To the authors’ knowledge, this study is the first to describe a large sample of adults 50 years and older with pertussis. Consistent with prior findings [22–28], our results suggest that under-reporting of pertussis cases occurs despite nationwide mandates to notify health authorities when cases are identified. We estimated the national incidence of diagnosed pertussis to be between 2.1 and 4.6 cases per 100,000 people across the two age groups (50–64 and ≥65) during the years 2006 to 2010. By contrast, the ratios of IMS to NNDSS incidence rates were greater than one in each of the five years analyzed supporting the hypothesis that pertussis infections in older adults occur more frequently than previously reported. Although larger than NNDSS estimates, we suspect our estimated incidence of diagnosed pertussis is still an understatement of the true rate of diagnosed pertussis cases as patients who tested positive for pertussis but did not have a private practitioner claim with an ICD-9-CM code for pertussis and patients that were only observed in the hospital setting were not accounted for in the incidence calculation. While infants and young children bear a well-documented burden caused by Bordetella pertussis, this disease’s direct financial and health burden is not exclusive to childhood as once previously thought. The increasing incidence of pertussis among the older adults in this study is consistent with studies in adolescents and younger adults, among whom the incidence of pertussis also appears to be increasing [11, 14–18].
Our observation that females with pertussis represented a significantly higher proportion of the total sample compared to males is consistent with results in younger patient populations [22, 23] and behavioral (i.e., treatment-seeking) literature [29, 30]. The age effect on the initial diagnosis of pneumonia in whooping cough suggests that susceptibility to pneumonia and/or a pneumonia diagnosis may increase as age increases. Regarding the finding of significantly longer episodes of care among patients whose initial pertussis confirmations were made through positive laboratory tests, we interpret that this is related to severity. That is, a sicker patient may be more likely to undergo laboratory testing.
The results of this study also suggest that pertussis is not always diagnosed on the first occasion that the patient presents for care. This finding has been previously reported [e.g., [27, 28]. Cough and acute bronchitis were the most common conditions diagnosed prior to the pertussis diagnosis. Over half (57%) of the sample had one or more visits with at least one pertussis-like diagnosis in the period immediately preceding the index pertussis diagnosis. These initial diagnoses may have been made while practitioners awaited laboratory confirmation of etiology or they may have been misdiagnoses. If the latter is true, misdiagnosed patients that remain untreated represent a potential source of transmission to others. Under diagnosis of pertussis in older patients may occur because healthcare providers continue to view pertussis as a childhood disease . These results underscore the need to heighten healthcare provider awareness of the occurrence of pertussis in adults, including older adults.
The average charges associated with a diagnosed pertussis episode of care ranged from $496 to $3,239 per patient in the pre-index period and $987 to $16,971 per patient in the post-index period. When the average episode of care charges from patients observed in the private practitioner setting ($1,834) are applied to the average annual incidence estimate between 2006 and 2010 (1,513 cases), the extrapolated value approaches 3 million U.S. dollars. Within some categories (e.g., visits, lab tests, medications) the observed charges were greater in the younger age group (i.e., 50–64 years old) compared to the older (i.e., 65+ years old) age group. This result was driven by a minority of patients in the 50–64 year old group with very high healthcare utilization (and thus charges). Additionally, there were more patients in the younger group with asthma which may have led to the observation of increased utilization of respiratory/nebulizer treatments and medications. It is also possible that physicians are more likely to order lab tests for younger patients or that younger patients delay seeking care for a longer period of time and as a result present with more severe cases necessitating additional healthcare resources.
The use of charges is a proxy for the direct medical costs and represents the upper bound of the direct cost of physician office and hospital encounters from the current study. Historical direct medical costs (i.e., 2002 US$) of adult pertussis cases in the U.S. obtained from published literature range from $181 to $5,310 per patient depending on a number of factors including geographic location, population and type of services examined [31, 32]. Additionally, while not studied here, the indirect costs for the patients and their caregivers should also be considered in the societal economic burden.
The results of this study are subject to at least four limitations. First, with the exception of the projected incidence rates of ICD-9-CM coded pertussis from private practitioners, the data analyzed in the current study may not be generalizeable to the entire nation as it is an unweighted convenience sample obtained for the purpose of describing pertussis episodes of care. Second, charges were used as a measure of a portion of the economic burden. Because charges for healthcare services are typically higher than reimbursed amounts, charges reflected in the current study are likely to be higher than the costs paid by health insurers and patients. Third, claims data can be inherently limiting because they are collected for billing and reimbursement purposes, rather than for research purposes. This particular limitation may have yielded an underestimate of the projected incidence of diagnosed pertussis if practitioners are reluctant to submit a claim with a pertussis code in absence of laboratory confirmation, or ‘under’ code for other reasons. In addition, if a diagnosis were confirmed by a laboratory result after a claim was submitted, a physician would not submit a second claim as reimbursement would not be affected. Finally, it is likely that the methods used to diagnose pertussis (i.e., clinical and/or laboratory evaluations) in this study are not always 100% sensitive and/or 100% specific potentially yielding imprecise results. The retrospective, observational nature of the study also should be considered when interpreting the results as it could make the results subject to selection bias. To our knowledge, potential biases may include an underrepresentation of patients receiving care when a reimbursement claim is not submitted (e.g., patients without health insurance), underrepresentation of patients consulting providers that do not submit medical claims electronically as well as any miscoding, none of which could be identified or corrected for in this analysis. Strengths of the study include the large sample size and the national representativeness of the weighted sample and the inclusion of patient data from multiple care settings (i.e., private practitioners, hospitals, and laboratories).