Background:Clostridioides difficile infections (CDIs) present and are transmitted in both community and healthcare settings. Patients who become colonized or infected during hospitalization may be discharged into the community. Asymptomatic spread and/or community-based transmission have also been posited as alternative sources for healthcare-onset CDI cases. The objective of our study was to determine whether individuals are at greater risk for developing a CDI if they have a family member that spent time hospitalized in the prior 90 days, even if the hospitalized family member had no prior diagnosis of CDI. Methods: We conducted a retrospective cohort study using the Truven Marketscan database from 2001 through 2017; both commercial claims and Medicare supplemental data were included. We categorized enrollees by age, sex, month, year, exposure to a family member with CDI, hospitalization, or high- or low-risk antibiotic use in the prior 90 days. We then subdivided these groups based on the total amount of time that other family members spent hospitalized in the prior 90 days: ≤4 days, 5–10, 11–20, 21–30, 41–50 or >50 days. Within each subgroup, we computed the incidence of CDI. We then used a stratified regression model (log-linear quasi-Poisson) to estimate the incidence of CDI in each enrollment bin. Finally, we repeated our analysis using all CDI cases, CDI cases with no prior CDI in the family, and cases without prior hospitalization. Results: Over the 17-year study period, >5.1 billion enrollment months were represented in our dataset. We identified 224,818 cases of CDI, 223,744 cases without prior CDI in a family member and 164,650 CDI cases where the case patient had no prior hospitalization. Table 1 depicts the estimated risk (incident rate ratios) associated with the amount of time that other family members spent hospitalized in the prior 90 days. There is a very clear dose–response curve, and the relative risk for CDI increase as the amount of time other family members spent hospitalized increased. Other risk factors included prior hospitalization, low- and high-risk antibiotics, age, female sex and exposure to a family member with CDI. Conclusions: Having a family member who has been hospitalized in the prior 90 days significantly increases the risk for CDI, even if the family member did not have CDI. The total amount of time other family members spent in the hospital is positively associated with the level of risk.

Funding: CDC Modeling Infectious Diseases (MInD) in Healthcare Network

Disclosures: None

Background: Simulations based on models of healthcare worker (HCW) mobility and contact patterns with patients provide a key tool for understanding spread of healthcare-acquired infections (HAIs). However, simulations suffer from lack of accurate model parameters. This research uses Microsoft Kinect cameras placed in a patient room in the medical intensive care unit (MICU) at the University of Iowa Hospitals and Clinics (UIHC) to obtain reliable distributions of HCW visit length and time spent by HCWs near a patient. These data can inform modeling efforts for understanding HAI spread. Methods: Three Kinect cameras (left, right, and door cameras) were placed in a patient room to track the human body (ie, left/right hands and head) at 30 frames per second. The results reported here are based on 7 randomly selected days from a total of 308 observation days. Each tracked body may have multiple raw segments over the 2 camera regions, which we “stitch” up by matching features (eg, direction, velocity, etc), to obtain complete trajectories. Due to camera noise, in a substantial fraction of the frames bodies display unnatural characteristics including frequent and rapid directional and velocity change. We use unsupervised learning techniques to identify such “ghost” frames and we remove from our analysis bodies that have 20% or more “ghost” frames. Results: The heat map of hand positions (Fig. 1) shows that high-frequency locations are clustered around the bed and more to the patient’s right in accordance with the general medical practice of performing patient exams from their right. HCW visit frequency per hour (mean, 6.952; SD, 2.855) has 2 peaks, 1 during morning shift and 1 during the afternoon shift, with a distinct decrease after midnight. Figure 2 shows visit length (in minutes) distribution (mean, 1.570; SD, 2.679) being dominated by “check in visits” of <30 seconds. HCWs do not spend much time at touching distance from patients during short-length visits, and the fraction of time spent near the patient’s bed seems to increase with visit length up to a point. Conclusions: Using fine-grained data, this research extracts distributions of these critical parameters of HCW–patient interactions: (1) HCW visit length, (2) HCW visit frequency as a function of time of day, and (3) time spent by HCW within touching distance of patient as a function of visit length. To the best of our knowledge, we provide the first reliable estimates of these parameters.

