There were 1179 papers identified through the search of databases with an additional 70 papers identified from other sources; after checking the abstract of the articles, 285 papers (21% of the original search) were obtained for full assessment. The reasons for exclusion are summarised in Fig. 1. There was heterogeneity across the studies in terms of population assessed, study locations, identification of pertussis cases and study designs, which precluded meaningful statistical pooling/meta-analysis of the available data (Table 1). The available data are synthesized into a narrative summary focusing on general trends within selected groupings.
Overall, 34 papers met the inclusion criteria. These papers are summarised in three sections: those that reported associations between pertussis and subsequent chronic health conditions/illnesses, those which reported associations between underlying chronic conditions/illnesses and subsequent risk of pertussis, and those that reported on the effect of pertussis on the complications or severity of underlying chronic health conditions/illnesses. One of the methods used for the diagnosis of pertussis is to measure immunoglobulin M (IgM), immunoglobulin A (IgA) and immunoglobulin G (IgG) antibodies against B. pertussis by enzyme-linked immunosorbent assays (ELISA) , which are faster and more accurate than traditional bacterial agglutination tests  and immunofluorescence tests . Although both culture and serology tests should be conducted for optimal diagnosis [32, 33], these have largely been superseded by polymerase chain reaction (PCR) assay [34, 35]. Nonetheless, culture and PCR lack sensitivity among those tested late after presenting with symptoms [36,37,38,39]. There is also wide variation in laboratory diagnosis of pertussis both within countries and between countries [40,41,42]. Since there is a lack of standardised testing for the disease we have included studies with both laboratory-confirmed and suspected/clinically diagnosed pertussis. A brief summary of the key results from each of the relevant studies is provided below, categorised where possible by comorbid condition and we have also distinguished between studies where pertussis was confirmed by laboratory testing or suspected/clinically diagnosed (e.g. diagnosed by a healthcare professional, and those studies which relied on patient self-recall of infection).
Associations Between Previous Pertussis and Subsequent Development of Chronic Health Conditions/Illnesses
There were four studies in infants and young children that suggested previous pertussis could lead to an increased risk of developing asthma or other respiratory conditions [43,44,45,46].
The only study relying on confirmed diagnosis of pertussis was conducted in the Netherlands. It reported that those born between 1 July 2005 and 1 February 2008 with a history of laboratory-confirmed pertussis in infancy (within the first 6 months) (n = 89) had a relative risk of 2.8 (95% confidence intervals [CI], 1.1–7.0) for “asthma symptoms” at toddler age (13–45 months) . The small sample size in this study might, in part, explain the wide confidence intervals associated with the estimate of effect.
Four additional studies investigated this association relying on survey-based parental confirmation of prior pertussis, and found comparable results. A study conducted between December 2000 and April 2001 among children (n = 273) aged 6–12 years in Palestine with wheezing (including asthmatics) in the previous 12 months reported that pertussis (which was reported by the parents as having occurred any time prior to the study) was an important determinant of asthma (odds ratio [OR] 4.13; 95% CI 1.76–9.67) . Another study conducted in the Middle East found that children aged 5–14 years (n = 209) with a history of pertussis (by self-report at any stage prior to the study) had higher odds of probable asthma (OR 3.39; 95% CI 2.56–4.50), allergic rhinitis (OR 1.23; 95% CI 1.06–1.42), atopic eczema (OR 1.28; 95% CI 1.06–0.55) or any of these diseases combined (OR 3.22; 95% CI 2.41–4.32) . A history of pertussis was also identified as an independent risk factor for asthma in Croatia (OR 8.14; 95% CI 4.14–16, P < 0.001) in a sample involving school children (n = 411) aged 6–18 years compared to non-asthmatic controls (n = 403) .
In contrast, one case–control study conducted in New Zealand in 1994 which included 399 children with asthma aged 7–9 years and 398 matched controls, reported no statistically significant association between pertussis infection and increased risk of asthma (OR 1.57; 95% CI 0.58–4.24) .
There were two studies examining potential associations between pertussis and atopic conditions, both of which relied upon self-report of pertussis [48, 49].
A study in the UK assessing childhood infections and atopic diseases in children (n = 2111) aged 10–14 years in 1964 found that there was an increased risk of eczema in patients who reported a history of pertussis after age 3 years (OR 2.0; 95% CI 1.2–3.5) . Conversely, another study undertaken in Italy in 1987 (children [n = 2226] aged 7–11 years) observed that a self-reported history of pertussis was not associated with increased risk of atopy . These contradictory findings, all based on self-reporting of pertussis, give weak evidence towards such an association.
