The Impact of Data Quality on Decision Support System in Healthcare - Essay Example

Table of Contents EXECUTIVE SUMMARY. 3 1.0.Introduction 4 1.2. Background. 5 1.2.1. Data quality in health care 5 What is clinical data quality? 5 1.3 Methodology. 7 Study selection 7 Data extraction and Quality assessment. 8 2.0 .Data quality issues. 8 Description 9 Positive factors 9 Potential problems. 11 Summary of evidence search and selection 15 3.0Improvement of Data Quality 19 Missing Data 19 Database Errors 19 Inaccuracy of Data 20 Modified Data 21 Dirty Data 22 Poor Classification Data 23 Poor Data Frameworks and Security 23 References. 25 EXECUTIVE SUMMARY. This is a report about the data quality in decision support system in health care. It is structured into three parts; the first part is about the background of data quality, DSS, E-health and their current situations, the problem statement and methodology. The main purpose of this part is to present the current situation. The second part of this report explains the issues of data quality in DSS. The purpose of this part is to justify and classify the information. All the information which includes the description of positive factors, potential problems we researched will be presented. Positive factors include right framework, precision of data, completed security, meticulous data recording and prefect classification. Potential problems issues include missing data, database errors, inaccuracy of data, dirty data, poor classification of data and poor framework and poor security. Some impacts of CDSS(Clinical Decision Support System) are also the significant details for analysis, such as increasing decision making, convenience of query to database, mistrust and rejection by decision makers and causing errors in diagnosis by doctors. In the last part, we classified those positive factors and potential problems by factors of system and personal levels. In addition, a table with specific data set of issue of data quality which is very clear and easy to understand on CDSS will be demonstrated. All the resources came from ECU library and academic reference search engine. The third part is the most important part in this report. The analysis of the effect of data quality on DSS in health care is explained. The main purpose of this part is to identify the problem and suggest resolutions. It includes ways to improve data quality, presenting solutions to potential problems and limitations of these solutions, identifying some types of DSS, explaining examples of successful cases and failures. In this part, some technologies and theories will be used to analyze the information. In the final part, all the deliverables will be reviewed. Furthermore, we will expand the analysis and image the future. 1.0. Introduction For many years, investigators were able to conduct research using electronically stored data collected through the health care system (Weiner, Lyman, Murphy& Weiner, 2007).More recently; electronic health supplies a professional assistance which is both efficiency and effectiveness to access patient records. These developments provide huge technical help for health care, such as online booking systems and home care systems supported by Internet technology (Scott, 2005). A large number of systems can be used, but decision support systems (DSS) is an excellent one. DSS is defined as a computer-based information system that stands by business or organizational decision-making activities. DSS includes management, operations and planning levels of an organization and help to make decisions, which may be rapidly changing and not easily specified in advance (Beraldi, Violi, & Simone, 2011). DSS helps doctors make decisions, such as diagnosing patient's condition through clinical tests. To facilitate this, the data in database must be accurate, timely, completely, consistently and available when needed. From the research, there were 10-50 percent of data found in health organizations are of poor quality (Kerr, Norris& Stockdale, 2008). Data quality was defined as a method that ensures that data is accurate before data is entered into a database (Kaufmann, 2009). So, high data quality will help health organization improve medical quality. However, there are some problems on DSS in E-health. There is a possibility of having inaccurate and inappropriate data at some points . Therefore, this report will discuss specific issues in depth and combine other methodologies such as a systematic review of the decision support systems. In addition, we will state how these issues can be analyzed and the methodologies can be applied in this project. 1.2. Background. 