Use of a clinical decision support system to improve hypoglycemia management.
Currently, over 10.9 million U.S. residents age 65 or older have a diagnosis of diabetes. An estimated 79 million U.S. residents age 20 or older have pre-diabetes (Centers for Disease Control and Prevention [CDC], 2011). In the United States, diabetes remains the 7th leading cause of death. Serious complications also are associated with diabetes, including heart disease, kidney failure, vision loss, and amputation (CDC, 2011). A current debate centers on whether practitioners should enforce stringent glycemic control. A recent meta--analysis (Murad et al., 2012) found stringent glycemic control carries an increased risk of hypoglycemia. Therefore, one of the concerns around stringent blood glucose control is the increased risk for hypoglycemic events. In the hospital setting, paper-based guidelines often promote normal blood glucose. However, adherence to the paper-based hypoglycemia treatment guidelines has been poor (Maynard, Huynh, & Renvall, 2008).
In this study, the research team implemented a nursing clinical decision support system (CDSS) in the form of electronic hypoglycemia management guideline advice. Interactive guidelines were embedded within the nursing documentation section of the electronic health record (EHR). The primary purpose was to improve nursing adherence to hypoglycemia management guidelines and improve compliance tracking. The secondary purpose was to determine the impact of change of work shift on guideline adherence.
Review of the Literature
Hypoglycemia is defined as blood glucose of less than 70 mg/dL (Pagana & Pagana, 2010). In the hospitalized patient, hypoglycemia is associated with adverse drug reactions, decreased caloric intake, failure to adjust medications that lower blood glucose, and missing meals due to diagnostic testing requiring the patient to be away from the unit (Anthony, 2007). More than 40% of patients experiencing one hypoglycemic episode will experience a second episode during the same hospitalization (Maynard et al., 2008).
Guidelines to manage hypoglycemic episodes have been developed to allow nurses to intervene immediately without having to contact the physician first. Nurses who continue to use the paper-based guidelines internalize a process and no longer review the document but rather rely on memory (Rycroft-Malone, Fontenla, Seers, & Bick, 2009). In addition, these guidelines typically are housed in policy and procedure manuals; as a result, nursing utilization and documentation have been sub-optimal. In a study examining nursing adherence to practice guidelines, Anthony (2007) found nurses had poor compliance with paper-based hypoglycemia treatment guidelines. In this descriptive study, 210 retrospective medical record reviews were completed; not one record was compliant with all requirements in the practice guideline. One proposed solution to this problem is to embed guidelines within the nursing documentation section of a patient's EHR.
Lyerla, LeRouge, Cooke, Turpin, and Wilson (2010) developed a CDSS with an embedded protocol to improve nursing adherence to guidelines preventing ventilator-assisted pneumonia. The embedded protocol in the CDSS prompted the nurse to position the angle of the bed correctly for ventilator-dependent patients and document the angle in the clinical record. The study was completed in three phases, with 105 observations of documentation and head-of-bed angles recorded at each phase. Participants included 42 patients and 33 registered nurses. Adherence to documentation and head-of-bed elevation was measured before intervention, 1-2 months after intervention, and 4-5 months after intervention. Significant improvements were noted in protocol adherence at I and 5 months after implementation.
In another study, Kwok, Dinh, Dinh, and Chu (2009) evaluated the use of an asthma protocol embedded into clinical documentation. In this study, 50 patients with asthma were compared to a historical control group of 50 patients with a discharge diagnosis of asthma. The 50 patients with asthma who presented in the emergency department over a 6-month period were assessed by physicians using the Asthma Clinical Assessment Form and Electronic decision support CDSS. Following implementation of the CDSS, the medical records of the study patients were compared to a control group. A significant improvement in documentation of key clinical parameters for asthma management was noted in the study group.
