Abstract
The mortality and morbidity rates in children with lower respiratory tract infection (LRTI) remain high.
To describe the number of bacteria that is associated with leukocytes in differential diagnosis of bacterial, mycoplasma, and viral LRTI in children.
Sputum smears were Gram stained for counting single-morphology bacteria associated with leukocytes. The differential diagnostic values of bacterial number were assessed in children with LRTI.
The area under the receiver operating characteristic (ROC) curve was 0.95 for bacterial number in the differential diagnosis of bacterial infection from mycoplasma and viral infections. The area under the ROC curve was 0.62 for procalcitonin and 0.94 for bacterial number in the differential diagnosis of bacterial infection from mycoplasma infection.
The number of bacteria associated with leukocytes in sputum was valuable and rapid in differential diagnosis of bacterial infection in children with suspected bacterial, mycoplasma, and viral LRTI.
Lower respiratory tract infection (LRTI) is a common disease in children, in whom mortality and morbidity resulting from the disease remain high. The main pathogens causing LRTI in children are bacteria, viruses, and Mycoplasma pneumoniae.1 Rapid differential diagnosis of different types of pathogens causing LRTI is helpful for precise treatment, selection of narrow spectrum–sensitive antibiotics, reduction of adverse drug reactions, and improvement in therapeutic effect.
Procalcitonin (PCT) increased in bacterial and mycoplasma infection but was not increased or only slightly increased in viral infections. PCT has been widely used as a biomarker for differential diagnosis between broadly bacterial infection and viral infection.2 However, PCT cannot effectively distinguish bacterial infection from M pneumoniae infection, and some severe viral infections are still indistinguishable from bacterial infections.3–6
Sputum culture is the routine method for laboratory diagnosis of bacterial LRTI. Sputum specimens are liable to be contaminated by oropharyngeal secretions. As a result, the growth of normal oropharyngeal flora will reduce the specificity of the results, and the growth of a large number of normal flora will lower the detection rate of pathogenic bacteria, especially fastidious bacteria. At the same time, dominating pathogens may be colonized in the upper respiratory tract, such as Streptococcus pneumoniae and Haemophilus influenzae, which were often misjudged as pathogens in the culture results.7,8
Sputum Gram stain is a rapid, simple, and low-cost method to establish the presence of an infection and to identify a microbiological etiology; however, its role in the diagnosis of LRTI remains controversial. The disadvantage of sputum smear is that it requires a highly skilled microbiologist to interpret the results. The qualified sputum specimens were screened by Gram-stain microscopy for culture.9,10
The positional relationship of bacteria with leukocytes and squamous epithelial cells can indicate whether the bacteria originated from the upper respiratory tract or the lower respiratory tract, which provides direction in the interpretation of culture results.11,12 However, the relationship between the number of bacteria associated with leukocytes and the occurrence of infection still needs further study. The purpose of this study is to analyze the values of bacterial number in rapid differential diagnosis of pediatric patients with respiratory tract infection caused by bacteria, viruses, and M pneumoniae, and to compare these values with PCT testing and sputum culture results.
Materials and Methods
Specimens
In 12 months, a total of 249 qualified sputum and PCT blood specimens were collected from hospitalized pediatric patients with LRTI in Zhongshan Hospital, a teaching hospital in southern China. FIGURE 1 presents the sex and age distribution of these 249 pediatric patients. All cases of LRTI adopted in this study were diagnosed as follows: 38 viral infections based on clinical symptoms, imaging and serology; 87 bacterial infections based on clinical symptoms, imaging, and bacterial culture; 102 M pneumoniae infections by clinical symptoms, imaging, serology, and nucleic acid testing; and 22 mixed infections based on clinical symptoms, imaging, serology, bacterial culture, and nucleic acid testing results.
Sex and age distribution of patients.
Microscopic Examination
Sputum smears were dried and fixed before Gram staining. The sputum smears were Gram stained for counting the single-morphology bacteria associated with leukocytes. The number of squamous epithelial cells and leukocytes in the ×100 (low-power) field were counted and their average values calculated. According to the criteria of sputum squamous epithelial cells <10 per ×100 field and leukocytes >25 per ×100 field,10,13 specimens were judged as acceptable sputum specimens. When more than one type of bacteria was associated with leukocytes, the most abundant bacteria were counted. The criteria for bacteria quantity grade are as follows: 0, no bacteria found; 1+, less than 1 per oil immersion field (OIF); 2+, 1 to 5/OIF; 3+, 6 to 30/OIF; and 4+, >30 per OIF.
