The Epidemiology of Nationally Reported Pertussis in the United States, 2000–2016

Abstract

Background

Despite successful vaccination programs, pertussis remains endemic in the United States, and increasing incidence has been reported. We used national surveillance data to describe pertussis epidemiology, including patient demographic characteristics, geographic distribution, and temporal trends.

Methods

We included cases reported through the National Notifiable Diseases Surveillance System between 1 January 2000 and 31 December 2016. Differences in case characteristics were compared using Pearson χ2. Average annual incidence (cases per 100 000 population) was calculated overall and by age (<1 year, 1–6 years, 7–10 years, 11–18 years, 19–39 years, 30–64 years, and ≥65 years) and geographic subgroup. Annual percent change was estimated using negative bionomial regression.

Results

During 2000–2016, 339 420 pertussis cases were reported. The majority were in white (88.2%) and non-Hispanic (81.3%) persons, 9.9% were hospitalized, and 0.1% were fatal; however, differences existed by age. Infants had the highest incidence (75.3/100 000 population), accounting for 88.8% of deaths. Incidence increased significantly over time (P = .0019), increases were observed for all groups except persons aged <1 year and 19–64 years. Elevated case counts among persons aged 7–10 and 11–18 years coincided with the aging of acellular-primed cohorts. Incidence varied by geographic region, with some similarities in disease cyclicity.

Conclusions

Increasing baseline incidence and changing age distribution of pertussis suggest a central role of the transition to acellular vaccines in the US disease resurgence. Continued monitoring of national data is important to evaluate and optimize pertussis prevention and control strategies.

pertussis, immunization, Tdap, DTaP, waning immunity

Pertussis is a highly contagious, cyclic, respiratory disease. Prior to the introduction of whole-cell pertussis vaccines in the United States during the 1940s, the annual number of reported cases often exceeded 200000. The introduction of childhood pertussis vaccines dramatically altered the epidemiologic landscape of pertussis, with reported cases declining to a nadir of just over 1000 (0.47/100000 population) by the mid-1970s [1]. However, despite substantial declines in disease since the prevaccine era, pertussis remains a significant health problem worldwide, with incidence increasing in several countries in recent years.

To protect against pertussis, vaccines are administered throughout the lifespan, beginning in early childhood. A 5-dose series is recommended for children between 2 months and 6 years of age. While whole-cell vaccine formulations were originally used in the United States, concerns over safety prompted the transition to less reactogenic, acellular vaccines (DTaP; diphtheria, tetanus toxoid, and acellular pertussis) between 1992 and 1997 [2]. DTaP coverage has remained relatively stable since vaccine introduction, with 93.7% of children aged 19–35 months receiving ≥3 doses in 2016 [3]. However, vaccine-induced protection wanes over time, necessitating additional booster doses of pertussis vaccines. In 2005, a single dose of Tdap (tetanus toxoid, reduced diphtheria toxoid, and acellular pertussis) vaccine was recommended for adolescents and adults, with preferred administration at age 11–12 years [4, 5]. While adolescent Tdap coverage is high (88.0%), improvements are still needed among adults (26.6% in 2016) [6, 7]. Because infants are at greatest risk for severe pertussis-related morbidity and mortality, Tdap was recommended for pregnant women in 2011 to protect infants in the first months of life through the passive transfer of maternal antibodies; this recommendation was expanded in 2012 to include Tdap during every pregnancy [2].

We analyzed national surveillance data to describe the epidemiology of reported pertussis in the United States. We examined the distribution of disease by patient age, race, ethnicity, and geographic region, as well as temporal trends during 2000–2016, with an emphasis on epidemiologic changes in recent years.