Clinical Decision Support System
An interactive guideline embedded within an electronic health record is an example of a CDSS. A clinical decision support system is a computerized program utilized within the health care setting to support decision making. A nursing CDSS is used within the context of nursing to support nursing decision making. CDSS programs are based upon if-then rules that tell the computer what actions to take given certain information (Kumar, Singh, & Sanyal, 2009).
Information is accessed and stored within the knowledge base component of the CDSS. Clinical decision support systems often are used to generate alerts, reminders, or advice. Information used and generated by a CDSS should be evidence based, current, and have the ability to be updated. In this study, the diabetic educator was responsible for maintaining current evidence associated with hypoglycemia management. As the management standards changed, the diabetic educator was charged to work with programmers to ensure nurses had access to current hypoglycemia management guidelines.
Study Design and Setting
The study took place at a small community hospital in the Midwest. Approval from the hospital's institutional review board was obtained prior to study initiation. Subjects were limited to hospitalized hypoglycemic subjects over age 18. Patients who experienced blood glucose below 70 mg/dL but were receiving intravenous (IV) insulin and following the IV insulin protocol were excluded from the study as the treatment protocol was different from the hypoglycemia protocol used in this study. Using an interrupted time series design, three different samples of 150 or more medical records of patients with diabetes who had at least one incident of blood glucose below 70 mg/dL were reviewed for hypoglycemia protocol implementation and documentation. The first review was completed 6 months prior to implementing the intervention; the second review was completed 6 months following implementation; and the last review was completed 7-12 months after implementation. The intent of the final review was to determine if guideline adherence rates persisted over time.
The purpose of this study was to develop and integrate a nursing CDSS for managing hypoglycemia within the EHR to facilitate adherence to the guideline. The CDSS involved a hypoglycemic management protocol embedded within the electronic nursing documentation section of the patient's EHR. The CDSS was activated when a blood glucose level of 70 mg/dl or lower was entered. The nurse then would be asked a series of questions, each one followed by advice in accordance with the programmed hypoglycemic management guidelines. For example, an initial question asked if the patient can swallow. If the answer was "yes," oral glucose would be included in the treatment recommendation. If the answer was "no," intravenous glucose would be listed as a recommended intervention. Additional information on the development and implementation of this CDSS is published elsewhere (Harrison & Lyerla, 2012).
Review of discharged electronic health records was conducted for patients who had a primary or secondary diagnosis of diabetes from February 2010 to August 2011. A data collection tool was developed and used by the researchers. Blood glucose documentation was reviewed in each record for evidence of hypoglycemia episodes (below 70 mg/dL). Episodes occurring within the intensive care unit were excluded because a different intravenous glucose protocol was utilized. All episodes and documented treatments were collected via the data collection tool.
Ten nurse volunteers tested the initial CDSS prototype. Nurses were provided hypoglycemia scenarios and asked to follow the steps provided by the CDSS. A debriefing session then was held to gather feedback to guide changes regarding the sequence of questions and associated recommendations. Once the final version of the CDSS had been identified, a training manual and instructional videos were created. Manuals and videos were placed on each unit as well as uploaded to the organization's intranet.
All data were entered, stored, and analyzed using SPSS v19 (Statistical Package for the Social Sciences--IBM, Armonk, NY). All subjects' identifying information was removed from the data collection tool before entering the data in a computer database. Data were collected via a chart audit tool created by the research team. Collected data included subject visit number, date and time of initial episode, recorded blood glucose value, level of consciousness, initial treatment (4 or 8 ounces of orange or apple juice, three graham crackers, 25 ml of 50% dextrose IV, 50 ml of 50% dextrose IV, or 1 mg glucagon intramuscularly), swallowing ability, IV access, initiation of IV fluids, need for follow-up blood glucose administration within 15 minutes, repeated follow-up actions if indicated, and physician notification. A compliance score was assigned to each observation for each of the data collection phases. A grace period of 5 minutes before or after the 15-minute follow-up mark was allowed. Observations were grouped into those that occurred within 1 hour of a shift change and those that did not. Chi-square statistics were used to determine significance of the intervention as well as a change of shift impact. Analysis of variance (ANOVA) was used to compare mean blood glucose values across each phase and t tests within phases. The statistical analysis concluded with a logistic regression model to control for variables found to be significant. All p values were two-tailed and were compared to a 0.05 alpha.
