Bhatia, H., Patel, N., Ivory, C., Stewart, P., Unertl, K. and Lehmann, C. (Feb, 2018) Medication compliance in pediatric inpatients—Time to the first dose. Online Journal of Nursing Informatics (OJNI), 22 (1), Available at http://www.himss.org/ojni
Objective: For most medical conditions, reducing the time between intent to treat and actual treatment is beneficial. The goal of this study was to determine the current state of the medication process at a university-affiliated children’s hospital. We also intended to investigate variations in order-to-administration intervals according to hospital location, medication, scheduled time, and patient age.
Method: We used the enterprise data warehouse (EDW) to collect medication process data using ordering, scheduling, and administration timestamps. We calculated the intervals for medication process components (o rdering to d ispensing and to a dministration) and analyzed the respective distributions.
Results: We identified an association of the medication process intervals with the order-type, verification requirements, patient unit, Anatomical Therapeutic Chemical (ATC) class of the ordered medication, scheduled hour, and patient age.
Conclusion: Meaningful information can be obtained from the analysis of medication process timestamps and computed intervals identifying areas for improvement. Institution-wide analytics of EDW repository data may measure the “health” of the medication process.
Medication not taken, or taken in an incorrect fashion, will negate the anticipated benefit to the patient. Thus, medication compliance is critical for both acute and chronic conditions. In pediatric inpatient settings, medications are generally administered by the nursing staff (or occasionally by parents). Compliance is therefore less dependent on patients and more dependent on the “ health” of the medication dispensing and administration process (Scharnweber, Mollenkopf, Fackler, Dover, & Lehmann, 2013). For most medical conditions, reducing the time between intent to treat and actual treatment is beneficial. Compliance in the medication process involves dispensing, delivering, and administering the treatment (as dose, form, route, etc.) and when it was intended by the provider (Johnson, Lehmann, & Council on Clinical Information Technology of the American Academy of Pediatrics, , 2013).
There is a lack of literature on compliance in pediatric inpatient settings (Bhatia et al., n.d.). However, there have been some studies on the related topic of medication errors (Slight et al., 2015). Kaushal (2001) pointed out that the medication “errors with potential for harm occurred most often in the youngest, most vulnerable patients cared for in the NICU (p. 2118)” and described contributory factors that included delivery systems, human interactions (with the system and other individuals), and work environment ( Lehmann & Council on Clinical Information Technology, 2015). These factors also have influence on the elapsed time for steps in the medication process.
In this paper, a high-level (IRB- approved) retrospective analysis of the medication process time-intervals demonstrates the possibility of reasoning, inferencing, and interpretation of the medication process using data from computerized provider order entry (CPOE) and electronic medical administration record (eMAR) systems.
An important component for the recovery of inpatients is receiving their medications as prescribed. The medication process for inpatients is an elaborate, collaborative effort between providers, pharmacists, and nurses and involves scheduling, preparation, and actual administration. In most hospitals, medication scheduling is a function performed by pharmacists and is conducted in consultation with the nursing staff responsible for administration. A pharmacist reviews medication orders entered by providers and schedules the administration of the respective medications according to the urgency of the individual orders. For STAT and NOW orders, the pharmacist confers with the involved nursing staff on scheduling and arranges to deliver the medication doses prior to the required administration time, while doses for the ROUTINE orders are dispensed with the next scheduled batch delivery to the patients’ respective units and are scheduled at the unit’s routine administration times according to the medication frequency.
The medication preparation step is handled either by the pharmacy or by the nursing staff member responsible for administering the dose, or both, depending on the requirements for the specific medication and the order. In many instances, the nurse may have to carry out essential preparation, such as priming an infusion line, just before administering the medication . In addition to the medication preparation, nurses may also spend time on essential checks to ensure a proper process and the accuracy of the dose (Alsulami, Choonara, & Conroy, 2014; Kim et al., 2006). Medication administration is a complex process that may be disturbed multiple times due to distractions, workflow issues, emergencies, or other interruptions, while requiring close collaborations, multiple handoffs, and redundant checking (Hughes & Edgerton, 2005; Carayon, Xie, & Kianfar, 2013; Lowry et al., 2014; Irwin et al., 2008).
