Cryptococcus neoformans gained notoriety in the 1980s, emerging as a common opportunistic infection in adults with acquired immunodeficiency syndrome (AIDS). With the introduction of highly active antiretroviral therapy (HAART) in 1996, the reported incidence has declined1,2; a surveillance study of 1491 cases of cryptococcal illness showed that the annual incidence dropped in Atlanta from 66 per 1000 patients with AIDS in 1992 to 7 per 1000 patients with AIDS in 2000.3 However, the disease has been responsible for morbidity and mortality in patients with other immunocompromising conditions and in apparently immunocompetent individuals for decades preceding the AIDS epidemic.3–8
Despite the many large studies conducted in adults, particularly in the HIV-positive population, few studies dedicated to pediatric patients exist. The studies that do exist are small case series, literature reviews, or large adult studies that include small numbers of pediatric patients. On the basis of these limited data, cryptococcosis is believed to be less common among children than adults.9 Given the dearth of information on cryptococcal illness in children, we used a large administrative database to describe the epidemiology and outcomes of children admitted to the hospital with cryptococcal infection (CI).
MATERIALS AND METHODS
Data Source and Study Population
Data for this study were obtained from the Pediatric Health Information System (PHIS) database, a comprehensive, comparative pediatric database containing clinical and financial details of more than 6 million patient cases. Specifically, the database contains the diagnosis and procedure codes and billed transaction/utilization data of inpatient and outpatient hospital encounters among the 42 PHIS owner children's hospitals nationwide. Member hospitals represent most of the major metropolitan areas across the United States with only one children's hospital representing each city. Hospitals access PHIS data through a virtual private network.
To protect patient confidentiality, medical record number, billing number, physician numbers, and zip code are all encrypted in the database. A third-party vendor loads, processes, and warehouses PHIS data that are submitted by member hospitals. The warehouse partner then applies 175 reliability and validity checks to each patient record. Submissions that do not meet the error threshold of less than 2% are returned to the hospital of origin for correction and resubmission.
The source population for this retrospective study included patients below the age of 19 admitted to 1 of 42 free-standing children's hospitals contributing to the PHIS database between January 1, 2003 and September 30, 2008. The patients were included if they had a discharge International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9) diagnosis code for cryptococcosis (117.5) or cryptococcal meningitis (CM) (321.0). Up to 21 discharge diagnosis codes were screened for each admission. Patients were divided into 2 groups: patients with a code for CM were classified as having meningitis, whereas patients with only a code for cryptococcosis were defined as nonmeningitis cryptococcosis (non-CM). All patients, regardless of site of infection, were classified as having CI.
All patients included in the final cohort were followed forward after their first admission to identify any subsequent hospitalizations for cryptococcosis. Every readmission with an ICD-9 discharge code for cryptococcosis or CM within the study period was evaluated.
Demographic information, including age, gender, race, and region of the country, were collected. Using Census defined regions we grouped the hospitals into 1 of 4 US regions: West, North-Central (Midwest), Northeast, and South (Appendix Table, Supplemental Digital Content 1, http://links.lww.com/INF/A617).
We manually reviewed all of the assigned ICD-9 diagnosis codes for each hospitalization (including all readmissions) to determine underlying medical conditions for each patient. The diagnosis codes deemed by the investigators to be most significant to the patients' immune status were used to categorize them into various immunosuppressive or chronic disease categories. Patients were categorized as having an immunocompromising or chronic condition if they had a discharge diagnosis code consistent with such an illness. Patients were categorized as immunocompetent if they lacked a discharge diagnosis code indicative of a chronic or immune compromising condition.
Use of corticosteroid and immunosuppressive medication was also considered as possible immunocompromising factors. Because corticosteroids may have been prescribed as a treatment modality for CM, prescription of corticosteroids on the first day of admission was used as a proxy indicator for steroid exposure before admission.