Of the 585 hypoglycemic episodes reviewed, 284 were reviewed in phase 1,150 were reviewed in phase 2, and 158 reviewed in phase 3. Although compliance was not as high as the researchers had hoped, the increase from 4% to 13% and then from 13% to 25% was statistically significant ([chi square] = 42.16, p [less than or equal to] 0.001) (see Figure 1).
Only 34% (n=68) of the 201 hypoglycemic episodes recorded in the study occurred during a shift change. However, 95% (n=191) of those occurred during the morning shift change. The percentage of hypoglycemic episodes occurring at shift change for phase 1 was 43%, 30% for phase 2, and only 23% for phase 3 (see Figure 2). The relationship between compliance level and shift change was not statistically significant ([chi square]=0.005, p=0.946).
The average blood glucose value in each of the three phases ranged from 56 to 57 mg/dl and was not statistically significant (ANOVA, f=0.146, p=0.864) (see Figure 3). However, across patients in phase 2, a statistically significant difference existed between blood glucose for those episodes where the protocol was followed (51 mg/dl) and those where it was not (58 mg/dl) (T test, t=4.85, p=0.010).
Blood glucose was controlled by its inclusion in the logistic regression analysis (see Table 1). Researchers found no greater likelihood for improved compliance as a result of different blood glucose values (0.973). The odds ratio for phase 2 indicated the protocol was almost twice as likely (OR=1.843) to be followed than in phase 1 and almost three times as likely (OR=2.818) in phase 3.
Clinical guidelines and protocols for managing hypoglycemia are used routinely in health care practice. Nursing adherence to these guidelines has been less than optimal (Anthony, 2007; Maynard et al., 2008). The purpose of this study was to improve nursing adherence to a clinical guideline on hypoglycemia management using CDSS and to improve the ability to track compliance. In this study, a statistically significant improvement was noted before and after implementation of the hypoglycemia management CDSS; however, the improvement was not significant from a compliance standpoint (25% compliance). Nurses documenting hypoglycemia continued using the first page of the documentation tool for recording the blood glucose value and at times narratively added a treatment within an open text box. Once the page with the recorded blood glucose value was completed, however, nurses often failed to continue to the second page to complete the area in which the hypoglycemia protocol CDSS was embedded. Additionally, no statistical difference was found in compliance levels when hypoglycemic episodes occurred around shift change. However, tracking compliance after implementing the new CDSS required navigation to only one screen, where all applicable components of the hypoglycemia guideline were located. This was much improved from the previous documentation that required navigating to multiple screens to capture compliance data. Compliance in completing the training was initially a concern for researchers. However, the study site reported 97% of staff did complete the training.
During the CDSS development, nurses at the study site opted to keep the original blood glucose reporting page that included a text box for the current blood glucose value and an open text box for the nurse to add a short narrative response. The CDSS for hypoglycemia management was added as a second page or link that was consistent with other areas of nursing documentation. No alerts were built into the system to require the nurse to go beyond the first page where text boxes were located.
Current evidence-based education on the management of hypoglycemia was made available to all staff nurses prior to implementing the new CDSS (Manchester, 2008). A PowerPoint presentation about the hypoglycemia protocol, newly developed CDSS (including screen shots), and case study examples were developed, recorded, and transferred to a DVD. Copies of the DVD were distributed to all nursing units. After nurses watched the DVD, a 10-item post-test was completed using an online educational platform. Training manuals including screenshots of the CDSS also were distributed to all the nursing units. Live training sessions were not conducted at the request of nurse leaders, who already had scheduled a number of training sessions related to other changes in the clinical documentation system.
