Quality Analysis

paper and assignment attached with chapter text 

Chapter 14

 

Quality Analysis

LEARNING OBJECTIVES

· 1. To describe how statistical process control can be used to monitor and improve the services provided by healthcare organizations.

· 2. To describe the use of total quality management and continuous quality improvement methods and models to enhance system performance.

· 3. To analyze service systems using run charts and control charts.

REAL WORLD SCENARIO

A multispecialty group practice is striving to serve its patients promptly. It has defined a late patient as one who is brought into a medical examination room more than 5 minutes after the scheduled appointment. Patients who wait in the waiting room more than 5 minutes after the scheduled time of their appointment are late patients. The practice has collected the following data over a

10

-day period (

Table 14-1

).

Number of Patients Who Waited

It should be noted that this multispecialty group practice gives approximately the same number of available appointments each day. Based on this, does the clinic need to redesign its patient care systems to better serve its patients? Is there a problem? If yes, describe it and recommend an approach. If no, explain your logic.

LEARNING OBJECTIVE 1: TO DESCRIBE HOW STATISTICAL PROCESS CONTROL CAN BE USED TO MONITOR AND IMPROVE THE SERVICES PROVIDED BY HEALTHCARE ORGANIZATIONS

This chapter calls attention to the methods and models to monitor, diagnose, and improve the outcomes associated with healthcare service systems. Quality analysis requires the analyst to define quality, gather data related to measures used to quantify quality, and analyze the data to arrive at conclusions related to the quality of system outcomes. These conclusions then become the basis for system change.

Table 14-1 Number of Patients Who Waited (Data: Clinic Records for

Jun

e

)

 4

 8

16

 

 

Day	Number of Patients Who Waited
1		12
2			16
	3	26
		4
5		8
6	17
	7	13
9	22
	10		18
Mean = 15.2
Standard Deviation = 6.40
Median = 16.0

As indicated in 

Figure 14-1

, service system outcomes are the product of the conversion of inputs (e.g., skills, resources) in accordance with a specific protocol. Skills and resources are used in accordance with some form of protocol to achieve desired outcomes. However, all outcomes may not be desired because of the malapplication or misapplication of the inputs. This is the classical systems model.

Figure 14-1 General Systems Framework

Every healthcare organization is a series of interrelated and interlocked service systems. Outcomes associated with one system are inputs for other systems. For example, the meals served by a hospital’s dietary department can be considered the outcome associated with the dietary subsystem in the hospital. When we look at the patient care subsystem, the meals served are one input that (should) contribute to intended patient care outcomes. The outcomes of one system are the inputs in other subsystems.

For our purpose, quality is defined as an intended system or subsystem outcome. It is the outcome intended by the service system and can be influenced by a change in the inputs and/or processes (e.g., protocol) used to convert inputs into the intended outcome. This is an important point as it is based on the premise that quality (or lack of quality) is a product of the inputs and processes used to convert inputs into system and subsystem outcomes.

Quality has many dimensions. What constitutes quality to patients may be different from what constitutes quality to physicians, nurses, therapists, and technicians. Quality analysis defines quality in measurable terms, monitors and measures it, and then subjects these data and information to analysis to improve quality.

Although quality has many dimensions, to manage and analyze quality requires that it be defined in measurable terms. In other words, the methods used to analyze quality require measurable definitions. Some refer to these as quality characteristics. For example, patients entering a clinic for an appointment to see their primary care physician do not want to wait. When they wait, they may ascribe less quality to their appointment than when they do not wait. We are able to count this quality characteristic (the expectation of not waiting). We can count those patients who do wait and do not wait as well as count waiting time, the time a waiting patient waits.

The first step in quality analysis is to identify quality indicators based on quality characteristics, such as waiting for an appointment. This identification must define these indicators in sufficient detail so they can be measured and counted.

Once the quality indicators are specified, data need to be collected. Because these data will be the basis for conclusions regarding the service system, the data acquisition must be reliable, valid, and include an appropriate amount of data gathered using scientific rules of sampling.