Funding: None

Disclosures: None

Background. Colonization pressure at the unit level is known to be a risk factor for Clostridioides difficile infections in hospitals. Because C. difficile colonization is not routinely detected in clinical practice, only patients identified as having C. difficile infection (CDI) are included in these pressure calculations. We used data from the University of Iowa Hospitals and Clinics (UIHC) to determine whether highly local CDI pressure, due to patients in nearby rooms, is more strongly correlated with CDI than unit-level CDI pressure. Methods: We designed a base logistic regression model using variables known to be risk factors for CDI: age, antibiotic/gastric acid suppressor use, low albumin, prior hospitalization, comorbidities. To the base model, we add 2 measures, mean colonization pressure (MCP) and sum colonization pressure (SCP) of CDI at the unit level to obtain new models. To the base model, we also added CDI colonization pressure by considering CDI cases at different distance thresholds from the focal patient. Distances between patient rooms were extracted from hospital floor plans. Results: Adding unit-level CDI colonization pressures to the base model improved performance. However, adding CDI colonization pressures due to roommates and due to patients at different distances improved the model much more (Table 1). The top (resp. bottom) row shows in-sample (resp. out-of-sample) C-statistics for the base model, the base model with unit-level MCP, the base model with roommate MCP, and the base model with MCP from patients are different distances added as separate features. C-statistics for the base model and the base model with unit CDI pressure (SCP and MCP) are compared in Fig. 1 with C-statistics from the base model with CDI pressure from patients at distances D = 0, 1, 2, 3, 4, 5, 10, 15, 20 hops (1 hop = 5–6 meters). Conclusions: Our results support the hypothesis that unit CDI colonization pressure is a risk factor for CDI. However, by incorporating spatially granular notions of distances between patients in our analysis, we were able to demonstrate that the true source of CDI pressure at the UIHC is almost exclusively attributable to roommates and patients in adjacent rooms.

Funding: None

Disclosures: None

Background: Patient sharing between hospitals has long been known to be a contributor to the regional transmission of hospital-acquired infections. This inter–healthcare-facility connectedness suggests that regional, as opposed to local, control and surveillance strategies should be favored. However, the absence of easily and universally accessible patient transfer data hampers researchers and public health agencies who wish to build accurate network models of interfacility transmissions. Medicare data offer only a biased subsample of the full patient transfer network, but because it is widely available, this data source has historically been used for inference and simulation studies. Objective: The purpose of this study was to determine whether Medicare data could successfully be recalibrated to more closely resemble 100% inpatient capture data. Methods: We used data from the Healthcare Cost and Utilization Project (HCUP) to construct 100% capture and Medicare-only patient sharing networks among hospitals in Iowa, Wisconsin, and Nebraska. We used matrix decomposition techniques on the Medicare-only networks along with hospital characteristics from the American Hospital Association (AHA) data for feature construction in a truncated Poisson regression model, and we used Monte Carlo integration to obtain predicted values. These predicted values served as calibrated Medicare-only networks. We split the patient transfer data into training and testing sets and computed the mean squared prediction error (MSPE) for the testing data. We also built an individual based model (IBM) using HCUP and AHA data to perform epidemic simulations that depended on a matrix of patient transfer rates between hospitals. We then compared epicurves from these IBMs resulting from 100% capture networks, Medicare-only networks, and our calibrated networks. Results: Our calibrated networks reduced the MSPE with respect to Medicare-only networks by 84%, 47%, and 88% for Iowa, Wisconsin, and Nebraska, respectively. Although the epicurves from Medicare-only networks differed considerably from that from 100% capture networks, our calibrated networks retained high fidelity to the curves obtained from 100% capture networks. Conclusions: Medicare-only networks greatly underestimate the number of patients transferred between hospitals. Our approach allows us to use Medicare data to estimate networks when 100% inpatient capture is unavailable.