Human immunodeficiency virus (HIV) status was determined by an ICD-9 code for HIV infection (042). Because of the sensitive nature of the illness, we were concerned that some patients with HIV infection were not recorded as such on their billing documents. Therefore, we also identified patients as HIV-positive if they were billed for antiretroviral medications during their stay.
Potentially therapeutic medications that were billed to each patient were reviewed for each hospital day for all admissions. The following medications were considered as therapy for CI: fluconazole, itraconazole, voriconazole, posaconazole, amphotericin B (conventional or lipid formulations), flucytosine, and interferon gamma-1b. Initial therapy was defined as the first documented time a patient received 2 consecutive days of an antifungal agent or agents. A patient was defined as receiving more than one antifungal agent as initial therapy if both medications were given simultaneously for at least 2 consecutive days.
Additionally, ICD-9 procedure codes were analyzed for each hospitalization to identify the number and type of procedures performed on each of the identified patients with CI.
Summary statistics were constructed using frequencies and proportions for categorical data elements and medians and interquartile ranges for continuous variables. The 2 groups were compared across all variables using the Fisher exact test for categorical variables and the Wilcoxon-Mann-Whitney U test for continuous variables. All analyses were conducted with Stata version 9.0 statistical software (College Station, TX).
Human Subjects Oversight
The conduct of this study was approved by the Committee for the Protection of Human Subjects at the Children's Hospital of Philadelphia.
Between January 1, 2003 and September 30, 2008, 65 patients from 27 of the 42 PHIS hospitals accounted for 84 hospital admissions associated with a diagnosis of CM or non-CM. This resulted in a CI admission rate of 6.2 per million hospitalizations. It was necessary to exclude 2 patients from the analysis due to missing medication data.
Twenty-four of the remaining 63 patients (38.1%) had an ICD-9 diagnosis code consistent with CM while 39 patients (61.9%) had an ICD-9 code consistent with non-CM. Of the 24 patients that were assigned the 321.0 code for CM, 23 of them were also assigned the 117.5 code for cryptococcosis. This suggests that typically patients with any form of CI are assigned the 117.5 code, while patients with CM will also be assigned the 321.0 code.
Demographic Data and Underlying Medical Conditions
Table 1 lists demographic data including age, gender, race, and Census region for all patients and separately for disease type. A majority of patients were male (60.3%) and 55.6% were white. The median age was 11.98 years (interquartile range: 6.11–15.14 years). The western region was most commonly represented (42.9%) while the Midwest region had the fewest admissions (11.1%).
Underlying medical conditions of patients in the study included HIV infection in 15.9% and malignancy in 22%. Only 20.6% of patients were immunocompetent. All patients who were identified as HIV-positive by ICD-9 codes were also billed for antiretroviral medications, supporting the accuracy of this code for identifying HIV infection. All patients who received immunosuppressive drugs or steroids at the onset of their hospitalization also had an ICD-9 code for an immunocompromising or chronic condition.
When comparing CM and non-CM patients, CM patients were older (P = 0.012), more commonly of black race (P = 0.016), and more likely to be HIV infected (P = 0.029). Non-CM patients were more likely to have cancer (P = 0.034).
Table 2 summarizes the initial antifungal regimens prescribed to patients during their first admission. Nine patients were not administered a medication with anticryptococcal activity on their first admission. Of these 9 patients, 7 had immunocompromising conditions and 2 of these patients ultimately died during their hospitalization. Most patients who were prescribed amphotericin were given lipid formulations of amphotericin. One patient received conventional amphotericin alone and 2 patients received conventional amphotericin in combination with flucytosine. Patients with CM were more likely than non-CM patients to receive amphotericin B and flucytosine together as initial therapy (P = 0.002). Fluconazole was prescribed as monotherapy to 22.2% of patients. These patients were more likely to have non-CM (P = 0.034). One patient received fluconazole for only 1 day of a 2-day hospital stay. On readmission, this patient again received fluconazole monotherapy, so he or she was counted as receiving fluconazole. Of note, 3 patients with non-CM were initially prescribed voriconazole alone.