The medication process in pediatric care carries increased safety risks due to the patient’s continuously changing physiology, the need for weight- based dosing, and the limited internal reserves to buffer the impact of disease and medication errors (Kaushal , 2001; Hughes & Edgerton, 2005). For certain diseases (e.g., serious infections), reducing the time between intent to treat and actual treatment has been shown to improve patient outcome and reduce readmissions (Weiss et al., 2014). Therefore, t he medication process requires balancing the opposing pressures of urgency and the required thoroughness (right patient, medication, dose, route, form, time, documentation, response, and patient education, etc.)( National Coordinating Council for Medication Error Reporting and Prevention , 201 5). The dichotomy becomes especially relevant for the STAT and NOW orders.
In this study, the time-interval between medication order and corresponding administration of the first-dose were studied as a proxy for the complete medication process from order to administration. As part of a larger medication compliance effort (unpublished work), the times to medication scheduling (Ord-Sched interval) and medication administration (Ord-Adm interval) for medication orders of pediatric inpatients at Vanderbilt University Medical Center (VUMC) were analyzed. The goal was to evaluate the “health” of the medication process by determining if the medication administration process is compliant with medication orders.
After obtaining institutional IRB approval, the enterprise data warehouse ( EDW) at VUMC was used to extract data for this exploratory and descriptive study. The EDW receives daily feeds from the VUMC’s operational systems, such as its order entry system (Horizon Order Entry) (McKesson, 2009 ) and the electronic medication administration record (eMAR) system. The data in the EDW are retained to support data analytics and research efforts. The datasets and objects that store patient demographics, encounters, medication orders, dispensing, and administration details were used.
The EDW uses an Oracle 11g database engine and is located behind a firewall. This research analysis was part of a larger effort to study medication compliance for pediatric inpatients at the VUMC, which was approved by the Vanderbilt IRB. Statistical package R [64-bit version 2.15.2 (2012-10-26)] (Murdoch, n.d.) for statistical analyses and plots was used.
The patient dataset was isolated based on the encounters for pediatric inpatients, including admissions and observational stays. Then, the medication orders were extracted for this set of patients during the period of the study – from July 1, 2010 through December 31, 2015. Orders were filtered to include only those served by pharmacy, excluding most of the non-medication orders. Orders not explicitly for medications (e.g., orders named “PHARMACY MESSAGE”) or for discontinuation of medications were also excluded. Additional filters were applied to exclude PRN orders and total parenteral nutrition orders.
The EDW also stores data pertaining to the location (hospital unit) for admitted patients at VUMC. Each medication order was linked with the corresponding patient location at the time the order was issued (order time). Any orders with the location other than pediatric inpatient units (like the pediatric emergency department) were excluded.
The order records contained “PRIORITY” descriptors. Order records with the “PRIORITY” descriptor of “STAT” or “NOW” were considered urgent; otherwise, the order was termed as the “ROUTINE” type.
For the resulting medication orders, the corresponding administration data were collected from the eMAR system. The first administered dose for each medication order was identified by linking the administration record with the earliest administration time. For each pair of orders and corresponding first-dose administration record, three timestamps were collected including the date and time: order-time, schedule-time, and administration-time. Using these three time-points the time-to-medication-administration (Ord-Adm) was computed as the interval between order-time and corresponding administration-time, time-to-medication-schedule (Ord-Sched) as the interval between medication order-time and corresponding schedule-time, and schedule-to-administration (Sched-Adm) as the interval between schedule-time and administration-time. These intervals satisfied the universal “medication process equation”:
Ord-Adm = Ord-Sched + Sched-Adm
The Ord-Sched interval represents the time taken by the pharmacy to formulate and dispense the medication prior to the scheduled time, which is assigned by the pharmacy in conference with the nursing staff. The Sched-Adm interval is the time for the nursing staff to retrieve and prepare the medication and administer it to the patient. All time-gaps are measured in hours on a continuous scale. Though the “ medication process equation” holds true for individual medication order instances, it may not hold up for respective statistical parameters. For example, the medians of Ord-Adm, Ord-Sched, and Sched-Adm may not follow the “medication process equation.” However, the statistical parameters are used to examine the underlying trends of the medication process from different angles and to draw indicative inferences.
