

fact, this approach clarifies the natural history of pulmonary

thromboembolism in that extension of the pulmonary

embolus itself rarely occurs without treatment, and treat-

ment is directed solely at prevention of extension of the

venous thrombus. This noninvasive approach can be used to

avoid pulmonary angiography in some patients with nondi-

agnostic lung scans. For critically ill patients, however, this

strategy may not be feasible because of concomitant heart

and lung disease and lack of cardiopulmonary reserve. In

addition, patients being considered for therapy other than

anticoagulation, such as thrombolytic therapy, cannot be

evaluated appropriately using this method.

Another completely noninvasive strategy included the com-

bined use of CT angiography, a ventilation-perfusion scan,

plasma D-dimer measurements, and in some patients compres-

sion ultrasonography of the lower extremities. In 247 patients

evaluated using this strategy, the 3-month risk of developing an

embolic event without therapy if none of these studies revealed

venous thromboembolism was only 1.7%. In general, the diag-

nostic protocols that combine these noninvasive studies either

have obtained all studies during the initial assessment and made

treatment decisions based on all study results altogether or have

obtained a single study at a time and continue to obtain addi-

tional diagnostic data if the results are not conclusive until a

definitive diagnosis is obtained or excluded.

d. CT pulmonary angiography—Imaging studies that

attempt to assess the pulmonary vasculature bed directly have

been evaluated for their role in the diagnostic workup of pul-

monary thromboembolism. The development of helical

(spiral) and electron-beam CT pulmonary angiography has

nearly replaced the ventilation-perfusion scan in many cen-

ters with its reported sensitivities in most studies of greater

than 80% and specificities of greater than 90% for diagnosing

pulmonary emboli in the main, lobar, and segmental pul-

monary arteries. The lower sensitivity of this imaging study

results in part from the poor performance in the diagnosis of

subsegmental emboli. The clinical impact of emboli in these

subsegmental arteries is unclear, and they may not pose the

same morbidity and mortality risks as emboli in larger seg-

ments. However, in patients with a limited cardiopulmonary

reserve, emboli in subsegmental arteries potentially could be

devastating. Newer imaging techniques that are available

today, including thin collimation (1–3-mm instead of 5-mm

slices) and faster acquisition timing have improved the evalu-

ation of these subsegmental vessels. An additional advantage

of these imaging studies is their ability to provide additional

information with regard to other disease processes within the

thorax that may be responsible for the patient’s clinical pres-

entation, such as pneumonic infiltrates, pleural disease, medi-

astinal lymphadenopathy, or parenchymal masses. CT

imaging requires the administration of intravenous contrast

material and a degree of patient cooperation with the ability to

lie still and breath-hold for approximately 25 seconds in some

protocols to obtain good-quality pictures of the vasculature.

In a small but important proportion of studies, the results are

not acceptable because of movement artifacts or inadequate

concentration of contrast material in the pulmonary arteries.

The exact role of these CT imaging techniques with the

diagnosis of pulmonary thromboembolic disease is continu-

ing to evolve, and these techniques are included in a number

of combined-strategy models. In fact, some investigators

have adopted CT imaging as the key starting diagnostic test

for all patients considered likely to have a pulmonary



CHAPTER 24

554

embolism. A meta-analysis evaluated the 3-month clinical

outcome in patients with suspected pulmonary embolism

managed solely based on the results of CT pulmonary

angiography. The rate of subsequent venous thromboem-

bolism after a negative CT result in untreated patients was

1.4%, which is similar to the rate seen in previous studies

with negative pulmonary angiograms. Because concern still

exists among many physicians about the overall sensitivity of

CT imaging but with acceptable specificity, the helical CT

can be used to diagnose pulmonary thromboembolism if a

thrombus is seen but may not be able to exclude significant

pulmonary artery thrombi if clots are not visualized in

patients with a high pretest probability of having this disease

state. Nevertheless, a diagnostic strategy based on clinical

probability, D-dimer, and CT pulmonary angiography

proved highly effective. In those in whom pulmonary

embolism was “excluded,” only 1.3% of untreated patients

had recurrent venous thromboembolism.