All studies which analysed the association between pertussis and cancer relied on self-reported prior experience of pertussis. As a result, and considering the contradictory findings explained below, the level of evidence for any such association is weak.
There were two studies based on a longitudinal cohort of individuals (n = 1142) born in Newcastle, UK in 1947. The authors reported that childhood pertussis was associated with higher cancer mortality during age 15–60 years (adjusted hazard ratio = 4.88, 95% CI 2.29–10.38) . In an abstract (based on the same study), the authors reported that either tuberculosis or pertussis was independently predictive of mortality between ages 18 and 60 years [adjusted hazard ratio (aHR) = 2.00 (95% CI 1.17–3.41) and 1.95 (95% CI 1.21–3.14), respectively] . However, the effect of pertussis on mortality was largely attributable to a higher risk of death from cancer. It was hypothesised that possibly pertussis and cancer may share some common risk factors that were not examined in the study (e.g. parental smoking or lower fruit and vegetable consumption during childhood). Other alternatives considered were poor immune function in the individuals affected, thereby increasing their susceptibility to pertussis and cancer, or that pertussis toxin itself might cause a relative increase in cell proliferation.
In contrast, three case–control studies found an inverse association between self-reported previous pertussis and a risk of developing some forms of cancers [52,53,54]. Becker et al., in a large multinational study (n = 12,585 cases and 15,416 controls aged 16–96 years) found that self-reported prior pertussis disease was associated with a 15% reduced risk in the subsequent development of non-Hodgkin lymphoma (NHL) . In an Italian study of 649 leukaemia cases between 1990 and 1999 (vs. 1771 controls) , Parodi et al. found that previous childhood pertussis was associated with lower risk of chronic lymphoid leukaemia (CLL) (OR 0.66, 95% CI 0.45–0.98) and acute myeloid leukaemia (AML) (OR 0.52, 95% CI 0.32–0.87). A later publication by Parodi et al. of 1193 NHL cases, diagnosed between 1990 and 1993, found that previous childhood pertussis was associated with lower risk (OR 0.74, 95% CI 0.62–0.88) of B cell lymphomas (1102 cases assessed vs. 1708 controls) .
Two case-controlled studies found no association between pertussis in childhood (either prior to age 14  or age 21  years) and solid tumour-related cancers [55, 56]. Albonico et al. showed no association between previous pertussis and the risk of developing carcinoma (malignant solid epithelial tumour [n = 410] vs. controls [n = 379]) in Switzerland between June 1993 and January 1994 . Stagnaro et al. showed no association between previous pertussis and the risk of multiple myeloma (213 cases vs. 1128 controls) in Italy between 1990 and 1993 . In addition, it should be noted that in both the Parodi papers described earlier [53, 54], there were no associations found between pertussis and some haematological cancers including chronic myeloid leukaemia and T cell lymphomas.
A study conducted in Switzerland compared survey data of those with multiple sclerosis (n = 606) in 1995 with data obtained from the Swiss Federal Health Office and from the Institute of Medical Statistics. It identified that a higher proportion of participants (86.0% vs. 56.7%) reported prior experience of pertussis below age 10 years in those who subsequently developed multiple sclerosis . However, another study reported no association (aOR 0.9, 95% CI 0.7–1.1) between self-reported previous childhood pertussis and diagnosis of multiple sclerosis (at age 15–56 years) in Danish cohorts born since 1940 (455 cases vs. 1801 controls) or since 1950 (182 cases vs. 690 controls) and 1975 . These contradictory findings, and the more robust design of the Danish study, argue in favour of a lack of association.
A Danish population-based study assessed the association between hospital-diagnosed pertussis and the long-term risk of epilepsy in a cohort born between 1978 and 2011. There were 4700 patients with pertussis diagnosis identified from the Danish National Patient Registry, of whom 90 developed epilepsy during the up to 15-year follow-up (versus 511 in the 47,000 members in the comparison cohort). Whilst the risk of epilepsy was increased in children with hospital-diagnosed pertussis compared with the general population (hazard ratio 1.7; 95% CI 1.3–2.1 at age 10 years) , the absolute risk was low [incidence rate 1.56 per 1000 person-years (95% CI 1.55–1.57)]. Interestingly, although the hazard ratio was not statistically different between the 1978–1997 and the 1998–2011 birth cohorts, the absolute risk of epilepsy was reduced among participants with previous pertussis diagnosis who were born from 1998 onwards and thus vaccinated with an acellular pertussis vaccine, compared to 1997 and before .