1.2.1. Data quality in health care Health care is a science which is based on clinical data and information. In addition, all the clinical processes and decision making is closely linked with the medical data as The Digital Information Management (DIM) has been widely applied in most healthcare institution (Mark et al, 2007). As a result, clinical data quality has direct influence on the stand or fall of final decision. Especially in today’s social environment which industry informatization level has been improving, it is more necessary to reduce data quality issues in management systems and improve the quality of data (Beraldi, violi & simone, 2011). In this part, we will list the most common data quality issues on CDSS in Healthcare. In addition, the impact of the issues on different areas will be explained. Furthermore, we will also analyze the reasons or the factors cause the issues. What is clinical data quality? There are two aspects that we can easily understand what the clinical data quality is. The first one is the quality of clinical data itself which according to (Wang & Guarascio, 1996) includes accuracy, integrity, and self-consistency. 1. Accuracy: Data accuracy means the clinical data of every situation must be reflected accurately so that the clinical data can be used by medical personnel 2. Integrity: Data integrity means the clinical data must keep integrated. Every missing data or errors will make the data lose the effect of reflecting situation 3. Self-consistency: Data cannot exist isolated any more when the data has been created. There are relationships between data and data. This relationship should be not only restricted by each other but also not contradiction. Data Accuracy, Data Integrity, and Data self-consistency are three attributes carried by clinical data itself. We identify the three attributes as absolute attribute of clinical data. While the data has quality itself, the process of data using can also has quality. The process of data quality consists of use quality, transmission quality, and storage quality (Wang & Guarascio, 1996). 1. Data Use Quality: Data use quality means clinical data must be use correctly and availably. Incorrect using of data leads to poor quality of conclusions. 2. Data Transmission Quality: Data transmission between different areas is more and more common due to the development of information age. It is necessary to keep the high and efficient quality of clinical data when sharing and transferring data. 3. Data Storage Quality: The clinical data should be storage in a safety medium which can be computer-based database. 1.3 Methodology. The project employed the systematic review methodology. A systematic review of the studies done between January 1976 and January 2011 specifically in MEDLINE were reviewed. The data was accessed through CINAHL, PubMed, PsycINFO, web of science and the Cochrane Database of Systematic Reviews. Study selection The researchers identified trials of CDSSs that have been implemented in real clinical settings that are used by health care specialists for decision making during provision of health care services. Reports from the study were expected to include clinical outcomes, healthcare processes, user workloads and efficiency, economy, relationship centered and implementation by the health care providers. Data extraction and Quality assessment. Data related to the study such as outcomes, comparators, and intervention characteristics were extracted by one of the group members and confirmed by another. The outcomes were independently categorized and evaluated. The researchers also identified issues relating to study setting, interventions and the outcomes that limited the applicability of evidence. 2.0 .Data quality issues. Table 1: Data quality issues Positive factors Impact on personnel Impact on CDSS Factors 1) Right Framework and strategy (Kerr, Norris & Stockdale, 2012) Increase the speed of decision making Make CDSS working smoothly System 2) Precision of data ( ClinExpOptom, 2009) Increase doctors confidence Reduce the mistakes and confusion of data which improve quality of data System personnel 3) Complete security Trust the data from DSS Prevent hacking and malicious modification System 4) Meticulous data recording and saving Convenience of query to database Simplify the management System Personnel 5) Perfect classification (Wang & strong, 1999) Systematic data seeking and using Convenient to seek and manage System Potential problems Impact on personnel Impact on CDSS Factors 1) Missing data (Fairclough et al, 2008) Clinical neglect of doctors misunderstanding and mistakes Incomplete database System Personnel 2) Database errors ( Berndt, Fisher, Hevner & Studnicki, 2001) Delay the decision time and patients diagnosis Break off the CDSS and make CDSS unavailable System Personnel 3) Inaccuracy of data (Wang &strong, 1999) Mistrust and rejection by decision makers Incomplete system data records Personnel 4) Modified data ( Wang &strong, 1999) Lose confidence in data Lose integrity and reliability of database System Personnel 5) Dirty data (Wang &strong 1999) Cause errors in diagnosis by doctors Distort the database unsafe System Content 6) Poor classification of data (Wang &strong, 1999) Misusing the data Huge accidents in healthcare Difficulty in management Poor context environment System 7) Poor framework and security ( Kerr, Norris &stockdale, 2012) Difficult to use Difficult to management and improve, also easy to be broken into System Description Positive factors 1 A perfect and suitable Framework and strategy will not only reduce the waste of time and HR but also increase the speed of decision making. In addition, for the DSS in Healthcare, a right framework and strategy will help DSS work smoothly and efficiently[Gen00]. Due to a framework and strategy for DSS will be affected by previous system building, database structure building, and availability of the strategy framework. Therefore, the systems factors may be the main factor affecting the framework and lead to success or failure. 