METHODS

Pertussis cases reported through the National Notifiable Diseases Surveillance System (NNDSS) [8] between 1 January 2000 and 31 December 2016 were included in the analysis. Cases were classified by reporting states according to the Council of State and Territorial Epidemiologists (CSTE) case definition. The pertussis clinical case definition requires cough of ≥2 weeks duration with paroxysms, inspiratory whoop, or post-tussive vomiting. A confirmed case was defined as acute cough illness of any duration with isolation of Bordetella pertussis from culture or a clinical case with either positive polymerase chain reaction (PCR) or epidemiologic linkage to a laboratory-confirmed case (Supplementary Table 1). (Cases in persons aged ≥11 years from Massachusetts with a single acute serum immunoglobulin G anti-pertussis toxin antibody level of ≥20µg/mL were also classified as confirmed.). Clinical cases with no laboratory confirmation or epidemiologic linkage to a laboratory-confirmed case were classified as probable. To increase the sensitivity of the case definition for infants with severe disease, apnea was added to the clinical case definition for infants aged <1 year in 2014, and the infant probable category was expanded to include acute cough illness of any duration with ≥1 clinical symptom noted above and either PCR confirmation or epidemiologic linkage to a laboratory-confirmed case [9]. No additional changes were made to the case definition during the study period. We used state and local health department determined classifications for this analysis and included all cases classified as confirmed, probable, or with an unknown status.

Age-specific analyses were conducted using the following age groups: <1 year, 1–6 years, 7–10 years, 11–18 years, 19–29 years, 30–64 years, and ≥65 years. We defined geographic regions according to the Morbidity and Mortality Weekly Report (MMWR) classifications. (New England: Connecticut, Maine, Massachusetts, New Hampshire, Rhode Island, Vermont; Mid-Atlantic: New Jersey, New York, Pennsylvania; E.N. Central: Illinois, Indiana, Michigan, Ohio, Wisconsin; W.N. Central: Iowa, Kansas, Minnesota, Missouri, Nebraska, North Dakota, South Dakota; S. Atlantic: Delaware, District of Columbia, Florida, Georgia, Maryland, North Carolina, South Carolina, Virginia, West Virginia; E.S. Central: Alabama, Kentucky, Mississippi, Tennessee; W.S. Central: Arkansas, Louisiana, Oklahoma, Texas; Mountain: Arizona, Colorado, Idaho, Montana, Nevada, New Mexico, Utah, Wyoming; Pacific: Alaska, California, Hawaii, Oregon, Washington.) Incidence rates were calculated using case data from NNDSS as numerators and population estimates from the Centers for Disease Control and Prevention’s (CDC’s) National Center for Health Statistics as denominators (bridged-race, 2016 post-censal vintage estimates for 2000–2016) and were expressed as cases per 100000 population. Pearson χ² was used for the comparison of proportions; proportions were calculated out of known responses. Annual percent change (APC) was estimated using negative binomial regression to examine cyclic trends over time. P values of <.05 were considered statistically significant.

RESULTS

Case Characteristics

A total of 339420 pertussis cases were reported through NNDSS between 1 January 2000 and 31 December 2016. Cases were predominately in white (88.2%) and non-Hispanic (81.3%) persons and classified as CSTE confirmed (74.1%) (Table 1). Overall, 9.9% of patients were hospitalized and 0.12% died. Infants aged <1 year comprised 15.1% of reported cases; 16.8% were in persons aged 1–6 years, 15.0% in persons 7–10 years, 27.7% in persons 11–18 years, 3.3% in persons 19–29 years, 19.7% in persons 30–64 years, and 2.4% in persons ≥65 years.

Table 1.

Characteristics of Pertussis Patients Reported in the United States Between 2000 and 2016

Characteristic	N	%a	Average Annual Incidence (per 100000 Population)
Age group, years
<1	50950	15.1	75.3
1–6	56396	16.8	13.8
7–10	50594	15.0	18.3
11–18	93190	27.7	16.3
19–29	10960	3.3	1.4
30–64	66257	19.7	2.8
≥65	8177	2.4	1.2
Unknown	2896	…	…-
Sex
Male	152103	45.3	5.9
Female	183873	54.7	6.9
Unknown	3444	...	...
Race
White	235157	88.2	5.7
Black	15006	5.6	2.1
Other	16477	6.2	4.6
Unknown	72780	...	...
Ethnicity
Hispanic	48247	18.7	5.8
Non-Hispanic	209674	81.3	4.8
Unknown	81499	...	...
Hospitalization status
Hospitalized	25667	9.9	0.5
Non-hospitalized	232493	90.1	4.5
Unknown	81260	...	...
Mortality
Died	294	0.1	...
Survived	249956	99.9	...
Unknown	89170	...	...
US regionb
New England	22925	6.8	9.4
Mid-Atlantic	34553	10.2	5.0
E.N. Central	65681	19.3	8.3
W.N. Central	41082	12.1	11.9
S. Atlantic	25142	7.4	2.5
E.S. Central	9537	2.8	3.1
W.S. Central	36259	10.7	5.9
Mountain	42204	12.4	11.5
Pacific	62037	18.3	7.3
Case classification
Confirmed	251074	74.1	...
Probable	87738	25.9	...
Unknown	608	...	...