The demographic data for 56,428 distinct pediatric inpatients and observation patients, who had a total of 110,435 inpatient encounters, were isolated. The patients were 54% male and 46% female. The analysis included 509,304 distinct orders along with associated unit, patient demographic information (gender and age) and corresponding first-dose administrations.
Figure 1: Ord-Sched, Sched-Adm, and Ord-Adm intervals (see the legend on the plot for details) for all medication orders for pediatric inpatients admitted at the VUMC, by Order-Type. The box-thickness is proportional to the number of medication orders (shown in the parentheses).
Figure 1 shows the box-plots of Ord-Sched, Sched-Adm, and Ord-Adm intervals for all the medication orders segregated by order-type: ROUTINE, NOW, and STAT. The plots for STAT and NOW orders show a narrower distribution for all three intervals – Ord-Adm, Ord-Sched, and Sched-Adm. For ROUTINE orders, Ord-Adm and Ord-Sched intervals have higher median and variance while the Sched-Adm interval seems comparable to those of NOW or STAT orders.
Figure 2: Ord-Sched, Sched-Adm, and Ord-Adm intervals (see the legend on the plot for details) for all medication orders for pediatric inpatients admitted at the VUMC, by Verify-Indicator. The box-thickness is proportional to the number of medication orders (shown in the parentheses).
Figure 2 shows the box-plots for Ord-Sched, Sched-Adm, Ord-Adm intervals for all medication orders arranged by administration verification status. As a general nursing practice, a medication requiring verification entails checking of the medication and its dose by a second nurse prior to administration to assure safety and accuracy (Kellett & Gottwald, 2015). The plots suggest that medications requiring verification have shorter Ord-Adm, Ord-Sched, and Sched-Adm intervals, with narrower variations. Interval medians for medications requiring verifications were about an hour shorter than those that did not require verification. All intervals consistently show smaller variation for orders requiring verification compared to those that do not (data not shown).
Figure 3: Ord-Sched, Sched-Adm, Ord-Adm intervals (legend on the plot for details) for all medication orders segregated by patient unit – NICU or Other. The box-thickness is proportional to the number of medication orders (shown in the parentheses).
Figure 3 shows the Ord-Sched, Sched-Adm, and Ord-Adm interval box-plots for orders comparing NICU locations with all other locations. At Vanderbilt, most NICU beds are collocated with the pediatric pharmacy. Ord-Sched intervals for NICU patients are shorter and have less variation, especially for STAT and NOW orders, compared to patients in other units, suggesting faster delivery of medications. Sched-Adm and consequently Ord-Adm intervals for NICU orders have wider variations and higher medians compared to the respective values for other units.
Figure 4: Ord-Adm intervals for all medication orders for pediatric inpatients admitted at the VUMC, by Anatomical Therapeutic Chemical (ATC) class of medicines. The box-thickness is proportional to the number of medication orders (shown in the parentheses).
Figure 4 shows the box-plots for Ord-Adm intervals by Anatomical Therapeutic Chemical (ATC) class (WHO Collaborating Centre for Drug Statistics Methodology, n.d.) of the ordered medications. The oncology drugs (“Antineoplastic and Immunomodulating Agents – ATC class ‘L’) have the largest Ord-Adm median, which would be expected due to the complexity associated with dispensing. The median Ord-Sched interval for this class is also the largest of all medication classes and the administration – Sched-Adm – interval has one of the largest medians and the highest variance, with the third quartile extending beyond 2 hours compared to ~1 hour for other classes.
Figure 5: Ord-Adm intervals for all medication orders for pediatric inpatients admitted at the VUMC, by scheduled-hour of the ordered medication. The box-thickness is proportional to the number of medication orders (shown in the parentheses) for the respective scheduled hours.
Figure 5 displays Ord-Adm intervals by the hour of the day when the medication was scheduled for administration. The variations and medians of the Ord-Adm interval for hours 6 a.m., 8 a.m. and 10 a.m. – and in a mirroring fashion also for corresponding p.m. hours – are higher (corresponding boxes filled with pattern). A corresponding lengthening is shown by the Ord-Sched intervals for these hours suggesting a slower turnaround in pharmacy from 6 to 10 (a.m. and p.m.) hours. The Sched-Adm, however, had highest medians and variances at 7 a.m., 9 a.m., and 11 a.m.
