Invasive Fungal Infections in Inpatient Solid Organ Transplant Recipients With COVID-19: A Multicenter Retrospective Cohort.

BACKGROUND: The prevalence and outcomes of COVID-19-associated invasive fungal infections (CAIFIs) in solid organ transplant recipients (SOTRs) remain poorly understood. METHODS: A retrospective cohort study of SOTRs with COVID-19 admitted to 5 hospitals within Johns Hopkins Medicine was performed between March 2020 and March 2022. Cox regression multilevel mixed-effects ordinal logistic regression was used. RESULTS: In the cohort of 276 SOTRs, 22 (8%) developed IFIs. The prevalence of CAIFIs was highest in lung transplant recipients (20%), followed by recipients of heart (2/28; 7.1%), liver (3/46; 6.5%), and kidney (7/149; 4.7%) transplants. In the overall cohort, only 42 of 276 SOTRs (15.2%) required mechanical ventilation; these included 11 of 22 SOTRs (50%) of the CAIFI group and 31 of 254 SOTRs (12.2%) of the no-CAIFI group. Compared with those without IFIs, SOTs with IFIs had worse outcomes and required more advanced life support (high-flow oxygen, vasopressor, and dialysis). SOTRs with CAIFIs had higher 1-y death-censored allograft failure (hazard ratio 1.65.116.4, P = 0.006) and 1-y mortality adjusting for oxygen requirement (adjusted hazard ratio 1.12.45.1, P < 0.001), compared with SOTRs without CAIFIs. CONCLUSIONS: The prevalence of CAIFIs in inpatient SOTRs with COVID-19 is substantial. Clinicians should be alert to the possibility of CAIFIs in SOTRs with COVID-19, particularly those requiring supplemental oxygen, regardless of their intubation status.

Teaching Everyone Everywhere All at Once: Leveraging Social Media to Implement a Multisite Fungal Diagnostics Curriculum.

Background: Environmental fungi are threats to personal and public health. Fungal in vitro diagnostics help diagnose invasive fungal infections (IFIs), but clinicians remain underinformed about their use and interpretation. Given the increasing use of social media to share infectious diseases-related content, we designed and implemented a multisite Twitter-based curriculum focused on IFIs and related diagnostics. Methods: Questions were posted through a dedicated Twitter account twice weekly over 8 weeks. We surveyed clinicians at 3 US academic centers before and after completion of the curriculum and interviewed a subset of participants. We undertook quantitative and qualitative evaluations and reviewed Twitter analytics. Results: We surveyed 450 participants. One hundred twenty-one participants (27%) completed the knowledge assessment precurriculum, 68 (15%) postcurriculum, and 53 (12%) pre- and postcurriculum. We found a significant increase (72% vs 80%, P = .005) in the percentage of correct answers in the pre- versus postcurriculum knowledge assessments. Perceived benefits included a well-executed curriculum that facilitated engagement with appropriately detailed tweetorials from a dedicated Twitter account. Perceived barriers included lack of awareness of tweetorial posts and timing, competing priorities, and the coronavirus disease 2019 pandemic. The Twitter account accrued 1400 followers from 65 countries during the 8-week period. Tweets with multiple-choice questions had a median of 14 904 impressions (interquartile range [IQR], 12 818-16 963), 798 engagements (IQR, 626-1041), and an engagement rate of 6.1% (IQR, 4.2%-6.6%). Conclusions: Educators can leverage social media to share content with a large audience and improve knowledge while being mindful of the barriers associated with implementing a curriculum on social media.

Progressive disseminated histoplasmosis: The experience in one non-endemic medical center.

Histoplasmosis, the most common endemic mycosis in North America, presents in a myriad of ways, spanning the spectrum from self-limiting pneumonia to progressive disseminated histoplasmosis (PDH). Toward better describing contemporary histoplasmosis syndromes, risks, and outcomes, this single-center retrospective cohort study was performed (2009-2019). The population who developed PDH was similar to that with other forms of histoplasmosis (OFH) except for higher rates of preexisting immunocompromising conditions (91.3% vs. 40%, P < .001) and a trend toward receiving more chronic immunosuppression (65.2% vs. 33.3%, P = .054) compared to those with OFH. Diagnosis was most frequently achieved by urinary or serum antigen positivity. People with PDH more frequently tested positive compared to those with OFH, but negative tests did not rule out histoplasmosis. Median time to diagnosis was prolonged among people with both PDH and OFH (32 vs. 31 days, respectively). Following diagnosis, people with PDH received more liposomal amphotericin (78.3% vs. 20%, P < .001). Subsequent survival at 90 and 365 days and treatment response were similar in both groups. Patients with PDH were more often hospitalized (95.7% vs. 60%, P = .006); however, once admitted, there were no differences in hospital length of stay or intensive care unit admission rate. The challenges of diagnosing histoplasmosis based on clinical presentation alone highlight the need for heightened awareness of these entities especially given the recent reports on expanded endemicity and delays in diagnosis.

