A Review on Decision Support Methods for Hospital Based Nurse Practitioners
Abstract— The aim of healthcare sector is to provide safe, efficient, and high- quality patient care in a cost-effective way. Thus, automated systems are in demand for clinical as well as administrative purposes. Clinical decision support system (CDSS) is such an automated application that analyses data, to help clinicians in making decisions for improving patient care. People know that, nurses are the assisted service providers to in-patients and emergency care units in every hospital or clinical department. In spite of a pretty big work and working hours, nurses have to take decisions some times during doctor’s absence. Lack of nursing expertise during their preliminary career stages or physical tiredness due to overwork, many times wrong decisions would be made from nurse’s end. As they are the direct contact point for every in-patient, development of specific decision support systems for nurses is highly relevant. Even though many CDSS have been created for doctors, now developed countries are promoting researchers for investigating and designing nurse’s decision support system (NDSS). That awareness about the need and application of specific NDSS in India, is the motivation of our ongoing research work. Designing specific NDSS could improve the nurse’s service, and thus socio economic importance. This paper mainly focuses on existing decision support system for nurses from previous 7 years, and to review about it. The research objective is to, solve specific barriers by investigating on advanced-applied intelligence techniques. Thus in the coming future, we would be able to develop specific NDSS on real data with the help of domain experts.
Keywords—CDSS; Nurse practitioners; Clinicians
Now a day the main focus of health care industry is, to adopt modern technologies for providing best care to their patients or customers. One of the technologies is the clinical decision support system (CDSS) which assist the clinicians in their work. Assistance may be in the form of : (1)confirming about the decisions that they have taken like asking to a second opinion, (2) giving steps to take a final clinical decisions, (3) alerting about a patient’s state, (4) giving treatment recommendation, (5) survival analysis for giving specific medicines, etc. Assist systems are mostly made using artificial intelligence technology. Clinical decision support is a technology to find meaningful information for clinical decisions and distribute the best possible patient care to the patients. CDSS consist of a medical expert knowledge, patient data, inference engine and the case specific advice. The medical knowledge consists of some rules which is related to disease, symptoms, drugs, dosage etc. The patient data consist of the current details ; history of patients i.e., the complete records of the patient. Using the medical knowledge, patient details are analysed and make a decision or future prediction drawing from the inference engine. In the case specific advice phase the system gives a suitable advice for the clinicians and which leads to possible patient care. Decision support can be paper-based, but computerised systems have the technological advancement of fast processing and can give useful information which maps to computerised decision rules or algorithms The CDS systems and nurses play a vital role in the hospital settings and also it helps for giving better patient care. The nurse practitioners also use the CDS system in the care of any emergency situations take place in the hospital, most of the time they take advice from the doctors through calls. But this may take long time and a lag occurs for giving care to the patients. For handling these emergency situations the CDS system will help nurses. This research work focus on the nurses in the hospital for utilizing the clinical decision support system in the case of emergencies and eliminating the existing barriers of handling the CDS system of the nurses13618.
II. LITERATURE REVIEW
This section discusses various existing decision support systems based on review of literature. The survey is based on the studies conducted from 2011-2017 focusing on the nurses for using the CDS systems. The researchers identified certain barriers for the nurses in the hospital for utilizing the CDS system. Having accurate information including updated evidence-based practice guidelines, accurate clinical pathways, and existing clinical algorithms, were the most common barriers 18.
Smith and colleagues (2011)did a qualitative analysis to reduce gastrointestinal events caused by non-steroidal anti-inflammatory drugs using a risk assessment CDS calculator. Their study focused on nurses and the Healthcare Cost and Utilization Project inpatient database. The researchers examined the two barriers :Incomplete patient information in the electronic health record ,Nurse practitioner’s lack of knowledge related to the availability of the risk assessment calculator16. There was no study taken place in 2012 related to the CDS system focusing on the nurses.
Stanik-Hutt (2013) did a systematic review were that nurse practitioners deliver safe and effective care to the patients and the nurses use hospital technology along with CDS systems to provide better care for the patients who is admitted in the hospital. This paper also discusses the pros and corns of using CDS technology, its demand is effectively increasing in the health care industry 5.
