the onset of clinical symptoms (24,33,45,46). The initial nodular interstitial spread occurs
without significant alveolar involvement. In order to be large enough to be appreciated on a
plain chest radiograph, however, some spread to the adjacent alveoli will have to have
occurred (47). Furthermore, while many studies report extraordinary high rates of classic
radiologic findings; this usually is a self-fulfilling prophecy as the radiologic findings were
often used as an inclusion criterion as well. Recent studies that did not rely on radiologic
criteria for inclusion found the classic X-Ray presentation in less than 50% of patients with
miliary TB (24,33). An additional 10% to 30% of patients have larger or atypical lesions.
Asymmetrical nodular pattern, coalescing nodules, mottled appearance, snowstorm appear-
ance, ground-glass appearance, and air-space consolidation have been described (3).
Conversely, other conditions that typically present with larger nodules such as alveolar
hemorrhage, lymphangitic cancers, or inhalational diseases can appear as early small nodules.
While most of the nodules observed in these diseases tend to be larger and more
heterogeneous than classic miliary TB, the overlap may be significant (48). Approximately
5% of patients have additional findings that may provide additional clues to the diagnosis.
Such findings include evidence of intrathoracic lymphadenopathy, pleural effusion,
parenchymal lesions and cavitations, thickening of interlobular septa, pneumothorax,
pericardial effusion, or other evidence of active or healed parenchymal TB.
Subtle miliary lesions are best appreciated in slightly underpenetrated films, but in many
cases visualization requires a high index of suspicion and review with an experienced chest
radiologist.
CT Scanning
CT scanning, especially with HRCT is more sensitive for miliary TB than plain chest
radiography. Numerous small (1–3 mm) nodules, distributed throughout both lungs, are easily
visualized. However, while sensitive, these findings are not necessarily specific. In series
correlating clinical and pathologic findings with HRCT, disseminated nodules were also found
in many other infections (Haemophilus influenzae,Mycoplasma pneumoniae,Candida albicans) and
noninfectious diseases (sarcoidosis, metastatic adenocarcinoma, lymphoma, amyloidosis,
hypersensitivity pneumonitis, and pneumoconiosis) (49–51). CT-guided needle biopsies may
help elucidate the diagnosis, but no data on the sensitivity of CT-guided invasive techniques
are available.
Microbiology
Smear and Culture
Smear and culture examination of expectorated or induced sputum, gastric lavage, pleural,
peritoneal, or pericardial fluid, cerebrospinal fluid, urine, pus, bronchoscopic secretions, peripheral
blood, bone marrow, liver biopsies, lymph node material, and transbronchial lung biopsy
specimens have all been used to confirm the diagnosis of miliary TB, but with varying results.
A recent review, however, came to the conclusion that “in the published reports, no
systematic pattern of diagnostic approach could be identified and invasive diagnostic
sampling appeared to be arbitrary and individualized, especially in the pediatric series” (3).
While it is indeed difficult to generate evidence-based recommendations for testing, recent
studies have helped establish several important testing paradigms (24,33).
Smears for acid-fast bacilli are generally not sensitive enough to rule out miliary TB as
samples at any site were only positive in a minority of patients (Table 2). However, the
probability of a positive smear increased with the number of sites sampled. Thus, when
present, samples of sputum, gastric aspirate, urine, pleural fluid, pericardial fluid, and ascites
should all be rapidly examined for the presence of acid-fast bacilli. Fluorochrome dye–based
stains may be more sensitive than conventional Ziehl–Nielsen staining (52). It should be noted
that neither of these traditional stains allows for distinction between tuberculous and
nontuberculous mycobacteria, but direct probes have been developed that allow for species
detection in smear-positive samples (53).
Miliary Tuberculosis in Critical Care 425