This retrospective interrupted time series design had several limitations. First, the study was conducted during the same time a number of other clinical documentation changes were occurring. This made staff education challenging as members already were involved in multiple training sessions. Additionally, although the CDSS was included as a second page to the documentation, the initial page for blood glucose value documentation remained unchanged. This may have been confusing to nurses; with no alert system in place, they could stop easily after documenting the blood glucose value, as had been done in the past. Only after the training videos were developed and distributed were the principal investigators notified that audio capabilities were limited on some of the nursing units. The study also had no control group. Finally, with one facility as the study site, results are not generalizable to other institutions.
The results of this study indicate a CDSS embedded within the nursing documentation section of the EHR can improve adherence to hypoglycemia management guidelines. However, to achieve maximum benefit, researchers must track the expected outcomes following CDSS implementation to determine if additional user training or system alterations are needed. Medical-surgical nurses routinely manage hypoglycemic events. This study demonstrated a CDSS could be used to assist nurses as they choose the most appropriate actions to take during a hypoglycemic event.
Finally, current government mandates require health care facilities to prepare reports indicating the meaningful use of their electronic health records. The Centers for Medicare & Medicaid Services established objectives for meeting the meaningful use criteria. One of the meaningful use core objectives includes the development of CDSS rules similar to those used in this study (U.S. Department of Health & Human Services, 2010). This study found a significant improvement in nursing guideline adherence following the implementation of a CDSS. Other health care facilities can use this as an exemplar in meeting this government mandate.
Adherence to current clinical guidelines is an important facet of quality improvement. Nurses need quick access to specific guidelines in a format that allows them to document in an accurate and efficient manner. The use of a CDSS can meet this need and also can provide a means for tracking documentation compliance easily. This study resulted in improved nursing compliance with hypoglycemia guideline documentation, although opportunity exists for continued improvement. As a result of the project, nurse leaders at the study site removed the open text box on the first page of the CDSS and added directions to proceed to the second page. Removal of the open text box may yield a higher compliance score in the future. Hypoglycemia guideline compliance now is measured and reported monthly during meetings of the site's quality council. In the future, the timing in presenting a new CDSS will be considered so training can be completed through a live in-service with demonstration and return demonstration rather than a video format.
Anthony, M. (2007). Treatment of hypoglycemia in hospitalized adults. The Diabetes Educator, 33(4), 709-715. doi:10.1177/0145721707303806
Centers for Disease Control and Prevention (CDC). (2011). 2011 national diabetes fact sheet. Retrieved from http://www. cdc.gov/diabetes/pubs/references11.htm
Harrison, R.L., & Lyerla, R (2012). Using nursing clinical decision support systems to achieve meaningful use. Computers, Informatics, Nursing, 30(7), 380-385. doi:10.1097/NCN.0b013eb813
Kumar, A.K., Singh, Y., & Sanyal, S. (2009). Hybrid approach using case-based reasoning and rule-based reasoning for domain independent clinical decision support in ICU. Expert Systems with Applications, 36(1), 65-71. doi: 10.1016/ j.eswa.2007.09.054
Kwok, R., Dinh, M., Dinh, D., & Chu, M. (2009). Improving adherence to asthma clinical guidelines and discharge documentation from emergency departments: Implementation of a dynamic and integrated electronic decision support system. Emergency Medicine Australasia, 21, 31-37. doi:10.1111/j.1742-6723.2008.01149.x
Lyerla, F., LeRouge, C., Cooke, D.A., Turpin, D., & Wilson, L. (2010). A nursing clinical decision support system and potential predictors on head of bed positioning for mechanically ventilated patients. American Journal of Critical Care, 19(1), 39-47. doi:10.4037/ ajcc2010836
Manchester, C.S. (2008). Diabetes education in the hospital: Establishing professional competency. Diabetes Spectrum, 21(4), 268-271. doi:10.2337/diaspect.21.4.268
Maynard, G., Huynh, M., & Renvall, M. (2008). Iatrogenic inpatient hypoglycemia: Risk factors, treatment, and prevention. Diabetes Spectrum, 21(4), 241-247. doi: 10.2337/diaspect.21.4.241
Murad, M.H., Coburn, J.A., Coto-Yglesias, F., Dzyubak, S., Hazem, A., Lane, M.A., ... Montori, V.M. (2012). Glycemic control in non-critically ill hospitalized patients: A systematic review and meta-analysis. Journal of Clinical Endocrinology and Metabolism, 97(1), 49-58. doi:10.1210/jc.2011-2100
Pagana, K.D., & Pagana, T.J. (2010). Mosby's manual of diagnostic and laboratory tests. St. Louis, MO: Mosby Elsevier.