n = 339420.

^aCalculated out of those with a known response.

^bNew England: Connecticut, Maine, Massachusetts, New Hampshire, Rhode Island, Vermont; Mid-Atlantic: New Jersey, New York, Pennsylvania; E.N. Central: Illinois, Indiana, Michigan, Ohio, Wisconsin; W.N. Central: Iowa, Kansas, Minnesota, Missouri, Nebraska, North Dakota, South Dakota; S. Atlantic: Delaware, District of Columbia, Florida, Georgia, Maryland, North Carolina, South Carolina, Virginia, West Virginia; E.S. Central: Alabama, Kentucky, Mississippi, Tennessee; W.S. Central: Arkansas, Louisiana, Oklahoma, Texas; Mountain: Arizona, Colorado, Idaho, Montana, Nevada, New Mexico, Utah, Wyoming; Pacific: Alaska, California, Hawaii, Oregon, Washington.

Case classification, race/ethnicity, hospitalization status, and case fatality varied by age group. Among all age groups <19 years, the proportion of cases classified as confirmed ranged from 78.5% to 82.6%; however, the proportion of confirmed cases was lower among the older age groups (ranging from 63.1% among patients aged 19–29 years to 44.3% among patients aged ≥65 years). The proportion of cases in black and Hispanic persons was significantly higher among infants aged <1 year (14.0% and 36.4%, respectively) compared to the other age groups (P < .0001); proportions decreased steadily with age (black cases, 5.2% among patients aged 1–6 years to 2.6% among patients aged ≥65 years; Hispanic cases, 21.2% among patients aged 1–6 years to 6.7% among patients aged ≥65 years). The proportion of cases hospitalized was highest among infants aged <1 year (49.9%), followed by adults aged ≥65 years (14.2%) (Table 2). Of 294 deaths reported, 261 (88.8%) occurred among cases aged <1 year and 14 (4.8%) occurred among cases aged ≥65 years (Table 1).

Table 2.

The Percentage of Pertussis Hospitalizations and Incidence of Hospitalized Pertussis Cases in the United States, by Age Group, 2000–2016

Age Group, Years	Percent Hospitalizations			Incidence of Hospitalized Pertussis
	Total Number of Hospitalized Cases (%a)	Annual Percent Change	P Value	Average Annual Incidence of Hospitalized Pertussis (per 100000 pop.)	Annual Percent Change	P Value
<1	19562 (49.9)	–4.4%	<.0001	28.9	–2.6	.0675
1–6	1494 (3.5)	–6.0%	<.0001	0.4	2.8	.0953
7–10	440 (1.2)	–5.3%	.0003	0.2	6.7	.0008
11–18	804 (1.2)	–4.3%	.0032	0.1	0.8	.6229
19–29	174 (2.0)	2.3%	.1914	0.02	3.0	.1705
30–64	2216 (4.2)	4.7%	<.0001	0.1	7.7	.0004
≥65	889 (14.2)	2.8%	.0009	0.1	12.6	<.0001
Unknown age	88 (6.5)	...	...	...	...	...
Overall	25667 (9.9%)	–6.5%	<.0001	0.5	–1.5	.2777

^aProportion of patients that are hospitalized, by age group; calculated out of those with known hospitalization status.

Average annual pertussis incidence was highest among infants aged <1 year (75.3/100000), exceeding incidence in the other age groups by 4.1–62.8 times (Table 1). Incidence was comparable among persons aged 1–6 years, 7–10 years, and 11–18 years, ranging from 13.8/100000 among patients aged 1–6 years to 18.3/100000 among patients aged 7–10 years. Among all adult age groups, average annual incidence was ≤2.8/100000.