Sputum Culture
All of the acceptable sputum specimens were immediately inoculated on sheep blood agar, chocolate agar, and MacConkey agar and incubated at 35°C in 5% CO2. The culture plates were read after 18–24 hours, and the negative culture plates were read again after extended incubation for 24 hours.
PCT Assay
Blood specimens were collected for PCT detection by electrochemiluminescence from patients during the period in which they had a high fever. The blood specimens were anticoagulated with EDTA-3K and sent to the laboratory for assay in time to process them. The clinical laboratory where PCT values were examined has achieved ISO15189 accreditation.
Statistical Analysis
Statistical analyses were performed using Stata software, version 16 for Windows. t testing was used for comparing the mean value of PCT in 2 different types of infection. Logistic regression analyses were performed to evaluate the power of PCT and the number of bacteria in differential diagnosis among bacteria, viruses, and mycoplasma infection. Their diagnostic capabilities were evaluated by receiver operating characteristic (ROC) analysis. The agreement between sputum culture and bacterial number was calculated using kappa statistics.
Results
Difference of PCT Concentrations among Patients with Bacterial, Mycoplasma, and Viral Infections, and the Differential Diagnostic Values
The distribution of PCT concentration among 249 patients with LRTI is shown in FIGURE 2. The mean values and ranges are presented in TABLE 1.
Distribution of PCT, Sputum Culture, and Numbers of Bacteria in Different Groups of Patients
| Cases | Mean PCT (SD) (ng/mL) | PCT range (ng/mL) | Bacteria, No. | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Culture + | Normal flora | |||||||||||
| 0 | 1+ | 2+ | 3+ | 4+ | 0 | 1+ | 2+ | 3+ | 4+ | |||
| Viral LRTI (n = 38) | 0.06 (0.02) | 0.02–0.15 | 1 | 3 | 0 | 0 | 0 | 8 | 14 | 11 | 1 | 0 |
| Bacterial LRTI (n = 87) | 0.44 (0.75) | 0.05–5.10 | 0 | 0 | 25 | 53 | 5 | 0 | 0 | 1 | 3 | 0 |
| Mycoplasma LRTI (n = 102) | 0.25 (0.36) | 0.03–1.74 | 2 | 3 | 0 | 0 | 0 | 51 | 25 | 15 | 6 | 0 |
| Viral + bacterial LRTI (n = 3) | 0.39 (0.55) | 0.04–1.03 | 0 | 0 | 1 | 2 | 0 | 0 | 0 | 0 | 0 | 0 |
| Viral + mycoplasma LRTI (n=10) | 1.18 (2.06) | 0.06–5.15 | 0 | 0 | 0 | 0 | 0 | 6 | 4 | 0 | 0 | 0 |
| Bacterial + mycoplasma LRTI (n = 9) | 0.78 (1.73) | 0.06–5.36 | 0 | 0 | 2 | 6 | 0 | 0 | 0 | 0 | 1 | 0 |
| Cases | Mean PCT (SD) (ng/mL) | PCT range (ng/mL) | Bacteria, No. | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Culture + | Normal flora | |||||||||||
| 0 | 1+ | 2+ | 3+ | 4+ | 0 | 1+ | 2+ | 3+ | 4+ | |||
| Viral LRTI (n = 38) | 0.06 (0.02) | 0.02–0.15 | 1 | 3 | 0 | 0 | 0 | 8 | 14 | 11 | 1 | 0 |
| Bacterial LRTI (n = 87) | 0.44 (0.75) | 0.05–5.10 | 0 | 0 | 25 | 53 | 5 | 0 | 0 | 1 | 3 | 0 |
| Mycoplasma LRTI (n = 102) | 0.25 (0.36) | 0.03–1.74 | 2 | 3 | 0 | 0 | 0 | 51 | 25 | 15 | 6 | 0 |
| Viral + bacterial LRTI (n = 3) | 0.39 (0.55) | 0.04–1.03 | 0 | 0 | 1 | 2 | 0 | 0 | 0 | 0 | 0 | 0 |
| Viral + mycoplasma LRTI (n=10) | 1.18 (2.06) | 0.06–5.15 | 0 | 0 | 0 | 0 | 0 | 6 | 4 | 0 | 0 | 0 |
| Bacterial + mycoplasma LRTI (n = 9) | 0.78 (1.73) | 0.06–5.36 | 0 | 0 | 2 | 6 | 0 | 0 | 0 | 0 | 1 | 0 |
LRTI, lower respiratory tract infection; PCT, procalcitonin; +, positive.