Histoplasmosis is the most common endemic mycosis in North America. This article summarizes the clinical features, risk factors, and outcomes in patients who developed disseminated disease compared to more localized forms of histoplasmosis.

Tuning sterol extraction kinetics yields a renal-sparing polyene antifungal.

Decades of previous efforts to develop renal-sparing polyene antifungals were misguided by the classic membrane permeabilization model1. Recently, the clinically vital but also highly renal-toxic small-molecule natural product amphotericin B was instead found to kill fungi primarily by forming extramembraneous sponge-like aggregates that extract ergosterol from lipid bilayers2-6. Here we show that rapid and selective extraction of fungal ergosterol can yield potent and renal-sparing polyene antifungals. Cholesterol extraction was found to drive the toxicity of amphotericin B to human renal cells. Our examination of high-resolution structures of amphotericin B sponges in sterol-free and sterol-bound states guided us to a promising structural derivative that does not bind cholesterol and is thus renal sparing. This derivative was also less potent because it extracts ergosterol more slowly. Selective acceleration of ergosterol extraction with a second structural modification yielded a new polyene, AM-2-19, that is renal sparing in mice and primary human renal cells, potent against hundreds of pathogenic fungal strains, resistance evasive following serial passage in vitro and highly efficacious in animal models of invasive fungal infections. Thus, rational tuning of the dynamics of interactions between small molecules may lead to better treatments for fungal infections that still kill millions of people annually7,8 and potentially other resistance-evasive antimicrobials, including those that have recently been shown to operate through supramolecular structures that target specific lipids9.

FDA Public Workshop Summary-Addressing Challenges in Inhaled Antifungal Drug Development.

Allergic bronchopulmonary aspergillosis and invasive fungal diseases represent distinct infectious entities that cause significant morbidity and mortality. Currently, administered inhaled antifungal therapies are unapproved, have suboptimal efficacy, and are associated with considerable adverse reactions. The emergence of resistant pathogens is also a growing concern. Inhaled antifungal development programs are challenged by inadequate nonclinical infection models, highly heterogenous patient populations, low prevalence rates of fungal diseases, difficulties defining clinical trial enrollment criteria, and lack of robust clinical trial endpoints. On September 25, 2020, the US Food and Drug Administration (FDA) convened a workshop with experts in pulmonary medicine and infectious diseases from academia, industry, and other governmental agencies. Key discussion topics included regulatory incentives to facilitate development of inhaled antifungal drugs and combination inhalational devices, limitations of existing nonclinical models and clinical trial designs, patient perspectives, and industry insights.

Neuroimaging of Cryptococcal Meningitis in Patients without Human Immunodeficiency Virus: Data from a Multi-Center Cohort Study.

BACKGROUND: A clearer understanding is needed about the use of brain MRI in non-HIV patients with cryptococcal meningitis. METHODS: Cerebral CT and MRI were studied in 62 patients in a multicenter study of cryptococcal meningitis in non-HIV patients. CT was performed in 51 and MRI in 44. MRI results are reported for the images read at NIH for 29 of the 44 patients. CT reports obtained from the original REDCap database were added to calculate the incidence of normal findings. RESULTS: CTs were read as normal in 24 of 51 (47%), MRIs were normal in 10% (three of 29). The most characteristic lesions of cryptococcal meningitis on MRI were small basal ganglia lesions representing dilated perivascular spaces in 24% and basal ganglia lesions with restricted diffusion (infarcts) in 38%. In the 18 patients who received contrast, contrast-enhancing lesions, likely representing masses of cryptococci and inflammatory cells, were found in the basal ganglia in 22% and elsewhere in the brain in 22%. Meningeal enhancement was seen in 56%, ependymal enhancement in 24%, and choroid plexus enhancement in 11%. Hydrocephalus was found in five (18%), though increased intacranial pressure was not detected. Suboptimal imaging (n = 6), lack of contrast administration (n = 11) and lack of follow-up, however, markedly limited the accurate assessment of abnormalities in multiple cases. CONCLUSION: MRI characteristics of non-HIV cryptococcal meningitis include hydrocephalus, meningeal and ependymal enhancement and basal ganglia lesions. Optimal imaging is, however, necessary to maximize the diagnostic and prognostic usefulness of MRI.