Funding: None

Disclosures: None

Background:Clostridioides difficile is a leading cause of healthcare-associated infections, and greater healthcare exposure is a primary risk factor for Clostridioides difficile infection (CDI). Longer hospital stays and greater CDI pressure, both at the hospital level and the level, have been linked to greater risk. In addition, symptoms associated with healthcare-associated CDI often do not present until a patient has been discharged. Our study objective was to estimate the extent to which exposure to different types of healthcare settings (eg, prior hospitalization, emergency department [ED], outpatient or long-term care) increase risk for hospital-onset CDI. Methods: We conducted a case-control study using the Truven Marketscan Commerical Claims and Medicare Supplemental databases from 2001 to 2017. Case patients were selected as all inpatient visits with a secondary diagnosis of CDI and no previous CDI diagnosis in the prior 90 days. Controls were selected from all inpatient admissions without any CDI diagnosis during the current admission or prior 90 days. A logistic regression model was used to estimate risk associated with prior healthcare exposure. Indicators were created for prior exposure to different healthcare settings: separate indicators were used to indicate transfer, exposure to that setting in the prior 1–30 days, 31–60 days and 61–90 days. Separate indicators were created for prior hospitalization, ED, outpatient clinic, nursing home or long-term care facilities (LTCFs), psychiatric or substance-abuse facility or other outpatient facility. We also included an indicator for prior exposure to a family member with CDI and prior outpatient antibiotics. Results: Estimates for selected variables (odds ratios) are presented in Table 1. Prior hospitalization, ED visits, outpatient clinics, nursing home and LTCFs were all associated with increased risk of secondary diagnosed CDI. Prior hospitalization and nursing home/LTCF conveyed the greatest risk. In addition, a ‘dose-–response’ relationship occurred for each of these exposure settings, with exposure nearest the admission date having the largest risk. Prior exposure to psychiatric , substance abuse, or other outpatient facilities were not risk factors for CDI. Having a family member with prior CDI and both low-risk and high-risk outpatient antibiotics were associated with increased risk. These factors also exhibited a ‘dose–response’ pattern. Conclusions: Exposure to various healthcare settings significantly increased risk for secondary CDI. Prior healthcare exposures occurring nearest to the point of admission conveyed the greatest risk. These results suggest that many hospital-associated CDI cases attributed to a current hospital stay may actually be acquired from prior healthcare settings.

Funding: CDC Modeling Infectious Diseases (MInD) in Healthcare Network

Disclosures: None

Background: Hospital-acquired infections are commonly spread through the movement of healthcare professionals (HCPs). Computational simulations provide a powerful tool for understanding how HCP behavior contributes to these infections, but how well they reflect the real world rests on a number of critical parameters. Our goal is to provide accurate, fine-grained estimates of real HCP movement and interaction parameters suitable for simulating the potential spread of pathogens over different types of inpatient facilities. Methods: We obtained a commercial data set with 44 million deidentified elements compiled from >27,000 HCPs from >30 job types. The data were collected over 27 months from >20 facilities of varying size using a proprietary electronic sensor system. Each observation recorded an HCP visiting 1 of 12,000 rooms (38% being patient rooms) and consisted of the entry and exit time stamps, hand hygiene behavior, and for many rooms, their (x, y) geometric coordinates within the facility. From these data, we can reconstruct the behavior (including location and hand-hygiene adherence) of each instrumented HCP across multiple shifts. Results: Distributions describing various aspects of HCP behavior (eg, arrival rates and dwell times) were derived using HCP job function, department or unit assignment, type of shift (day vs night), time of day, facility size, and staffing of facility. In a similar fashion, we constructed HCP cross-table transition probabilities using job type, room type, department type, unit type, and facility type. These distributions were used to generate reasonable HCP movement and behavior patterns in a simulation environment. Distributions of dwell time were, for the most part, heavy tailed, but they varied by type of job and facility: dwell times over all facilities, job types, and room types averaged ∼339 seconds (SD, 495 seconds), with a mean of maximums by job type of ∼37,168 seconds. However, these distributions differ within job type but across facilities (ie, nurses in 1 facility averaged 397 seconds, but 277 seconds in another) and within facility but across job type. For example, physicians averaged 292 seconds, whereas nurses averaged 397 seconds and physical therapists averaged 861 seconds. Conclusions: Our results provide a unique resource for disease modelers who wish to build meaningful simulations of the transmission of hospital-acquired infections. The scale and diversity of the data gave us the unique capability to provide, with confidence, distinct parameter sets for different types and sizes of healthcare facilities across a wide range of situations.

