Diagn Microbiol Infect Dis. 2013 May;76(1):10-5.

Molecular Detection and Quantification of Pertussis and Correlation with Clinical Outcomes in Children

*J P DeVincenzo1, *Cliff Guyton1, Harrison Rea1, Evan Elmore1, Shivam Patel2, Luke Wynn1, Lisa Harrison3, Chadi M El Saleeby4, and **Bindiya Bagga3, 5, 6, 7

1Memphis University School, Memphis, TN; 2White Station High School, Memphis TN; 3Division of Infectious Diseases and Department of Pediatrics, University of Tennessee Health Sciences Center, Memphis, TN; 4Division of Pediatric Infectious Diseases and Department of Pediatrics, Harvard Medical School, Massachusetts General Hospital for Children, Boston, MA; 5Infectious Diseases, St Jude Children’s Research Hospital, Memphis, TN; 6Le Bonheur Children’s Hospital Memphis, TN; 7The Children’s Foundation Research Institute, Memphis, TN

*Co-first authors

**Corresponding author



Pertussis is an under-recognized serious infection. Conventional cultures are insensitive and of limited utility after antibiotic exposure. We corroborated the utility of real time PCR as a diagnostic tool in pertussis and investigated its role as a prognostic tool by evaluating its benefit in the quantification of pertussis bacterial load. All pertussis-positive PCR tests (n= 104) submitted over five years were collected for retrospective study. PCR cycle threshold was compared to quantitative culture in 43. Compared to PCR, the sensitivity of culture was 41%. Our PCR assay reliably quantified bacterial load and was quantitatively reproducible. Higher bacterial load correlated with longer duration of hospitalization (p=. 0003) and multivariate logistic regression models demonstrated this association to be independent. The study confirmed PCR as a superior diagnostic tool in pertussis. PCR quantification of bacterial load at initial diagnosis predicts later clinical disease severity, suggesting a potential benefit of PCR as a prognostic tool in pertussis.

PMID: 23490009



Pertussis is an under-recognized serious infection with a wide clinical spectrum that continues to be a disease of global importance despite vaccination strategies. Diagnosis is important because proper and early treatment can control and prevent community epidemics. Culture has traditionally been regarded as the gold standard for laboratory diagnosis of infection with Bordetella pertussis, but low sensitivity can limit its diagnostic utility. The sensitivity of culture-based diagnosis of pertussis decreases with disease progression (Cimolai et al., 1996), increase in age of the patient (Cimolai et al., 1996), and prior antibiotic exposure. These factors may be overcome by using PCR as a diagnostic tool for pertussis. Overall, the diagnosis of B. pertussis infections by nucleic acid amplification-based methods has been shown to be both highly sensitive and specific (Muller et al., 1997). Nonetheless, little is known of the bacterial load of pertussis and its importance, if any, as a predictor of disease severity. In real time PCR assays, the bacterial load can be measured by cycle threshold (CT) value, which is the point at which the amplification becomes exponential. The CT value is inversely proportional to the log of the starting concentration of the target DNA. The bacterial load of pertussis as a marker of disease severity has not been well described. We undertook this study to 1) explore the utility of real time semi-quantitative PCR assay as a prognostic tool in childhood pertussis by studying the relationships between clinical disease severity and PCR-based quantification of bacterial load, and to 2) confirm its role as a reliable diagnostic tool in this disease by evaluating its performance to conventional culture.

In our retrospective study, we confirmed PCR to be superior and significantly more sensitive than the conventional diagnostic technique (culture) in detecting pertussis. Our PCR assay was reproducible (figure1b) and correlated well with clinical features of pertussis, immunization status and peripheral blood lymphocytosis (figure 2a,b). Our other and more important goal was to evaluate if the bacterial load as quantified by the PCR assay could predict disease severity. In vitro correlation of PCR quantity with culture quantity demonstrated that our semi-quantitative pertussis PCR assay reliably quantified bacterial load (figure 1a). We determined length of hospitalization measured continuously (in days) and dichotomously (< 4 days; ≥ 4 days) as a marker for disease severity. Multiple possible factors, namely age, sex, prematurity, immunization status, CT value (bacterial load), total white blood cell count, and absolute lymphocyte count were tested for their association with the outcome variable – length of hospitalization. In univariate analyses, the independent variables of younger age and lower CT values (higher bacterial loads) were both associated with greater disease severity (increased length of hospitalization) measured continuously (P = 0.0002 and 0.0003, respectively). This same association was found when length of hospitalization was used as a dichotomous variable (< 4 days; ≥ 4 days, P = 0.0007 and 0.0027, respectively). Higher bacterial load (lower CT value) measured at pertussis diagnosis predicted an increased disease severity (more prolonged duration of hospitalization; figure 3b). To determine whether bacterial load was acting independently to predict disease severity, we conducted multivariate analyses and found that greater bacterial load at the time of diagnosis (quantified by CT value) was independently and significantly associated with greater length of hospitalization, measured both continuously and dichotomously (P = 0.0090 and 0.0496, respectively). A CT decrease of 3 units (≈10-fold higher bacterial load) predicted approximately 0.5 days longer duration of hospitalization.