Coronavirus Disease 2019-Associated Pulmonary Aspergillosis: A Noninvasive Screening Model for Additional Diagnostics.

Background: Coronavirus disease 2019 (COVID-19)-associated pulmonary aspergillosis (CAPA) is likely underdiagnosed, and current diagnostic tools are either invasive or insensitive. Methods: A retrospective study of mechanically ventilated patients with COVID-19 admitted to 5 Johns Hopkins hospitals between March 2020 and June 2021 was performed. Multivariable logistic regression was used for the CAPA prediction model building. Performance of the model was assessed using the area under the receiver operating characteristic curve (AUC). Results: In the cohort of 832 patients, 98 (11.8%) met criteria for CAPA. Age, time since intubation, dexamethasone for COVID-19 treatment, underlying pulmonary circulatory diseases, human immunodeficiency virus, multiple myeloma, cancer, or hematologic malignancies were statistically significantly associated with CAPA and were included in the CAPA prediction model, which showed an AUC of 0.75 (95% confidence interval, .70-.80). At a screening cutoff of ≥0.085, it had a sensitivity of 82%, a specificity of 51%, a positive predictive value of 18.6%, and a negative predictive value of 95.3%. (The CAPA screening score calculator is available at www.transplantmodels.com). Conclusions: We developed a CAPA risk score as a noninvasive tool to aid in CAPA screening for patients with severe COVID-19. Our score will also identify a group of patients who are unlikely to have CAPA and who therefore need not undergo additional diagnostics and/or empiric antifungal therapy.

Food and Drug Administration Public Workshop Summary-Development Considerations of Antifungal Drugs to Address Unmet Medical Need.

Pressing challenges in the treatment of invasive fungal infections (IFIs) include emerging and rare pathogens, resistant/refractory infections, and antifungal armamentarium limited by toxicity, drug-drug interactions, and lack of oral formulations. Development of new antifungal drugs is hampered by the limitations of the available diagnostics, clinical trial endpoints, prolonged trial duration, difficulties in patient recruitment, including subpopulations (eg, pediatrics), and heterogeneity of the IFIs. On 4 August 2020, the US Food and Drug Administration convened a workshop that included IFI experts from academia, industry, and other government agencies to discuss the IFI landscape, unmet need, and potential strategies to facilitate the development of antifungal drugs for treatment and prophylaxis. This article summarizes the key topics presented and discussed during the workshop, such as incentives and research support for drug developers, nonclinical development, clinical trial design challenges, lessons learned from industry, and potential collaborations to facilitate antifungal drug development.

Invasive fungal infections after respiratory viral infections in lung transplant recipients are associated with lung allograft failure and chronic lung allograft dysfunction within 1 year.

BACKGROUND: Respiratory viral infections (RVI) are associated with chronic lung allograft dysfunction (CLAD) and mortality in lung transplant recipients (LTRs). However, the prevalence and impact of secondary invasive fungal infections (IFIs) post RVIs in LTRs have not been investigated. METHODS: We performed a single center retrospective study including LTRs diagnosed with 5 different respiratory viral pathogens between January 2010 to May 2021 and evaluated their clinical outcomes in 1 year. The risk factors of IFIs were evaluated by logistic regression. The impact of IFIs on CLAD stage progression/death was examined by Cox regression. RESULTS: A total of 202 RVI episodes (50 influenza, 31 severe acute respiratory syndrome coronavirus-2, 30 metapneumovirus, 44 parainfluenza, and 47 respiratory syncytial virus) in 132 patients was included for analysis. Thirty-one episodes (15%) were associated with secondary IFIs, and 27 occurred in LTRs with lower respiratory tract infection (LRTI; 28% from 96 LRTI episodes). Aspergillosis was the most common IFI (80%). LTRs with IFIs had higher disease severity during RVI episodes. In multivariable analysis, RVI with LTRI was associated with IFI (adjusted odds ratio [95% confidence interval (CI)] of 7.85 (2.48-24.9). Secondary IFIs were associated with CLAD stage progression/death after accounting for LRTI, pre-existing CLAD, intensive care unit admission, secondary bacterial pneumonia and underlying lung diseases pre-transplant with adjusted hazard ratio (95%CI) of 2.45 (1.29-4.64). CONCLUSIONS: This cohort demonstrated 15% secondary IFI prevalence in LTRs with RVIs. Importantly, secondary IFIs were associated with CLAD stage progression/death, underscoring the importance of screening for fungal infections in this setting.

