Using data to improve medical practice by measuring processes and outcomes of care.

BACKGROUND: The purpose of this article is to help clinicians expand their use of data to improve medical practice performance and to do improvement research. Clinical practices can be viewed as small, complex organizations (microsystems) that produce services for specific patient populations. These services can be greatly improved by embedding measurement into the flow of daily work in the practice. WHY DO IT?: Four good reasons to build measures into daily medical practice are to (1) diagnose strengths and weaknesses in practice performance; (2) improve and innovate in providing care and services using improvement research; (3) manage patients and the practice; and (4) evaluate changes in results over time. It is helpful to have a "physiological" model of a medical practice to analyze the practice, to manage it, and to improve it. One model views clinical practices as microsystems that are designed to generate desired health outcomes for specific subsets of patients and to use resources efficiently. This article provides case study examples to show what an office-based practice might look like if it were using front-line measurement to improve care and services most of the time and to conduct clinical improvement research some of the time. WHAT ARE THE PRINCIPLES FOR USING DATA TO IMPROVE PROCESSES AND OUTCOMES OF CARE?: Principles reflected in the case study examples--such as "Keep Measurement Simple. Think Big and Start Small" and "More Data Is Not Necessarily Better Data. Seek Usefulness, Not Perfection, in Your Measures"--may help guide the development of data to study and improve practice. HOW CAN A PRACTICE START TO USE DATA TO IMPROVE CARE AND CONDUCT IMPROVEMENT RESEARCH?: Practical challenges are involved in starting to use data for enhancing care and improvement research. To increase the odds for success, it would be wise to use a change management strategy to launch the startup plan. Other recommendations include "Establish a Sense of Urgency. (Survival Is Not Mandatory)" and "Create the Guiding Coalition. (A Small, Devoted Group of People Can Change the World)." SUMMARY: Over the long term, we must transform thousands of local practice cultures so that useful data are used every day in countless ways to assist clinicians, support staff, patients, families, and communities.

Results of a regional study of modes of death associated with coronary artery bypass grafting. Northern New England Cardiovascular Disease Study Group.

BACKGROUND: It is well known that surgeon-specific in-hospital mortality rates for coronary artery bypass grafting vary, but this aggregate measure does not suggest specific opportunities for improvement. METHODS: We performed a regional prospective study of 8,641 consecutive patients undergoing isolated coronary artery bypass grafting by all of the 23 cardiothoracic surgeons practicing in northern New England during the study period. Mode of death was assigned by an end points committee using predetermined definitions. Surgeons were ranked according to risk-adjusted mortality rates and grouped in terciles, and cause-specific mortality rates were determined. RESULTS: The mortality rate was 3.3% in the lowest surgeon mortality tercile and 5.8% in the highest tercile. Fatal heart failure accounted for 80.0% of the difference in aggregate mortality rates, ranging from 1.9% in lowest surgeon mortality tercile to 4.0% in the highest tercile (p < 0.001). Rates of other causes did not differ significantly across surgeon mortality terciles. Differences in rates of fatal heart failure could not be explained by differences in preoperative left ventricular dysfunction or other patient characteristics. CONCLUSIONS: Most of the difference in observed mortality rates across surgeons is attributable to differences in rates of heart failure.

Looking at care from the inside out: a conceptual approach to geriatric care.

Today, managing care from the "outside in" is the predominant model for changing health care. The risk of this outside-in approach is that the health care system may lose sight of the people and communities for which it serves and cares. In this article, an "inside-out" model for viewing health care in a geriatric population is presented from the perspective of patients and providers, placing the provider in a proactive rather than reactive role. By focusing attention on the outcomes or value a patient is experiencing, providers are challenged to consider new ways of managing care.

Building measurement and data collection into medical practice.

Clinicians can use data to improve daily clinical practice. This paper offers eight principles for using data to support improvement in busy clinical settings: 1) seek usefulness, not perfection, in the measurement; 2) use a balanced set of process, outcome, and cost measures; 3) keep measurement simple (think big, but start small); 4) use qualitative and quantitative data; 5) write down the operational definitions of measures; 6) measure small, representative samples; 7) build measurement into daily work; and 8) develop a measurement team. The following approaches to using data for improvement are recommended. First, begin with curiosity about outcomes or a need to improve results. Second, try to avoid knee-jerk, obstructive criticism of proposed measurements. Instead, propose solutions that are practical, goal-oriented, and good enough to start with. Third, gather baseline data on a small sample and check the findings. Fourth, try to change and improve the delivery process while gathering data. Fifth, plot results over time and analyze them by using a control chart or other graphical method. Sixth, refine your understanding of variation in processes and outcomes by dividing patients into clinically homogeneous subgroups (stratification) and analyzing the results separately for each subgroup. Finally, make further changes while measuring key outcomes over time. Measurement and improvement are intertwined; it is impossible to make improvements without measurement. Measuring and learning from each patient and using the information gleaned to test improvements can become part of daily medical practice in local settings.