Ariosto’s (2014) did a quantitative analysis on CDS alerts based on the opiate allergies in a medical centre. They did this study on in patients in the hospital. The researchers explained four barriers to utilizing CDS alerts: First, the high number of insignificant and inappropriate alerts led to nuisance alerting and high override rates, the second barrier to fully utilizing CDS was the incorrect system setup of allergy alerts in the electronic health record, the decision support algorithm used to explain opiate allergies was improperly defined, another barrier described was inappropriate data intake and inaccurate patient information in the electronic health record9. Benson and colleagues conducted a quantitative study involving a nurse practitioner–led rapid response team in a 350-bed teaching hospital. The study method was a pre and post implementation survey comparison focusing on CDS database queries to identify sepsis. The researchers identified two barriers to CDS: The lack of available clinical patient information in the electronic health record, The CDS algorithm not supporting current evidence-based practice as outlined in the Surviving Sepsis Guidelines 10.
Knoble and Bhusal(2015) conducted qualitative research studying CDS electronic algorithms by means of mobile technology. The sample targeted nurse practitioners in a heterogeneous set- ting that included hospitals in Nepal. The authors identified four barriers: Software flaws in the CDS algorithms, an application error, hardware technical challenges including inadequate touch technology, unreliable battery life, which obstructed use of the CDS system15.A meta synthesis by Miller and colleagues(2015) of nine qualitative studies included nurse practitioners providing direct patient care in inpatient and outpatient settings. The authors examined the difficulty of integrating CDS into clinical work. The meta synthesis identified five barriers: Usability and user interface challenges, lack of integration between the electronic health record and the CDS system, immature CDS algorithms, lack of interoperability, Patient safety.17.
A randomized controlled trial by Eldredge (2016) involved 23 providers in an urban area, including seven nurse practitioners, which focused on CDS utilization at the point of care. A pre and post trial survey was developed for data collection. The outcome of this small clinical trial found two barriers to adoption: lack of training with the CDS tool, lack of time to use the tool at the point of care11. Oh and colleagues conducted a randomized controlled trial to study CDS text-based electronic alerts focusing on acute kidney injury. The research occurred in the hospital setting and the blended sampling (98 participants) included nine midlevel practitioners. The researchers identified four barriers: alert fatigue, delayed alerting resulting in lack of timeline, lack of ability to target alerts to specific groups.De Wit and colleagues (2016) conducted a qualitative, pharmacy-led study to examine drug-related and medication-error CDS alerts. The study took place in a Dutch hospital setting with a blended sample that included nurse practitioners.14 they found five barriers to CDS-generated alerts: alert fatigue, incorrect CDS content, CDS inefficiencies, reliance on manual input of patient details, and inappropriate timing of the CDS alert notifications
Based on the above nine studies Osheroff (2017) has provided the necessary framework to successfully design and implement CDS systems. Table 1 defines all the barriers of nurses for using CDS system. The framework, also known as the CDS Five Right.
• Right information
• Right people
• Right format
• Right channel
• Right time
The CDS right information includes clinical knowledge, evidence-based practice guidelines, clinical pathways, and clinical algorithms. The CDS right people considers the people who need information for clinical decision making and consists of physicians, nurse practitioners, nurses, pharmacists, clinical staff, and patients. The CDS right format describes decision sup- port interventions, such as alerts, data, prompts, order sets, and informational buttons. The design and implementation of decision support through the CDS right channels incorporates the Internet, electronic health records, patient portals, workstations, and mobile technology systems. Finally, the CDS right time pinpoints the timing of CDS in the workflow along with the right time to guide key decisions or actions. Use of the CDS Five Rights framework can assist in determining optimal CDS system utilization, as well as identifying barriers to use618.
CDS alerts/nuisance alerts
CDS systems algorithm: improperly defined/not current with evidence-based practice
Inaccurate patient information/lack of clinical patient information
Inappropriate timing of CDS
Lack of training
Lack of knowledge
Reliance on manual input
Incorrect CDS content
Lack of interoperability
Table 1: Common barriers of nurses for using CDS system
III. WORK PLAN
This section explains the process concept of our proposed NDSS using advanced machine learning systems. Design and development of such system will be conducted in 5 different phases.
1. Knowledge acquisition-
The research work started with gathering of domain knowledge from medical experts including nurses. Individual meetings, group elicitation, literature review etc are the planned acquisition methods. Knowledge acquisition was done in Manak Hospital, Kerala. Doctors and nurses were interviewed after making them aware the need of such system in Kerala. Existing decision support methods will be reviewed and documented regularly, while one phase of reviews has already been over. Knowledge acquisition and Data collection helped the authors to select experimental data and departments (Gastroenterology and Urology) for the pilot study. Thus this phase has been completed.
