To the Editor—I read with interest the article by Dr. Schleihauf and colleaguesReference Schleihauf, Fathima and Pettipas ¹ on the number of nasopharyngeal (NP) specimens for diagnosis of respiratory virus by polymerase chain reaction (PCR). Their point is well taken and I agree that the diagnostic yield of respiratory viral PCR testing is not increased beyond 3. In intubated hospitalized adults with undiagnosed viral pneumonia, lower respiratory tract may be preferable to upper respiratory tract sampling. Recently, I had a patient who demonstrated the critical importance of specimen source in accurate diagnosis.

A 64-year-old woman presented with fever, chills, myalgias, dry cough, and shortness of breath. The patient had chronic obstructive pulmonary disease, atrial fibrillation, and recent contact with her sick grandson. She was in respiratory distress, her temperature was 38.7°C, her pulse was 127 beats/min (irregularly irregular), and her respiratory rate was 22 breaths/min. Physical examination was unremarkable except for bilateral conjunctival injection and expiratory wheezes. Laboratory studies included a white blood cell count of 11.5 K/mm³ (neutrophils=88%, lymphocytes=4%, and monocytes=7%), with a creatinine level of 1.1 mg/dL. Serum transaminases were unremarkable. Chest radiograph was clear with a questionable left lower lobe infiltrate. Procalcitonin (PCT) was 0.72 ng/mL. NP rapid influenza test was negative. NP PCR respiratory viral panel was negative for respiratory viruses. She was started on azithromycin, her respiratory status improved, but she remained febrile (temperature, 40.4°C). Her respiratory status deteriorated on hospital day (HD) 3, and she was placed on noninvasive positive pressure ventilation and transferred to the respiratory intensive care unit. Repeat NP respiratory viral panel was again negative. Repeat PCT was 2.36 ng/mL and on HD 2 ceftriaxone was added. Repeat chest radiograph showed increased interstitial markings, but no segmental/lobar infiltrates. Chest computed tomography showed diffuse lower lobe air space opacities. Ceftriaxone and azithromycin were discontinued and she was started on vancomycin, piperacillin/tazobactam, and doxycycline. Her creatinine level was now 4.55 mg/dL. She was intubated on HD 6, and bronchoscopy was performed. Bronchoalveolar lavage fluid showed 594 nucleated cells (many “smudge cells”) and abundant red blood cells. Direct fluorescent antibody for Pneumocystis pneumonia was negative. Gram stain showed few polymorphonuclear leukocytes with no organisms and cultures were negative. Respiratory viral panel PCR performed on bronchoalveolar lavage fluid was positive for adenovirus and antibiotics were discontinued. Adenovirus antibody titer was elevated at 1:256 (normal <1:8) and serum quantitative adenovirus PCR was highly positive with 288,000 copies/mL. She defervesced on HD 6, but on HD 8, she developed loose, watery stools positive for Clostridium difficile. She was successfully treated for C. difficile diarrhea with metronidazole and vancomycin (Figure 1). Although there were no segmental/lobar infiltrates on chest radiograph to suggest bacterial pneumonia, her PCT was elevated and empirical antibiotics were given. Respiratory viral PCR performed on bronchoalveolar lavage fluid was positive for adenovirus and bronchoalveolar lavage fluid cytology showed adenovirus cytopathologic effects, which are large intranuclear basophilic inclusions resulting in a smudged appearance (“smudge cells ”) pathognomic for adenovirus infection.Reference Retalis, Strange and Harley ^{2 ,} Reference Cibas and Ducatman ³

FIGURE 1

Hospital course in case study. BAL, bronchoalveolar lavage; C. difficile, Clostridium difficile; CXR, chest radiograph; IV, intravenous; PCR, polymerase chain reaction; PO, by mouth; RVP, respiratory viral panel.

The need to consider the validity of a sampling source has been reported previously. During the 2009-2010 influenza pandemic, a middle-aged immunocompetent man was hospitalized with a severe viral pneumonia thought to be due to influenza A. Multiple NP specimens were negative for influenza. He rapidly deteriorated and died of severe/prolonged hypoxemia. At autopsy, lung specimens were PCR positive for influenza A(H₁N₁).Reference Cunha, Syed, Mickail and Strollo ⁴ Although we agree that the optimal number of NP PCR specimens for viral pneumonia diagnosis is 3, there are exceptions. With viral pneumonia of undetermined etiology in intubated patients and repeatedly negative NP PCRs, lower respiratory tract testing may be more reflective of the pulmonary pathogen than an upper airway source.

This case also merits comment from a antibiotic stewardship standpoint since without a definite diagnosis there was no clinical rationale for empirical antibiotics.Reference Cunha, Varughese and Mylonakis ^{5 ,} Reference Cunha ⁶ PCT is unhelpful in bacterial pneumonia diagnosis and may be elevated with renal insufficiency, as in this case.Reference Cunha ^{7 ,} Reference Cunha, Syed and Strollo ⁸ There were no clinical findings to suggest bacterial coinfection/pneumonia. Unnecessary antibiotic costs aside, from an antibiotic stewardship perspective there is the additional cost of potential antibiotic adverse effects, which should be carefully considered when initiating empirical antibiotic therapy. In this case, C. difficile diarrhea may have been avoided.Reference Cunha, Varughese and Mylonakis ⁵

Antibiotic stewardship lessons from this case are clear. First, accurate diagnosis is needed in order to treat accurately.Reference Cunha, Varughese and Mylonakis ⁵ Second, in hospitalized adults adenovirus pneumonia, when not localized, may mimic influenza pneumonia in radiographic appearance—that is, adenovirus in this patient mimicked influenza pneumonia.Reference Retalis, Strange and Harley ² Third, when influenza or other viral pneumonias are complicated by bacterial coinfection, the clinical findings of bacterial pneumonia are readily recognizable on chest radiograph—for example, focal segmental/lobar infiltrates superimposed on bilateral patchy interstitial infiltrates of the underlying viral pneumonia.Reference Cunha ^{7 ,} Reference Cunha, Syed and Strollo ⁸ In this case, it was assumed she had influenza that may be complicated by bacterial pneumonia, and empirical antibiotic therapy was given on the basis of a potential predisposition of influenza to bacterial coinfection/pneumonia.Reference Oosterheert, vanLoon and Schuurman ^{9 ,} Reference Hernes, Hagen, Quarsten, Bjorvatn and Bakke ¹⁰ However, this case of adenoviral pneumonia had no clinical findings that suggested bacterial coinfection/pneumonia. Elevated PCT levels are often misleading and are often due to other infectious and noninfectious disorders.Reference Cunha ^{7 ,} Reference Cunha, Syed and Strollo ⁸

Last, from an antibiotic stewardship perspective, empirical antibiotics should be selectively prescribed for the shortest possible duration. Unnecessary empirical antibiotic therapy may have adverse pharmacoeconomic and clinical consequences—for example, in this case C. difficile diarrhea.Reference Cunha, Varughese and Mylonakis ⁵ In conclusion, it cannot be overstated that accurate diagnosis is essential for accurate therapy. In hospitalized adults with viral pneumonia of unknown etiology, if 3 NP PCR specimens are negative, results should be interpreted in light of the specimen source—that is, is the specimen test result reflective of the infection source?