Building a quality future.

How can healthcare leaders stay ahead of the curve? What can they do to see what the future holds and to secure a place for their employees and their organizations? They must begin doing today what they need to do to survive tomorrow. Furthermore, they must take wise action today or there will be no tomorrow. This article looks into the future and connects it with what we must see and do today. The article begins with a glimpse of the future and with an exploration of what people really want from health and healthcare. Next, it examines what appear to be inexorable megatrends and healthcare trends that are sweeping through society. This leads us to consider the quality and value imperatives that must be faced to secure a stake in the healthcare delivery. We will discuss a model for managing care for individual patients and small populations by focusing on where patients, populations, and caregivers meet--at the front lines of patients care. We conclude with some advice on how to build sustainable organizations by exploiting the inevitable.

Use of the internal mammary artery graft in Northern New England. Northern New England Cardiovascular Disease Study Group.

BACKGROUND: There is evidence that patients who receive an internal mammary artery graft (IMA) during coronary artery bypass surgery have increased long-term survival. However, an IMA is not used in all patients. METHODS AND RESULTS: We studied the use of IMA grafts among 7944 patients undergoing initial, isolated coronary artery bypass surgery in Maine, New Hampshire, and Vermont from 1992 to 1995. Overall, the IMA graft was used in 82% of patients; of these, 97.2% had left IMA grafts. The use of the IMA graft varied considerably by patient and disease factors. Women received an IMA graft significantly less often (76% versus 85% in men, P<.01). Older patients (> or =75 years) were less likely to receive an IMA graft (67% versus 86%, P<.001). Smaller BSA was also associated with lower rates of IMA grafts in both sexes; however, men and women with BSA <1.8 m2 received an IMA graft at about the same rate. In general, more sick and more urgent patients had lower rates of IMA use. Patients with left ventricular ejection fraction <40% received an IMA less often than those with an ejection fraction > or =60% (77% versus 85%, P<.01). Patients with a greater number of diseased coronary vessels received an IMA more often (one, 78%; two, 82%; three, 85%). IMA use varied significantly by priority of surgery, with elective patients receiving an IMA 88% of the time, urgent 83%, and emergent 51% (Ptrend<.01). The use of the IMA graft varied from 42% to 95% among individual surgeons. Surgeons were consistent in their patterns of IMA graft use for specific risk groups. All surgeons had lower rates of IMA use among older patients, lower rates of IMA among women, and lower rates of IMA use among emergent or urgent patients. However, "low-use" surgeons had consistently lower rates of use within these patient groups. The overall rate of IMA graft use increased from 76% in 1992 to 86% in 1995 (Ptrend<.001). IMA graft use increased in all five centers and in all patient subgroups. The largest increases in use were seen among women (from 69% to 83%), among patients older than 75 years (from 55% to 75%), and in emergent patients (from 40% to 72%). CONCLUSIONS: This regional prospective study of IMA graft use in initial coronary artery bypass surgery describes substantial variability in patient groups receiving an IMA as well as increasing IMA graft use over time. It also suggests that the practice patterns of surgeons are an important determinant of IMA use. These data indicate that even more patients could benefit from the use of this technique.

Transforming patient feedback into strategic action plans.

Patients' perceptions provide valuable insight into areas for improvement and opportunities for strategic planning. Using both quantitative and qualitative research methods, the topics of what drives patient satisfaction, what delights patients, and what disappoints patients were examined. A case study approach was used to develop strategic recommendations for two market segments. For primary care patients, recommendations revolve around "provider caring" and "choice." For specialty patients, recommendations concentrate on "provider caring," "provider competence," and "office wait time."

Clinical process cost analysis: a promising tool for clinical improvement.

In today's environment, health care organizations are expected to provide the best possible care at the lowest possible cost. Neither aspect can be considered independently, but correlating the two with traditional systems of cost analysis is very difficult. This article presents a new method for linking costs to the process of care that also promises to be a powerful tool for clinical improvement and redesign.

Continually improving the health and value of health care for a population of patients: the panel management process.