CT pulmonary angiography combined with CT venogra-

phy of the pelvis and lower extremities to evaluate the venous

system for the source of the embolus during the same study

(ie, CTA-CTV) was investigated in the PIOPED II trial. While

the addition of venography increased the sensitivity of this

combined CT imaging study, the authors concluded that

both CT angiography alone or combined with CT venogra-

phy had high predictive values when concordant with the

pretest probability of disease. However, when these imaging

studies were discordant with the pretest probability, addi-

tional studies were needed before excluding venous throm-

boembolic disease from the differential diagnosis for the

patient’s clinical signs and symptoms. Therefore, the addi-

tion of CT venography to diagnostic algorithms has not been

strongly recommended based on the little it adds to predic-

tive value of CT angiography and clinical probability. MRI of

both the pulmonary vasculature and the venous system of

the pelvis and lower extremity also may have a future role in

the workup of this disease. The advantage MRI studies of the

vasculature would be the avoidance of iodine-based contrast

material and its associated risks, including anaphylactic reac-

tions and renal impairment.

4. Diagnostic approach to pulmonary thromboem-

bolism—Many algorithms optimizing diagnostic strategies

have been proposed. A logical approach is to assess clinical

pretest probability based on symptoms and signs and then

apply tests that increase or decrease the probability of disease

(posttest probability). In this way, the likelihood of pul-

monary embolism can be estimated and the risks of treat-

ment compared. Table 24–5 assigns low, moderate, and high

clinical estimates for disease based on clinical findings for

patients evaluated in an emergency department setting.

Validation of these data confirmed that these groups had 8%,

28%, and 74% likelihood of pulmonary embolism.

The clinician must decide whether these probabilities are

a sufficient basis for decisions about whether or not to treat

a patient. If a more certain diagnosis is warranted—and this

is almost always the case with clinical information alone—

another diagnostic test is applied. The first diagnostic study

in many algorithms in the past was the ventilation-perfusion

radionuclide scan, but this has now been replaced in many

centers by CT pulmonary angiography, especially in patients

with preexisting obstructive or parenchymal lung disease

likely to result in uninterpretable scans. Table 24–6 compares

posttest probability using typical sensitivity and specificity

values for various imaging results for selected clinical pretest

probabilities. If posttest probability is sufficiently high to jus-

tify starting treatment or sufficiently low to justify withhold-

ing treatment, further diagnostic tests are not indicated. If

more diagnostic certainty is desired, then noninvasive stud-

ies—compression ultrasonography or pulmonary angiogra-

phy—can be performed. For decision-making purposes, the

pulmonary angiogram is assumed to have 100% specificity;

that is, a negative study excludes pulmonary embolism. The

probabilities of pulmonary embolism when assessed by com-

pression ultrasonography, D-dimer, and other diagnostic

tests are shown in Table 24–7. These tests can be applied

sequentially to support the diagnosis or to exclude the diag-

nosis; although to be certain, these estimates are derived

from a number of studies of different populations. While

they are not strictly independent estimates, the combined

probabilities are still likely to have value. Any decision to

begin or withhold treatment must take into account the risk

of treatment compared with its potential benefit especially in

the ICU patient population.

Treatment

A. Anticoagulation—Anticoagulation is the major therapy

for deep venous thrombosis and pulmonary embolism.

Heparin, either unfractionated heparin (UFH) or low-

molecular-weight heparin (LMWH), is most often given for

4–5 days, overlapping with oral anticoagulant therapy begin-

ning on day 1. Oral anticoagulant agents, mainly in the form

of vitamin K antagonists such as warfarin, are given for a

minimum of 3 months. There are a number of other treat-

ment schedules depending on the underlying cause of the

event, the history of previous thromboembolic events, and

the patient’s underlying medical history. Heparin therapy

should be started immediately before any definitive tests are

performed if there is a strong clinical suspicion and no con-

traindications exist. Heparin also should be started once a

diagnosis of deep venous thrombosis or pulmonary

embolism is confirmed, if not started prior to the diagnosis.

Anticoagulants do not directly affect existing thrombi, but if

given in sufficient amounts, they can prevent further clot

propagation until the patient’s inherent fibrinolytic system

can begin to break down the thrombus or embolism.

In the absence of contraindications, initial treatment

should be with heparin in the form of continuous intra-

venous infusion of UFH or subcutaneous LMWH. To begin

therapy with continuous UFH, an infusion of heparin is

given at a dose that achieves and maintains a stable activated

partial thromboplastin time (aPTT) of 1.5–2.5 times control

when measured at 6-hour intervals. The aPTT then can be



PULMONARY DISEASE

555

measured at approximately daily intervals once this goal

range has been achieved. Failure to achieve an aPTT at least

in the lower level of this range within the first 24 hours of

therapy has been associated with recurrent venous throm-

boembolism in patients with deep venous thrombosis.