Statistical process control provides the manager with the tools needed to monitor, diagnose, and (potentially) improve the outcomes associated with healthcare service systems, such as systems involving patient care, general administration, or both. These methods are most commonly found associated with continuous quality improvement (CQI) and total quality management (TQM) methods and initiatives. These methods call attention to processes used to convert inputs into outcomes. Instead of focusing on who does what (e.g., job descriptions) in what order to create an intended outcome, these methods focus on the “overall conversion process” and the outcomes associated with current systems.

Systems and subsystems do not always produce the same outcome. Again, this is a very important point. Outcome variability occurs in any system or subsystem. As such, even though we define quality as the intended system or subsystem outcome, we recognize that outcomes will vary. A certain amount of variation in system outcome is natural and expected. This leaves the analyst with the responsibility to discern whether any outcome variability is natural variability or (conversely) a signal of the system producing outcomes that fail to meet expectations.

Statistical process control recognizes that all systems produce variation in their outcomes to some degree. The vital analytic task is to discern whether the variation is within acceptable limits (of variation), or an indication that the system used to create the intended outcome needs to be revised. In this regard, statistical process control provides analytic approaches to assist the analyst to decide whether outcomes are within acceptable limits and whether the variation constitutes appropriate grounds for system modification or is a false-positive—a potentially false signal that system modification may be needed.

Once appropriate quality measures are identified, statistical process control provides the basis to statistically describe actual measurements and compare actual measurements with intended goals. It also provides the context to monitor and evaluate outcome variations. This gives the analyst the ability to identify natural system variation and variation that signals the need for formal intervention.

LEARNING OBJECTIVE 2: TO DESCRIBE THE USE OF TOTAL QUALITY MANAGEMENT AND CONTINUOUS QUALITY IMPROVEMENT METHODS AND MODELS TO ENHANCE SYSTEM PERFORMANCE

CQI and TQM are methods used to diagnose and improve systems performance. They focus on adapting inputs and conversion processes to realize new or revised system outcomes. These methods and models must be a part of the repertoire of the quality analyst.

CQI and TQM focus on what is done (i.e., tasks), by whom (i.e., responsibility), and in what sequence (i.e., process). They identify modification that will improve the system outcome. The names CQI and TQM demonstrate their orientation to “quality,” even though both can be used to improve service efficiency and effectiveness as well as service quality. The use of these techniques in health care has been influenced by many factors, including quality improvement as an accreditation requirement in hospitals and other healthcare providers. In some instances, corporate healthcare systems have adopted TQM as the company-wide approach to identify and improve patient care in all aspects of operations. They are very appropriate techniques to improve systems that involve many servers.

In health care, TQM represents a significant change in orientation involving the provision of clinical care and may represent a true management innovation. It focuses the attention of clinicians and managers on the total process of providing patient care. It de-emphasizes focus on specific individuals and departments in the functionally organized bureaucracy. It recognizes the responsibilities assigned to specific individuals as part of the total process or system. It is a systems-oriented model that builds the capacity to analyze, design, or redesign and implement with some important modifications and considerations.

TQM acknowledges the difference between clinical quality and service quality and offers methods to examine the processes used to provide services to patients. Quality is viewed as a system outcome that can and must be controlled and managed. As a management method, it forces systems to specify indicators of quality, monitor themselves against these indicators, and identify and correct extreme variations. TQM strives to improve the quality of patient care from both a clinical and service perspective.

TQM is a formal methodology designed to improve the operation of a system. It focuses on outcomes and strives to identify strategies to change conversion processes to improve service outcomes. As such, the quality of service can be improved using TQM. Efficiency can be improved using TQM. In other words, TQM is a systems improvement methodology. As methods to analyze and design systems, TQM is based upon specific concepts and skills. TQM has a specific language and requires users and those required to train and support users to be able to use specific techniques.