Figure 6 shows the Ord-Adm intervals by patient age. The median intervals and the variations increase with the patients’ age. The Ord-Sched intervals have a smaller variance for neonates (age < 31 days) with the median near 0 hours. The Sched-Adm intervals show widest variation for this age group.
Figure 6: Ord-Adm intervals for all medication orders for pediatric inpatients admitted at the VUMC, by patient age-group. The box-thickness is proportional to the number of medication orders (shown in the parentheses) for the respective age groups.
The medication process for inpatients is an elaborate, collaborative effort between providers, pharmacists, and nurses. After receiving an order from a provider, a pharmacist interprets a provider’s intent (e.g., the pharmacist decides whether to dispense a liquid or a tablet), confirms the availability in the formulary, and dispenses the dose based on the patient’s parameters. If necessary, the pharmacist consults with the ordering provider for substitutions (e.g., the specific ordered medication is not available). The pharmacist then schedules the medication delivery after conferring with the responsible nurse, who then administers the medication as per the order.
Effective team work is important to assure appropriate medication delivery and administration as intended by the ordering provider. The analysis of medication process intervals, showing the variations in durations of medication dispensing and administration, can generate pointers leading to further investigation of the underlying causes of observed delays or conditions of excellence. The order-to-administration (Ord-Adm) time interval was analyzed by dissecting it into two components – order-to-schedule (Ord-Sched) and schedule-to-admin (Sched-Adm) intervals. The Ord-Sched interval reflects pharmacy actions including dispensing and scheduling of the medication, and Sched-Adm reflects nursing actions including preparing and administering medications. For the first medication doses, the relationship Ord-Adm = Ord-Sched + Sched-Adm holds true and is referred to as “medication process equation” in this paper.
Overall, the medication process at Vanderbilt’s Children’s H ospital (VCH) appears to be healthy based on the intervals of Ord-Sched and Sched-Adm. From the box-plots of medication administration intervals by order-type (Figure 1), it can be inferred that a bulk of Ord-Adm time stems from the Ord-Sched interval for ROUTINE orders. Given that all ROUTINE orders are scheduled to be delivered with the “next-batch” medication delivery, a sizable padding to the scheduling and the timing of the order in relation to the next scheduled batch may add time to the earliest orders resulting in a wide spread. For routine orders, delays to optimize pharmacy workflow processes appear to be acceptable. Also, the median Sched-Adm interval median and variance is comparable across all order-types, unlike the medians for the Ord-Sched intervals. This indicates that the nursing workflow does not change considerably for STAT and NOW medications, but workflow for pharmacists does.
The median Sched-Adm intervals are similar for all of the order types (STAT, NOW, or ROUTINE), suggesting that the administration processes are very similar irrespective of the order type. Relatively shorter median Sched-Adm intervals for ROUTINE orders also indicate medication availability and managing a known and expected event. Relatively longer median Sched-Adm intervals for STAT and NOW orders may reflect the need to scramble for the medication (run to pharmacy), or an unforeseen and unanticipated event requiring additional processing time.
Surprisingly, the box-plots of the Ord-Adm, Ord-Sched, and Sched-Adm, by Verify status in Figure 2, suggest that the respective intervals for the orders requiring verification by another nurse are shorter than those without – as apparent from the respective medians and variance. Empirically, any order requiring verification (by a fellow nurse) before administration is expected to take longer than if no verification is required. Plots in Figure 2, however, indicate otherwise. It was observed that the respective median intervals are still shorter for the orders with verif ication compared to those without – except the Ord-Sched for the ROUTINE orders (data not shown). The observed difference was calculated for the median Sched-Adm intervals, specifically for the NOW and STAT orders, which have wider variance. One reason for the reduced median intervals for the doses with ‘verify-indicator’ could be that these orders are given priority by the nurses and pharmacists. However, this finding will require further research.
