Thursday, March 31, 2011

The AURI Cycle

Historically, quality improvement in healthcare has been done on the basis of expert consensus. One of the most supported models in that vein is the Plan-Do-Study-Act cycle endorsed by the IHI. The cycle begins when a quality improvement intervention is identified. Once the intervention is identified, you plan how you will study the effect of the intervention, including data collection methods. The intervention is then enacted and data collection begins in a small sample population. Once data collection is complete, it is analyzed and compared to predictions to identify lessons learned. The intervention is modified based on what was learned, and the cycle is repeated with the modified intervention.

This PDSA process can be extremely costly, time consuming and involve many false-starts. In an era of expanding health information systems, a new model based on Clinical Knowledge Management is possible. Over the last several years, I have developed a new data-driven model; the CKM process, when applied to quality improvement, produces the Analyze-Understand-Redesign-Implement cycle.

I have had many successes using the AURI cycle applied to various QI projects. However, the AURI cycle is only possible when you have a robust health information system in place, and it is continually capturing and classifying data.

Analyze: Retrieve and evaluate data and information to create actionable knowledge with regard to your QI question. This step employs multiple statistical methods and data-analytic approaches.

Understand: The "reality check" of the new knowledge. New knowledge is presented to stakeholders to determine what is feasible and to identify barriers to success.

Redesign: The alteration of current processes, both major and minor, based on the understanding of the new knowledge.

Implement: The education plan, roll-out of new processes, and modifications to the health IT applications to ensure the continued capturing and classification of relevant data. Due to the necessity of altering HIT applications, the QI initiatives often affect a large sample of patients.

The reason large scale interventions and subsequent extensive data collection and analysis is possible is the continuous capturing and classifying of data in robust health information systems. Incorporating the ability for the health IT applications to capture the redesign process into the implementation phase is critical to successful AURI cycles.

Where robust HIS is available, the AURI cycle of quality improvement is a more efficient and cost effective model than the PDSA cycle. However, due to the large number of patients affected, it must be implemented strategically.

Monday, March 28, 2011

Hans Roslings on Data, Information, and Knowledge

This TEDtalk by Hans Rosling first debuted in 2006, but the points he makes about data, information, and knowledge are priceless.

As you can see from this presentation, the ability to visually display data allows for in-depth contextualization, and the subsequent formation of new knowledge. Although the knowledge Dr. Rosling presents does not reach the threshold of actionable knowledge, it is easy to see how one could ask the right questions from a global health perspective to create actionable knowledge from the data available using his data visualization teachniques.

To further explore the new knowledge Dr. Rosling is creating, I highly recommend viewing his other TEDtalks. You can see his full bio and links to additional presentations here.

Thursday, March 24, 2011

Who is the End User?

It is generally accepted that information systems should be designed to benefit the end user. However, it is often unclear who the end user is. The answer to this question seems even more unclear in health information systems.

Thomas Goetz, executive editor or Wired and author of "The Decision Tree: Taking Control of Your Health in the New Era of Personalized Medicine" articulates a great argument that the end user of health information systems should be the patient.

I've held this same opinion for quite some time. Watching this TED talk causes me to think back to my studies of the Kimball Group's method of data warehouse implementation, outlined in their book, The Data Warehouse Lifestyle Toolkit. A large portion of this book was dedicated to end user identification and needs assessment. 

While there are many stakeholders in health care, and any one of them may be an end user for an individual application, the ultimate beneficiary of all health care information systems should be the patient. In the field of Clinical Knowledge Management, this means the focus must be maintained on ensuring that knowledge for improved decision making is made available not only to clinicians, but also to patients.

Monday, March 21, 2011

CKM and Central Line Infections

A recent article from the CDC reports that "Compared to 2001, approximately 58% fewer bloodstream infections occurred in 2009 in ICU patients with central lines." (see full Vital Signs article)

This improvement has largely been based on the acceptance and implementation of the "Central Line Bundle." According to the IHI, the bundle consists of:
  • Hand Hygiene
  • Maximal Barrier Precautions During Insertion
  • Chlorhexidine Skin Antisepsis
  • Optimal Catheter Site Selection
  • Daily Review of Line Necessity
This bundle may seem unrelated to clinical knowledge management, but it is an excellent example of the potential effects of clinical knowledge management, even when the process is unintentionally used.

This bundle was originally created during a research study at the Johns Hopkins University in 1998, published in 2004, performed by Peter Pronovost and others. The bundle compiled evidence-based practices for preventing central line infections into a single, digestible checklist. While the researched might not have set out to use the clinical knowledge management process, their success is a direct reflection on how well the CKM process was executed. Let me walk you through each step of the process, and how it was completed in this study.

Capturing and Classifying: The capturing and classifying of medical data, information, and even knowledge occurs on almost a daily basis through the publication of peer-reviewed medical journals and websites. The evidence on best practices to reduce central line infections was available to the researchers at the time of their literature review.

Retrieving: The identification and collection of articles that contained data, information, and knowledge about best practices to prevent central line infections represents the retrieval component of the CKM process.

Evaluating: The thoughtful analysis of which components in the literature review represented the greatest successes and were necessceary to include in their intervention embodies the evaluation step of the CKM process.

Sharing: The creation of the new knowledge represented by the checklist was not enough to initiate change; the researchers had to share their results in a meaningful way. They did this through a well-designed education and implementation plan.