A large longitudinal study of two British birth cohorts (the 1970 British Cohort Study [born 5–11 April 1970] and the National Child Development Study [born 3–9 March 1958]) followed from birth to ages 30 and 42 years, respectively, assessed the association of previous childhood pertussis (ascertained during periodic home interviews) and type 1 diabetes mellitus (DM) . There were 100 type 1 DM cases among the 16,820 individuals assessed. Self-reported previous pertussis was associated with increased risk of type 1 DM onset after age 10 years (aOR 2.59; 95% CI 1.56–4.30). Of note, infant vaccination against pertussis appeared to decrease this risk (aOR 0.63; 95% CI 0.42–0.94).
One study assessed the association between pertussis before age 17 years and the risk of myopia among UK Biobank participants of white ethnicity aged 40–69 years who underwent autorefraction (n = 91,592) during 2006–2010. After adjusting for potential confounding variables (age, sex, birth order and educational qualifications) a self-reported history of pertussis was associated with increased risk of myopia (OR 1.39; 95% CI 1.03–1.87) .
A study in Germany involving 4181 adults aged 18–65 years in 1998 found that self-reported previous (during lifetime) pertussis (OR 1.69; 95% CI 1.36–2.09) was associated with increased prevalence of any anxiety disorder and that those subjects with both infectious diseases and anxiety disorders reported lower levels of both mental and physical quality of life, compared with subjects with only one or neither condition .
Inflammatory Bowel Disease
No association between self-reported childhood pertussis and development of inflammatory bowel disease (Crohn's disease or ulcerative colitis) at age 10–95 years was observed in a Danish case–control study (267 cases and 267 matched control) undertaken for the period 2003–2004 (combined OR, 0.56; 95% CI 0.27–1.20) .
Associations Between Underlying Chronic Conditions/Illnesses and Subsequent Risk of Pertussis
A study in Australia (the 45 and Up Study), which included 263,094 adults aged over 45 years recruited between January 2006 and December 2008 , reported that after adjustment for age, sex and other factors, adults with pre-existing asthma were more likely to be diagnosed with pertussis (based on notification record) compared to those without asthma (RR 1.64; 95% CI 1.06–2.55). Capili et al.  in a US-based study (223 PCR confirmed pertussis cases in 2004–2005 matched against 5537 negative controls) also reported that asthma was associated with an increased risk of pertussis (OR 1.73; 95% CI 1.12–2.67; P = 0.013), representing a 17% population attributable risk percentage of asthma for the risk of pertussis. In a Canadian study conducted by De Serres et al. during an outbreak in 1998, there was increased prevalence of asthma in patients at least 12 years old (n = 664) with clinical pertussis (30% of cases laboratory confirmed), compared to the general population, but the strength of the association was not assessed. Nonetheless, while patients with asthma did not assess their pertussis illness as more severe than patients without asthma, pertussis in patients with asthma resulted in increased duration of cough (P = 0.004) and increased number of disturbed night sleep (P = 0.03) .
A review of pertussis cases (n = 15,942) with cough onset in 2011–2015 notified in the seven US Emerging Infections Program Network states found at least one underlying medical condition in 32.6% (168/515) of hospitalised cases. This increased to 87.2% (102/117) among hospitalised adults aged 21 years or older. Among adolescents and adults aged 12–20 years, 43.5% (10/23) had a history of asthma. More than a quarter of hospitalised adults 21–64 years old and elderly adults aged 65 years old and above with pertussis had a history of asthma (26.3% and 26.8%, respectively) .
A retrospective analysis of the US administrative claims databases covering the 2006–2014 period recorded a relative risk of diagnosed pertussis of 3.96 (95% CI 3.81–4.12) in persons with pre-existing asthma (incidence rate 0.274 per 1000 person-years) compared to persons without asthma (incidence rate 0.069 per 1000 person-years) . In addition, a larger proportion (2.02% vs. 1.44%) of patients with pertussis and pre-existing asthma than patients with pertussis but no asthma had at least one pertussis-related hospitalisation in the 45 days, 3 months and 6 months period following pertussis diagnosis. Patients with pertussis and pre-existing asthma accrued greater all-cause adjusted costs compared with matched patients ($1193 more in a 45-day period, $1301 more in a 3-month period and $1639 more in a 6-month period; all P < 0.0001).
Chronic Obstructive Pulmonary Disease (COPD)
The review of 2011–2015 pertussis cases (n = 15,942) notified in the seven US Emerging Infections Program Network states described earlier in more detail also reported that 14.5% (11/76) of adults 21–64 years old and 26.8% (11/41) of elderly adults (aged 65 years or older) had a history of COPD .