2. Dataset creation-
Based on the pilot study, attributes for the data collection will be fixed in consultation with medical domain experts. This would help to design the schema of dataset correctly. Then data will be collected to create a dataset for experiments. This will be cleaned after duplicate and null value deletion, noise removal, outlier detection, missing data identification and normalization. The cleaned data will be formatted based on the schema, which would be suitable for any machine learning algorithm.
3. Review of decision support systems –
Advanced soft computing methods especially machine learning will be studied, as it has already proven its success in different domains. Suitable algorithms will be selected to make models of NDSS.
4. Training ; Testing- An advanced soft computing based decision support system will be built based on the cross validation training on collected dataset. Hyper parameters will be tuned and best suitable model will be selected, to be used as decision support system for nurses for the above selected departments. Generalization testing will be done to confirm the performance stability of the system.
Health information technology is increasing day by day due to vast number of ongoing research in that area. CDS play a role at the point of patient care. The nurse practitioners faces some barriers for utilizing the CDSS in hospital. The objective of this work is to solve all these barriers by making an efficient CDSS for nurse practitioners..
The work is supported by MANK hospital in Kerala. We express our sincere thank to Dr. Joseph Mank( Uro Surgeon) who provided insight and expertise that greatly assisted the research. We thank Mr.Libin Abraham (nursing co-ordinator ,AIIMS Raipur )for his valuable comments that greatly improved the manuscript.
1. HealthIT.gov. What is clinical decision support? https://www.healthit.gov/ providers-professionals. Accessed March 10, 2017.
2. Corrigan JM. Crossing the Quality Chasm. Building a Better Delivery System. Washington, DC: National Academies Press Publications; 2001.
3. Greenes RA. Clinical Decision Support. The Road to Broad Adoption. 2nd ed. London: Academic Press Publications; 2014.
4. American Association for Nurse Practitioners. NP Facts. https://www.aanp. org/images/documents/about-nps/npfacts.pdf. Accessed March 10, 2017.
5. Stanik-Hutt J, Newhouse RP, White KM, et al. The quality and effectiveness of care provided by nurse practitioners. The Journal for Nurse Practitioners. 2013;9(8): 492–500.
6. Osheroff JA, ed. Improving Medication Use and Outcomes With Clinical Decision Support. Chicago, IL: HIMSS Publishing; 2009.
7. Mertz L. From annoying to appreciated: turning clinical decision support systems into a medical professional’s best friend. IEEE Pulse. 2015;6(5): 4–9.
8. Troseth MR. Elsevier clinical practice model achieves milestone successes to advance clinical decision support and evidence-based practice. Computers, Informatics, Nursing: CIN. 2014;32(7): 303–304.
9. Ariosto D. Factors contributing to CPOE opiate allergy alert overrides. AMIA Annual Symposium Proceedings/AMIA Symposium. 2014;2014: 256–265.
10. Benson L, Hasenau S, O’Connor N, et al. The impact of a nurse practitioner rapid response team on systemic inflammatory response syndrome outcomes. Dimensions of Critical Care Nursing: DCCN. 2014;33(3): 108–115.
11. Eldredge JD, Hall LJ, McElfresh KR, et al. Rural providers’ access to online resources: a randomized controlled trial. Journal of the Medical Library Association.
12. Oh J, Bia JR, Ubaid-Ullah M, et al. Provider acceptance of an automated electronic alert for acute kidney injury. Clinical Kidney Journal. 2016;9(4): 567–571.
13. Cato K, Hyun S, Bakken S. Response to a mobile health decision-support system for screening and management of tobacco use. Oncology Nursing Forum. 2014;41(2): 145–152.
14. De Wit HA, Hurkens KP, Mestres Gonzalvo C, et al. The support of medication reviews in hospitalised patients using a clinical decision support system. Springer plus. 2016;5(1): 871.
15. Knoble SJ, Bhusal MR. Electronic diagnostic algorithms to assist mid-level health care workers in Nepal: a mixed-method exploratory study. International Journal ofMedical Informatics. 2015;84(5): 334–340.
16. Smith MA, Hampton C, Rodier L, Taylor A, Taylor-White C. Informatics: detecting risks in elderly patients taking NSAIDs. Nurse Practitioner. 2011;36(4): 51–53.
17. Miller A, Moon B, Anders S, Walden R, Brown S, Montella D. Integrating computerized clinical decision support systems into clinical work: a meta-synthesis of qualitative research. International Journal of Medical Informatics. 2015;84(12): 1009–1018.
18. Cindy Borum,.Barriers for hospital based nurse practitioners utilizing clinical decision support systems.Computer ,Informatics, Nursing(CIN):2017;