Pertussis remains a significant cause of morbidity and mortality in young children, particularly infants <2 months of age who are too young to be vaccinated [1–3]. Interventions to prevent pertussis in this age group – including immunization of pregnant and postpartum women, close contacts of young infants, and health care workers – have been introduced in some countries [4–8]. Although maternal immunization with pertussis-containing vaccines is one of the most effective strategies, it is not routine in many developing countries. To support introduction of routine maternal pertussis immunization in low- and low-middle income countries, high-quality data on infant pertussis epidemiology are needed [9]. In Latin America, surveillance studies of infant pertussis have been reported from Mexico [10], Costa Rica [11], Panama [12], Argentina [13], and Brazil [14]; however, similar data from other countries in the region, including many in Central America, are lacking [15].

Many Latin American countries conduct pertussis surveillance using the World Health Organization (WHO) or the US Council of State and Territorial Epidemiologists (CSTE) case definitions for pertussis [16, 17] (Table 1). However, during recent pertussis epidemics in the United States (US) and the United Kingdom, observational studies found that the clinical presentation of laboratory-confirmed pertussis in young infants was often atypical when viewed in the context of these published case definitions [18–20]. Given these observations, a significant proportion of infant pertussis may be unrecognized and the burden underestimated in low and low-middle income countries.

To begin to address this issue, we used data from a surveillance study of acute respiratory illness (ARI) hospitalizations in Guatemala [21], where maternal pertussis vaccination is not routine, and identified pertussis among hospitalized infants aged <2 months by testing nasopharyngeal and/or oropharyngeal specimens using a real-time polymerase chain reaction (rt-PCR) assay for Bordetella pertussis. Using data from parental interview and medical charts, we characterized the epidemiologic and clinical characteristics of pertussis in this vulnerable population.

During the surveillance period, there were 355 infants aged <2 months who were hospitalized with ARI in Santa Rosa and Quetzaltenango. Of these, 323 infants (91%) were enrolled into the overall surveillance study, and a NP/OP swab specimen was obtained from 319 (90%). A total of 301 aliquots (94% of the NP/OP specimens collected) were available for rt-PCR testing for B. pertussis, and only data from these subjects were used in subsequent analyses. Of these 301 subjects, 186 (62%) were enrolled prior to February 2011 based on the ARI case definition (although 160 of these also met the WHO IMCI clinical criteria). After February 2011, 22 subjects (7%) were enrolled based on the ARI definition alone, 23 (8%) were enrolled based on the WHO IMCI criteria alone, and 70 (23%) met both the ARI and WHO IMCI case definitions for enrollment. Out of the 301 infants with a NP/OP swab specimen available for testing, 11 (3.7%) tested positive for B. pertussis (hereafter referred to as the pertussis-positive infants) and one (0.3%) tested positive for B. parapertussis.

This is the first report of the epidemiologic and clinical characteristics of laboratory-confirmed infant pertussis from Guatemala. Using rt-PCR we identified 11 infants with laboratory-confirmed pertussis among a group of 301 infants <2 months of age hospitalized with ARI in two departments of Guatemala. A high proportion of these pertussis-positive infants had severe disease with complications, including seven (64%) that required ICU admission and two (18%) deaths. These rates of complications are similar to those reported for hospitalized infants with pertussis from a variety of geographic regions [11, 19, 20, 25–28], including several countries in Latin America, such as Mexico [10], Costa Rica [11], Panama [12], Argentina [13], and Brazil [14]. Notably, these same countries have experienced a resurgence of pertussis in recent years, with a number of infant hospitalizations and deaths, and have since introduced routine maternal pertussis immunization into their national vaccination programs [17].

Based on our clinical data, only one of the pertussis-positive infants met the clinical criteria specified by the WHO case definition for pertussis, which requires a cough duration of ≥14 days at the time of presentation [29]. However, two other pertussis-positive infants who did not meet the WHO clinical criteria were diagnosed with pertussis by a physician. In our study, the majority (80%) of pertussis-positive infants actually had a cough duration of ≤7 days at the time of hospitalization. This finding is similar to what has been reported for infant pertussis by investigators in other countries [12, 18–20, 27, 28, 30]. Infants with pertussis may also present with other atypical manifestations (i.e. not captured by the WHO case definition for pertussis) such as apnea and seizures [31], with the latter being associated with death in one recent series [32]. We did not have data on apnea, but 18% of the pertussis-positive infants in our series had seizures. Together, these data suggest that pertussis may be an important under-recognized cause of severe respiratory disease in young infants, and should not be excluded from the differential diagnosis based on the absence of classic clinical features. Furthermore, infant pertussis is likely to be substantially underreported in countries that conduct surveillance using the WHO clinical criteria.

To begin to address this problem in the US, the Council of State and Territorial Epidemiologists (CSTE) updated its pertussis surveillance case definition in 2014 to better capture disease in infants <1 year of age. The updated CSTE definition no longer requires a minimum duration of cough for reporting probable pertussis cases in this age group, and also added apnea to the list of associated clinical features that would satisfy the revised case definition in young infants [33]. Since we did not have data on the “typical” associated features of pertussis (paroxysmal cough, whoop, or post-tussive emesis) for the pertussis-positive infants with cough duration of <7 days, or data on apnea for any of the infants in the study, we cannot assess the performance of the new CSTE case definition in our sample. Importantly, the CSTE definition requires laboratory confirmation with culture for infants <1 year of age to be classified as confirmed cases. Therefore, even infants hospitalized with severe pertussis whose disease is confirmed with PCR, but who did not have a cough for ≥14 days, would only be classified as probable cases according to CSTE criteria.