The retrospective analysis of US administrative claims databases (described in more detail earlier) found that the incidence of diagnosed pertussis was higher among patients with COPD (0.176 per 1000 person-years than among matched patients (0.069 per 1000 person-years), with a relative risk of 2.53 (95% CI 2.40–2.68) . In addition, a larger proportion (4.08% vs. 2.33%) of patients with pertussis and pre-existing COPD than patients with pertussis but no COPD had at least one pertussis-related hospitalisation in the 45 days, 3 and 6 months period following pertussis diagnosis. Patients with pertussis and pre-existing COPD accrued greater all-cause adjusted costs compared with matched patients ($3694 more in a 45-day period, $4173 more in a 3-month period and $6154 more in a 6-month period; all P < 0.0001).
A case–control study in Iranian adults (mean age 67.7 years) found 92.2% of 90 patients with COPD vs. 51.1% of 90 matched controls without COPD (P < 0.001) had serological indication (anti-PT IgG) of recent B. pertussis infection .
Asthma or COPD
A chart review survey of hospitalised laboratory-confirmed and clinically diagnosed pertussis cases (n = 464) notified in 2010–2014 in the seven US Emerging Infections Program Network states identified 165 (35.6%) cases with at least one underlying condition, with higher rates reported among adults aged 21 years or older (89.2%), including a substantial proportion (31.4%) with a history of asthma and/or COPD . The authors concluded that the high rates of asthma or COPD suggest a potential risk factor for severe pertussis requiring hospitalisation, although the study did not test the hypothesis statistically.
A population-based case–control study of pertussis cases (n = 142) identified in 2004 and 2005 in the USA found similar rates of atopic conditions (eczema, hay fever, atopic dermatitis and allergic rhinitis) among cases (37.4%; 53/142) and corresponding controls (39.4%; 112/284) . The OR for atopic conditions predicting risk of pertussis was 0.91 (95% CI 0.58–1.40), suggesting that these conditions did not increase susceptibility to pertussis.
The previously described review of 2011–2015 pertussis cases (from seven US Emerging Infections Program Network states) found that among those hospitalised with pertussis 52.6% (40/76) of adults aged 21–64 years and 39.0% (16/41) aged 65 years or older were obese . The Australian 45 and Up Study described earlier also examined the association between body mass index (BMI) and pertussis. Adults with a high BMI (RR 1.52 [95% CI 1.06–2.19]; BMI > 30 kg/m2 vs. BMI < 25 kg/m2) were more likely to be notified as a pertussis case . It should be noted that a subsequent analysis of data from the same cohort  did not observe that the risk of pertussis hospitalisation was statistically significantly higher for patients with obesity but the point estimate suggested that the risk ‘may well be increased’.
The review of 2011–2015 pertussis cases (from seven US Emerging Infections Program Network states) described earlier also found other underlying conditions were common in adults aged 21–64 years (n = 76) and 65 years or older (n = 41) including having a potentially immunocompromising condition (25% and 9.8%, respectively), being diabetic (19.7% and 31.7%) having some form of cardiac condition (9.2% and 29.3%), renal disease (10.5% and 9.8%) or a neurological condition (5.3% and 9.8%) .
A study that investigated the bacterial aetiology of community-acquired pneumonia (CAP) in hospitalised adults in Norway from May 2007 through December 2008 found that less than 1% (2/324) were positive for pertussis . No relationship was found between CAP and pertussis infection. It was concluded that routine testing for pertussis would have no place in these groups of patients. Another study, which assessed the clinical and epidemiologic characteristics of pertussis among children aged 1–59 months hospitalised with severe pneumonia in seven African and Asian countries between August 2011 and January 2014, identified pertussis by PCR in 1.3% (53/4200) cases and 0.2% (11/5196) controls . Among the African cohort, pertussis was identified in 3.7% of 137 in-hospital deaths among those aged 1–5 months; the pertussis-infected pneumonia case fatality ratio was 12.5% (95% CI 4.2–26.8%; 5/40). No pertussis-positive cases aged 1–5 months were identified at the Asian sites involved. In comparison, 23% (3/13) died among the pertussis-positive cases aged 6–59 months.
Three studies investigated the impact of smoking on the risk and burden of pertussis. The review of 2011–2015 pertussis cases (from seven US Emerging Infections Program Network states) described earlier identified that 31.6% of adults 21 years of age and older hospitalised with pertussis were current or past smokers . An analysis of the Australian 45 and Up Study also found that smoking was a risk factor for pertussis-related hospitalisations (aOR 2.37 [95% CI 1.11–5.06] ever versus never) . These findings were reinforced by the results of the retrospective analysis of notified pertussis cases among adolescent and adults 12 years of age and older in the 1998 outbreak in Québec, Canada. De Serres et al. found that smokers had increased duration of paroxysmal cough (P = 0.004), increased prevalence of sinusitis in pertussis (P = 0.008) and somewhat increased number of nights disturbed by coughing (P = 0.4) .