Treatment recommendations for non-HIV associated cryptococcal meningoencephalitis including management of post-infectious inflammatory response syndrome.

Cryptococcal meningoencephalitis (CM) continues to cause major morbidity and mortality in a range of patients such as those immunosuppressed from HIV and with biologic immunosuppressants, including treatments of autoimmunity, malignancies, and conditioning regimens for transplantation. It is currently the most common cause of non-viral meningitis in the United States. Infections in previously healthy patients also develop with autoantibodies to granulocyte-macrophage colony stimulating factor or with monogenetic defects. In all populations, mortality and significant long-term morbidity occur in 30-50% despite therapy, and immune reconstitution and post-infectious inflammatory response syndromes complicate management. To help with these difficult cases, we present here a practical tutorial of the care of a range of patients with CM in the absence of HIV/AIDS.

Development of an Interferon-Gamma Release Assay (IGRA) to Aid Diagnosis of Histoplasmosis.

Establishing diagnosis of latent and active histoplasmosis is challenging. Interferon gamma-release assays (IGRAs) may provide evidence of latent and active infection. An enzyme-linked immunospot (ELISpot) assay was developed using yeast cell lysate (YCL) antigen prepared from a representative North American Histoplasma capsulatum strain. Assay parameters were optimized by measuring responses in healthy volunteers with and without Histoplasma infection. Assay performance as an aid for diagnosing histoplasmosis was assessed in a prospective cohort of 88 people with suspected or confirmed infection, and 44 healthy controls enrolled in two centers in North America (2013 to 2018). Antigen specificity of IFN-γ release was demonstrated using ELISpot and enzyme-linked immunosorbent assay (ELISA). Antigen-evoked, single-cell mRNA expression by memory T cells was shown using flow cytometry. The area under the receiver operating characteristic curve (AUC) was estimated at 0.89 (95% confidence interval [CI]: 78.5% to 99.9%). At optimal cutoff, sensitivity was 77.2% (95% CI: 54.6% to 92.2%) and specificity was 100% (95% CI: 89.7% to 100%). Sixteen of 44 healthy volunteers (36.4%) from a region of hyperendemicity had positive responses, suggesting detection of previously unrecognized (latent) infection. The ELISpot assay is sensitive and specific as an aid to diagnose H. capsulatum infection and disease, supporting proof of concept and further development.

Statistical inference on shape and size indexes for counting processes.

Single-index models have gained increased popularity in time-to-event analysis owing to their model flexibility and advantage in dimension reduction. We propose a semiparametric framework for the rate function of a recurrent event counting process by modelling its size and shape components with single-index models. With additional monotone constraints on the two link functions for the size and shape components, the proposed model possesses the desired directional interpretability of covariate effects and encompasses many commonly used models as special cases. To tackle the analytical challenges arising from leaving the two link functions unspecified, we develop a two-step rank-based estimation procedure to estimate the regression parameters with or without informative censoring. The proposed estimators are asymptotically normal, with a root-n convergence rate. To guide model selection, we develop hypothesis testing procedures for checking shape and size independence. Simulation studies and a data example on a hematopoietic stem cell transplantation study are presented to illustrate the proposed methodology.

Coronavirus Disease 2019-Associated Pulmonary Aspergillosis in Mechanically Ventilated Patients.