Another notable finding from our study was that nearly half (45%) of the pertussis-positive infants had a documented fever within the first 24 hours of hospitalization. It is possible that the fever in the pertussis-positive infants in our study is attributable to concomitant or secondary bacterial or viral infections – for example, three of the five infants with fever had radiographic infiltrates (two meeting the WHO radiological criteria for pneumonia) and two had evidence of RSV infection by PCR. We did not have additional data on the course of fever to speculate on the potential etiologies of fever in these infants. However, the contribution of pertussis infection to their clinical presentation and the course of their illness should not be overlooked. Although fever is generally thought to be an uncommon manifestation of pertussis in this age group, especially compared to common viral pathogens, there is considerable variability in the prevalence of fever that has been reported in descriptive studies of infant pertussis. For example, in two recent community-based surveillance studies of infant pertussis in Pakistan and Zambia, no infants with PCR-confirmed pertussis had fever [34, 35]. In contrast, in other analyses of hospitalized infants who are found to have laboratory-confirmed pertussis, the prevalence of fever has been much higher – in one such study in South Africa it was nearly 25% [36], and in a study of pertussis-positive infants admitted to pediatric ICUs in the United Kingdom, 44% of infants had fever [19]. Finally, in a secondary analysis of data from a randomized controlled trial of maternal influenza immunization in South Africa – for which the surveillance definition for respiratory illness required the presence of fever – nearly a quarter (24.3%) of infants with laboratory-confirmed pertussis had fever [37]. Notably, in 2011 the Global Pertussis Initiative (GPI) proposed age-stratified clinical case definitions to account for the unique features of pertussis in younger children, stipulating the presence of “cough and coryza with no or minimal fever” in children 0–3 months of age [38]. While this definition improves upon the WHO clinical criteria for pertussis (which did not account for age), systematically excluding infants with fever from confirmatory laboratory testing for B. pertussis infection may lead to underestimation of the pertussis burden in young infants. This includes the important burden of pertussis co-infection with other respiratory pathogens, and the burden of pertussis in critically ill infants. Indeed, one recent study specifically identified fever as a predictor of more severe pertussis in children [39].

There are several limitations to our study. Since our sample was limited to hospitalized infants – who were therefore by definition severe cases – we cannot comment on the clinical characteristics of non-hospitalized (i.e. less severe) pertussis in this population, or on the background incidence of pertussis in Guatemala during the surveillance period. As a result, it is difficult to assess the potential impact of various control strategies, including routine maternal pertussis immunization, in our population. Similarly, our ARI case definition – which required either fever or an abnormal white blood cell count – may also bias toward the most severe pertussis cases. Since we obtained NP/OP swab specimens from only 90% of eligible infants (or 99% of enrolled infants), and of these only 93% were available for this analysis, it is possible that this introduced selection bias since the infants that did not undergo testing may be systematically different from those that did. However, we note that this rate of testing (93% of enrolled infants) is similar to the proportion of enrolled subjects for whom laboratory data were available in previously published analyses of other respiratory pathogens in this surveillance cohort (e.g. RSV [40] and human metapneumovirus [41]). The small number of cases in our study and the lack of data on the symptoms of early (i.e. coryza) or “typical” pertussis (i.e. paroxysmal cough, whoop, and post-tussive emesis) also limited our ability to identify characteristics that might be predictive of pertussis as a cause of ARI. Also, since apnea was not part of the clinical criteria used for enrollment, and this symptom is now recognized as a common manifestation of pertussis in young infants [31], we may have missed infants with pertussis that may differ from the infants included in our analysis. Ultimately, improved characterization of severe infant pertussis in diverse settings is essential to guide efforts to decrease morbidity and mortality in this vulnerable age group.

Some of the observed differences in the hospital course and clinical outcomes between the pertussis-positive and pertussis-negative infants may be due to unmeasured differences in their pre-hospital course – for example, the infants that tested positive for pertussis may have presented later in the course of their ARI compared to pertussis-negative infants, as evidenced by the longer duration of fever prior to presentation, which could partly explain the observed differences in clinical severity (e.g. higher case fatality rate). However, other than fever duration there were no significant differences in the types of symptom at the time of presentation between pertussis-positive and pertussis-negative infants, indicating that the differences in clinical outcomes were likely attributable to pertussis itself. Similarly, it is possible that some of the pertussis-positive infants in our study had an alternative diagnosis underlying their clinical presentation and B. pertussis was identified incidentally. However, the 11 pertussis-positive infants differed from the 290 pertussis-negative infants in several important parameters, including admission WBC, admission ALC, proportion requiring ICU admission, and case fatality rate, which is consistent with other reports of infants hospitalized with pertussis mono-infection [18, 42]. A systematic difference in these parameters between the pertussis-positive and pertussis-negative infants would not be expected if the assay were simply identifying asymptomatic carriage of the organism.

In this sample of infants hospitalized with ARI in Guatemala we found that pertussis-positive infants had high rates of complications, with higher mortality compared to pertussis-negative infants. Furthermore, most of the pertussis-positive infants would not have been captured by the existing WHO clinical case definition, which suggests that the burden of pertussis in young infants is likely to be underestimated in developing countries such as Guatemala. Improvements in current case definitions would enhance pertussis surveillance and guide implementation of strategies to prevent severe disease in young infants.