Medical breakthroughs aren’t the only advancements assisting a patient’s return to better health these days. Information Technology has infiltrated the medical world and created opportunities for doctors and other medical professionals to do a better job of helping patients by offering tools which aid in the speed and accuracy of their work. An area which still has a lot of opportunity for improvement, though, is the speed and accuracy in the diagnosis and treatment of patients, particularly in complex medical cases. Business intelligence in the medical world is quickly growing and the desire to use these tools is not only making life easier for the medical community but it may save lives. In this paper, I explore the need and potential for Decision Support Systems (DSS) in the medical world. My research will explain the benefits of using DSS to aid medical professionals in the diagnosis and treatment of patients, the growth potential in using this technology beyond current use today, the impacts a medical facility in utilizing this tool, and the challenges to utilizing DSS tools for this purpose.
The diagnosis of a patient is often a difficult task and treatment plans become even harder. According to Delaney, Kostopoulou, & Munro, “Diagnostic error accounts for the greatest proportion of claims against GPs in the UK (63%) and for a third of negligent adverse events in the US primary care (34%).” In regards to insurance billing, studies also show “that patients get the wrong diagnosis as much as 20% of the time and also receive the wrong treatment half of the time. 35% of doctors and 42% of patients report errors in their own care or that of a family member (s)” (Nilson, 2009). According to Karen Nilson, “Assigning a wrong diagnostic code may be enormously costly, time-consuming, and ineffective for patients; more critically, it is a threat to patient safety and quality. It comes down to time, money, expense, costs, research, training, physician and hospital visits, treatment’s, surgeries, drugs, disabilities, death, quality healthcare, higher medical and life insurance premiums, lack of or no health insurance, heartache, malpractice, and just feeling down right ill! Not only to mention, ‘do I trust you?’” Obviously, the impact of misdiagnosis is huge.
As a former employee of the Mayo Clinic, I am quite familiar with the manual effort still required to process patient data. Medical records have shown some improvement over the last decade by moving from paper forms to electronic formats. Medical charts have gone either entirely digital (including the use of electronic clipboards via the use of tablets by medical personnel) or medical charts and paperwork and scanned into an electronic chart, i.e. Cyberchart which allows for medical documentation updates becoming real-time and accessible to other medical professionals almost instantly.
Medical billing software has grown into integrated Practice Management software which provides tools for patient appointment scheduling, insurance verification, insurance claim (1500) billing, accounts receivable and general ledger support, as well as collection functionality. These systems can communicate electronically using technology such as Electronic Data Interchange (EDI) with insurance companies as well as with internal bolt-on systems such as electronic chart software via remote communication functionality or middleware. These advances have had a large impact already on the way that the business office side of a medical practice runs.
The question then becomes, if we have the medical documentation and billing world working pretty well, why do we still have the misdiagnosis of patients and organizations still at risk? The answer is simple – doctors and medical professionals are human and subject to human experience and error. In particular complex medical cases, doctors are often left to make medical decisions based on a best guess. What if that could change? What if doctors could validate or research diagnosis and treatment plans in other cases, in order to more quickly and accurately diagnose and treat a current patient? One could argue that they could do that now in reviewing other cases. My challenge to that is the labor-intensive nature of researching other cases manually would not only place additional responsibilities on already in demand medical professionals but the likelihood of them actually doing that is very slim. Additionally, how do medical professionals begin to know where to look? How would a doctor in one state know that another doctor had treated another patient with similar symptoms a couple of months ago? They wouldn’t. Simply put, doctors aren’t computers and aren’t capable of processing and compiling data like a computer can. Software procedures can be put in place to guide a system to book an appointment, verify insurance, compile data from a medical chart and bill the insurance company. But technology can’t diagnose a patient, can it? My research shows that through the use of business intelligence, DSS can aid in the diagnosing of a patient by assisting a medical professional in providing relevant data from other cases which has a 96% accuracy rate (Graber & Mathew, 2007).