Distribution of PCT, Sputum Culture, and Numbers of Bacteria in Different Groups of Patients
| Cases | Mean PCT (SD) (ng/mL) | PCT range (ng/mL) | Bacteria, No. | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Culture + | Normal flora | |||||||||||
| 0 | 1+ | 2+ | 3+ | 4+ | 0 | 1+ | 2+ | 3+ | 4+ | |||
| Viral LRTI (n = 38) | 0.06 (0.02) | 0.02–0.15 | 1 | 3 | 0 | 0 | 0 | 8 | 14 | 11 | 1 | 0 |
| Bacterial LRTI (n = 87) | 0.44 (0.75) | 0.05–5.10 | 0 | 0 | 25 | 53 | 5 | 0 | 0 | 1 | 3 | 0 |
| Mycoplasma LRTI (n = 102) | 0.25 (0.36) | 0.03–1.74 | 2 | 3 | 0 | 0 | 0 | 51 | 25 | 15 | 6 | 0 |
| Viral + bacterial LRTI (n = 3) | 0.39 (0.55) | 0.04–1.03 | 0 | 0 | 1 | 2 | 0 | 0 | 0 | 0 | 0 | 0 |
| Viral + mycoplasma LRTI (n=10) | 1.18 (2.06) | 0.06–5.15 | 0 | 0 | 0 | 0 | 0 | 6 | 4 | 0 | 0 | 0 |
| Bacterial + mycoplasma LRTI (n = 9) | 0.78 (1.73) | 0.06–5.36 | 0 | 0 | 2 | 6 | 0 | 0 | 0 | 0 | 1 | 0 |
| Cases | Mean PCT (SD) (ng/mL) | PCT range (ng/mL) | Bacteria, No. | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Culture + | Normal flora | |||||||||||
| 0 | 1+ | 2+ | 3+ | 4+ | 0 | 1+ | 2+ | 3+ | 4+ | |||
| Viral LRTI (n = 38) | 0.06 (0.02) | 0.02–0.15 | 1 | 3 | 0 | 0 | 0 | 8 | 14 | 11 | 1 | 0 |
| Bacterial LRTI (n = 87) | 0.44 (0.75) | 0.05–5.10 | 0 | 0 | 25 | 53 | 5 | 0 | 0 | 1 | 3 | 0 |
| Mycoplasma LRTI (n = 102) | 0.25 (0.36) | 0.03–1.74 | 2 | 3 | 0 | 0 | 0 | 51 | 25 | 15 | 6 | 0 |
| Viral + bacterial LRTI (n = 3) | 0.39 (0.55) | 0.04–1.03 | 0 | 0 | 1 | 2 | 0 | 0 | 0 | 0 | 0 | 0 |
| Viral + mycoplasma LRTI (n=10) | 1.18 (2.06) | 0.06–5.15 | 0 | 0 | 0 | 0 | 0 | 6 | 4 | 0 | 0 | 0 |
| Bacterial + mycoplasma LRTI (n = 9) | 0.78 (1.73) | 0.06–5.36 | 0 | 0 | 2 | 6 | 0 | 0 | 0 | 0 | 1 | 0 |
LRTI, lower respiratory tract infection; PCT, procalcitonin; +, positive.
Individual values of procalcitonin (PCT) in the different patient groups. LRTI, lower respiratory tract infection.
The PCT values were significantly different (t testing, P < .05) between patients with viral infection and broadly bacterial infection (including bacterial infection, mycoplasma infection, and mixed infection of bacteria + virus, bacteria + mycoplasma, and virus + mycoplasma). When PCT was taken as a biomarker for differential diagnosis of broadly bacterial infection from viral infection, the area under the ROC curve was 0.89 (FIGURE 3A). At the cutoff value of 0.07 ng/mL for PCT, the sensitivity and specificity for differential diagnosis were 85.78% and 73.68%, respectively. However, although PCT was used as a biomarker for differential diagnosis of bacterial infection from viral and mycoplasma infections, the area under the ROC curve was only 0.68 (FIGURE 3B).