Brainstorming

Using TQM requires all workers to think, manage, and work. It does not segregate by level or role within the organization. Brainstorming is a technique used to facilitate group thinking. It is frequently used with a facilitator. Brainstorming is the process of collecting ideas from all members of a group without rendering any judgment or evaluation of the ideas. Participants are free to offer anything either new or something that builds upon the comment of others. Brainstorming usually separates discussion of the ideas from the presentation of the ideas. Group process techniques suggest that results be recorded on a neutral space, such as a flip chart. Participants should focus on the neutral space to guard against any negative or positive reaction to any specific idea. This includes individual reactions that may involve subtle body language. Brainstorming can be used to identify problems and develop solutions to problems. A typical question could be, “How could we improve the quality of service rendered to our patients?”

Consensus Building Techniques

Brainstorming puts ideas on the table for the group to consider. Different perspectives often conflict. Processes to identify ideas can only be considered successful when the process facilitates and builds consensus and ownership of the ideas. Consensus building techniques are used to avoid individualized argument and confrontation, maintain the substance of the ideas as the focus of the process, and identify those ideas that the group can agree are important and relevant. Under the heading “nominal group techniques” a manager can find numerous techniques and games a facilitator can use to build and identify group consensus. Consensus is not total agreement or the agreement of a simple majority. It is the willingness of the group to own the idea. This is a very important distinction. Typically, nominal group processes use multiple rounds of voting and discussion until the group (not just a majority in the group) is comfortable that their ideas have been heard and the group’s ideas are ones that they can own and embrace. As discussion evolves, it is sometimes necessary to halt activities and add to the group representatives from other parts of the organization. It is essential that the group include experts from those parts of the organization included or implied in the desired improvement. The membership of the group as well as the facilitator’s ability to insure a neutral environment is essential for the success of brainstorming.

Force Field Analysis

Force field analysis (FFA) is used to create group consensus as well as to examine problems and issues the group feels merit improvement. It requires the group to identify “driving forces” that the group believes are causing the need to change or “the problem.” It also requires the group to identify the “restraining forces” that are impeding the ability to change. Users generally believe that working to eliminate a restraining force is more successful than enhancing a driving force. The group then lists the desired improvement at the top and then identifies “driving forces” and “restraining forces.” Typically, specific driving forces are linked with specific restraining forces. The technique forces participants to broaden their thinking about a specific improvement and begin to identify and build a strategy (

Figure 14-2

).

Figure 14-2 Example of Force Field Analysis

Force field analysis is problem oriented. It requires the group to specify the problem and list the forces that are causing the problem (driving forces) and those forces that are preventing it from being solved (restraining forces). It is important that the group concentrate its attention on forces that are under the control of the organization. As a technique it can elicit many or a small number of forces. Unlike the fish bone chart discussed in the following, which imposes a classification approach on the group, a force field analysis provides the group the ability to analyze a problem from many perspectives.

The

Fish Bone Chart

Often referred to as a cause-and-effect diagram, the fish bone chart is used to portray the group’s thinking of the factors that are contributing to a specific problem or current operation. Fish bone charts force comprehensive thinking and a comprehensive diagnosis of the problem. Groups prepare fish bone charts as part of the process used to analyze the problem. Unlike force field analysis, which identifies driving and restraining forces, a fish bone chart portrays all the factors related to a specific outcome. Two common frameworks exist. One arranges the factors that contribute to the problem using the four categories: Equipment and Supplies, Procedures, Policy, and People. The other framework uses the four categories: Methods, Machinery, Manpower, and Materials.

Using one of these frameworks, the quality analyst indicates the specific factors, such as a specific policy or work procedure that could be the cause of the problem. Some may see the fish bone chart as a comprehensive causal map that indicates the factors that are combined to create a specific outcome. The previous figure indicates that the purpose of the analysis is to “identify the amount of time a patient waits before being seen by an attending physician.” The fish bone chart identifies the factors that influence this outcome, using four categories:

· 1. Procedures

Procedures of New Patient Processing

Procedures for Existing Patients

Procedures for Medically Ordered Return Visits

· 2. Equipment and Supplies

Computer Systems and Appointment Software

Medical Record

Telephone

· 3. People

Job Description and Role of the Receptionist

Job Description and Role of the Nurse

Job Description and Role of the Physician

· 4. Policy

First In First Out (FIFO) Patient Processing

All Medicaid Patients Referred to Local Emergency Department

Ability to Pay Determined before Service

Policy on Serving Patients with Outstanding Bills

Figure 14-3 Example of a Fish Bone Chart

Although this list is not comprehensive, each of these factors contributes to the outcome listed in 

Figure 14-3

. Used appropriately, the fish bone chart depicts the factors that influence the specified outcome. Stated another way, the specified outcome would not occur unless all the factors listed in the fish bone chart interacted to cause it.