Standard heparin dose-adjustment protocols or nomo-

grams for deep venous thrombosis and pulmonary

thromboembolism are highly desirable and have been

shown to increase therapeutic efficacy and reduce bleeding

complications. One weight-based nomogram is shown in

Table 24–8. A bolus of UFH at 80 units/kg of body weight is

given, followed by 18 units/kg per hour. For a 60-kg adult, this

corresponds to 4800 units as a bolus followed by 1080 units/h

or about 26,000 units/day. Concern has been expressed that

Table 24-7. Sensitivity, specificity, likelihood ratios, and posttest probabilities of pulmonary embolism from selected

studies of compression ultrasonography and D-dimer.

Instructions:

For a patient with suspected pulmonary embolism, estimate pretest probability and find posttest probability at intersection of pretest probability

(columns) and test results (rows).

aPTT (s)

Rate Change

(units/kg/h)

Additional Action Next aPTT

<35 (<1.2 × normal) +4 Rebolus with 80 units/kg 6 hours

35–45 (1.2–1.5 × normal) +2 Rebolus with 40 units/kg 6 hours

46–70 (1.5–2.3 × normal) 0 None 6 hours

71–90 (2.3–3.0 × normal) –2 None 6 hours

>80 (>3 × normal) –3 Stop infusion for 1 hour 6 hours

∗

During the first 24 hours, repeat aPTT every 6 hours. Thereafter, obtain aPTT every morning unless it is outside the

therapeutic range.

Modified from Hyers TM: Venous thromboembolism. Am J Respir Crit Care Med 1999;159:1–14.

Table 24–8. Body weight-based dosing of intravenous unfractionated heparin.

Initial dosing

Loading: 80 units/kg

Maintenance infusion: 18 units/kg/h using 25,000 units in 250 mL D

W (100 units/mL)

Obtain aPTT before and 6 hours after starting heparin

Subsequent dose adjustments based on aPTT measured at 6-hour intervals

∗

Pretest Probability

8% 28% 50% 78%

Compression Ultrasonography (for Patients with Symptoms of DVT)

Sensitivity Specificity Test Result Likelihood Ratio Posttest Probability

95% 95% Positive 19.00 62% 88% 95% 98%

Negative 0.05 0.40% 2% 5% 12%

Compression Ultrasonography (for Patients without Symptoms of DVT)

62% 97% Positive 21.00 65% 89% 95% 98%

Negative 0.39 3% 13% 28% 53%

D-Dimer

85–100% 45–68% Positive 1.6–2.7 12–19% 38–51% 62–73% 82–88%

Negative 0.09–0.22 0–1.80% 3–8% 8–18% 20–39%



CHAPTER 24

556

the maintenance dose of heparin is frequently underesti-

mated, resulting in recurrent deep venous thrombosis or pul-

monary thromboemboli. The mean heparin requirement in

several studies was approximately 1300 units/h (about 31,000

units/24 h) compared with older studies suggesting that

1000 units/h usually was adequate. The dose is adjusted

upward or downward as needed based on the aPTT.

Although this nomogram should be strictly used only when

the same reagent is employed as in the development of the

nomogram, it should prove useful as a guide in all hospitals.

LMWH differs from standard UFH in its pharmacoki-

netics, bioavailability, and anticoagulant activity and has

found a role in the treatment of both deep venous throm-

bosis and pulmonary embolism. These heparin fractions

have greater bioavailability when given subcutaneously,

longer duration of action, allowing for once- or twice-daily

dosing, and predictable anti-factor Xa activity. The antico-

agulation effects of LMWH can be correlated with body

weight for dosing purposes, and this diminishes the need

for following laboratory parameters to ensure adequate

anticoagulation. In direct comparisons, LMWH has been

associated with fewer major bleeding complications, less

thrombocytopenia, and a lower incidence of osteoporosis

than UFH. All LMWH formulations crossreact with UFHs

and cannot be used as an alternative form of anticoagulation

in patients with heparin-induced thrombocytopenia syn-

drome. Dosage adjustments may be necessary in patients

with morbid obesity and renal insufficiency. Each LMWH

formulation has its own distinct pharmacokinetic profile,

so data on one form cannot be readily extrapolated to

another form in the same class. In the treatment of deep

venous thrombosis and pulmonary thromboembolism, the

2004 American College of Chest Physicians (ACCP) con-

sensus recommendations are as follows: LMWH treatment

administered subcutaneously in doses adjusted to body

weight or dose-titrated intravenous UFH for at least the

first 5 days of therapy; LMWH is preferred over UFH in

patients with “nonmassive” pulmonary emboli, whereas

UFH is preferred in patients with severe renal impairment.