Funding: None

Disclosures: None

Background: In the field of public health, network models are useful for understanding the spread of both information and infectious diseases. Collecting network data requires determining network boundaries (ie, the entities selected for data collection). These decisions, if not made carefully, have potential outsized downstream effects on study findings. In practice, collaboration and coordination between healthcare organizations are often dictated by historical or geopolitical boundaries (eg, state or county boundaries), which may distort the underlying network under study, and thereby affect the reliability and/or accuracy of the network model. Objective: We compared natural communities in a patient-sharing network with those induced by geopolitical boundaries. Methods: Using data from the Healthcare Cost and Utilization Project (HCUP), we constructed a patient-sharing network among hospitals in California, splitting the data into a training set and a holdout set. We performed edge-betweenness clustering on the training set, and with the holdout set, we compared the resulting partition with the partition by counties using modularity. We also clustered contiguous counties that might function more cohesively together than individually. We performed spatially constrained hierarchical clustering on the network constructed from total patient flow between pairs of counties. The results were again compared via modularity on the holdout set to the county partition. Lastly, we built an individual-based model (IBM) using HCUP and American Hospital Association data to perform epidemic simulations. For each of several counties, we implemented this model to estimate the proportion of patients infected over time. We then reran the individual-based model using the entire state while dividing the results into corresponding counties. Results: In total, 680,485 patients transferred between 374 hospitals in 55 counties from 2003 to 2011. The out-of-sample modularity for the edge-betweenness clustering partition was 464% higher than that of the county partition. Aggregating the counties into half as many contiguous clusters was 319% higher, and aggregating them into 6 clusters was 489% higher (Fig. 1). The epicurves from the individual-based model ranged from small to significant deviations between state versus county boundaries (Fig. 2) . Conclusions: Collecting network data using externally imposed boundaries may lead to inaccurate network models. For example, counties serve as a poor proxy for their underlying communities, resulting in poor overall disease spread simulation results when county boundaries are allowed to drive network construction. These issues should be considered when building coordination partnerships such as the Accountable Communities for Health.

Funding: None

Disclosures: None

Background: Mobility patterns of healthcare workers (HCWs) (ie, the spatiotemporal distribution of patient rooms they visit) have a significant impact on the spread of healthcare acquired infections (HAIs). Objective: In this project, we used fine-grained data from a sensor deployment at the medical intensive care unit (MICU) in the University of Iowa Hospitals and Clinics (UIHC) to study the mobility patterns of HCWs and their impact on HAI spread. Methods: We analyzed 10 days of data from a 20-bed MICU sensor deployment. For parameters t1 and t2, each pair of rooms i and j is assigned a weight W(i, j) representing the number of times an HCW spends at least t1 seconds in room i followed by at least t1 seconds in room j, within t2 seconds of each other. W(i, j) is a measure of HCW traffic going from room i to room j; we study the correlation between W(i, j) and the distance between rooms i and j. Additionally, we perform 2 disease-spread simulations: (1) a base simulation, obtained by replaying observed HCW mobility traces and (2) a perturbed simulation, which is the same as the base simulation, except that we replace each HCW who visits a room by a random available HCW. Thus, the perturbed simulation removes correlations in the observed HCW mobility traces. Results: We computed W(i, j) for all room pairs i, j for parameters t1 = 30 seconds and t2 = 1,800 and 3,600 seconds. For nurses, there was a strong negative correlation of between pairwise room distance and the weights W(i, j) (−0.768 for t2 = 1,800; −0.711 for t2 = 3,600), The more distant 2 rooms were, the less they shared nurse traffic. This was not true for physicians (correlation = −0.027 for t2 = 1,800; −0.014 for t2 = 3,600). Figure 1 shows a weight versus distance scatter plot for nurses for t1 = 30 and t2 = 1,800. This spatial correlation has positive implications for disease spread; the base simulation, which preserves these spatial correlations, has between 12% and 55% fewer mean infected patients (>100 replicates) for different simulation parameters compared to the perturbed simulation. Conclusions: Our results, based on fine-grained data, show a “naturally emerging” cohorting behavior of nurses, where nurses are more likely to visit rooms close to each other within a 30–60 minute time window, than rooms further away. Through simulations, this behavior provides substantial protection against disease spread.

Funding: None

Disclosures: None