Fish bone charts are used to break a problem into its component parts. They focus the attention of the group on the problem and require the group to construct a comprehensive diagnosis of the problem. They force the group to consider the many potential causes of the problem, not just a few that may quickly come to mind. Starting with the outcome, they create a retrospective map of the factors that make the outcome. By accomplishing this, the fish bone chart calls attention to specific factors that can be changed to change the outcome.

Pareto Chart

s

A Pareto chart is a vertical histogram that lists the most common problems or problem causes in descending order from the leftmost margin of the chart. Pareto charts are based upon the belief that, in general, “80% of the trouble comes from 20% of the problems.” The purpose is to focus on the major problems or the major causes of the problem (

Figure 14-4

).

Like all histograms, these charts are efficient approaches to visually present relative frequencies of events. They are easy to read as long as the number of bars is kept to a minimum and labels are used to present the scale, the title, and the legend for each individual bar.

Figure 14-4 Example of a Pareto Chart

Run Charts and Control Charts

Run charts are used to illustrate patterns of data collected over time. They are intended to indicate patterns. Run charts can be used to monitor a system over time and plot occurrence against the average or desired average or desired level of service to monitor performance. They are also used to identify when a system is not in compliance with a desired outcome or process indicator. Over the long term, the run chart provides a data image of the system in operation. Such a chart can indicate when the system is functioning within acceptable limits and when the overall average changes. In general, functioning systems should yield data points above and below the average. When a system begins to consistently yield data points above the average, this may mean that the average is shifting up. Conversely, when the system begins to demonstrate a pattern of data points consistently below the average, this may mean that the average is shifting down.

Run charts are used to monitor systems. They report the status of a system as well as trends that have or may be developing. When desired levels of service are added (in place of the average), run charts provide the ability to visually inspect and evaluate the system. Control charts take the approach one step further. A control chart adds to the data plot control limits. In some instances these limits are based on standard deviation. When standard deviations are used, a line that represents +1.96 and −1.96 standard deviations is added to the chart. Deviations beyond this line represent, by definition, abnormal events. Consider the following data in 

Table 14-2

.

Table 14-2 Data for a Control Chart Number of Medically Complicated Births, Durham Hospital

 25 

 17 

		Month	Number of Medically Complicated Births
		Jan uary		19
		Feb ruary		27
		Mar ch		20
		Apr il		16
			May		18
	June				25
	Jul y		22
	Aug ust		24
		Sep tember				17
		Oct ober
		Nov ember		15
		Dec ember
		Total		245
Mean		20.4

In this example (see 

Figure 14-5

), the bold horizontal center line is the plot average (i.e., 15.2 patients). The solid lines at the boundary of the data are the 95% confidence interval (i.e., average +1.96 and −1.96 standard deviations; 1 standard deviation for these data is 6.4 patients).

The Flow Chart

A general systems flow chart is used to describe and analyze work processes. Such charts create an image of the steps used in a work process and the decisions that create branches in the work processes. 

Chapter 3

 provides instructions and examples of general systems flow charts.

In a TQM environment, the general systems flow chart is used to describe what is the current process and system. A second chart is then prepared by the group to describe how the system or process should function. The two charts are then compared to indicate what changes must be made. As with most applications of general flow charts, the chart is used for analysis as well as design or redesign. Other types of flow charts are also used. A workflow analysis chart is used to describe each step in a work process by workstation. For example, a patient’s bill is followed between and among the many desks or workstations in the business office. A workflow analysis chart is used to identify redundant operations and to develop work flows that are natural loops through a well-arranged office. Typically, a simple chart is drawn to indicate how an office is physically arranged. Flow lines are added to indicate how work flows between desks.