Correlation between quantitative bacterial DNA load in respiratory secretions and the number of organisms detected by quantitative culture has been shown in other respiratory pathogens such as pneumococcus and Moraxella catarrhalis (Greiner et al., 2003; Greiner et al., 2001). A predictive role for respiratory bacterial pathogen load and disease severity has also been demonstrated in patients with Mycoplasma pneumoniae infection (Nilsson et al., 2010). A previous study revealed lower nasopharyngeal B. pertussis DNA loads in adults compared to children (Nakamura et al., 2011). Our study looked at bacterial loads within a diverse pediatric population presenting with pertussis and systematically correlated B. pertussis bacterial load in upper respiratory secretions with a clinically meaningful measure of disease severity, namely length of hospitalization and known clinical laboratory parameter traditionally associated with clinical pertussis, namely peripheral blood lymphocytosis.

Our results suggest that higher bacterial load (lower CT value) at presentation (figure 3b), as quantified by the pertussis PCR assay, correlated with an increased length of hospitalization. Additionally, patients with higher bacterial load had higher counts of peripheral blood lymphocytes (figure 2c,d) and were younger in age (figure 3a). These data demonstrate a statistically significant association between clinical disease severity and bacterial load in patients with pertussis, an association that was maintained even after a multivariate logistic regression analysis. This association implies that the pertussis bacterial load quantification at initial diagnosis by real time PCR assay may have a clinically useful role in predicting severity of illness, or even the need for hospitalization. Consequently, serial measures by semi quantitative PCR and the rate of change of quantitative bacterial loads over the disease course may better elucidate the dynamics of this bacterial pathogen.

To conclude, our results imply that, in pertussis, where the spectrum of illness is wide and the phenotype of the disease is variable and dependent on a multitude of factors, a semi-quantitative PCR assay may serve as a reliable molecular tool in clinical decision-making with both diagnostic and prognostic utilities.



  1. Cimolai N, Trombley C, O’Neill D. Diagnosis of whooping cough: a new era with rapid molecular diagnostics. Pediatr Emerg Care 1996;12:91-3
  2. Greiner O, Day PJ, Altwegg M, Nadal D. Quantitative detection of Moraxella catarrhalis in nasopharyngeal secretions by real-time PCR. J Clin Microbiol 2003;41:1386-90
  3. Greiner O, Day PJ, Bosshard PP, Imeri F, Altwegg M, Nadal D. Quantitative detection of Streptococcus pneumoniae in nasopharyngeal secretions by real-time PCR. J Clin Microbiol 2001;39:3129-34
  4. Muller FM, Hoppe JE, Wirsing von Konig CH. Laboratory diagnosis of pertussis: state of the art in 1997. J Clin Microbiol 1997;35:2435-43
  5. Nakamura Y, Kamachi K, Toyoizumi-Ajisaka H, Otsuka N, Saito R, Tsuruoka J, Katsuta T, Nakajima N, Okada K, Kato T, Arakawa Y. Marked difference between adults and children in Bordetella pertussis DNA load in nasopharyngeal swabs. Clin Microbiol Infect 2011;17:365-70.
  6. Nilsson AC, Bjorkman P, Welinder-Olsson C, Widell A, Persson K. Clinical severity of Mycoplasma pneumoniae (MP) infection is associated with bacterial load in oropharyngeal secretions but not with MP genotype. BMC Infect Dis 2010;10:39.



Bindiya Bagga, M.D.

Assistant Professor, Department of Pediatrics,

Division of Infectious Diseases,

University of Tennessee Health Science Center

Le Bonheur Children’s Hospital Memphis,

50 N. Dunlap, Memphis, TN 38103



Fig 1-1Figure 1. In vitro correlation of PCR quantity with culture quantity and reproducibility of PCR assay. Panel (a) compares the bacterial load as measured by CT value to that measured as log10 colony forming units. Panel (b) shows the comparison between CT-1 (CT value of first PCR run) with CT-2 (CT value on the duplicate run of the same sample). The dashed curved lines indicate the 95% CI of the slopes of the regression line.


Revised Fig 2-2Figure 2. PCR, peripheral blood lymphocytosis and disease severity. Panel (a): Compared to PCR negative patients, those with PCR-positive pertussis had a higher mean percent lymphocyte count (P < 0.0001). Panel (b): Compared to PCR negative patients, those whose samples were PCR-positive had greater absolute mean lymphocyte counts (P = 0.0016). Panel (c) compares the bacterial load as measured by cycle threshold (CT) value to the peripheral blood percent lymphocyte count. Panel (d) shows the comparison between bacterial load as measured by CT value and peripheral blood absolute lymphocyte count. Higher bacterial load correlated with increased percent lymphocyte count (P = 0.0061) and increased absolute lymphocyte count (P = 0.0082).

Revised Fig 3-2Figure 3. PCR, age and disease severity. Panel (a) shows the comparison between bacterial load as measured by cycle threshold (CT) and age. Increasing age correlated with decreasing bacterial load (increasing CT value, P = 0.0013; R2=0.096). Panel (b) shows the comparison between bacterial load as measured by CT value and length of hospitalization. Higher bacterial load (lower CT value) correlated with increased length of bacterial stay (P = 0.0003).

Multiselect Ultimate Query Plugin by InoPlugs Web Design Vienna | Webdesign Wien and Juwelier SchönmannMultiselect Ultimate Query Plugin by InoPlugs Web Design Vienna | Webdesign Wien and Juwelier Schönmann