Perspective: Improving Patient Safety with Business Intelligence


March / April 2006


Improving Patient Safety with Business Intelligence

Patient safety has always been a serious matter in healthcare. After all, it is called health “care” for a reason. What has changed over the past few years is the degree of scrutiny applied to clinical organizations regarding patient safety records and the public’s demand for improvement. Along with cost, efficiency, and access, patient safety has become one of the most urgent issues facing healthcare today.

Life on this sort of stage is not easy. And it is only going to get worse. New organizations watching over patient safety are forming on an almost daily basis. Several of the larger ones, who have tremendous clout, are voicing their disappointment and outrage over clinical failures more loudly than ever, as well as pointing fingers directly at individual provider organizations. Consumers are becoming better informed and payers, employers, and governmental agencies that are related to healthcare are becoming more sophisticated in their analysis, and more vocal in their demands. They all want action. Now. From you.

We believe business intelligence can help. You can use your data to better prepare to defend yourself. But more importantly, you can use that same data to proactively improve your processes and your decisions, leading to improved patient safety. But before you can use it, your data needs to be clean, focused, organized, and accessible by the right people, or it will become just another nasty pile of work for you and your organization to deal with. Sadly, having data in this clean, organized state is not always the case.

The Nature of Patient Safety and Patient Safety Data
Three facts must be kept in mind when discussing patient safety and patient safety data. These are:


  • Nature. People are going to have accidents and become sick, and people will die despite the best efforts of clinicians. All healthcare professionals we know have had this realization pounded into their heads throughout their education and into their professional lives.
  • Human Nature. As human beings, we view the standard in healthcare safety as absolute, total perfection. Therefore, it is in our nature that we tend to understand patient safety in negative terms (i.e. deviations from perfection). The stakes are too high to do otherwise. It would be reprehensible to say a certain level of failure is acceptable. It is not.
  • Nature of Patient Safety Data. Because safety statistics are negative by nature, data is hard to come by. Any numbers that are reported are viewed as criticism of the person, the process, and the organization. Understanding this data is critical to improvement, but difficult to obtain reliably and consistently due to the natural reluctance to admit failure.


Nobody can install a system that will solve these problems. But you can use systems to develop a base of evidence to support continuous improvement efforts to prevent as many patient safety problems as possible. That is what business intelligence is all about.

Patient Safety Numbers
That said, our starting point is to look at the numbers. A widely publicized figure is the number of deaths caused by medical errors: as many as 98,000 lives each year. This is one of a broader category of unnecessary deaths that the healthcare field refers to as “iatrogenic deaths.” Some of the top medical errors causing iatrogenic deaths in the U.S. are:


  • Medication prescription and/or medication administration errors
  • Nosocomial infections (hospital-acquired infections)
  • Medical device defects
  • Human errors in using medical devices
  • Surgical errors


Experts must estimate these numbers, since precise data is hard to collect. According to the HealthGrades Second Annual Patient Safety in American Hospitals Report (May 2005), hospital-acquired infections worsened by approximately 20% from 2000 to 2003 and accounted for 9,552 deaths and $2.6 billion — almost 30% of the total excess cost related to the patient safety incidents.

The most commonly noted causes for medical errors are:


  • Nursing shortages
  • Environmental factors (noise, lighting, disruption)
  • Overworked staff (forced overtime due to shortages)
  • Poor communication between patient and providers and among providers
  • Mislabeled medication and equipment
  • Improper or lack of hand-washing
  • Systematic errors/clerical errors
  • Slow or no access to medical records


To make matters worse, some of the leading healthcare organizations that tally errors and work with doctors and hospitals to improve quality say that care providers routinely under-report their errors and mistakes, probably most often due to liability concerns. Such low reporting contributes to medical errors because it makes it difficult to find solutions to the problems that created the errors in the first place.

One Example: Nosocomial Infections
To illustrate the various ways business intelligence can help improve patient safety, let’s drill down into one example — controlling nosocomial, or hospital-acquired, infections. This example is a composite of numerous interviews with healthcare professionals, as well as our own past experiences working with infection control information.

In the past several years, hospitals have made great strides in their efforts to prevent nosocomial, or hospital-acquired, infections. Yet, according to the Center for Disease Control (CDC), between 5% and 10% of patients admitted to hospitals in the United States became infected. While many nosocomial infections are simply unavoidable, it is estimated that about 25% of these infections could be prevented by healthcare workers taking proper precautions while caring for patients.