Figure 14-5 Example of Control Chart Using Standard Deviations

A development chart is used to assign or analyze responsibility for different process steps in the overall work process. Typically such a chart lists the name of the worker across the top horizontal axis of the chart. Under each name is the list of responsibilities each work is assigned in the process. The chart is used to analyze current responsibilities as well as revise work assignments.

The

Scatter Diagram

Scatter diagrams are used to illustrate the relationship between two variables or process characteristics. At best, such charts can suggest associative properties. Such charts provide no basis to conclude a casual relationship. Any scatter diagram should report the correlation coefficient between the two variables.

Scatter diagrams create a cloud of data. The cloud can suggest a negative, positive, or no correlation between the two variables or characteristics. By providing a picture, they are very efficient at calling attention to the relationship between variables. Consider the following example of a scatter diagram (

Figure 14-6

).

This chart indicates the graphical relationship between the total number of visits and the total number of patients who waited over a 20-day period.

Figure 14-6 Example of a Scatter Diagram

The previous eight items constitute the primary methods and language of TQM. Each technique provides a unique ability. To reiterate:

Consensus Building Techniques

Force Field Analysis

Technique	Use
Brainstorming	To generate ideas
To create group ownership and consensus
To help create group consensus
Fish Bone Chart	To determine the components of a problematic outcome
Pareto Chart	To illustrate frequency of outcomes
Run and Control Charts	To illustrate trends over time
Flow Charts	To illustrate steps and sequence
Scatter Diagram	To illustrate (potential) relationships

TQM is a collection of formal management methods, concepts, and models. Many practical guides indicate different steps to follow to use TQM. In general, all can be subsumed under the following approach.

Prepare the Organization for Total Quality Management

The organization must state the purpose of TQM as being to improve the quality of services rendered to patients. If TQM is perceived as a negative or punitive process, employees will resist its use. All must be prepared to use TQM and understand its language and methods. TQM is not something others do in the organization. Success requires that TQM methods become the quality management approach used throughout the organization.

Assign Responsibilities

As a management method, TQM is only as good as the teams that use it. Teams are assigned responsibility for a specific problem or outcome. The membership of each team must be comprehensive to ensure that the team possesses sufficient expertise to understand and address its responsibility.

Identify Indicators of System Processes and Outcomes

Under this step teams are required to analyze specific processes of the organization, define what is meant by “quality,” and develop measurable indicators of system performance. To develop indicators, teams must know and express the goals of “their” system and be able to assess the degree to which these goals are being achieved. Indicators are quantitative. They are the product of counting.

Teams use brainstorming, consensus building techniques, flow charting, and force field analysis to develop their understanding of the system being studied. Fish bone charts may be prepared to develop understanding of the factors that contribute to a specific indicator or outcome. For example, hospital-wide indicators could be:

· • 

Average

length of stay by diagnosis

· • Hospital readmission rate within 30 days

· • Expenses per patient day

· • Revenue per patient day

· • Uncompensated care per patient day

· • Appropriateness of posthospital placement

· • Number of patients awaiting transfer to a nursing home

· • Full-time equivalent (FTE) personnel per patient day

For specific subsystems in a hospital, indicators could be the number of:

· • Surgical deaths

· • Postoperative surgical infections

· • Units of blood used

· • Medication errors

· • Medical records awaiting signature by attending physician

· • Imaging procedures repeated

· • STAT laboratory tests

· • Meals served per patient day

· • Hospital bills unresolved after 30, 60, 90, and 120 days

· • Safety incidents

Indicators should be specified by the team responsible to diagnose and enhance system quality. Management should not mandate indicators. The team should be composed of individuals who possess the expertise to know the best and most appropriate indicators of system performance.

Collect Data on the System or Work Process

At this phase in the TQM process, data are collected to indicate system performance. Frequency and level are important. Pareto charts are constructed. Run and control charts illustrate trends over time. Scatter charts illustrate relationships. Flow charts describe how the system currently functions. In some instances, the team develops a questionnaire or data protocol to be used to collect data. For example, in analyzing surgery, the TQM team authored the following questionnaire:

· 1. Does the patient arrive ready for surgery? If no, what does the patient need?