Interferon gamma-1b was given to only 1 patient, a 10-year-old boy with a combined immunodeficiency and a ventricular septal defect, who also received amphotericin B and fluconazole. This patient was discharged to home after a 21-day hospital stay.
Surgical procedures were performed in 84% of patients during their first hospitalization. The most common procedures were biopsies of the bronchus or lung (8 patients), chest tube insertion (4), or esophagogastroduodenoscopy with biopsy (5). Invasive pulmonary procedures were performed in 30.8% of non-CM patients versus 8.3% of CM patients (P = 0.034), whereas CNS procedures were more common (20.8% vs. 5.1%, P = 0.067) in patients with CM.
The median follow-up time after first admission was 1085 days (range: 119–2079 days). Twelve patients in the study were readmitted and assigned a cryptococcal ICD-9 code during the readmission. Five of the 12 patients had non-CM, of which 4 had an underlying malignancy and 1 had pulmonary hypertension. All 5 of these patients required only 1 additional admission for non-CM, and all were discharged to home.
Among the 7 patients with CM that were readmitted, 1 had hereditary progressive muscular dystrophy and required 3 admissions, 2 HIV-infected patients required 3 admissions each, and 1 patient with an aortic arch anomaly and hypogammaglobulinemia required 6 admissions. Two other HIV-infected patients and 1 immunocompetent patient required a second hospital stay for CM.
The majority of patients (82.5%) were discharged to home after a median length of stay of 19 days (interquartile range: 11–28). The patients with CM were more likely to require an intensive care unit stay (70.8% of patients with CM vs. 17.9% of non-CM, P < 0.001), and were more likely to receive mechanical ventilation (50% of CM patients vs. 15.4% of non-CM, P = 0.004).
The in-hospital mortality rate for this cohort was 9.5%. Of the 6 deaths, 5 occurred during the patient's first hospitalization for cryptococcal illness. When evaluated by type of infection, the in-hospital mortality rate was 7.7% for patients with non-CM and 21.5% for patients with CM. Only 1 patient with HIV died during their hospitalization. This patient died during the third hospitalization for CM, a hospitalization complicated by Staphylococcus aureus septicemia and end-stage renal disease. Of note, 2 patients in our study had both CM and a hematologic malignancy, and both of these patients died during their admissions.
To our knowledge, this is the largest study of CI in children reported. We found an incidence rate of 6.2 cases of CI per million hospitalizations with an in-hospital mortality rate of 9.5%. Patients with CM comprised 38% of patients in the study, whereas 62% had cryptococcosis of an unspecified site. In this cohort, 20.6% were immunocompetent, 63.5% were HIV-negative with immunocompromising conditions, and 16% were HIV-positive. Most patients received some combination of fluconazole, amphotericin B, and flucytosine in their treatment regimens; however, 9 patients received no antifungal medications.
Previous reviews of the literature or case series have only reported on 23 or fewer pediatric patients.7,10–15 Various studies have highlighted larger numbers of pediatric patients with CI, but these studies dealt exclusively with children with AIDS.16–19
The median age of our patients was 12 years, consistent with previous reports of median ages of 9.8 and 13 years.12,18 We found that children with CM were statistically significantly older than children with non-CM. In addition, patients in our study were more commonly male (60%), which is consistent with prior data in children revealing 62.5% to 76.9% of patients to be male.12,13 Why there is a male predominance in CI and why patients with CM tend to be older are interesting questions requiring further evaluation.
Surprisingly, 55.6% of our patients were white. A population-based study of 2 large US metropolitan areas reported that 58% of cryptococcal illnesses in patients of all ages occurred in African American patients,3 compared with only 27% in our study. This discrepancy may reflect the population of patients presenting to free-standing children's hospitals as compared with the aforementioned adult studies in urban areas. Among the patients admitted to PHIS member hospitals in the period of our study, 65% were white and 20.5% were black. However, 46% of the patients with CM in our study were black, compared with only 15% of patients with non-CM. Whether this suggests that black patients are more susceptible to CM if infected with cryptococcosis or that they are more likely to present with advanced disease is unclear. Further investigation is necessary to determine if a true racial predilection exists for CM.