2 Truthful and exact data will not only increase the doctors’ clinical confidence but also reduce the errors and confusion of data on DSS in healthcare. Therefore, both personal and system reason may affect the data quality[Gar05]. For the personal factors, when people inputting and sharing the data and information, wrong data and confused data will be recorded by different people. As a result, personal mistakes will make the data not truthful which make the data quality poor. Meanwhile, for the system reasons, the recording of data can lose integrity and authenticity during data transporting into database. This will be caused by system faultiness and transport errors. 3 Good security of DSS leads to a high quality of data. In addition, a high quality of system security will increase the sense of trust of DSS by decision makers[Ind99]. Moreover, a good security system can prevent from external hacking and data intercept. Therefore, system factors such as high/poor security quality and uncompleted/completed security system will cause the data quality High/ low. 4 Popular and easy to understand data will need both personal and systematic efforts. Meticulous data recording and saving can increase the convenience of data management and querying. In addition, clearly and easy understanding data recording will help system to save accurately. Therefore, for personal reason, good personal recording skills will make the data quality high. Systematic factor, a perfect system context environment and easy context input process will contribute to high quality data[Eis09]. 5 Classifying different levels of data will help people manage data easily and efficiently which means categorize the data into different areas. For instance, the data can classified as different security levels such as patients data, medicine data, and medical technology data[Cel11]. In addition, the patients’ data can also categorized by different department such as neurology, surgery, and internal medicine. A high quality data is not only about the quality of data itself, but also about the efficient and speed of data querying for decision making[Gen00]. As a result, detailed and clear classification of data will contribute to a high quality data. The perfect classification system can be mainly affected by system factors. For instance, a high quality of classification system building and late management system building can be the main factors (Wang & Guarascio, 1996). Potential problems. 1 Obviously, missing data can lead to poor data quality. The problem of missing data may not only increase the clinical mistakes and misunderstanding of decision makers in healthcare, but also make DSS database incomplete which renders data unusable[Ind99]. Both personnel and systematic factors can cause this problem. For instance, personnel mistakes such as careless data inputting and recording can have an impact on data quality. Meanwhile, the poor quality of system context environment building and errors in data transporting can also lead to problem of missing data (Wang & Guarascio, 1996). 2 Database error which regard as a big problem in healthcare, especially during DSS running. This problem may render the system unavailable and unusable. As a result, decision making will not be preceded. In addition, query errors can make the efficient of decision making decreased. Clearly, both personal and systematic factors can contribute to this problem[Ker07]. For the personal reason, unsuitable system operation can directly cause the database error. While the personal reason may not be the main factor, incomplete DSS database building and poor database in structure can directly lead to poor data quality on DSS. 3 Inaccurate data can obviously reduce the confidence of decision makers and the efficient of decision making. Furthermore, the inaccuracy data can make the incomplete database recordings. The main factor that causes this issue is personal mistakes. For example, mistakes in inputting data and sharing data can lead to data inaccuracy[Kon09]. 4 Modified data can be caused by both personal and system factors. The modified data may lead to data bug and errors. The main reason of this issue is unsuitable personal operation. In addition, hacking is one aspect of personal mistakes which make data unusable. Moreover, poor quality of security of DSS can directly cause the data security issues and unauthorized database system access[Pay00]. 