BACKGROUND: Coronavirus disease 2019 (COVID-19)-associated pulmonary aspergillosis (CAPA) occurs in critically ill patients with COVID-19. Risks and outcomes remain poorly understood. METHODS: A retrospective cohort study of mechanically ventilated adult patients with COVID-19 admitted to 5 Johns Hopkins hospitals was conducted between March and August 2020. CAPA was defined using composite clinical criteria. Fine and Gray competing risks regression was used to analyze clinical outcomes and, multilevel mixed-effects ordinal logistic regression was used to compare longitudinal disease severity scores. RESULTS: In the cohort of 396 people, 39 met criteria for CAPA. Patients with CAPA were more likely than those without CAPA to have underlying pulmonary vascular disease (41% vs 21.6%, respectively; P = .01), liver disease (35.9% vs 18.2%; P = .02), coagulopathy (51.3% vs 33.1%; P = .03), solid tumors (25.6% vs 10.9%; P = .02), multiple myeloma (5.1% vs 0.3%; P = .03), and corticosteroid exposure during the index admission (66.7% vs 42.6%; P = .005), and had lower body mass indexes (median, 26.6 vs 29.9 [calculated as weight in kilograms divided by height in meters squared]; P = .04). Patients with CAPA had worse outcomes, as measured by ordinal severity of disease scores, requiring longer time to improvement (adjusted odds ratio, 1.081.091.1; P < .001), and advancing in severity almost twice as quickly (subhazard ratio, 1.31.82.5; P < .001). They were intubated twice as long as those without CAPA (subhazard ratio, 0.40.50.6; P < .001) and had longer hospital stays (median [interquartile range], 41.1 [20.5-72.4) vs 18.5 [10.7-31.8] days; P < .001). CONCLUSION: CAPA is associated with poor outcomes. Attention to preventive measures (screening and/or prophylaxis) is warranted in people with high risk of CAPA.

Outcomes of SOT Recipients With COVID-19 in Different Eras of COVID-19 Therapeutics.

BACKGROUND: Few reports have focused on newer coronavirus disease 2019 (COVID-19) therapies (remdesivir, dexamethasone, and convalescent plasma) in solid organ transplant recipients; concerns had been raised regarding possible adverse impact on allograft function or secondary infections. METHODS: We studied 77 solid organ transplant inpatients with COVID-19 during 2 therapeutic eras (Era 1: March-May 2020, 21 patients; and Era 2: June-November 2020, 56 patients) and 52 solid organ transplant outpatients. RESULTS: In Era 1, no patients received remdesivir or dexamethasone, and 4 of 21 (19.4%) received convalescent plasma, whereas in Era 2, remdesivir (24/56, 42.9%), dexamethasone (24/56, 42.9%), and convalescent plasma (40/56, 71.4%) were commonly used. Mortality was low across both eras, 4 of 77 (5.6%), and rejection occurred in only 2 of 77 (2.8%) inpatients; infections were similar in hypoxemic patients with or without dexamethasone. Preexisting graft dysfunction was associated with greater need for hospitalization, higher severity score, and lower survival. Acute kidney injury was present in 37.3% of inpatients; renal function improved more rapidly in patients who received remdesivir and convalescent plasma. Post-COVID-19 renal and liver function were comparable between eras, out to 90 d. CONCLUSIONS: Newer COVID-19 therapies did not appear to have a deleterious effect on allograft function, and infectious complications were comparable.

Impact of Changes of the 2020 Consensus Definitions of Invasive Aspergillosis on Clinical Trial Design: Unintended Consequences for Prevention Trials?

BACKGROUND: Consensus definitions for the diagnosis of invasive fungal diseases (IFDs) were updated in 2020 to increase the certainty of IFD for inclusion in clinical trials, for instance by increasing biomarker cutoff limits to define positivity. To date, there is a paucity of data as to the impact of the revised definitions on clinical trials. METHODS: In this study, we sought to determine the impact of the new definitions on classifying invasive aspergillosis (IA), the most common invasive mold disease in immunocompromised patients. We reclassified 226 proven and probable IA cases plus 139 possible IFD cases in the Aspergillus Technology Consortium (AsTeC) and in an antifungal prophylaxis trial (BMT CTN 0101) using the new criteria. RESULTS: Fewer cases met the more stringent diagnostic 2020 criteria after applying the reclassification criteria to define probable IA. Of 188 evaluable probable cases, 41 (22%) were reclassified to 40 possible IA and 1 probable IFD. Reclassification to possible IFD occurred in 22% of hematologic malignancy (HM) patients, 29% of hematopoietic cell transplant (HCT) patients, and in no lung transplant (LT) patients. Date of diagnosis was established a median (range) of 3 (1-105) days later in 15% of probable IA cases using the new criteria. Applying the new definitions to the BMT CTN 0101 trial, the power to detect the same odds ratio decreased substantially. CONCLUSIONS: The updated IA consensus definitions may impact future trial designs, especially for antifungal prophylaxis studies.

Cryptococcus gattii Species Complex as an Opportunistic Pathogen: Underlying Medical Conditions Associated with the Infection.