Trends Over Time

While overall pertussis incidence was cyclic, ranging from a low of 2.7/100000 in 2001 to a high of 15.4/100000 in 2012, incidence increased significantly over time (P = .0019) (Figure 1). Between 2000 and 2008, the average incidence of pertussis in nonpeak years (2000–2003, 2007, and 2008) was 3.4/100000. However, between 2009 and 2016, the average incidence in nonpeak years (2009, 2011, 2015, and 2016) was 5.9/100000, representing a 1.7-fold increase (Figure 1). Rates during peak years ranged from 8.7/100000 to 15.4/100000.

Figure 1.

Overall annual US incidence of reported pertussis, 2000–2016. Abbreviation: Tdap, tetanus toxoid, reduced diphtheria toxoid, and acellular pertussis.

In addition to temporal fluctuations, there was significant variability in pertussis incidence by geographic region, with the highest average annual incidence in the W.N. Central (11.9/100000) and Mountain (11.5/100000) regions (Table 1). Over time, the highest annual regional incidence alternated between the Mountain, Pacific, New England, and W.N. Central regions, ranging from 8.4 to 39.0/100000. The lowest annual rates of pertussis were consistently reported from the S. Atlantic and E.S. Central regions, never exceeding 7.0/100000 (Supplementary Figure 1c). Epidemic peak intervals varied regionally, ranging from 2 to 7 years, and distinct patterns were observed with some notable similarities in the cyclicity of disease (Supplementary Figures 1a–c). The E.S. Central and S. Atlantic regions showed increasing incidence over time, but disease cyclicity was limited (Supplementary Figure 1c).

Throughout the study period, pertussis incidence remained highest among infants aged <1 year (Figure 2). While rates of disease among the other age groups were similar during the early years, changes emerged around 2008 (Figure 2). Between 2000 and 2006, adolescents aged 11–18 years generally had the second highest annual incidence, ranging from 6.1/100000 in 2001 to 26.1/100000 in 2004. However, beginning in 2007 and continuing through 2014, the second highest annual incidence was reported among children aged 7–10 years (range, 8.5/100000 in 2007 to 58.5/100000 in 2012). In 2015 and 2016, adolescent (11–18 years) pertussis rates once again surpassed rates in children aged 7–10 years, but the absolute differences were small (from 2.3 to 3.1/100000). While annual incidence increased in all age groups during the study period, increases were only significant among persons aged 1–6 years, 7–10 years, 11–18 years, and ≥65 years (Figure 2).

Figure 2.

Annual US incidence of reported pertussis by age group, 2000–2016. *P < .05.

Overall, the proportion of patients hospitalized for pertussis decreased significantly over time, from 20.6% in 2000 to 6.7% in 2016 (P ≤.0001). Significant declines were observed in all age groups <18 years of age; the largest decline occurred among patients aged 1–6 years (APC = –6.0%, P < .0001) (Table 2). Among patients aged 30–64 and ≥65 years, the proportion hospitalized increased significantly over time (APC = 4.7% and 2.8%, respectively, P < .001 for both) (Table 2). This was accompanied by significant increases in the incidence of hospitalized pertussis in these age groups (APC = 7.7% and 12.6%, respectively). Among the other age groups, incidence of hospitalized pertussis increased significantly only among those aged 7–10 years (APC = 6.7%, P = .0008).