A medical chart today consists of a lot of data: patient history, symptoms, lab results, hospital stays, treatment plans, prescriptions, and diagnosis (past & present) to start (Delaney, Kostopoulou, & Munro, 2008). This data is invaluable to the diagnosis and treatment of the current patient but also can be invaluable to other patients as well. Through Business Intelligence, this data can be queried and compiled for the use of others. A medical claim form today contains medical codes for diagnosis called ICD-9 (International Classification of Diseases) and medical procedure codes called CPT (Current Procedural Terminology) (McNamee, 2001). These codes are industry standards for medical billing and could provide the starting foundation for an industry-wide DSS structure.
In order to go down the path of an industry-wide DSS, we would need to have more industry specific guidelines than what we currently have today. ICD-9 and CPT are what are used for billing purposes because of the need to be consistent in what procedures or diagnosis are given. When it comes to the application of the code, what actually is going on can be quite different. If we were to look a surgery, for instance, the surgery for removal of a gall-bladder may be one CPT procedure code; however, what actually happens on the operating table, may be much different. The procedures a doctor follows for the removal of gall-bladder removal for an eighty-year old man may be very different from that of a five-year old girl. To capture this level of data, we’d have to rely on more than codes, we’d need to be able to get to the notes or medical chart of each patient. How these notes are compiled and the terminology used then becomes very important. To maximize the use of the medical data, the accuracy of the data is just as important as in any other DSS.
The structure design for a DSS system, whether it is for use in a particular facility or across the medical industry would require a lot of thought and organization. The first step would be to consider what physicians would want to report on. This would include detailed facts down to symptoms, lab results, x-ray readings, history, etc. Determining what would be the most valuable information I would guess would differ depending on the medical expertise and needs. This is by far the biggest decisions to be made in the design of the DSS.
Introduction of a DSS would change way that business is currently being conducted. Master data such as symptoms, patient characteristics, and historical medical events would all have to be standardized. For organization specific DSS, ownership would need to reside with someone at a Medical Director’s level who can make decisions about what is and isn’t reported within the tool.
Once a structure for the medical DSS was created, the impact to an organization would be in the details and require an overhaul of how medical documentation is being maintained. Likely the implementation would result in an increase in jobs as the aggregation of data becomes so important. Consistency in information is something that medical professionals would struggle with and therefore would require additional knowledge of new standards. Impacts to patients would also be felt down to the completeness and accuracy of data being supplied to the medical professionals. Medical offices would need to strive for complete data from patients including support for what isn’t going on with the patient as well as what is. Accuracy of data could be driven by utilizing tablets or computers with selective options to drive consistent answers. The obstacle around this approach becomes the “what isn’t available for selection” that a patient wants to share with his or her physician. Other things to consider would be how the data is organized within the chart. No longer would open notes be the best option for medical documentation. Categories of information would be standardized so that the queries to pull information are known – i.e. one wouldn’t want to pull information as a symptom of the patient for diagnosis when it actually belonged in the family history section of the chart.
Once consistent data is available, a table structure would need to be created to hold data which is being queried from medical charts. The method and frequency by which the data would be extracted would be important. Data could be synchronized weekly, monthly, daily, or hourly depending on need. There would also need to be a tie-out process to ensure that correct data is properly aligning to the tables for future presentation. Inconsistent data could pose a lot of risk for a patient so the scrutiny with which this is performed in imperative.
After data is confirmed to be consistent and ties-out decisions would next need to be made around the presentation of the information. Medical professionals do not want to have to have a degree in IT in order to manipulate the data for their purposes. The user interface design would have to make data easy to access, provide options for filtering and searching for data as well as easy to self-navigate. In a smaller organization, utilizing a tool such as Excel or Access may suffice but for larger implementations, my suggestion would be a web interface using common navigation tools used on any website.