5 Dirty data is another issue in healthcare. Dirty data which means inappropriate data or incorrect data can be mixed up in existing data and database. As a result, the data can be distorted and its authenticity lost. Moreover, the decision makers may be misunderstood and misled[Wan96]. Furthermore, the distorted data may cause big medical negligence in potential risks. Dirty data may be caused by systematic and context factors. The system reasons such as poor or lack of verifying system and incomplete authorization database can lead to unsafe data quality environment, which can cause system broken into by hackers. While the system factors can be the main reason affects the data quality, the complex and unclear context environment may also cause dirty data problems. Context ambiguity may cause normal data cannot be recognized or identified. As a result, the data will become errors or unidentified codes in the database[Cel11]. Summary of evidence search and selection Table 2: Summary of evidence search and selection Database searched Search terms /key words Articles found Articles with relevant issues CINAHL Systematic review Booth, A., Clarke, M., Ghersi, D., & Moher, D. (2011). An international Registry of systematic- review protocols. Lancet. Kawamoto K, H. C. (2005). Improving clinical practice using clinical decision support systems: a systematic review of trials to identify features critical to success. BMJ. Morgan, K. (2009). Data Quality Relational Database Design and Implimentation. 3 PubMed Systematic review Data quality in health care systems Barnett G, O., Cimino J, J., Hupp J, A., & Hoffer E, P. (1987). An evolving diagnostic decision-support system. . JAMA. Beraldi, P., Violi, A., & De Simone, F. (2011, february 17). A decision support system for strategic asset allocation:. Decision Support Systems , pp. 549 - 561. Berndt, D., & Fisher, J. (2001). Healthcare Data Warehousing and Quality Assurance. Cella, H. (2011). The Impact of missing Data on Estimation of Health -related Quality of life Outcomes: An analysis of a Randomized Longitudinal Clinical Trial. Eisenberg, J. M. (2009). Quality Research for Quality Healthcare : The Data Connection. Agency for Healthcare Research and Quality. Garg A, X., Adhikari N, K., H, M., Rosas-Arellano M, P., PJ, D., & J, B. (2005). Effects of computerized clinical decision support systems on practitioner performance and patient outcomes: a systematic review. JAMA. hunt, D. L., Haynes, R. B., & Smith, K. (1998). effects of computer- based clinical decision support systems on physician performance and patient outcomes: a systematic review. JAMA , pp. 39 - 46. InduShobha, N., Chengalur, S., Donald, P. B., & Harold, P. L. (1999). The impact of Data Quality Information on Decision making. IEEE transactions on Knowledge and Data Engineering . Mark, W. G., Jason, L. A., Shawn, M., & Michael, W. (2007). Electronic health records: High Quality Electronic Data for High quality Clinical Research. British Computer Society. Shiffman, R., Liaw, Y., Brandt, C., & Corb, G. (1999). Computer-based guideline implementation systems: A systematic review of functionality and effectiveness. . J Am Med Inform Assoc . 10 PsycINFO Data quality in health care Healthcare information systems Wang, R., Strong, D., & Guarascio, L. (1996). Beyond Accuracy: What data Quality Means to Data Consumers. Journal of Management Information Systems , 12 (4), 5- 33. Scott, R. (2005). E-Health:current Situation and Examples of Implimented and Beneficial E-Health Applications . Journal of Telemedicine and Telecare , 121-130. 2 MEDLINE Data quality Decision support systems. Gen00: , (Gendron, 2000), Gar05: , (Garg A, Adhikari N, H, Rosas-Arellano M, PJ, & J, 2005), Ind99: , (InduShobha, Chengalur, Donald, & Harold, 1999), Eis09: , (Eisenberg, 2009), Cel11: , (Cella, 2011), Ker07: , (Kerr & Stockdale, Data Quality Information and decision making , 2007), Kon09: , (KonradPesudoys, 2009), Pay00: , (Payne T, 2000), Wan96: , (Wang, Strong, & Guarascio, 1996), Ber01: , (Berndt & Fisher, 2001), Gendron, M. S. (2000). Data Quality in the Healthcare Industry. ProQuest Dissertations and Thesis . Jason, S. (2009). Hospital Factor Associated with Clinical data Quality: Health policy. Karolyn, A. K. (2008). the strategic management of Data Quality In healthcare. Washington: SAGE publications. Kerr, K. (2006). The institutionalization of data Quality in the NewZealand Health sector. Kerr, N. K., & Stockdale, R. (2007). Data Quality Information and decision making . 18th Australian Conference on information systems (ACIS 2007), (pp. 45- 54). Queensland. Kerr, N. K., & Stockdale, R. (2008). The strategic management of data quality in healscare . Health Informatics Journal , 240 - 259. KonradPesudoys, R. (2009). Data quality and clinical decision making :do we trust machines blindly? journal compilation 2009 optometrists Association Australia , 173-175. NextGen Healthcare Information Systems. (2012). ONLINE DEMONSTRATION. Retrieved from NEXTGEN HEALTHCARE: https://www.nextgen.com/Community/VirtualOffice/DemoLogin.aspx?RequestId=4b49aa5f Payne T, H. (2000). Computer decision support systems. chest. 