Of the 56 persons enrolled in this study, 1 did not provide a serum sample at baseline and was excluded from analysis. Fifty-five percent of the enrollees were male, and the median age was 48 years (interquartile range 33–63 years) (Table 1). Most (87%) persons were non-Hispanic whites, which reflected the demographics of the study-site sample. Patients with lower respiratory syndromes had a median of 3 additional inclusion criteria. Individual criteria were evaluation by a chest radiograph (82%), administration of an antimicrobial drug (80%), dyspnea on exertion (56%), pleurisy (44%), and multiple visits for the same condition (42%).

Of the patients enrolled in our prospective study from select ambulatory care settings within the disease-endemic region, 29% were diagnosed serologically as having coccidioidomycosis. Even if one takes into account the wide 95% CI (16%–44%), this number demonstrates a high proportion of CAP caused by this infection. Furthermore, although the serologic tests used for diagnosis are highly specific for coccidioidal infection (12,13,16–18), another study emphasized that in the first weeks of primary illness these tests frequently show negative results (15). In the current study, 1 of 13 participants not initially serologically positive had coccidioidal antibodies in a second serum sample. Twenty-seven persons serologically negative at enrollment did not return for retesting; thus, additional coccidioidal infections may have been identified in this group. Our results will likely provide an underestimate of the incidence in this group of patients, further strengthening the conclusion that valley fever is a common cause of CAP in persons exposed to Coccidiodes in a disease-endemic area.

The high frequency of valley fever as a cause of CAP found in this study is consistent with previous estimates of coccidioidomycosis as a dominant cause of CAP with exposure in disease-endemic areas. A similar estimate of 25% to 30% has been obtained retrospectively at the Southern Arizona Veterans Administration Health Care System in Tucson, Arizona (7). Conversely, a diagnosis of valley fever requires laboratory testing. That this practice may not be uniform among clinicians was shown in a retrospective analysis of physician-specific diagnoses at primary care clinics in Tucson, Arizona, in which the rate of diagnosing coccidioidomycosis varied between 0% and 25% among physicians within the same group practice (7). Similar differences might also account for the increasing case rate associated with patient age that was reported in a recent analysis of 2001 Arizona state statistics (8). Case rates for persons >44 years of age were nearly twice those for persons 21–44 years of age. In our study in which all persons were uniformly evaluated for valley fever, all age groups had similar rates (27.3%–30.0%). Furthermore, although not detailed in our results, severity of illness in terms of respiratory symptoms was less in elderly subjects. We interpret the differences between the state statistics and those of our study as indicating that older persons who develop an illness are more likely to have an exact diagnosis determined, underscoring underreporting of illness in some patient groups such as young adults.

A corollary to the high frequency of coccidioidomycosis seen in this study is that persons anywhere with CAP and a history of recent travel to south-central Arizona or other regions where coccidioidomycosis is highly endemic would be expected to have a similarly high risk. For this reason, obtaining a travel history for any patient with CAP is essential for early and accurate diagnosis of this disease, as well as for other regional problems such as severe acute respiratory syndrome (SARS), hantavirus pneumonia, and avian influenza. Although the Infectious Diseases Society of America practice guidelines for CAP currently recommend obtaining a complete travel history only in patients with refractory pneumonia (19), we recommend that the guidelines be revised to recommend obtaining a travel history at the first evaluation.

Analysis of multiple symptoms at baseline showed several characteristics associated with coccidioidal infections. Both a shorter length of exposure in a disease-endemic region and a greater frequency of radiographic abnormalities were seen in persons with valley fever compared with those without valley fever. These associations were also evident in a previous report from a university health center (5). Symptoms of myalgia and reduced productivity were also evident with coccidioidal infection. However, none of these associations, alone or in combination, were of a sufficient magnitude to assist clinicians in the initial diagnosis. Therefore, our findings, as in the previous study (5), emphasize that laboratory testing at the initial physician visit is essential to identify patients with symptoms of CAP that are caused by valley fever.

A high proportion (81%) of persons with valley fever were prescribed an initial course of antimicrobial drugs. Of these, 12 patients, 3 of whom were diagnosed with valley fever, received 2 courses of these drugs. Although diagnosis of valley fever by serologic methods is frequently delayed by 3–5 days, use of antimicrobial drugs could still be avoided or stopped earlier in patients whose illness is determined to be caused by Coccidioides species.

The inclusion criteria used in this study were designed both to be broadly inclusive and to select patients with more severe illness. As such, they differ in some respects from commonly used entry criteria for clinical trials of new antimicrobial drugs as treatment for CAP. For example, we chose pleuritic chest pain and dyspnea at rest and fever as an entry requirement. By using these entry criteria, we found that 3 of the 12 patients with valley fever who underwent radiographic examination had normal radiographs, which is consistent with results of a previous study (3), but did not adhere to Infectious Diseases Society of America or American Lung Association definitions of pneumonia (19). When comparing our findings to those of other studies, the way in which patients were selected should be taken into account.

Several limitations of our methods deserve emphasis. Because our inclusion criteria were not standard, comparison of our results to those of other studies of CAP is difficult. Also, we did not have a diagnosis for patients without valley fever. Since this is a relatively small study, additional expanded studies may be useful, especially to extend observations to other groups such as children, the elderly, those requiring hospitalization, and residents elsewhere within disease-endemic regions. Future studies are also needed to determine best practices for management of primary coccidioidal infection and possible therapy with specific antifungal treatment. Such a high proportion of CAP caused by Coccidioides species should provide further impetus to conduct those studies.