Median medication process intervals for patients in NICU and other units (Figure 3) show that Ord-Sched intervals have a marginally shorter median time with narrower variance, indicating a more efficient pharmacy process for NICU patients. The fact that, at VUMC, a majority of the NICU beds are co-located with the pediatric pharmacy may play a role. Each of the three intervals are further analyzed by order-type, which show patterns consistent with the plots in Figure 3 – with longer Ord-Adm and Sched-Adm times for NICU patients (data not shown). The longer Sched-Adm intervals in the NICU may be attributed to the complexity of the NICU patients and their treatments, and the compound tasks to be performed for the administration of medications in critically ill neonates.
The oncology drugs (class ‘L’ – antineoplastic and immunomodulating agents) showed the largest median Ord-Adm interval. The Ord-Sched plots by ATC class show a similar pattern. The Sched-Adm interval plots reflecting nurse administration indicate a uniform median time of less than half-hour, albeit showing a much wider variation for oncology drugs. This observation suggests a greater variability for nurses to prepare and administer oncology drugs (Chen & Lehmann, 2011), which may reflect the increased safety measures required. The wider variations in the Ord-Sched intervals of some other ATC classes may require further investigation.
The distribution of the median Ord-Adm intervals by scheduled hour of the medication can show bottlenecks in the medication process. The median Ord-Adm intervals in Figure 5 are measurably longer, with much wider variations, for orders scheduled during 6th , 8th , and 10th hours on the clock (thus likewise for 18th, 20th and 22nd hours). Though the generation of orders during periodic rounding is logical explanation for 10th and 22nd hours (also apparent in the volume of orders expressed in the thickness of the bar), the rationale for a similar pattern during 6th , 8th , 18th, and 20th hour is unclear. The corresponding Ord-Sched intervals show similar patterns. Exploration with pharmacy personnel revealed that the pharmacy is short-handed during these hours of the day and, the repetition schedules of medications (q8, q12, etc.) fall during these hours.
The Sched-Adm intervals by scheduled hour indicate that the variation in administration is quite steady across all the “ waking” hours. However, the Sched-Adm intervals are longer for 7th, 9th, and 11th hours (as suggested by respective medians and variance). The shift-change during the 7th and 19th hour explains the lengthening of Sched-Adm time during those hours, but further validation and investigation may be needed for the 9th and 11th hours. A possible explanation of the prolonged intervals for the 9th hour may be the scheduling of daily doses at this time resulting in additional workload for nurses. Future studies have to validate this.
Ord-Adm by patient age-group (Figure 6) show the median and the variation marginally increase with the patients’ age. Further analysis shows that Ord-Sched intervals are shortest for the neonates (age < 31days), while they vary widely for older patients. The smaller Ord-Sched interval for the neonates can be explained by the fact that most of the NICU beds are co-located with the p harmacy (neonates account for many patient-days in the NICU), which may influence the medication dispense and delivery times favorably. The longer Sched-Adm time for neonates can be explained by the extra care and the related complexity involved in medication administration for these very young and fragile patients. The median Sched-Adm time of a half-hour or less for all ages indicates an efficient administration process. The p erformance of nurses appears consistent across all the age-groups.
Apparently, Figure 6 box-plots may not reconcile well with those for Ord-Adm plots by patient unit (NICU vs. other u nits) shown in Figure 3, since Figure 3 suggests that the Ord-Adm interval for NICUs is longer. However, not all the neonates admitted to VCH are in the NICU – as corroborated by the respective counts of the medication orders (Figure 3 has NICU count of 87,840 vs. 116,261 for neonates). However, this dichotomy may need further investigation to ascertain the difference caused by the additional patients in the age-group analysis.
We also plotted the Ord-Adm, Ord-Sched, and Sched-Adm across other dimensions like patient demographics (Gender, Race, and Ethnicity), Nursing Staff Status, and Provider Privileges (Attending, Referring, and Surgical) (Data not shown). No discernible association of these dimensions with the medication process intervals was noticed.
In summary, this study demonstrated the utility of the health data repository to assess the health of the medication process in a healthcare institution. This study was further able to show that the existing medication process resulted in overall compliance with the medication orders, although our analysis showed pockets of potential improvement. The significance of this study lies in the demonstration of a new technique that allowed the determination of the health of the medication process – a function vital for hospitals to perform their mission. Next steps would include quality improvement projects for identified challenges such as medication administration at shift change and subsequent repeat measurement. Further, we anticipate using this method to determine effect of speedy administration of high value medications on duration of hospitalization and readmission.
