People with epilepsy who use multiple hospitals; prevalence and associated factors assessed via a health information exchange

Abstract

Objective: Hospital crossover occurs when people seek care at multiple hospitals, creating information gaps for physicians at the time of care. Health information exchange (HIE) is technology that fills these gaps, by allowing otherwise unaffiliated physicians to share electronic medical information. However, the potential value of HIE is understudied, particularly for chronic neurologic conditions like epilepsy. We describe the prevalence and associated factors of hospital crossover among people with epilepsy, in order to understand the epidemiology of who may benefit from HIE.

Methods: We used a cross-sectional study design to examine the bivariate and multivariable association of demographics, comorbidity, and health service utilization variables with hospital crossover, among people with epilepsy. We identified 8,074 people with epilepsy from the International Classification of Diseases, Ninth Revision (ICD-9) codes, obtained from an HIE that linked seven hospitals in Manhattan, New York. We defined hospital crossover as care from more than one hospital in any setting (inpatient, outpatient, emergency, or radiology) over 2 years.

Results: Of 8,074 people with epilepsy, 1,770 (22%) engaged in hospital crossover over 2 years. Crossover was associated with younger age (children compared with adults, adjusted odds ratio [OR] 1.4, 95% confidence interval [CI] 1.2-1.7), living near the hospitals (Manhattan vs. other boroughs of New York City, adjusted OR 1.6, 95% CI 1.4-1.8), more visits in the emergency, radiology, inpatient, and outpatient settings (p < 0.001 for each), and more head computerized tomography (CT) scans (p < 0.01). The diagnosis of "encephalopathy" was consistently associated with crossover in bivariate and multivariable analyses (adjusted OR 2.66, 95% CI 2.14-3.29), whereas the relationship between other comorbidities and crossover was less clear.

Significance: Hospital crossover is common among people with epilepsy, particularly among children, frequent users of medical services, and people living near the study hospitals. HIE should focus on these populations. Further research should investigate why hospital crossover occurs, how it affects care, and how HIE can most effectively mitigate the resultant fragmentation of medical records.

Keywords: Epidemiology; Epilepsy; Health information exchange; Health information technology; Health services research.

Figure 1. Patient timelines in a de-identified data set

The timelines for three example patients are represented in green, blue, and red. For each patient, the date of the initial diagnosis of 345.x or 780.39 was used as a start date to begin observation. However, use of anonymized, date-shifted dates prevents knowledge of when in “Study Year 1” this date actually occurred. Thus we were able to follow each patient only for two subsequent years. Note that each patient’s year 1 does not typically correspond to a study year, and that a given patient’s year 1 is typically not the same as another patient’s year 1.

Figure 2. Age Distribution of 8074 People with Epilepsy and the Subgroup of 1770 (22%) with Hospital Crossover

Histogram of age versus number of people with epilepsy who visited any of the seven Manhattan hospitals in the NYCLIX network over one year. Bin width is 2 years. Each bar is color coded to show the proportion of patients in each bin that had hospital crossover (blue) versus those who did not (orange). There are two peaks in the prevalence of epilepsy and in the prevalence of hospital crossover: one in children, and another in mid-adulthood (35-55 years of age). “Hospital crossover” means “visited more than one of the seven NYCLIX hospitals over two years”.

Figure 3. Percentage with Hospital Crossover versus Frequency of Health Service Utilization over Two Years for Six Services, among 8074 People with Epilepsy

Each dotplot shows the percentage of people with hospital crossover (x-axis) among those who used each service a specified number of times (y-axis), for six services (panels). For each plotted point, grey text provides the number of people with epilepsy, the number with hospital crossover, and the calculated percentage. For example (top left, topmost dot), 63% (198 of 312) of the people with epilepsy who visited the ED eleven or more times had visited two or more hospitals. The group with the most utilization for each type of service is highlighted in red. The visual trend suggesting more crossover at higher levels of utilization was statistically significant for ED visits (ρ=0.99 [95%CI 0.89, 1] p=0.001, Cochran-Armitage test for trend), inpatient visits (ρ=0.94 [0.34, 1] p=0.02), radiology days (ρ=0.98 [0.73, 1] p=0.003), and head CTs (ρ=0.99 [0.82, 1] p=0.002). Hospital crossover was associated with one or more outpatient visits, compared to none (unadjusted OR 4.9 [4.26, 5.64], p<0.001, Chi-square test), though the trend was not significant for more outpatient visits (ρ=0.86 [-0.08, 0.99] p=0.06). 20-30% of people had hospital crossover at each level of MRI use, without a clear trend (ρ=0.43 [-0.73, 0.95] p=0.47). “Hospital crossover” means “visited more than one of the seven NYCLIX hospitals over two years”.

Figure 4. Prevalence of Comorbidities among 6304 People with Epilepsy With Hospital Crossover and 1770 Without Hospital Crossover

The prevalence (expressed as a percentage) of each of 33 comorbidities is indicated by an ex (x) for people with hospital crossover and a circle (•) for those without. The symbols are colored according to statistical significance, grey for not significant, black for significant (chi-square test, p < (0.05 / 33), i.e. Bonferroni correction for multiple comparisons). The comorbidities are divided thematically into neurologic (top panel), psychiatric (middle panel), and medical (bottom panel).