Study design
To understand the original cluster and assess our primary seasonal hypothesis, we performed a retrospective cohort study that included patients during the cluster, and the period before and after the cluster. All HM patients with IMI microbiologically-confirmed at the University of Washington Medical Center (UWMC) and Seattle Cancer Care Alliance (SCCA) between June and August 2014 defined the initial cluster. To elucidate the baseline frequency of IMIs prior to the cluster, we also reviewed all cases of IMIs in HM patients between January 2012 and May 2014. Following the cluster of IMI in 2014 and associated control interventions, in order to determine where these efforts affected risks for IMIs, we prospectively monitored for additional cases over the next 12 months, ending in August 2015. All patients with an IMI that were not related to an underlying HM, were not included in these analyses.
In a second component of the study, all HM patients with other clinically important invasive filamentous fungal infections (IFIs) (e.g. Aspergillus spp., Fusarium spp.) diagnosed between January 2013 and August 2014 were also identified; this time frame included both the initial cluster period and a referent year (calendar year 2013). Data on Posaconazole use, either for prophylaxis or treatment, was collected during the period pre and post-cluster identification, to assess center-based changes in usage.
Setting
The SCCA/Fred Hutchinson Cancer Research Center is a National Cancer Institute-designated Cancer Center that sees approximately 75,000 patients annually, including high-risk hematopoietic cell transplant (HCT) recipients. Generally, patients commute from their permanent residence to the center or stay locally at center-based transition housing/community housing near the center. Those requiring inpatient care are admitted to the UWMC, a large 450-bed academic tertiary medical center which incorporates adult hematology and oncology units with over 100 inpatient beds.
Allogeneic HCT recipients in the initial 2 years of the study receive fluconazole as primary antifungal prophylaxis to day + 75 post transplantation. HCT recipients with high grade graft-versus-host disease (GVHD) are recommended to be given posaconazole prophylaxis unless undergoing treatment for another fungal infection [9]. Patients undergoing autologous HCT are given fluconazole until neutropenia and mucositis resolve. Prior to the cluster identification, patients with a HM at risk of prolonged neutropenia (e.g. acute myeloid leukemia [AML] and myelodysplastic syndrome [MDS]) also received fluconazole prophylaxis.
Patients admitted for treatment are managed on floors with single rooms that include high efficiency particulate air (HEPA) filtration. In the outpatient clinics, floors where HCT recipients are cared for are also HEPA filtered. Wearing a surgical mask in the outpatient department is not routine unless patients are known to have a contagious respiratory pathogen (e.g. influenza).
Participant eligibility/case definitions
All cancer patients who were admitted to the hospital’s HCT and HM cancer units during the cohort periods of interest were included in these analyses. IFIs were defined as per the revised European Organization for Research and Treatment of Cancer/Invasive Fungal Infections Cooperative Group and the National Institute of Allergy and Infectious Diseases Mycoses Study Group (EORTC/MSG) Consensus Group definitions [10]. Histopathological, cytopathological and/or direct microscopic evidence of tissue invasive hyphal elements or culture of mold from sterile tissue were defined as proven IMIs. Cases were considered probable if they included a host factor, clinical criteria and mycological criteria. For the purposes of this study, we created an additional category, “possible with positive PCR” for IMI cases, which included those patients that fulfilled host and clinical criteria and were also positive by a laboratory-developed pan-fungal and/or Mucorales-specific PCR assays. Final analysis counted proven, probable and “possible with positive PCR” infections as cases. Patients who had only possible infection (i.e. absent mycological criteria or PCR results) were not counted as cases.
Site of infection was broadly categorized into sinus, pulmonary, cutaneous, cerebral or other. A disseminated infection referred to patients with infection involving two or more non-contiguous sites [11]. Onset of fungal infection was defined as the date at which the patient met the criteria for proven, probable or “possible with positive PCR” definition. Mortality was defined as all-cause mortality within 90 days of fungal infection. For the purposes of the analyses, summer months were considered June, July and August; fall as September, October, November; winter as December, January, February; and spring as March, April and May.
Data sources
Cancer inpatient data, including unique patients and total inpatient-days were collected from center-based clinical databases. All patients with microbiologically-confirmed mucormycosis at the UWMC and SCCA from January 2013 – August 2015 and a second cohort of hematology/oncology patients with other clinically important IFIs from January 2013 - August 2014 were identified using laboratory records, database review and electronic medical records.
In order to address potential associations with local construction patterns, publically available construction permit data from the Seattle-King County records were assessed [12]. For the purposes of the analyses construction projects were considered active for the entire duration of the permit. Clinical cases were also compared to publically available local temperature and precipitation data gathered from Seattle Weather Service Forecast Office (WSFO) Sand Point Station (King County, WA) [13]. In order to link both sources to clinical outcomes, mean temperature and total precipitation were included in analyses. Data collection and all analyses were approved by the Fred Hutchinson Cancer Research Center (FHCRC) Institutional Review Board.
Diagnostic methods
At our center, fungal cultures are routinely performed on all pulmonary specimens in immunocompromised hosts and on non-pulmonary specimens upon request. Diagnostic bronchoalveolar lavages (BAL) are commonly performed in patients with HM being treated with chemotherapy or those following a HCT with clinical and/or radiologic findings consistent with a pulmonary infection. Fungal cultures and Platelia™ aspergillus enzyme immunoassay (Bio-Rad Laboratories. Hercules, CA) for galactomannan (GM) are performed on all BAL samples in patients suspected to have a fungal infection; serum GM is performed on request. Any fungi growing in culture and resembling filamentous fungi are identified using conventional morphology [14] and/or DNA sequence analysis. All BAL specimens also undergo routine cytologic review for hyphal elements by center-based pathologists. In addition to standard testing from airway specimens, all lung biopsies and autopsies from patients with HM are routinely assessed for fungal infections by both histopathology and culture.
In addition, laboratory developed pan-fungal and Mucorales-specific PCRs are performed on pulmonary and non-pulmonary specimens upon request. The pan-fungal PCR is performed using published primers [15] targeting fungal ribosomal RNA genes. For the PCR specific for the Mucorales, samples are assessed using published broad range fungal primers [15] targeting the ITS2 region of the ribosomal RNA genes followed by a nested PCR with a mixture of five laboratory developed Mucorales specific primers, designed to identify Rhizopus oryzae, Rhizopus microsporus or azygosporus, Mucor spp., Rhizomucor spp. and Lichtheimia corymbifera. The analytical sensitivity of this test is 1 genome per PCR reaction. PCRs are assessed on BAL fluid, biopsy, and autopsy samples by request.
Statistical analysis
The incidence rates of IMI were estimated by dividing the number of incident IMI cases developed in cohort subjects by the inpatient-time at-risk contributed by the overall cohort; exact 95% confidence intervals (CI) were estimated based on a Poisson distribution. Time at-risk was calculated using time during inpatient admission where patients contributed days at risk from date of admission to the date of discharge or death. Associations of season, mean monthly temperature, and total monthly precipitation with incidence rates were assessed using a Poisson regression model, with data aggregated in 1-month intervals. The number of IMI events was the dependent variable and the logarithm of the number of patient-days was included as an offset variable. Season, temperature, and precipitation were the independent variables, each included in separate univariable models. Because observed incidence rates did not increase linearly with increasing temperatures but instead showed a sharp increase at approximately 20 degrees C, we chose to model temperature as a binary variable indicating temperatures above 20 degrees C versus 20 degrees C or lower. Precipitation was modeled as a continuous variable. Model estimates were presented as incidence rate ratios (IRR) with 95% CIs. Posaconazole use was compared between early and later time periods using a chi squared test.