Given a widespread adoption of the electronic health record (EHR) systems (Gabriel, Furukawa, Jones, King, & Samy, 2013) due to the incentives offered through the American Recovery and Reinvestment Act (Office of the National Coordinator for Health Information Technology (ONC) (n.d. ), this approach should be available to most institutions. This study demonstrated the value of healthcare data that already exist within current systems. Some challenges in the medication process (e.g., a bottleneck in the medication process between 6 a.m. and 10 a.m. ; or medications of a specific class take longer to prepare, dispense, and administer) were identified that will be used to take appropriate steps such as observational studies and quality improvement efforts to improve the processes. T o improve hospital processes, outcome measures have to be observed and measured. This study demonstrated a series of outcome measures for the medication process that are easy to obtain and can be compared across hospital locations and even institutions.
This study evaluated a novel approach to determine the "health" of the medication process. This approach of measuring medication process intervals will allow others to determine the impact of projects aimed to improve medication delivery in inpatient settings through pre/post measurements.
Considering the data bulk, and the number of dimensions, this study was limited to a very high-level analysis. This study provided an understanding of delays and bottlenecks in the medication process. From the data, the origins and causes for these findings can only be deduced or speculated. These results will require further observational studies and interviews with stakeholders.
This study analyzed the intervals between different steps in the medication process. Unfortunately, there were no standards to compare these data to because this type of work has not been done before. For future work, the results in this study may provide a benchmark for comparison.
A detailed analysis at varying depths could reveal additional signals and may either address some of the unexplainable observations (e.g. longer Sched-Adm times for medications of ATC class ‘G’, ‘M’, or ‘S’) of observed variations or spur additional questions to investigate further.
It is critical for a health care institution to assure the effectiveness and safety of the medication process and ensure that the medication orders result in medication administration. It has been shown that for certain disease conditions, reducing the time between intent to treat and actual treatment can influence patient outcomes (Weiss et al., 2014). This study demonstrated that rich information can be obtained from the analysis of medication process timings at various steps. Analyzing the medication process intervals from various perspectives can provide clues to the scope for improvements along different dimensions.
One of the purposes of this paper was to demonstrate the use of health care data to enable institution-wide analytics at various levels. The EDW provided the widest possible breadth of patient data for an institution (VUMC). It can thus be concluded that research analytics for the medication process are possible at various depth levels . To our knowledge, this is the first institution-wide study to analyze the medication process for an extended period of time for both patients and medications.
Alsulami, Z., Choonara, I., & Conroy, S. (2014). Paediatric nurses’ adherence to the double-checking process during medication administration in a children’s hospital: an observational study. Journal of Advanced Nursing, 70(6), 1404–1413. https://doi.org/10.1111/jan.12303
Bhatia, H. L., Patel, N. R., Ivory, C. H., Stewart, P. W., Unertl, K. M., & Lehmann, C. U. (n.d.). Pediatric Inpatient Medication Compliance - What are we missing? American Medical Informatics Association (AMIA) Retrieved from https://knowledge.amia.org/59310-amia-1.2741865/t001-1.2745946?qr=1
Carayon, P., Xie, A., & Kianfar, S. (2013). Human Factors and Ergonomics. Evidence Reports/Technology Assessments, No. 211. Rockville (MD): Agency for Healthcare Research and Quality (US). Retrieved from http://www.ncbi.nlm.nih.gov/books/NBK133393/
Chen, A. R., & Lehmann, C. U. (2011). Computerized Provider Order Entry in Pediatric Oncology: Design, Implementation, and Outcomes. Journal of Oncology Practice, 7(4), 218–222. https://doi.org/10.1200/JOP.2011.000344
Gabriel, M. H., Furukawa, M. F., Jones, E. B., King, J., & Samy, L. K. (2013). Progress and challenges with the implementation and use of electronic health records among critical access hospitals. Office of the National Coordinator for Health Information Technology. Retrieved from https://healthit.gov/sites/default/files/cahdata_brief12.pdf
Hughes, R. G., & Edgerton, E. A. (2005). Reducing Pediatric Medication Errors: Children are especially at risk for medication errors. AJN The American Journal of Nursing, 105(5), 79–84.