· 2. Is the surgical suite ready when the surgical team is scheduled to occupy it? If no, what is preventing their use of the room?

· 3. How long do patients have to wait?

· 4. Is the surgical team performing the surgery in keeping with existing time estimates? If no, is there a problem with materials, machinery, manpower, or materials?

· 5. What are the problems?

These questions could be the product of a previous force field analysis or analysis of the problem using the fish bone chart. Sometimes historical records may yield data. In other instances, the TQM team arranges to have the data collected.

Evaluate the System or Work Process: Formal Evaluation Begins

The team, based upon its previous work and data, evaluates the operation of the current system or work process using the chosen measurable indicators. Throughout, the team considers many ways to improve the system or work process. Those parts of the system or work process that cause the most problems are focused upon. Pareto charts are used. Control charts are used to evaluate system performance against averages and determine whether performance is changing and/or is within acceptable limitations.

Improvements Are Designed and Implemented

Teams design improvement strategies. Fish bone charts may be used. Flow charts may be used to identify exactly where the improvement should be made and the specific improvement to be made.

Evaluation Continues to Determine Whether the Change Worked

Once begun, TQM is a continuous process of modification and evaluation followed by modification and evaluation. When a team has completed one set of responsibilities, the responsibilities of the team can be revised and enlarged.

Overall, TQM provides a formal approach to analysis, design, or redesign and implementation within a systems context. Numerous methods exist to facilitate TQM. By focusing the attention of the team “on the system,” TQM is a formal management method designed to use the expertise included in the team to analyze the current system and recommend improvements. In principle, TQM is not revolutionary. It is a recognized best practice associated with systems management. It provides a formal method for analysis, design, and implementation. It focuses attention on goals and indicators of performance. It requires teams of first-line experts to analyze a system or work process before suggesting a fix or improvement. By focusing on process, TQM does reorient management in most healthcare organizations. Most healthcare organizations are organized based upon function (e.g., laboratory, medical records, nursing, rehabilitative services, housekeeping, etc.). Labor has been divided by a common expertise. Most healthcare organizations are not organized by service or product line (e.g., maternity care, cardiac care, etc.). As such, TQM enhances the ability to coordinate services that require the coordinated interplay of many service stations located in many functionally organized departments. TQM focuses on the process of providing service. Under TQM, the needs of the department are secondary to the needs created by these processes. Lastly, as a philosophy, TQM challenges everyone to always do better. It holds everyone responsible to look for ways to improve services and provides an outlet for their input and wisdom. Whether TQM dilutes the authority, power, and prerogative of senior management is an open question. It clearly empowers work groups to analyze and improve system performance.

TQM is based upon the actions of the group or team. TQM is not done alone in the privacy of an office facing a computer screen. It requires the ability to work effectively with groups. Frequently these groups have diverse backgrounds, values, and perspectives. Some professions seem to always argue, even on the TQM team. The ability to use TQM and help others use TQM requires the skill to work within a diverse group and facilitate a group. TQM can also require the ability to analyze, display, and present data. Use of statistics can become important. As TQM has become popular, many have presented different twists and subtle refinements. Whether significant difference exists is a matter of belief and perspective. For example, CQI stresses that TQM is a continuous process of improvement. It also stresses the need to change the culture of the organization. Whether CQI and TQM are significantly different is a matter of opinion.

In the process of collecting and analyzing data, TQM teams are doing a form of applied research. In some instances, they are using data from a sample, not a population. When this is the case, teams may need advice concerning the appropriate way to extract a sample from a larger population and how to generalize the sample to the population. However, TQM is not scientific research. For example, a TQM team found that in 100 surgeries randomly selected out of

11

45 done in a month, 32 began more than 30 minutes late. The results were disputed by the surgeons as being based upon too small a sample taken from only 1 month. The dispute and the results both have merit. It is beyond dispute that the 32 cases actually did begin more than 30 minutes late. The 32 late cases are a reality. They occurred. As such, there is an opportunity to improve the process of surgery. This is beyond dispute. Whether 32% of all surgeries began or will begin 30 minutes late, however, as a general conclusion or interference based on this data, may require additional studies with larger samples. However, fixing the system to correct for late surgeries remains an opportunity. The only point of dispute is the severity of the problem. There should be no debate as to the existence of late-starting surgeries. Remember a cardinal principle of TQM, “If it ain’t broke, improve it anyway.”