Today's primary care provider faces the challenge of caring for individual patients as well as caring for populations of patients. This article offers a model--the panel management process--for understanding and managing these activities and relationships. The model integrates some of the lessons learned during the past decade as we have worked to gain an understanding of the continual improvement of health care after we have understood that care as a process and system.

Improving health care, Part 4: Concepts for improving any clinical process.

BACKGROUND: One promising method for streamlining the generation of "good ideas" is to formulate what are sometimes called change concepts-general notions or approaches to change found useful in developing specific ideas for changes that lead to improvement. For example, in current efforts to reduce health care costs by discounting provider charges, the underlying generic concept is "reducing health care costs," and the specific idea is "discounting provider charges." Short-term gains in health care cost reduction can occur by pursuing discounts. After some time, however, limits to such reduction in costs are experienced. Persevering and continuing to travel down the "discounting provider charges" path is less likely to produce further substantial improvement than returning to the basic concept of "reducing health care costs." THE HIP REPLACEMENT CASE: An interdisciplinary team aiming to reduce costs while improving quality of care for patients in need of hip joint replacement generated ideas for changing "what's done (process) to get better results." After team members wrote down their improvement ideas, they deduced the underlying change concepts and used them to generate even more ideas for improvement. Such change concepts include reordering the sequence of steps (preadmission physical therapy "certification"), eliminating failures at hand-offs between steps (transfer of information from physician's office to hospital), and eliminating a step (epidural pain control). CONCLUSION: Learning about making change, encouraging change, managing the change within and across organizations, and learning from the changes tested will characterize the sustainable, thriving health systems of the future.

Improving health care, Part 3: Clinical benchmarking for best patient care.

BACKGROUND: Benchmarking, which shows that a much better way of doing something may be possible, stimulates local interest in changing and in making changes previously thought not possible. A PLANNING WORKSHEET: The Worksheet has five basic steps: Identify measures, determine resources needed to find the "best of the best," design a data collection method and gather data, measure the best against own performance to determine gap, and identify the best practices producing best-in-class results. CASE EXAMPLE--BOWEL SURGERY: The Accelerating Clinical Improvement Bowel Surgery Team at Dartmouth-Hitchcock Medical Center (Lebanon, NH) was formed in November 1994 to improve the care of patients with diagnosis-related group (DRG) 148 or 149. Consulting two large, administrative databases and the medical literature, the team found that a substantial gap existed between the bowel surgery delivery process and the best results, as far as they were known, among comparable organizations. After flowcharting the delivery process, the team identified the high-leverage steps: same-day services, general surgery clinic, and routine care. The team then planned three successive PDCA (plan-do-check-act) cycles: utilization of same-day services for all elective surgery patients, establishment of a standardized preoperative bowel preparation, and utilization of pre- and postoperative routine care orders. These improvement cycles resulted in a reduction in length of stay from 9.66 to 8.29 days. Implementation of a critical pathway resulted in a further reduction to 5.04 days. CONCLUSION: Benchmarking can play an integral role in clinical improvement work and can stimulate wise clinical changes and promote measured improvements in quality and value.

Improving health care, Part 2: A clinical improvement worksheet and users' manual.

BACKGROUND: Small tests of change can be conducted in everyday clinical practice, thereby turning the health care delivery team into reflective practitioners who can learn from, and improve on, their work. CLINICAL IMPROVEMENT WORKSHEET AND USERS' MANUAL--CASE STUDY: The worksheet has been designed as a simple tool for applying clinical improvement to the core clinical delivery process. A carpal tunnel surgery (CTS) team was formed to improve outcomes and reduce costs for patients and to promote improvements in quality and value. The team wanted to determine if surgical patients treated with local anesthesia in an ambulatory setting have superior satisfaction with care, comparable clinical and functional outcomes, and lower medical (and social) costs. For the first time, standardized assessments of patient case mix, treatment processes, and health outcomes were designed into the delivery process by gathering data from the patient and from the surgeon presurgery and 4 weeks and 12 weeks postsurgery. Results for the first 49 of 50 to 100 consecutive patients show improved outcomes and reductions in costs, from $937 to $405 per patient. LESSONS LEARNED: Even though CTS was selected to be a quick and noncontroversial opportunity, considerable effort had to be expended to ensure that all clinicians and other affected staff would understand and support "the new way". RECOMMENDATIONS: "Ramp up" improvements as time passes, more and more change trials are conducted and their complexity increases. To ease implementation of changes, teams can diagram core process "components" and attach measures, use flowcharts to plan and monitor implementation and use change management thinking to help sharpen the plan and anticipate problems.