Irwin, D., Vaillancourt, R., Dalgleish, D., Thomas, M., Grenier, S., Wong, E., … Gaboury, I. (2008). Standard concentrations of high-alert drug infusions across paediatric acute care. Paediatrics & Child Health, 13(5), 371–376.
Johnson, K. B., Lehmann, C. U., & Council on Clinical Information Technology of the American Academy of Pediatrics. . (2013). Electronic Prescribing in Pediatrics: Toward Safer and More Effective Medication Management. Pediatrics, 131(4), e1350–e1356. https://doi.org/10.1542/peds.2013-0193
Kaushal R, B. D. (2001). Medication errors and adverse drug events in pediatric inpatients. JAMA: The Journal of the American Medical Association, 285(16), 2114–2120. https://doi.org/10.1001/jama.285.16.2114
Kellett, P., & Gottwald, M. (2015). Double-checking high-risk medications in acute settings: a safer process. Nursing Management, 21(9), 16–22. https://doi.org/10.7748/nm.21.9.16.e1310
Kim, G. R., Chen, A. R., Arceci, R. J., Mitchell, S. H., Kokoszka, K. M., Daniel, D., & Lehmann, C. U. (2006). Error reduction in pediatric chemotherapy: computerized order entry and failure modes and effects analysis. Archives of Pediatrics & Adolescent Medicine, 160(5), 495–498. https://doi.org/10.1001/archpedi.160.5.495
Lehmann, C. U., & Council on Clinical Information Technology. (2015). Pediatric aspects of inpatient health information technology systems. Pediatrics, 135(3), e756-768. https://doi.org/10.1542/peds.2014-4148
Lowry, S. Z., Ramaiah, M., Patterson, E. S., Brick, D., Gurses, A. P., Ozok, A., … Gibbons, M. C. (2014). Integrating electronic health records into clinical workflow : an application of human factors modeling methods to ambulatory care (No. NIST IR 7988). National Institute of Standards and Technology. Retrieved from http://nvlpubs.nist.gov/nistpubs/ir/2014/NIST.IR.7988.pdf
McKesson. (2009). Health Systems Use McKesson’s CPOE Solution to Enhance Patient Outcomes, Improve Medication Safety | McKesson. Retrieved April 11, 2017, from http://www.mckesson.com/about-mckesson/newsroom/press-releases/2009/hea…
Murdoch, D. (n.d.). The R-project for statistical computing. Retrieved April 11, 2017, from https://cran.r-project.org/bin/windows/base/old/2.15.2/
National Coordinating Council for Medication Error Reporting and Prevention . (201 5). Recommendations to Enhance Accuracy of Administration of Medications [Text]. Retrieved April 13, 2015, from http://www.nccmerp.org/recommendations-enhance-accuracy-administration-…
Office of the National Coordinator for Health Information Technology (ONC) (n.d.). Meaningful Use Definition and Meaningful Use Objectives of EHRs | Providers & Professionals | HealthIT.gov. Retrieved March 25, 2015, from http://www.healthit.gov/providers-professionals/meaningful-use-definiti…
Scharnweber, C., Mollenkopf, N. L., Fackler, J., Dover, G. J., & Lehmann, C. U. (2013). Utilizing electronic health record data to determine the health of the medication process after the relocation of a children’s hospital. Studies in Health Technology and Informatics, 192, 210–214.
Slight, S. P., Berner, E. S., Galanter, W., Huff, S., Lambert, B. L., Lannon, C., … Bates, D. W. (2015). Meaningful Use of Electronic Health Records: Experiences From the Field and Future Opportunities. JMIR Medical Informatics, 3(3), e30. https://doi.org/10.2196/medinform.4457
Weiss, S. L., Fitzgerald, J. C., Balamuth, F., Alpern, E. R., Lavelle, J., Chilutti, M., … Thomas, N. J. (2014). Delayed antimicrobial therapy increases mortality and organ dysfunction duration in pediatric sepsis. Critical Care Medicine, 42(11), 2409–2417. https://doi.org/10.1097/CCM.0000000000000509
(WHO Collaborating Centre for Drug Statistics Methodology. (n.d.) - Structure and principles. ( Retrieved March 10, 2015, from http://www.whocc.no/atc/structure_and_principles/
Haresh L. Bhatia, PhD
Dr. Bhatia is a senior outcomes research manager at Express Scripts Holdings Co., St. Louis, Mo . He received his Ph.D. in biomedical informatics from Vanderbilt University School of Medicine, Nashville, Tenn. , as a fellow of the National Library of Medicine , in 2015. He has extensive experience in informatics and data architecture and worked in the industry for over two decades prior to getting his doctorate. Dr. Bhatia serves on the b oard of the St. Louis chapter of The Data Warehouse Institute (TDWI). He also has an adjunct faculty appointment at the University College, Washington University, of St. Louis, Mo .