10 3.0 Improvement of Data Quality Missing Data Missing data mainly occur at the point of data collection. It can be due to negligence, mistakes or failure to know what kind of data is required (Kmietowicz, 2004). The first initiative in preventing incidences of missing data is training all the officers who are involved in data collection. This training should be aimed at helping them understand the nature of data that is required for each case. At the data collection points, there should be specific data elements which are defined in a uniform manner for every patient (Kerr, 2006). This means that each of those elements must be collected and recorded for every patient. These details should be availed in a manual either in soft or hard copy that will act as reference information for every data collection procedure. This information may be different for every data collection point depending on the intended purpose. Those involved in collection of data should be dedicated members of staff since failure to do it properly can result in poor diagnosis. Therefore health institutions should have well defined procedures on what data needs to be collected for every patient. Before this data is used or saved in the data base, it should be cross checked to ensure that there is no missing data Database Errors These are errors that occur when the data is being stored in the databases. Once an error has been done on this data, it can be propagated to other systems. One of the methods in which such errors can be prevented is by having a system that will give a feedback when the data is incomplete, has missing entries, values that are out of range or inconsistent entries. This feedback will be sent to the person entering the data into the database and will be able to correct the error immediately[Ber01]. Data audits into the database can also help in minimizing database errors. This can be done on entries that are highly subjective to errors. Data quality assurance procedures are also important and must be observed by the data entry staff. Training these staff on the procedures of data entry can also help in minimizing these errors. Once the database has been created, it is important to protect that data from unauthorized access. This is because some people may access that data and make intentional errors on the data and this may create errors in the database. Protection of databases is therefore important in ensuring that intentional errors are not made on the data (Berndt et al., 2001). Inaccuracy of Data Inaccuracy of data results during initial data entry, due to data decay or during data movement. To prevent inaccuracy during data entry, it is important to ensure that those involved in data entry have adequate typing skills. They should also avoid rush when entering data. It is also important to ensure that there are minimal or no distractions during data entry and that the exercise is carried out in a comfortable place. Data decay happens where values may be accurate when they are being entered but they become inaccurate with time. These are data such as contacts, marital status and number of dependents among others. To avoid this, it is important to keep re-verifying them every time a patient visits to ensure their continued accuracy. Inaccuracy during data movement occurs during extracting, transforming and loading data into another source. This happens if the system is too complex for the users (Shankaranarayan, Ziad& Wang, 2003). To avoid this, it is important to ensure that those involved in these data handling processes have adequate knowledge of the systems and also to verify the data before using it. Transmission of inaccurate data to the users may result in wrong decisions which may affect the end results. Modified Data One of the methods that can prevent modification of the stored data is to ensure limited access of that data. This can be achieved by ensuring that each user of that data has an individual account through which he or she will log into so that he or she can make changes into the stored data. After the changes, the person should log out to ensure that no other person accesses that data. Those authorized to access and change this data should only work with their passwords which they should not share with any other person. These passwords should also be changed at different intervals to ensure their security. Another measure for preventing data modification is to conduct data audit. This involves keeping track of all the changes made to the data in the electronic records. The audits will be to confirm whether it is only the authorized deletions, additions or alterations that have been made on the