Medication reconciliation, a process by which a provider obtains and documents a thorough medication history with specific attention to comparing current and previous medication use, can prevent medication-related errors and harm. An important barrier to effective medication reconciliation is the unreliability of patients' own reports about their medication use and incomplete provider history-taking. The James J. Peters VA (JJP VA) is an original participating facility in the Bronx RHIO, a health information exchange organization that in October 2008 implemented secure clinical data information exchange among the largest healthcare providers in the Bronx. JJP VA providers can access key clinical data, including medication use, from non-VA facilities with patients' consent. We posit that routine access to medication information from outside the VA has the potential to prevent harmful drug discrepancies and combinations, to prevent adverse drug events (ADEs), and reduce inpatient costs in veterans.
1) To adapt a medication reconciliation process to include information from a regional health information exchange (RHIO), 2) To conduct a controlled trial of a medication reconciliation process at the time of hospital admission either enhanced or not enhanced with data from a regional health information exchange, and examine effects on transition drug risk, adverse drug events, and total inpatient costs, and 3) To identify system and provider factors that impede and facilitate adoption of the information exchange tool for routine use by VA providers.
We will conduct qualitative and quantitative studies. First we will perform cognitive task analysis (CTA) to examine providers' decision-making and to map RHIO tool functions while providers interface with the tool, with interview questions to identify factors that facilitate or hinder use. Second we will enroll patients admitted to 2 inpatient units at the JJP VA, and assign them to receive usual medication reconciliation or RHIO-enhanced medication reconciliation with a notification to physicians about the patient's non-VA services that can be viewed in the RHIO. We will measure transition drug risk and ADEs by medical record review and patient interview, and measure inpatient costs with administrative cost data from national VA datasets. Third, we will conduct CTA with pharmacists and physicians at the end of the quantitative study who are expert in use of the tool to characterize differences in cognitive processes from prior to the quantitative study. This will supply information on system and provider factors that impede and facilitate adoption of the tool for routine use, and on potential improvements in the tool.
Aim 1 -- Development and refinement of intervention (qualitative).
Findings: Structured observations of providers' decision-making while using CPRS and the HIE tool indicated that use of HIE could feasibly be incorporated into the medication reconciliation process for veterans admitted to a VA hospital. This finding led to development of a study-specific medication reconciliation procedural guide that included instructions for accessing the systems, time-saving tips, taking a medication history with patients, and documenting findings. This formative work supported testing the use of HIE to improve medication prescribing outcomes in a controlled quantitative trial (Aim 2) and examining barriers to use of the HIE in a qualitative study (Aim 3).
Aim 2 -- Controlled trial of an HIE-enhanced medication reconciliation intervention.
Description of enrollees. 188 intervention and 193 control patients were enrolled who were admitted to one of 4 units (3 medical/surgical and 1 psychiatric unit) at the JJP (Bronx) VA. Overall 97% of enrollees were male, and the average age was 61 years. 63% self-identified as black race and 27% self-identified as Hispanic ethnicity. 36% had a service-connected condition. There were no statistically significant differences in demographic characteristics between intervention and control groups. Overall 68% were admitted to a medical/surgical unit and 32% were admitted to an inpatient psychiatric unit. There was a trend toward more intervention patients admitted to a psychiatric unit (36%) than controls (28%; p=.07).
Findings on availability of information to detect discrepancies in the HIE for intervention patients (potency of intervention). Among intervention patients, 19.8% and 24.2% had non-VA inpatient and outpatient medication data, respectively, available in the Bronx RHIO. This frequency increased to 29.4% during a time when a pharmacy benefits company temporarily participated in the RHIO (Surescripts). It is unfortunate that this was only a temporary participation because 13% of discrepancies in intervention patients were rated as detectable solely as a result of RHIO access during Surescripts participation, as opposed 2% at other times (p=0.007). Overall, there were 13 discrepancies detected solely as a result of RHIO access in the intervention group, of which 7 had potential to harm, extrapolating to 6.9 preventable discrepancies and 3.7 opportunities to prevent harm per 100 patients.
Findings on comparison between intervention and control patients. There were 1.25 and 1.13 discrepancies recorded per intervention and control patient, respectively (p=.44). 31% of discrepancies had potential to harm (no group difference). More discrepancies with potential to harm were unintentional in intervention than control patients (65% vs. 50%; p=.06), indicating that the HIE enhanced the potential for medication reconciliation to prevent unintentional harmful discrepancies as compared to control. However, in unadjusted analyses there was no difference between intervention and control in our outcomes of transition drug risk score (primary outcome), adverse drug events (secondary outcome), or medication prescribing quality.