When considering the entire PHIS database, only 22% of all admissions were at western hospitals and 27.5% of all admissions were to North-Central hospitals. However, our data show that 43% of patients with cryptococcosis were from the West and 11% were from the North-Central region. Cryptococcus neoformans is ubiquitous in nature, but Cryptococcus gattii has a predilection for tropical and subtropical climates with emerging endemicity in the Pacific Northwest.20 We do not know which species of Cryptococcus was predominant in our patient population. It is possible that some of the patients from the western states reflect more C. gattii infection.
In adults, 10% to 44% of patients are reported to have no immunocompromising illnesses.1,4,21 Our data showing 21% of patients to be immunocompetent are similar to previous studies in children,7,13 but differs from a recently reported series by Severo et al (48%).15 More patients with CM were found to be immunocompetent, consistent with previously reported data in adults.4,21 Only 10 patients (15.9%) in our study were HIV-infected. In adult series in the post-HAART era, this number ranges from 60% to 89%.1,3 One prior study in children found that 28% of patients with CI have AIDS.15 It is possible that the lower rates of HIV-positive status in children with cryptococcosis reflect the decrease in opportunistic infections in the HIV population with the advent of HAART.
The conditions predisposing our patients to cryptococcal illness mimic the case reports in children in the literature,13 suggesting that patients with malignancies, transplant recipients, and patients with immunologic disorders are at an increased risk of acquiring cryptococcal illness.4,8,21,22 It is noteworthy that of the 14 patients with malignancies, 9 suffered from acute lymphoblastic leukemia. This is consistent with the 1992 study of Leggiadro et al in immunocompromised children.12
In 2000, the Infectious Diseases Society of America (IDSA) delineated specific guidelines for the treatment of CI in adults based on the site of involvement and the host's immune status. Controlled trials have not been performed in pediatric populations, so we explored whether the children in our study were treated in accordance with published guidelines created for adults. For immunocompetent hosts who are asymptomatic with isolated pulmonary disease, observation alone may be appropriate.23 Of the 9 patients in our study who did not receive antifungal therapy, only 2 patients were immunocompetent and possibly met criteria for “watchful waiting.” In general, patients with non-CM were more likely to be initially treated with fluconazole alone and patients with CM were more likely to receive amphotericin B and flucytosine, consistent with the IDSA guidelines.
Ten of 11 patients in our study with CM who were HIV-negative but otherwise immunocompromised received the therapy recommended for adults. However, of the 29 immunocompromised patients with non-CM infection, only 7 received therapy consistent with the guidelines. It seems that immunocompromised HIV-negative children with non-CM are often treated outside of the established guidelines for adults.
The guidelines did not list voriconazole as a recommended treatment option, but this medication was prescribed as initial therapy to 3 patients in this study, all of which were immunocompromised. Although voriconazole may have activity against Cryptococcus, its effectiveness as a first-line agent in children has not been established. Further study is necessary to establish the appropriateness of voriconazole as a first- or second-line therapeutic option.
In February of 2010, the IDSA released new Clinical Practice Guidelines for the Management of Cryptococcal Disease. These new guidelines divide patients into 3 groups: HIV-infected, organ transplant recipients, and non-HIV infected and nontransplant hosts. These updated guidelines also comment on the treatment of pediatric patients.24 Treatment decisions made for the patients in our study were potentially based on the 2000 guidelines for adults with cryptococcal illness, and thus the aforementioned comparisons were made against the 2000 guidelines. It will be interesting to see the way by which the new guidelines will effect the management of children with cryptococcosis.