In addition to heparin, warfarin can be started on day 1

unless there are contraindications to its use. Warfarin, an oral

anticoagulant that inhibits synthesis of vitamin K–dependent

coagulation factors, becomes an effective anticoagulant only

after disappearance of previously synthesized circulating

coagulation factors. Thus several days are needed for war-

farin to have an antithrombotic effect, whereas its anticoag-

ulant effect occurs sooner. In the past, concern has been

raised about a potential hypercoagulable state induced by

warfarin because synthesis of proteins C and S, naturally

occurring anticoagulants, is also inhibited by this agent. This

problem is encountered very rarely.

Almost all patients can be started with a single oral dose

of warfarin at 5 mg/day. The goal is to achieve an interna-

tional normalized ratio (INR) of 2.5 (range 2.0–3.0). This

corresponds roughly to a prolongation of the prothrom-

bin time to 1.3–1.5 times normal (using the usual tissue

thromboplastin assay employed in laboratories in North

America). Nomograms for warfarin dosing suggest that the

INR measured at least 15 hours after the first dose is helpful

in deciding on subsequent doses. If the first INR is greater

than 1.5, a very low maintenance dose (1 mg) is probably suf-

ficient; an INR of between 1.2 and 1.3 calls for a dose of 2–3

mg/day. All other patients should receive a second oral dose

of 5 mg and should be monitored by continued daily INR

measurements. Heparin can be discontinued after 4–5 days if

the INR is greater than 2.0 for two consecutive days.

The dose of warfarin should be adjusted according to the

prothrombin time. Initially, the prothrombin time should

be measured daily until a stable INR is achieved. Thereafter,

twice-weekly measurements followed by weekly measure-

ments should be adequate. A number of drugs interact with

warfarin, both increasing and decreasing its effectiveness.

Antibiotics in particular may decrease bacterial flora of the

gut that are responsible for a significant amount of vitamin

K synthesis. In the ICU, other drugs that may prolong the

prothrombin time by potentiating the action of warfarin

include aspirin, nonsteroidal anti-inflammatory drugs

(NSAIDs), omeprazole, and amiodarone, as well as antimi-

crobial agents such as most cephalosporins, erythromycin,

fluconazole, and metronidazole. On the other hand, barbi-

turates, rifampin, and carbamazepine may reduce the effect

of warfarin on the prothrombin time by increasing the rate

of its metabolism in the liver. Warfarin and other oral vita-

min K antagonists are contraindicated during pregnancy

because of the potential for abnormal fetal development.

The optimal duration of anticoagulation therapy for deep

venous thrombosis and pulmonary embolism has not been

determined for all clinical situations and depends on under-

lying predisposing risk factors. Studies have shown that

heparin followed by 3 months of warfarin results in an

acceptably low (<5%) frequency of recurrent deep venous

thrombosis in patients who have a transient and identifiable

risk factor. This applies to patients with a first event whose

predisposing risk factor, such as immobilization from a bro-

ken bone, surgery, or trauma, has resolved. For patients with

a first episode of idiopathic venous thromboembolism (no

identifiable risk factor), at least 6 months of anticoagulation

therapy is recommended. On the other hand, patients in

whom risk factors are long term and poorly reversible, such

as those with chronic congestive heart failure or hypercoagu-

lable states, should receive a longer period of anticoagulation

therapy on the order of 12 months or even indefinitely. In

patients with a hypercoagulable state associated with malig-

nancy, the effectiveness of anticoagulation is highly variable.

Patients unable to receive warfarin should be given either

subcutaneous UFH at a dosage sufficient to prolong the

aPTT more than 1.5 times control (adjusted-dose subcuta-

neous heparin) or daily LMWH.

The major complication of UFH therapy is bleeding,

occurring in approximately 5% of patients (ranges from 1% in

those with low risk for bleeding to 10% in those with high risk).

LMWH has imposed a lower risk of major bleeding events