Action: The change in practice by clinicians that occurred after the implementation of the checklist is the action component of the CKM process.

Just like the CKM process requires you to constantly be re-evaluating the effectiveness of your action, the researchers reevaluated their intervention. When the intervention was shown to be effective, the researchers shared this new knowledge with the medical community at large through the publication of their results. As others around the country implemented this checklist as the result of Johns Hopkins success, the results were analyzed and published. The communication of these results spurred even more hospitals and clinicians to implement the checklist and eventually led to endorsement by the CDC.

This continued utilization of the CKM process, although unknown to those using it, has greatly contributed to the 58% reduction in central line infections reported by the CDC. Imagine how much greater that reduction would be if the process had been used intentionally.

Saturday, March 19, 2011

The Process of CKM

The process of clinical knowledge management involves the capturing, classifying, retrieving, evaluating and sharing of clinical data and information in a way that creates actionable knowledge and provides the context for effective care decisions and therapeutic actions.

But what does this really mean?
Capturing: The process of gathering data while it is being used for a separate purpose
Joe Smith visits his PCPs office for an annual exam. While there, his weight is measured and entered into the electronic medical record.
Classifying: The policies and procedures that determine where, how and when data is stored
The office policy is to measure and record weight in kilograms to one decimal point. The weight is a mandatory field in the electronic intake form and must be completed to finalize the nursing documentation.
Retrieving: The act of collecting data in an organized way from wherever it has been stored
A researcher writes a program to query the electronic medical record and gather all recorded weights at the office of Joe Smith’s PCP.
Evaluating: The purposeful analysis of retrieved data to generate information and create knowledge
The researcher determines that the average weight of patients at the office of Joe Smith’s PCP increased 3 kg after the opening of a new ice cream shop.
Sharing: The communication of actionable knowledge in a manner that allows for effective care delivery
The researcher presents his findings at a staff meeting with all the physicians at the office along with the nutritional information provided on the ice cream shop’s website.
Action:  The purposeful activities resulting from the knowledge that has been shared.
The physician group meets with the owner of the ice cream shop, and agrees to sponsor a nutritionist to work with the owner to develop lower calorie options.
Since the weights will continue to be captured at each visit, the process of clinical knowledge management can be repeated to determine the effect of the physician’s intervention.

The above example may seem straightforward, easy, and even trivial. But that’s because it used a single data element. When examining actual clinical questions with dozens of data elements and millions of datum, the process is far from straightforward and easy. And the generation of actionable knowledge for a real clinical problem is rarely trivial.

IT, Informatics, Information Systems, & CKM

But what does this mean for clinicians and administrators who are trying to make decisions?
It means that health IT is not sufficient. Even clinical informatics is not enough. Decision makers need the right hardware, software, infrastructure and personnel trained in HIT, clinical informatics, and clinical knowledge management. Robust health care information systems are required.
Health information technology (HIT) is the hardware and software that is responsible for the collection and presentation of clinical data and information. HIT is continually occurring in the health care setting that utilizes an electronic medical record. HIT is the foundation of clinical knowledge management because it involves the capturing and classifying of data.

Clinical informatics is the process of contextualizing the data stored in health IT applications. The contextualization creates clinical information that is presented to clinicians or decision makers when interacting with software, i.e. displaying an abnormal lab value as red and a normal lab value as blue.  Clinical informatics is difficult to achieve without an electronic medical record and occurs when users retrieve stored data. Clinical informatics forms the supporting pillars of clinical knowledge management as it involves the retrieval and evaluation of data.

Health information systems are the combination of people, processes and technology that facilitate health care decision making by the creation, storage and sharing of clinical knowledge. Health information systems utilize HIT and clinical informatics, but only occur when organizations purposely set out to facilitate decision making based upon sufficient information to create actionable knowledge. Simply installing an electronic medical record application does not create a health information system. Health information systems are the complex structure that allow for the efficient execution of clinical knowledge management.

With robust health information systems in place clinical knowledge management can be leveraged to improve the delivery of health care.

Data, Information, and Knowledge

In order to understand the process of clinical knowledge management, you must first know the difference between data, information, knowledge, and actionable knowledge.
Data is isolated facts. This could be the time, temperature, heart rate or any other number of discrete measurements. Information is data put into context. For instance, is the time EST, CST, or PST? Is the temperature a body temperature or the ambient temperature? Was the heart rate measured on a neonate or an adult?
Once data is put into context to create information, it can be combined with other information to create knowledge. For instance, noon in Tokyo occurs thirteen hours before noon in New York City. While a body temperature of 90F is definitely low, an ambient temperature of 90F may be high or low.  A heart rate of 130bpm in a newborn is average while a heart rate of 130bpm in an adult is fast.
However, none of this knowledge allows you to make a decision, let alone a good decision. It has to be combined with more information to create actionable knowledge. The threshold for reaching actionable knowledge depends on the question.
Some questions are easy: when do I dial into the conference call? You need the time of the meeting, the time zone the meeting is in, your time zone and the time difference between the two time zones. Some questions are hard: is the patient’s heart rate a cause for concern? You need the patient’s current heart rate, their historic heart rate, their age, the normal values of heart rate for their age, their current activity level, the rhythm of their heart beats, the shape and pattern of the heart beats on an EKG and their clinical presentation. 
But whatever the complexity of the question, generating actionable knowledge will improve decision making.