Neal R. Patel, M.D., MPH
Dr. Neal Patel is a professor of clinical pediatrics at Vanderbilt University School of Medicine, and serves as a co-medical director of pediatric critical care services and chief medical informatics officer (CMIO) of Vanderbilt University Health System. Dr. Patel is a recipient of the Vanderbilt Five Pillar Leader Award for leadership in service, quality, growth / finance, innovation, and promotion of the staff and faculty commitment. Dr. Patel received his M.D. from the University of Southern California and was trained at the Children’s Hospital, Los Angeles. He received his MPH from Vanderbilt University in 2000. Dr. Patel has faculty appointments in the Department of Pediatrics, the Department of Anesthesia, and the Department of Biomedical Informatics.
Catherine H. Ivory, Ph.D., R.N.-BC
Dr. Cathy Ivory is an assistant professor at the Vanderbilt University School of Nursing. Dr. Ivory’s clinical focus is inpatient obstetrics and perinatal nursing. She served as the p resident of the Association of Women’s Health, OB and Neonatal Nurses (AWHONN) in 2014, representing over 300,000 nurses who care for women and neonates. Dr. Ivory is also a board-certified informatics nurse and holds a secondary appointment in the Dep artment of Biomedical Informatics at the Vanderbilt University School of Medicine. Dr. Ivory’s r esearch interests include using data science methods to quantify nursing’s role in patient care, patient safety and outcomes; barriers and facilitators of normal childbirth; and implementation science. Additionally, Dr. Ivory is a senior nurse scientist for the VA Quality Scholars Fellowship P rogram and coordinates obstetrics quality improvement projects for the Tennessee Improving Perinatal Quality Collaborative (TIPQC).
Phillip W. Stewart, D.Ph.
Dr. Phillip Stewart is an informatics pharmacist at Vanderbilt University Medical School and is the residency program director for the pharmacy informatics program.
Kim M. Unertl, Ph.D.
Dr. Kim Unertl is an assistant professor of biomedical informatics at the Vanderbilt University School of Medicine. She works with Dr. Michael DeBaun as part of the Vanderbilt-Meharry-Matthew Walker Center of Excellence in Sickle Cell Disease Care. Dr. Unertl’s interests include the study of interaction between clinical workflow and health information technology. Her primary research interest is improving the fit between technology and work practices through the development of health information technology design and implementation strategies. Dr. Unertl received her Ph.D. in biomedical informatics from Vanderbilt University in 2009.
Christoph U. Lehmann, M.D., FAAP, FACMI
Dr. Chris Lehmann is a professor of pediatrics and biomedical informatics at Vanderbilt University Medical School. Dr. Lehmann has served on the board of the American Medical Informatics Association since 2008 and currently serves as the organization’s secretary. In 2010, Dr. Lehmann was inducted as a fellow into the American College of Medical Informatics and in 2012 he became a vice president of the International Medical Informatics Association (IMIA), while also being in charge of the IMIA yearbook. Dr. Lehmann was appointed medical director of the Child Health Informatics Center for the American Academy of Pediatrics. He serves on the Examination Committee of the American Board of Preventive Medicine, Subcommittee for Clinical Informatics. In 2009, Dr. Lehmann conceived and launched the journal of Applied Clinical Informatics, devoted to original research and commentary on the use of computer automation in the day-to-day practice of medicine, and has served as the editor-in-chief since its inception. Dr. Lehmann co-edited “Pediatric Informatics,” the first book on this subject, in 2009.