Rycroft-Malone, J., Fontenla, M., Seers, K., & Bick, D. (2009). Protocol-based care: The standardization of decision-making? Journal of Clinical Nursing, 18, 1490-1500. doi:10.1111/j.1365-2702.2008.026 05.x
U.S. Department of Health and Human Services, Centers for Medicare & Medicaid Services. (2010). Medicare and Medicaid programs; electronic health record incentive program; final rule. Federal Register, 75(144). Retrieved from http://www.gpo.gov/fdsys/ pkg/FR-2010-07-28/html/2010-17207.htm
Anthony, M. (2008). Relationships among nurse staffing, adherence to practice guidelines, and patient outcomes in the treatment of hypoglycemia. Quality Management in Healthcare, 17(4), 312-319. doi:10.1097/01.QMH.0000338552.13136.1e Roberta L. Harrison, PhD, RN, is Associate Professor, Southern Illinois University-Edwardsville, School of Nursing, Edwardsville, IL.
Sonia L. Stalker, MSN, CDE, APN-BC, is Diabetes Specialist, Anderson Hospital, Maryville, IL.
Rochelle Henderson, PhD, is Lecturer, Public Administration and Policy Analysis, Southern Illinois University-Edwardsville, Edwardsville, IL.
Frank Lyerla, PhD, RN, is Associate Professor, School of Nursing, Southern Illinois University-Edwardsville, Edwardsville, IL.
Note: The authors received the AMSN Phillips Healthcare Research Grant for the research reported in this article.
TABLE 1. Logistic Regression Controlling for Blood Glucose Regression Odds Coefficient (B) S.E. Reference Category Phase 1 Ratio Phase 2 0.611 * 0.193 1.843 Phase 3 1.036 * 0.177 2.818 Blood Glucose -0.028 * 0.011 0.973 Constant 0.250 * 0.597 1.025 * Statistically significant p < 0.05 FIGURE 1. Protocol Compliance by Phase Phase 1 Phase 2 Phase 3 (n = 284) (n = 151) (n = 150) Not Followed 96% 87% 75% Followed 4% 13% 25% FIGURE 2. Episodes Occurring at Shift Change and Protocol Compliance Phase 1 Phase 2 Phase 3 Not Followed 43% 30% 20% Followed 42% 32% 34% Overall 43% 30% 23% FIGURE 3. Blood Glucose Average and Protocol Compliance Phase 1 Phase 2 * Phase 3 No 56 58 57 Yes 53 51 54 Overall 56 57 56 * Statistically significant p < 0.05
|Printer friendly Cite/link Email Feedback|
|Title Annotation:||CNE SERIES|
|Author:||Harrison, Roberta L.; Stalker, Sonia L.; Henderson, Rochelle; Lyerla, Frank|
|Article Type:||Disease/Disorder overview|
|Date:||Jul 1, 2013|
|Previous Article:||Thou shalt not fall! Decreasing falls in the postoperative orthopedic patient with a femoral nerve block.|
|Next Article:||Optimizing adolescent transition to adult care for sickle cell disease.|