The Cryptococcus gattii species complex has often been referred to as a primary pathogen due to its high infection frequency among apparently immunocompetent patients. In order to scrutinize the immune status of patients and the lineages of etiologic agents, we analyzed patient histories and the molecular types of etiologic agents from 135 global C. gattii cases. Eighty-six of 135 patients had been diagnosed as immunocompetent, although some of them had underlying medical issues, and 49 were diagnosed as immunocompromised with risk factors similar to those seen in Cryptococcus neoformans infection. We focused on the 86 apparently immunocompetent patients and were able to obtain plasma from 32 (37%) to analyze for the presence of autoantibodies against the granulocyte-macrophage colony-stimulating factor (GM-CSF) since these antibodies have been reported as a hidden risk factor for C. gattii infection. Among the 32 patients, 25 were free from any known other health issues, and 7 had various medical conditions at the time of diagnosis for cryptococcosis. Importantly, plasma from 19 (76%) of 25 patients with no recognized underlying medical condition showed the presence of GM-CSF autoantibodies, supporting this antibody as a major hidden risk factor for C. gattii infection. These data indicate that seemingly immunocompetent people with C. gattii infection warrant detailed evaluation for unrecognized immunologic risks. There was no relationship between molecular type and underlying conditions of patients. Frequency of each molecular type was related to its geographic origin exemplified by the overrepresentation of VGIV in HIV-positive (HIV+) patients due to its prevalence in Africa. IMPORTANCE The C. neoformans and C. gattii species complex causes cryptococcosis. The C. neoformans species complex is known as an opportunistic pathogen since it primarily infects immunocompromised patients. C. gattii species complex has been referred to as a primary pathogen due to its high infection frequency in apparently immunocompetent people. We analyzed 135 global cases of C. gattii infection with documented patient history. Eighty-six of 135 patients were originally diagnosed as immunocompetent and 49 as immunosuppressed with similar underlying conditions reported for C. neoformans infection. A significant number of C. gattii patients without known underlying conditions possessed autoantibodies against granulocytes-macrophage colony-stimulating factor (GM-CSF) in their plasma, supporting the presence of GM-CSF antibodies as a hidden risk factor for C. gattii infection. No relationship was found between C. gattii lineages and the underlying conditions except for overrepresentation of the molecular type VGIV among HIV+ patients due to the prevalence of VGIV in Africa.

High-value laboratory testing for hospitalized COVID-19 patients: a review.

We present here an evidence-based review of the utility, timing, and indications for laboratory test use in the domains of inflammation, cardiology, hematology, nephrology and co-infection for clinicians managing the care of hospitalized COVID-19 patients. Levels of IL-6, CRP, absolute lymphocyte count, neutrophils and neutrophil-to-lymphocyte ratio obtained upon admission may help predict the severity of COVID-19. Elevated LDH, ferritin, AST, and d-dimer are associated with severe illness and mortality. Elevated cardiac troponin at hospital admission can alert clinicians to patients at risk for cardiac complications. Elevated proBNP may help distinguish a cardiac complication from noncardiac etiologies. Evaluation for co-infection is typically unnecessary in nonsevere cases but is essential in severe COVID-19, intensive care unit patients, and immunocompromised patients.

Aspergillus lentulus: An Under-recognized Cause of Antifungal Drug-Resistant Aspergillosis.

Aspergillus lentulus is a drug-resistant species that is phenotypically similar to A. fumigatus. It was discovered as a cause of azole-breakthrough disease, consistent with in vitro resistance. Clinical labs can misidentify the species as fumigatus based on phenotypic typing. We describe 4 recent cases and provide a brief review.

Recurrent Events Analysis With Data Collected at Informative Clinical Visits in Electronic Health Records.

Although increasingly used as a data resource for assembling cohorts, electronic health records (EHRs) pose many analytic challenges. In particular, a patient's health status influences when and what data are recorded, generating sampling bias in the collected data. In this paper, we consider recurrent event analysis using EHR data. Conventional regression methods for event risk analysis usually require the values of covariates to be observed throughout the follow-up period. In EHR databases, time-dependent covariates are intermittently measured during clinical visits, and the timing of these visits is informative in the sense that it depends on the disease course. Simple methods, such as the last-observation-carried-forward approach, can lead to biased estimation. On the other hand, complex joint models require additional assumptions on the covariate process and cannot be easily extended to handle multiple longitudinal predictors. By incorporating sampling weights derived from estimating the observation time process, we develop a novel estimation procedure based on inverse-rate-weighting and kernel-smoothing for the semiparametric proportional rate model of recurrent events. The proposed methods do not require model specifications for the covariate processes and can easily handle multiple time-dependent covariates. Our methods are applied to a kidney transplant study for illustration.