Aim 3 -- Summative qualitative evaluation of intervention.
Findings: Participants consisted of 8 primary care physicians, 2 specialists, 1 pharmacist, and 1 physician assistant. Seven had prior experience using the RHIO and 5 were naive users. Themes were identified as follows: Many participants wished for more comprehensive non-VA data in the RHIO (e.g. outpatient medications; discharge summaries) to justify the time for using it. Some participants wished for improved integration between the current entirely separate CPRS and RHIO interfaces. Some participants suggested usability improvements to the RHIO health record itself, including permitting formatted cutting and pasting into CPRS. Some participants indicated that misremembered RHIO passwords prevented regular use. Some participants indicated barriers outweighed usefulness and would not use the RHIO routinely for care. Others felt that the non-VA information in the RHIO was important and would use it despite the barriers.
Our study suggests that HIE had a small impact on improving medication prescribing and reducing adverse events in hospitalized veterans. There are several possible reasons for this. First, a minority of veterans with known non-VA service use had non-VA medication information available through HIE that providers had not already obtained by performing a thorough non-VA medication history directly with the patient. Second, the HIE that we used, though mature, had substantial gaps in completeness of non-VA prescribing information, especially outpatient prescriptions. This is because much non-VA outpatient prescribing information is the proprietary property of pharmacy benefits management companies such as Surescripts and in this case this information was unavailable to HIE users for most of the study. Third, medication reconciliation itself is not a "high potency" intervention, since only about 5% of identified medication discrepancies result in any harm, and usually the harm is small. Thus the vast majority of medication discrepancies identified by medication reconciliation have little potential to affect the health of the patient (e.g., a discrepancy in dosage of vitamin D), and for this reason providers often ignore or give low priority to the findings of medication reconciliation in their decision-making. Of note, our study is concordant with other studies that show relatively small direct health benefits from HIE.
Nevertheless, there are potentially harmful medication problems that are preventable only because of HIE access (estimated from our study at 3.7 per 100 veteran admissions). In addition, medication information is only one type of information that is becoming available in HIEs that may benefit veterans, such as diagnoses, labs, reports, clinical summaries, and clinical alerts. This supports continuing efforts to test and implement HIE for VA patients and providers.
In sum, health information exchange (HIE) has the potential to benefit veterans whose care is fragmented between VA and non-VA systems. Our findings from this project provide an understanding of the effect of real-time, actual-practice review of veteran's non-VA medication use -- enabled by a HIE tool -- on adverse drug events in veterans. The importance of study findings, both negative and positive, is underscored by the acceptance of 2 project-based abstracts at the 2015 VA HSRD annual meeting and 1 project-based abstract accepted at the 2015 AMIA annual meeting.
We conclude that additional work is required to achieve optimal feasibility and effectiveness of incorporating HIE access and use into every day work flow for VA providers. To improve use, HIE tools should minimize steps for access and viewing, optimize integration with CPRS, and maximize capture of non-VA data to justify providers' viewing time. In addition, VA dual care policies should be updated to provide guidance about VA providers' access and viewing of non-VA health information available in HIEs.
- Boockvar KS, Ho W, Pruskowski J, DiPalo KE, Wong JJ, Patel J, Nebeker JR, Kaushal R, Hung W. Effect of health information exchange on recognition of medication discrepancies is interrupted when data charges are introduced: results of a cluster-randomized controlled trial. Journal of the American Medical Informatics Association : JAMIA. 2017 Nov 1; 24(6):1095-1101.
- O'Kula SS, Gottesman E, Jones S, Signor D, Hung WW, Boockvar KS. Transitional Care Outcomes in Older Spanish-English Bilingual Veterans. Journal of the American Geriatrics Society. 2016 May 1; 64(5):1132-3.
- Hung WW, Morano B, Moodhe N, Boockvar K. Regional Health Information Organization (RHIO): its potential uses to improve veteran health care. Federal practitioner : for the health care professionals of the VA, DoD, and PHS. 2011 Jan 1; 28(3):33-36.
- Hung WW, Morano B, Moodhe N, Boockvar K. RHIOs have potential to improve healthcare of older adults. [Newsletter]. 2011 Jun 2; 2(3).