Several studies have discussed mortality in patients with cryptococcal illness, but to our knowledge, no study has specifically looked at in-hospital mortality in both HIV-infected and noninfected individuals. Therefore, in-hospital mortality rate of 9.5% of our study cannot be compared with previous studies in adults or children. A case series of 9 pediatric HIV-negative immunocompromised patients with extrapulmonary cryptococcosis reported no fatalities,12 whereas a series of 13 HIV-positive children with extrapulmonary cryptococcosis reported 3 fatalities (24%).16 In adults, the mortality rate ranges from 4% to 55%, depending on host immune status.4,8,21 Pappas et al described mortality by site of infection in HIV-negative adults, describing 16% mortality with CNS disease, 5% with pulmonary disease alone, and 24% with other sites of infection.4 This is higher than our reported in-hospital mortality rates of 12.5% with CM and 7.7% with non-CM.
The primary strength of our study is that it is a large, multicenter study drawing from a national pediatric database. To date, the identified cohort of pediatric patients is the largest reported and is nationally representative of the experience of inpatient management for CI at free-standing children's hospitals.
Numerous limitations are inherent when conducting retrospective analyses of a large administrative dataset. First, our reliance on ICD-9 billing codes for identifying the cases is susceptible to misclassification errors. The validity of ICD-9 coding for cryptococcal illness has not been established, however, ICD-9 codes have been found to be highly specific for other fungal diseases.25 This suggests that although we may have underestimated the number of admissions for CI, the identified cases should be true CIs. Furthermore, our findings (ie, increased number of invasive CNS procedures in patients categorized as CM) lend some internal validity to our differentiation of patients into CM and non-CM groups. Moreover, the lack of a date of diagnosis associated with the assignment of an ICD-9 billing code makes it impossible to critically evaluate outcomes that are time sensitive, such as time from diagnosis to in-hospital death, duration of hospital stay after cryptococcal diagnosis is made, or duration or choice of antifungal therapy after the diagnosis is determined.
Second, because of the lack of inpatient antifungal dosing data and outpatient treatment information, we were not able to fully evaluate the therapeutic regimens and ultimate outcomes for each patient. Because of gaps in the treatment data, we were also unable to specify if medications were prescribed as induction or consolidation therapies. Therefore, our ability to make evidence-based recommendations regarding optimal therapeutic interventions is limited. Furthermore, billing data were used to define the type, timing, and duration of antifungals given to each patient during their inpatient stay. It is possible that errors in billing data could result in the conclusion that a patient did not receive appropriate therapy, when in fact appropriate therapy was administered.
Finally, we cannot determine the variant of C. neoformans encountered by the patients in our study. C. neoformans var gattii is now considered its own species (C. gattii). It is known to affect the immunocompetent and may cause a more protracted course of illness with a poorer outcome. C. neoformans var grubii is known for its proclivity to cause illness in the immunocompromised. Given that the specific variety of C. neoformans may influence the epidemiology, prognosis, and treatment regimens, these data would be useful to gather in a future pediatric study.26
This is the largest study of CI in a pediatric population. Our data support our hypothesis that the illness is not common in hospitalized children, regardless of immune status. Surprisingly, only 16% of the patients were HIV-infected, suggesting that far more immunocompetent and non-HIV immunocompromised pediatric patients are diagnosed with CI. Furthermore, data suggesting a potential male predominance of CI and increased risk of CM in black patients need further exploration. Future studies will be necessary to further elucidate the risk factors and most effective treatment for children with CI.
1. Tintelnot K, Schar G, Polak A; European Confederation of Medical Mycology. Epidemiological data of cryptococcosis in Austria, Germany and Switzerland: part of the ECMM survey in Europe. Mycoses.