Multiple peaks occurred nationally during the study period. Despite differences in the magnitude of rates, notable changes were observed in the age distribution of reported cases over time. The epidemiology of disease during the 2004 and 2005 peak years (annual incidence, 8.8 and 8.7/100000, respectively) was characterized by low but sustained levels of pertussis among children aged 1–9 years, with the highest proportion of cases among adolescents aged 11–16 years (Figure 3A). Six years later in 2010, overall incidence was comparable (8.9/100000); however, the single year age distribution of cases had shifted. Increased case counts were observed among persons aged 7–10 years, which coincided with the aging of the first acellular-primed birth cohorts, with a peak in cases occurring among children aged 10 years (11.4% of pertussis patients aged 1–19 years); among adolescents, a decrease in the number of cases was observed with increasing age (Figure 3B). The largest national peak occurred in 2012, when overall incidence exceeded 15.0/100000, almost double the incidence of previous epidemic years. Case counts continued to be most elevated among children aged 7–10 years (Figure 3C). However, unlike previous epidemic years when the number of adolescent cases decreased with age, a second peak was observed among 13- and 14-year olds, who were among the leading edge of the acellular-primed birth cohorts. While the 2014 age distribution was similar, case counts were additionally elevated in 15- and 16-year olds (Figure 3D). This trend continued in 2016, with elevated reporting among adolescents through 17 years of age (Figure 3E).

Figure 3.

Annual US incidence of reported pertussis by single year of age during select years. Note: magnitude varies between figures. The bar across the bottom indicates the type of childhood vaccine (acellular vaccine [aP] vs whole-cell vaccine [wP]) a child would have received based on age; the hashed bar indicates the time period during the transition when children would have received a mixture of both vaccine types. For example, in 2004, adolescents aged ≥12 years would have received 5 doses of whole-cell vaccine. Tdap was introduced for adolescents and adults (preferred administration at 11–12 years of age) in 2005. Abbreviation: Tdap, tetanus toxoid, reduced diphtheria toxoid, and acellular pertussis.

DISCUSSION

Despite the accomplishments of the US immunization program, pertussis remains endemic, and reported cases have been increasing. This analysis of 17 years of national surveillance data highlights this gradual but steady increase in cases. Furthermore, as these data have shown, baseline pertussis incidence has increased nearly 2-fold when comparing the later to earlier part of our study period, with incidence in peak years increasing in magnitude. In 2012, overall incidence reached 15.4/100000, representing >48000 reported pertussis cases, the largest number since the 1950s [10]. Numerous factors have been recognized as credible contributors to the resurgence, including the use of more sensitive and rapid diagnostic tests (eg, PCR); heightened recognition and reporting; molecular changes in the organism, especially modifications that could enable B. pertussis to evade vaccines; and waning of vaccine-induced immunity [10, 11]. While emerging data suggest the causes are likely multifactorial, numerous studies have highlighted the key role that waning acellular vaccine immunity plays in the United States and other countries that have transitioned away from whole-cell vaccination programs [12–17].

In addition to an overall increase in reported disease, our analysis also revealed progressive changes in the age distribution of pertussis cases over time, reflecting the accumulation of susceptible individuals in key age groups. Throughout the study period, pertussis incidence remained highest among infants aged <1 year, the age group that accounted for the largest proportion of pertussis hospitalizations and deaths. This is consistent with pertussis epidemiology worldwide[11]. However, school-aged children and adolescents have made increasing contributions to the growing burden of disease in the United States. Neither vaccination nor prior pertussis infection confer life-long protection against disease, and booster doses of pertussis-containing vaccines are necessary to maintain adequate protection. As our analysis has shown, pertussis epidemics in 2004 and 2005 were driven by increases in disease among adolescents, prompting the introduction of Tdap for routine adolescent use in 2005 [4]. Targeted vaccination of adolescents appeared a promising strategy in the early years following Tdap introduction, greatly reducing the burden of adolescent disease in the United States [18]. By 2008, however, the age distribution began to shift as an increase in pertussis began to emerge among children aged 7–10 years. With DTaP coverage consistently high, this increase was unexpected and was the first indication of limitations in acellular vaccine performance, raising concerns about durability of protection. Stratification of national data by single year of age illustrates the close correlation of this trend with the aging of the first acellular-primed birth cohort from 1998. These data show that as this acellular-primed cohort continued to age, subsequent increases were also observed in adolescents, particularly 13- and 14-year olds, a group that should have been well protected by increasing Tdap coverage. Numerous studies have demonstrated a shorter duration of protection from acellular vaccination as compared to whole-cell vaccination, noting the important role of the childhood priming series on duration of protection by the Tdap booster dose [15, 17, 19]. Although pertussis is thought to be underreported among adult age groups, ongoing surveillance, with a particular emphasis on young adults, will be critical to detect age-specific increases as acellular-primed cohorts age into adulthood [20, 21].