Infection Control Practitioners (ICPs) generally have all inpatient lab results sent directly to them for review, which makes it relatively simple to identify most patients with a possible nosocomial infection (not all positive culture results are due to hospital-acquiredinfection). However, cultures are not always done on every patient with an infection, making the group of people with nosocomial infections particularly hard to identify.

For example, we were monitoring hip-replacement patients for post-surgical site infections. One of the limitations in our data meant we could only view patient census information by the service or unit to which they were admitted. The culture result reports received from the lab also had the unit identified, but not every patient admitted to the orthopedic unit was there for a hip replacement. There was no way to obtain a list of hip-replacement surgeries by patient. Conversely, when the orthopedic unit was full, a hip-replacement patient could be boarded on another unit. Therefore, we didn’t have high confidence that we had identified all of the patients in our focus group, which meant our report data could not be deemed reliable.

Identifying patient populations for infection control surveillance could be done using business intelligence, especially with patient data such as in a registry, which is a database of information about certain patient populations, usually used for disease management, care management, and quality measurement purposes. Instead of relying on a patient’s physical location within the hospital as a starting point for your investigation, you could retrieve the exact data you need. Plus, it would be possible to slice, dice, sort, and sum populations by date, time, provider, location, etc. This would allow you to identify which patients came into the hospital with an infection versus those who truly may have a nosocomial infection.

The Bigger Picture
Due to the volume of patients in larger hospitals, many infection control departments do not have the time or resources to monitor every patient for nosocomial infections. Often, they will monitor certain targeted groups centered around procedures, medical equipment, or even certain doctors or staff in the hospital that might be carriers of such pathogens. Using clean, accurate and complete data, organized for heavy-duty analysis, it is feasible to monitor all inpatients, identify the causes of infection, and even point toward preventative actions.

Don’t let the word “business” in business intelligence get in the way of using business intelligence’s potential in other forms of intelligence and surveillance. Business intelligence opens up many doors in the world of infection control surveillance, for example. For instance, hospitals would be able to quickly identify trends and patterns in infections, and zero in on the sources of the “roughly 25%” of nosocomial infections that are avoidable. This would lead to a better understanding of how to prevent them and reduce the costs associated with them.

While we can measure medical errors and the causes for these errors, we cannot make correlations between the cause and effect without solid evidence. Business intelligence enables us to understand the correlations between the metrics and delineates the cause-and-effect relationships among the metrics.

Currently, few healthcare organizations research outcome and process improvement strategies mainly because it is not easy to do. Plus, leaders are hesitant to walk into the fire for fear of what they might find. However, this is an opportunity for you to get ahead of the curve. With President Bush’s current initiatives on improving quality and patient safety, eventually all healthcare facilities are going to have to comply, or else face consequences and pressure from governing organizations.

With business intelligence you will not have to cross your fingers and pray that the decisions you make or the solutions you implement will be the right choices. Your decisions will be the right ones because they will be backed up by your owndata from your hospital or healthcare facility. By using business intelligence to drive decision- making, you will decrease the variation in healthcare delivery and thereby decrease costs and reduce errors.

Donald Berwick, M.D., CEO of the Institute for Healthcare Improvement, on the concern about rising healthcare costs, stated, “One (strategy) is to cut cost, but that will lead to shortcuts that may harm patient safety. The smart way to cut costs is to improve safety and quality.”

Next Steps
The potential value of business intelligence in improving patient safety is tremendous. Business intelligence practices, processes and technologies are similar in many respects to those used in surveillance systems. The first step is to determine which patient safety topics might be best helped by using business intelligence capabilities, and then map out a strategy for using these powerful resources in your patient safety effort.

Scott Wanless is a member of Greenbrier & Russel’s Business Intelligence Practice, which is a business and technology consulting and training firm. He has more than 20 years of experience in business intelligence strategic planning, business intelligence application development, IT and business goals alignment, and financial analysis across numerous industries including healthcare, laboratory research, insurance, lending, manufacturing, retail and state government. Wanless can be reached at

Jodi McManaway is a member of Greenbrier & Russel’s Advanced Technologies Practice, which is focused on developing Web-based solutions to meet business needs.ÝShe has more than 12 years of combined IT and healthcare experience including clinical healthcare, health plans, clinical systems development, and clinical research and analysis.