2. Palella FJ Jr, Baker RK, Moorman AC, et al. Mortality in the highly active antiretroviral therapy era: changing causes of death and disease in the HIV outpatient study. J Acquir Immune Defic Syndr
3. Mirza SA, Phelan M, Rimland D, et al. The changing epidemiology of cryptococcosis: an update from population-based active surveillance in 2 large metropolitan areas, 1992–2000. Clin Infect Dis
4. Pappas PG, Perfect JR, Cloud GA, et al. Cryptococcosis in human immunodeficiency virus-negative patients in the era of effective azole therapy. Clin Infect Dis
5. Baddley JW, Perfect JR, Oster RA, et al. Pulmonary cryptococcosis in patients without HIV infection: factors associated with disseminated disease. Eur J Clin Microbiol Infect Dis
6. Vilchez RA, Irish W, Lacomis J, et al. The clinical epidemiology of pulmonary cryptococcosis in non-AIDS patients at a tertiary care medical center. Medicine (Baltimore)
7. Dromer F, Mathoulin-Pelissier S, Fontanet A, et al. Epidemiology of HIV-associated cryptococcosis in France (1985–2001): comparison of the pre- and post-HAART eras. AIDS
8. Kerkering TM, Duma RJ, Shadomy S. The evolution of pulmonary cryptococcosis: clinical implications from a study of 41 patients with and without compromising host factors. Ann Intern Med
9. Mofenson LM, Brady MT, Danner SP, et al. Guidelines for the prevention and treatment of opportunistic infections among HIV-exposed and HIV-infected children: recommendations from CDC, the National Institutes of Health, the HIV Medicine Association of the Infectious Diseases Society of America, the Pediatric Infectious Diseases Society, and the American Academy of Pediatrics. Morb Mortal Wkly Rep Recomm Rep.
10. Emanuel B, Ching E, Lieberman AD, et al. Cryptococcus meningitis in a child successfully treated with amphotericin B, with a review of the pediatric literature. J Pediatr
11. Siewers CM, Cramblett HG. Cryptococcosis (Torulosis) in children: a report of four cases. Pediatrics
12. Leggiadro RJ, Barrett FF, Hughes WT. Extrapulmonary cryptococcosis in immunocompromised infants and children. Pediatr Infect Dis J
13. Sweeney DA, Caserta MT, Korones DN, et al. A ten-year-old boy with a pulmonary nodule secondary to Cryptococcus neoformans
: case report and review of the literature. Pediatr Infect Dis J
14. Huang KY, Huang YC, Hung IJ, et al. Cryptococcosis in nonhuman immunodeficiency virus-infected children. Pediatr Neurol
15. Severo CB, Xacier MO, Gazzoni AF, et al. Cryptococcosis in children. Paediatr Respir Rev
16. Leggiadro RJ, Kline MW, Hughes WT. Extrapulmonary cryptococcosis in children with acquired immunodeficiency syndrome. Pediatr Infect Dis J
17. Gonzalez CE, Shetty D, Lewis LL, et al. Cryptococcosis in human immunodeficiency virus-infected children. Pediatr Infect Dis J
18. Abadi J, Nachman S, Kressel AB, et al. Cryptococcosis in children with AIDS. Clin Infect Dis
19. Gumbo T, Kadzirange G, Mielke J, et al. Cryptococcus neoformans
meningoencephalitis in African children with acquired immunodeficiency syndrome. Pediatr Infect Dis J
20. Chayakulkeeree M, Perfect JR. Cryptococcosis. Infect Dis Clin North Am
21. Lui G, Lee N, Ip M, et al. Cryptococcosis in apparently immunocompetent patients. Q J Med
22. Husain S, Wagener MM, Singh N. Cryptococcus neoformans
infection in organ transplant recipients: variables influencing clinical characteristics and outcome. Emerg Infect Dis
23. Saag MS, Graybill RJ, Larsen RA, et al. Infectious Diseases Society of America. Practice guidelines for the management of cryptococcal disease. Clin Infect Dis
24. Perfect JR, Dismukes WE, Dromer F, et al. Clinical practice guidelines for the management of cryptococcal disease: 2010 update by the Infectious Diseases Society of America. Clin Infect Dis
25. Prasad PA, Coffin SE, Leckerman KH, et al. Pediatric antifungal utilization: new drugs, new trends. Pediatr Infect Dis J
26. Kidd SE, Hagen F, Tscharke RL, et al. A rare genotype of Cryptococcus gattii
caused the cryptococcosis outbreak on Vancouver Island (British Columbia, Canada). Proc Natl Acad Sci USA