![Receiver operating characteristic (ROC) curves. A, Procalcitonin (PCT) in the differential diagnosis of broadly bacterial infection from viral infections (area under the curve [AUC] = 0.89). B, PCT in the differential diagnosis of bacterial infection from viral and mycoplasma infections (AUC = 0.68). C, Bacterial number (ROC area = 0.95) and PCT (ROC area = 0.68) in the differential diagnosis of bacterial infection from viral and mycoplasma infections. D, Bacterial number (ROC area = 0.94) and PCT (ROC area = 0.62) in the differential diagnosis of bacterial infection from mycoplasma infections.](https://oup.silverchair-cdn.com/oup/backfile/Content_public/Journal/labmed/55/1/10.1093_labmed_lmad035/1/m_lmad035_fig3.jpeg?Expires=1747904073&Signature=Mv1sU41eqGvcmS4nxoFaQQjpX1AqomzRcuhEESBRtGl~BAWrJmFTS6fHB6-w3A1PAri3aidwmYYPnw0-ZFKlyoBUHYu6cZNWd7IEtr98nIOqFleKEbAF-gqnsWgpwaFMyb59tn8pSjdHIqSCvZDejIHYhMrAYxyq--zCa34SZF~2BQOeXDwG~n3jirjMeE9~VlpQhEt0j-~-8umgUFNMr3drpYSyXxuNFSRxbfmsUmgBECnYvaQ3aaMS~Ts6vBjyGKaAkLcB0b3oa2AWBwbX3laYzmb~E5tSwy5n3ruGbL3hNpioh8IC5ChQWVD6l84P3b0vrgduJ1lbstuq~leNwQ__&Key-Pair-Id=APKAIE5G5CRDK6RD3PGA)
Receiver operating characteristic (ROC) curves. A, Procalcitonin (PCT) in the differential diagnosis of broadly bacterial infection from viral infections (area under the curve [AUC] = 0.89). B, PCT in the differential diagnosis of bacterial infection from viral and mycoplasma infections (AUC = 0.68). C, Bacterial number (ROC area = 0.95) and PCT (ROC area = 0.68) in the differential diagnosis of bacterial infection from viral and mycoplasma infections. D, Bacterial number (ROC area = 0.94) and PCT (ROC area = 0.62) in the differential diagnosis of bacterial infection from mycoplasma infections.
The Distribution and Differentially Diagnostic Values of Bacterial Number
All of the sputum specimens were screened using smear microscopy, and sputum specimens were judged as acceptable specimens when squamous epithelial cells were <10 per low magnification (×100) and leukocytes >25 per ×100 field. FIGURE 4A shows the microscopic appearance of an unacceptable specimen under the ×100 field. The sputum specimen smears were Gram stained for counting the single-morphology bacteria accompanied with leukocytes (FIGURE 4B).12TABLE 1 shows the distribution of bacterial number in the sputum of patients with LRTI.
Photomicrographs of sputum Gram stains. A, Large proportion of squamous epithelial cells at low power (×100). B, Moraxella catarrhalis associated with leukocytes at high power (×1000).
The area under the ROC curve was 0.95 when the bacterial number was recorded and examined as a biomarker for differential diagnosis of bacterial infection from viral and mycoplasma infections (FIGURE 3C). The sensitivity and specificity for bacterial infection diagnosis were 5.05% and 100%, respectively, at the cutoff point of 4+ for bacterial number, 70.71% and 95.33% at the cutoff point of 3+, 100% and 78% at the cutoff point of 2+, and 100% and 45.33% at the cutoff point of 1+. Binary logistic regression analysis was used to assess the bacterial number as a risk factor associated with bacterial infection. The analysis results showed that bacterial number (OR, 116.77; P < .001) was significantly associated with bacterial infection among patients with bacterial and viral infection; and in patients with bacterial, viral, and mycoplasma infections, bacterial number (OR, 18.00; P < .001) was significantly associated with bacterial infection.
PCT and Bacterial Number in Discriminating Bacterial Infection from Mycoplasma Infection
The mean PCT of 87 patients with bacterial LRTIs and 102 patients with mycoplasma LRTIs were 0.44 ng/mL and 0.25 ng/mL, respectively, which were statistically significantly different (t test; P < .05). The logistic-regression analyses showed that PCT (OR, 1.26; P = .69) was insignificantly associated with bacterial infection in differential diagnosis between bacterial infection and mycoplasma infection; however, bacterial number (OR, 14.89; P < .001) was significantly associated with bacterial infection. The area under the ROC curve was 0.62 for PCT and 0.94 for bacterial number (FIGURE 3D). The sensitivity and specificity for differential diagnosis of bacterial infection from mycoplasma infection was 57.47% and 60.78%, respectively, at the cutoff level of 0.15 ng/mL for PCT, and 100% and 79.41%, respectively, at the cutoff level of 2+ for bacterial number.