Consistent with known pertussis epidemiology, we found the highest proportion of hospitalizations among infants aged <1 year and adults aged ≥65 years, groups most likely to suffer from severe pertussis-related symptoms and complications. Significant declines in the proportion of hospitalizations were observed among all patients aged <19 years. While these declines could represent a true decrease in disease severity or be an artifact of clinical management, a significant increase in the incidence of hospitalized pertussis cases was observed among children aged 7–10 years, reflecting the increase in overall incidence in this age group. In an era of waning immunity, a significant proportion of disease is occurring among vaccinated individuals. Because vaccination can modify the severity of illness, a decrease in disease severity that reduces the likelihood of hospitalization may not be unexpected [22, 23]. Interestingly, and contrary to the younger age groups, we found an increase in both the proportion of hospitalizations and incidence of hospitalized cases among persons aged ≥30 years. This may be a reflection of improved case ascertainment in this group, resulting from inclusion of pertussis in the differential for cough illness in adults. Further evaluation of these trends is warranted.

While the majority of cases reported during the study period were in white (88.2%) and non-Hispanic (81.2%) persons, a disproportionately higher number of patients aged <1 year were black or Hispanic compared to the other age groups. Hispanic ethnicity has been documented to increase an infant’s risk of pertussis, and a higher proportion of Hispanic patients have been observed among infants aged <1 year [24–27]. While the reasons behind this disparity are not fully understood, research suggests that Hispanic children may live in larger households and have more close contacts that could place an infant at increased risk for pertussis [24, 25, 28]. Understanding age-specific racial and ethnic disparities among pertussis cases is important for targeting prevention and control strategies appropriately, especially among populations where disparities in access to healthcare exist.

As observed in other countries that have transitioned to acellular pertussis vaccines, the epidemiology of pertussis in the United States has been characterized by significant epidemic peaks in recent years. While overall trends show peaks nationally every 2–5 years, our analysis of MMWR regions suggests that some US regions cycle differently than the nation overall and that similarities in cyclicity exist among regions that are not necessarily contiguous. Also, we have shown that the S. Atlantic and E.S. Central regions of the United States contributed approximately 10% of total cases and consistently reported lower annual incidence rates during the study period, with very limited cyclicity, despite accounting for 25% of the US population. Whether this accurately describes the real disease burden in this region or is a reflection of differences in surveillance infrastructure is not well understood. Further characterization of geographic trends, including stratification by age and smaller geographic areas, will be important to increase our understanding of disease transmission patterns in the United States and potentially identify susceptible populations as a way of predicting future outbreaks.

In the setting of a pertussis resurgence, national surveillance data provided an important early clue pointing to problems with duration of acellular vaccine immunity in the United States. Despite known limitations of nationally reported data, such as the incompleteness of key demographic and clinical variables and the potential underreporting of adult disease, NNDSS remains the most nationally representative source of data for monitoring US pertussis trends [29]. While we believe the overall decrease in pertussis incidence following the 2012 peak is the result of cyclic variation, the epidemiology of pertussis is clearly evolving. Maintaining high-quality and timely surveillance will help us to identify any new developments in disease trends and monitor the impact of new pertussis prevention and control strategies, particularly of maternal vaccination during pregnancy.

Supplementary Data

Supplementary materials are available at Clinical Infectious Diseases online. Consisting of data provided by the authors to benefit the reader, the posted materials are not copyedited and are the sole responsibility of the authors, so questions or comments should be addressed to the corresponding author.

Notes

Acknowledgments. We thank Amanda Faulkner, Christine Miner, Anna Acosta, and Nong Shang.

Disclaimer. The findings and conclusions in this report are those of the author(s) and do not necessarily represent the official position of the Centers for Disease Control and Prevention.

Financial support. This work was supported by the Centers for Disease Control and Prevention.

Potential conflicts of interest. All authors: No reported conflicts. All authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Conflicts that the editors consider relevant to the content of the manuscript have been disclosed.