records. Computer generated audits can be able to capture the modifications that were made on the data (Bowen, Fuhrer & Guess, 1998). Another method would be to ensure that the date and time on the computer systems is always correct and signals any changes made to the date and time. This is because people may make changes to the data and back date the time of those changes to show that they were out of duty at that particular time. Basically a tight security system should be maintained on the stored data. Dirty Data This is incomplete, inaccurate and erroneous data. Dirty data originates from data entry processes and is most of it is caused by human error. This can be avoided by accurately defining the data entry requirements. These are requirements such as what data is to be collected, where it is coming from and how often it should be collected. During transformation of data, there should be procedures in place for validating that data and for reporting any error in it. Then the data should be securely stored with limited access to ensure that it is not altered. Establishing effective data management policies and procedures at every data handling point is also very important (Fletcher, 2004). Every person handling data should be responsible for any action on it. This could help prevent duplication of records and missing data. Critical control means should be put at every data entry point to ensure that data is clean before it enters the database from where it circulates to the intended users. Data cleansing can also be adopted which will be routinely conducted in time intervals. This involves cross checking all the records both manual and electronic to ensure that no data is missing, incomplete or erroneous. It may be a costly undertaking but it protects the patients from wrong diagnosis. Poor Classification Data To ensure that data is properly classified, the hospital should establish a data classification scheme that will be applied in the entire organization depending on how critical and sensitive the data is. For example there may be public, confidential or top secret data. For electronic records, classification of data can be automated where there is a system that will automatically classify that data according to the level of sensitivity assigned to that data. High skills are also required for data classification. Data classification staff should be well trained on how to assign levels to data and they should be committed to achieve proper data classification (Institute of Medicine, 2000). For example hospital data may be classified according to age into the class of infants, school going age and adults. It may also be classified depending on the residence areas of the patients. This therefore requires collaboration between data storage administrators and the end user on the classification criteria. The doctors who are the end users should state how they would like the data to appear for use and agree with the system administrators. This will ensure that the users receive data in the correct format which easy for them to use. Once classified, data should also be protected from unauthorized access to prevent alterations which may end up mixing it up. Poor Data Frameworks and Security Data quality framework is a tool for assessing the quality of data in an organization (Wang et al.. 1996). The main purpose of this framework is to identify areas of poor quality or inefficiencies which may affect organizational performance (Kerr, Norris & Stockdale, 2007). To ensure an effective data quality framework, the hospital must have a data quality policy, data quality management and a data quality control. The data quality policy should state the direction of the hospital in regard to quality of data products and should be formally expressed by top management. Data quality management should be a management function whose responsibility is to determine and implement the data quality policy. The data quality control involves techniques and activities required in attaining the required quality of data. Such a system setting offers an effective framework for ensuring data quality (Wang, Veda, Christopher, 1995). Data security can be achieved by having a system for monitoring data movement to ensure that the data is in safe use and is not being used by unauthorized users. Every data user should however take responsibility of ensuring safe handling of the data and maintaining its privacy and security. The data handlers should ensure privacy of their log in details and ensure that they log out after every use. References. Gen00: , (Gendron, 2000), Gar05: , (Garg A, Adhikari N, H, Rosas-Arellano M, PJ, & J, 2005), Ind99: , (InduShobha, Chengalur, Donald, & Harold, 1999), Eis09: , (Eisenberg, 2009), Cel11: , (Cella, 2011), Ker07: , (Kerr & Stockdale, Data Quality Information and decision making , 2007), Kon09: , (KonradPesudoys, 2009), Pay00: , (Payne T, 2000), Wan96: , (Wang, Strong, & Guarascio, 1996), Ber01: , (Berndt & Fisher, 2001), Read More