Decision Support System technology is growing rather rapidly in the medical world but at this point, we have yet to execute on an industry-wide tool. According to Miller, “The prospects for adoption of large-scale diagnostic systems are better now than ever before, due to enthusiasm for implementation of the electronic medical record in academic, commercial, and primary care settings” (2011). Smaller scale implementations are currently being tested across various organizations.
The Isabel Diagnostic Checklist System (IDCS) is a tool that is now available over the web and offers physicians clinical decision support by providing a “checklist of likely diagnoses and drugs for a patient’s clinical features” (Isabel, 2012). The tool has gone mobile as well and is available via phone applications. Isabel also uses “statistical natural language processing software” and is “able to handle unstructured data” (Isabel, 2012). (Isabel, 2012)
Other DSS tools that have been utilized for some time are QMR, Illiad, Dxplain, and DiagnosisPro which in some cases have been aimed at finding resolution to clinical conundrums but do not have the large scale data that could be of benefit globally. Isabel proponents have highlighted that in the past, “clinicians balk, when presented with a system that seems to devalue their clinical education and experience” (Isabel, 2012). It would be important to emphasize that a DSS tool of this nature would not be implemented to replace qualified medical professionals but rather assist them in diagnosing and treating of patients as a knowledge-based tool.
To maximize the benefit to doctors and patients globally, I think industry-wide tool should be sponsored. The benefits seem never ending; timely and adequate diagnosis, proper treatment plans, decreased risk of malpractice suits for doctors, increased accuracy in medical billing claims, decreased out of pocket expense to patients, readily available data for medical studies to support advancements, sharing of experiences and pitfalls, and ultimately an increase in the wellbeing of patients. Additionally real-time data availability to groups like the Center for Disease Control (CDC) could also benefit from faster response times in disease outbreaks.
If considering for global benefit, standards we need to be implemented globally. One would need to compile data from various types of organizations, regions across the globe, and etc. Additionally, clear ownership would need to be assigned – who will govern over this and has decision rights around the functionality and availability of data. This task alone for a global tool would be significant to say the least. In the past, legislation has typically driven medical compliance around things such as insurance privacy (Health Insurance Portability & Accountability Act) which is overseen by US Department of Health and Human Services (HHS) (HHS, 2012). Governance around a global or US regional DSS would need to be supported by a group such as HHS as I don’t believe full ownership would ever be there otherwise. Medical profession associations wouldn’t be able to support a global model and insurance companies, despite having this information, have yet to show interest and I don’t see them wanting to take this on. The cost in doing so would be quite expensive to them and would lack priority. HHS is interested in doing what is best for the health and safety of people and helps with patient advocate topics already.
There are some challenges presented with the idea of a global roll-out of such a tool. Funding would certainly be difficult as this is something which would require significant resourcing and support. Legal obligations would come into play as well. Those who are working on the code behind the tool could have real life impact on patients if reporting was inaccurate and thus put patients at risk.
Data privacy is probably the largest challenge in this kind of an implementation. Would the general public be willing to share private data with other doctors, researchers and medical professionals? Would a waiver have to be signed? I would guess that most patients would be willing to share information as long they also stood to benefit from the information provided it didn’t show any personally identifying data such as name or social security number. The larger conflict gets into potential misuses of data – I could see insurance fraud likely to increase out of this idea. Another risk is that medical professionals simply opt out from utilizing the tool. Organizations who don’t have funding to change up medical documentation and procedures or simply would rather funnel their funds into other medical priorities wouldn’t benefit from the tool and neither would others in that area.
The bottom line for me is there are still advances to be made medically through the support of DSS. While it would be nice to have a federally funded global tool which we could standardize and utilize across the nation, we are still very far from that happening. These smaller tools that are currently being built by organizations are showing us what opportunities there are to be captured and will eventually drive the medical industry towards more standards, naturally. Additionally software growth in areas like SNLP will help to drive the accessibility of imperfect data which at this point, is the best we can probably hope for. Over time the health of the global population and the liability of medical professionals will stand to benefit from these technologies but the realization of the benefits is still a long way off.