Current issues in multispecialty group practice: the Lahey Hitchcock Clinic merger experience.

The merger of two large not-for-profit multispecialty group practices in 1995 forced the two founding organizations into a deeper understanding of their heritage and mission. Common features of Lahey Hitchcock and similar organizations of their vintage are discussed. Issues addressed include governance, desirable leadership characteristics and growth, "incentivizing change," and workforce planning.

A hemodynamic comparison of Omniscience and Medtronic Hall aortic prostheses.

BACKGROUND AND AIMS OF THE STUDY: Preliminary reports have suggested hemodynamic disparities between single leaflet tilting disc prostheses. We tested the hypothesis that similarities in prosthetic design may not necessarily indicate hemodynamic equivalence. MATERIALS AND METHODS: In a retrospective analysis, we compared the functional characteristics of two single tilting disc (Omniscience and Medtronic Hall) prostheses implanted in the aortic position in 30 patients matched for valve size and left ventricular systolic function. RESULTS: Echocardiographic and Doppler analysis blinded to patient and valve type indicated similar quantitative left ventricular ejection fractions (64% +/- 6% for both) and cardiac outputs (4.8 +/- 1.2 vs. 4.6 +/- 1.2 l/min, p = 0.65) in the Onmiscience and Medtronic Hall groups, respectively. Transprosthetic instantaneous peak gradients were greater for Omniscience than for Medtronic Hall valves (44 +/- 8 vs. 35 +/- 11 mmHg, p < or = 0.02), as were the mean values (24 +/- 6 vs. 18 +/- 6 mmHg, p = 0.01). Even when 21 and 23 mm prostheses were analyzed separately to allow for unequal sewing ring diameters in the smallest valve sizes (Omniscience = 19 mm, Medtronic Hall = 20 mm), higher gradients were noted in the Omniscience prostheses. Effective orifice areas were smaller in the Omniscience than Medtronic Hall prostheses whether data from all prostheses (0.92 +/- 0.11 cm2 vs. 1.09 +/- 0.18 cm2, p < or = 0.05), or only data from 21 and 23 mm valves (0.94 +/- 0.11 cm2 vs. 1.10 +/- 0.18 cm2, p < 0.05) were included. Similarly, the dimensionless obstructive index, a parameter independent of left ventricular flow and annular size, was reduced in the Omniscience valves, indicating greater obstruction, whether all valves (0.31 +/- 0.04 vs. 0.36 +/- 0.07, p < or = 0.01) or only 21 and 23 mm valves (0.31 +/- 0.04 vs. 0.36 +/- .04, p < 0.001) were analyzed. CONCLUSION: Despite prosthetic design similarities, resting hemodynamic assessment indicates greater stenosis in Omniscience than Medtronic Hall valves when placed in the aortic position. The long term effects of these differences in terms of ventricular diastolic and systolic function and mass regression requires further evaluation.

Using qualitative and quantitative patient satisfaction data to improve the quality of cardiac care.

BACKGROUND: In early 1993 leaders within the Hitchcock Clinic and cardiac services section at Dartmouth-Hitchcock Medical Center (Lebanon, NH) formed the Cardiac Services Improvement Group (CSIG) as a pilot program for patient-centered quality improvement (QI) at the sectional level. CSIG PROGRAM: For open heart surgery (OHS) and percutaneous transluminal coronary angioplasty (PTCA), a flowchart was constructed of sequential patient experiences. Content analysis of focus group discussions resulted in six key patient-defined quality characteristics: comfort, caring, certainty, convenience, communication, and cost. Linking of patient comments to points on the patient experience flowchart made it possible to determine where particular quality characteristics were most relevant. A patient satisfaction survey with questions that were specific to a patient experience and to a quality characteristic was mailed to 100 consecutively discharged OHS and PTCA patients; 35 of the 50 patients in each group responded. EVALUATING SUCCESS: Analysis of the survey results led to the formation of two QI teams. One team began work on development of a critical pathway for discharge preparation and identified marker questions to track and monitor pathway success in subsequent surveys. Another team began the development of protocols for more effective pain management during and after the PTCA procedure. A repeat patient satisfaction survey that took place from November 1994 to March 1995 suggested a global improvement in patient satisfaction for all patient experiences and all key quality characteristics. CONCLUSION: The CSIG pilot program of patients-based quality measurement and management at the sectional level has been successful in fostering QI team formation and has been associated with a positive deflection in patient-based quality measures. Additional sections of the medical center have initiated similar projects, beginning with process definition and focus groups.