Effect of Age on Pertussis Burden
The Australian 45 and Up Study reported that the pertussis incidence in the cohort was 94 per 100,000 (95% CI 82–108), which did not differ by age group, although hospitalisations increased with older age (2.2, 8.5 and 13.5 per 100,000 person-years in age groups 45–64, 65–74 and 75 years or older, respectively). However, it was acknowledged that passive notification and variations in laboratory testing may have contributed to underestimation of true rates in the cohort . An additional analysis  based upon the same study sample reinforced the conclusion that age was a significant factor in pertussis-related hospitalisations (aOR 5.4 [95% CI 1.6–18.2] and 8.9 [95% CI 2.3–34.7] in those aged 65–74 years and 75 years or older, respectively). The influence of age was also mentioned as a factor that appeared to increase the burden of pertussis in the analysis of cases notified during the 1998 outbreak in Québec, Canada (although in this study only 30% of cases were confirmed pertussis) . The study by Mbayei et al., which looked at pertussis-related hospitalisations (n = 515) in the USA between 2011 and 2015, also found that patients aged 65 years or older were at increased risk of both hospitalisations and intensive care admissions (relative risk [RR], 4; 95% CI 3–5.4 and RR, 1.8; 95% CI 0.67–4.9) respectively, compared to other age groups .
Impact of Pertussis on Progression of Chronic Conditions/Illnesses
In the review of 2011–2015 pertussis cases (from seven US Emerging Infections Program Network states) described earlier, 13.2% of 21–64-year-olds and 24.4% of those aged 65 years or older were hospitalised with admission diagnosis of asthma or COPD exacerbation . Harju et al.  assessed respiratory infections, including pertussis, as triggers of asthma exacerbations in Finland in 1999. Among the 103 asthmatics and 30 control subjects assessed there were 30 PCR-positive pertussis cases: five healthy controls (16.7%), 15 mild asthmatics (28.3%) and 10 moderate asthmatics (20%). Pertussis-positive individuals had lower forced expiratory volume/forced vital capacity (77.1% vs. 80.7%, P = 0.012) and more asthma symptoms than pertussis-negative cases (P = 0.053) . In addition, in the retrospective analysis of cases in the 1998 pertussis outbreak in Québec, Canada, 93% of pertussis cases who were previously using bronchodilator aerosol had to increase their medication because of pertussis . Finally, the retrospective analysis of US administrative claims databases (described in more detail above) found that the proportion of patients with at least one all-cause hospitalisation in the 45 days, 3 and 6 months following pertussis diagnosis was higher in the asthma cohort (6.92%, 5.47% and 4.51%, respectively) than in the no-asthma cohort (2.60%, 2.98% and 1.92%, respectively) .
As noted earlier, both the retrospective analysis of 2011–2015 notified cases in the seven US Emerging Infections Program Network states and the review of cases notified in the 1998 outbreak in Québec, Canada identified significant increase in signs of asthma or COPD exacerbation among pertussis cases [66, 67]. In addition, as for asthma, the retrospective analysis of US administrative claims databases (described in more detail earlier) found that the proportion of patients with at least one all-cause hospitalisation in the 45 days, 3 and 6 months following pertussis diagnosis was higher in the COPD cohort (12.24%, 10.21% and 8.46%, respectively) than in the non-COPD cohort (6.41%, 5.54% and 4.95%, respectively) .
Although the case–control study undertaken in Iran among patients with COPD and controls without COPD found increased signs of pertussis infection among patients with COPD, no association was observed between B. pertussis infection and severity of COPD .
A small study of adults (n = 26) with acute exacerbations of chronic bronchitis conducted in Switzerland between 2000 and 2002  identified B. pertussis infection by serology in eight (31%) of the patients. The study showed that pertussis infection was associated with a shorter duration of cough (mean 15 vs. 41 days, P = 0.04) compared with cough of unknown aetiology. Cough of at least 21 days duration was present in three (43%) of seven patients with evidence of Bordetella infection compared to 17 (94%) of 18 controls (P = 0.012). The shorter duration of cough in those with pertussis was somewhat unexpected and it was postulated that this was probably due to pertussis reinfection, which is common in adults, leading to comparatively mild symptoms in these cases owing to the partial immunity from previous infection.