Comparison between Sputum Culture and Bacterial Number
The sputum-culture results of 249 pediatric patients were as follows: 52 cases of S pneumoniae, 33 cases of H influenzae, 13 cases of Moraxella catarrhalis, 1 case of Staphylococcus aureus, 1 case of Klebsiella pneumoniae, 3 mixed cases of S pneumoniae and H influenzae, and 146 cases of normal flora. The agreement between sputum culture and bacterial number was calculated using kappa statistics. The kappa value was 0.67 at the cutoff level of 2+ for bacterial number.
Discussion
Viruses, bacteria, and mycoplasma are the dominant pathogens that cause LRTI infection in children.1,14 The rapid differential diagnosis of infections caused by these 3 different pathogens is of great significance for the diagnosis and treatment of LRTI in children. PCT has been a useful component of a comprehensive clinical assessment and supports treatment decisions in pediatric LRTI.4 Blood marker PCT values increase in bacterial and mycoplasma LRTI in children.15
In the findings of this study, the mean PCT of 87 cases of bacterial LRTI was 0.44 ng/mL, and the range was 0.05–5.1 ng/mL. The mean PCT of 102 mycoplasma cases was 0.25 ng/mL, ranging from 0.03 to 1.74 ng/mL. As a result, PCT cannot effectively distinguish bacterial infection from mycoplasma infection. On the other hand, the detection of pathogen is particularly important for diagnosis of patients with bacterial LRTI when their PCT concentration is less than 0.5 ng/mL.2
Although the PCT value increased, it was uncertain whether it was bacterial infection or mycoplasma infection according to PCT level. There was a statistically significant difference in mean PCT between mycoplasma and bacterial LRTI (t test; P < .05); however, the area under the ROC curve was 0.62, close to 0.5, which could not discriminate the 2 infections effectively. And the area under the ROC curve was 0.94 for bacterial number in the differential diagnosis of bacterial infection from mycoplasma infection. The results of logistic-regression analyses revealed that bacterial number (OR, 18; P < .001) was significantly associated with bacterial infection among patients with bacterial, viral, and mycoplasma infection. The area under the ROC curve was 0.95 for bacterial number in the differential diagnosis of bacterial infection from viral and mycoplasma infections. The sensitivity and specificity for bacterial infection diagnosis were 100% and 78%, respectively, at the cutoff point of 2+ for bacterial number.
Sputum-bacteria culture is a conventional method for diagnosis of bacterial LRTI.16–18 Bacteriological examination of the specimens obtained from the lower respiratory tract is challenging for medical microbiologists. Distinguishing whether the isolated pathogenic bacteria represent true causes of LRTI or are only colonized bacteria of the upper respiratory tract is often difficult.
Sputum specimens are contaminated by the oropharyngeal flora, which may include potential pathogens. Also, dominating pathogens may be colonized in the upper respiratory tract, such as S pneumoniae and H influenzae, which were often misjudged as pathogens in the culture results.7,8 Sputum bacterial culture is more time-consuming to process, generally needing 2 days to report, which is not suitable for rapid diagnosis. However, if qualified sputum specimens can be collected, it is possible to quickly predict potential pathogens by detecting bacteria associated with leukocytes in sputum.
The kappa value was 0.67 for the agreement between sputum culture and bacterial number in this study. Kappa values between 0.61 and 0.80 are considered to reflect good agreement.19 The results of our kappa analysis result showed that bacterial number is in good agreement with sputum bacteria culture. Bacterial number is more advantageous than sputum cultures for rapid diagnosis of bacterial infections; however, antimicrobial susceptibility tests are still needed for the acquired resistant strains. The deficiency of this study is that there is still a lack of criterion standard methods for the diagnosis of bacterial LRTI.
Conclusions
PCT has 2 shortcomings for the differential diagnosis of LRTI. First, viral infection is indistinguishable from bacterial infection when PCT level is low; second, bacterial infection cannot be differentially diagnosed from mycoplasma infection while the PCT value is increased.
In our cohort, the bacterial number was significantly different among patients with bacterial, mycoplasma, and viral infections. The results of ROC curve analysis showed that bacterial infection can be effectively differentially diagnosed from mycoplasma and viral infection via bacterial number. At the same time, the agreement between sputum culture and bacterial number was good. In conclusion, the number of bacteria associated with leukocytes in sputum was valuable information, which was rapidly determined in the differential diagnosis of bacterial infection in children with suspected cases of bacterial, mycoplasma, and viral LRTI.
Abbreviations
Conflicts of Interest Disclosure
The authors have nothing to disclose.