LEARNING OBJECTIVE 3: TO ANALYZE SERVICE SYSTEMS USING RUN CHARTS AND CONTROL CHARTS

This section addresses how to use basic time series analysis, run charts, and control charts as methods to monitor outcomes associated with either patient care or management systems.

Time Series Analysis

A time series analysis is a chart that presents a specific occurrence over a period of time. In a time series analysis, “time” is always the horizontal axis on the chart. The purpose of a time series analysis is to show the amount and direction of change over the time period included in the analysis. For example, consider the data in 

Table 14-3

.

Presented as a traditional time series analysis, these data demonstrate month-to-month variation (

Figure 14-7

).

Table 14-3 Number of Medically Complicated Births with Moving Range, Durham Hospital

Month

 19 

 

 27 

 20 

 16 

May

 18 

 25 

 7 

 22 

 24 

 2 

 17 

 7 

 25 

 8 

 15 

 17 

 2 

245 

 20.4

Birth	Moving Range
Jan
Feb		8
Mar			7
Apr	4
		2
Jun
Jul	3
Aug
Sep
Oct
Nov	10
Dec
Sum	60
Average	5.5

Figure 14-7 Example of Time Series Analysis I

Example of Time Series Analysis I

As stated, time series analysis is used to describe how occurrences change over a period of time. Time can be presented in any unit, such as days, weeks, months, quarters of years, and years. Time series analyses help to describe occurrences and portray the variation in the occurrences. For example, 
Figure 14-7
 demonstrates the number of medically complicated births at a hospital over a 12-month period. Medically complicated births ranged from a low of 15 in

November

to a high of 27 in

February

. On average, using the mean, 20.4 medically complicated births occurred at this hospital over this time period.

Although this average describes the pattern of occurrence, it does not indicate that the hospital experienced 20.4 medically complicated births every month. The average of 20.4 medically complicated births indicates that during these 12 months, on average, the hospital experienced this number of medically complicated births. Averages only describe the middle of a data distribution.

Using traditional statistical approaches, the next step is to describe the data in 
Table 14-3
 with its calculated 95% confidence interval using standard deviations. We know that in most distributions 95% of the data lie between the mean or average ± 1.96 standard deviations. In this case, our (sample) standard deviation is 4.1 births. As such, to further describe these data, we indicate that 95% of the time the number of medically complicated births was between the average (20.4) and + and − 8 births (e.g., standard deviation of 4.1 × 1.96). Expressed this way, our 95% confidence interval is 12.4 and 28.4 medically complicated births per month.

When this 95% confidence interval is added to the time series analysis, it can be seen that the number of medically complicated births reported in each month falls within this 95% confidence interval. All values were either more than 12.4 births or less than 28.4 births. In no month during this time period did the number of medically complicated births fall outside the range of the 95% confidence interval (

Figure 14-8

).

Figure 14-8 Example of Time Series Analysis II

Example of Time Series Analysis II

When time is involved, a time series analysis reveals more than traditional statistics (e.g., averages, standard deviations, and 95% confidence interval) about the occurrences being reported in the data. The calculated statistics (e.g., mean, standard deviation) can hide a very critical feature of the original data. The statistics alone can hide the month-to-month variation in the data. If there were no month-to-month variation in the data, however, then either approach would be acceptable. If variation is present, the time series analysis is the more accurate presentation. It preserves the evidence of the data. Preserving the evidence of the data is one of the strongest attributes of a time series analysis.

Analytic Methods: The Run and Control Chart

In an ideal world there would be zero medically complicated births. In the real world, however, medically complicated births do occur. As such, a hospital must be prepared to respond to their occurrence. If, using the previously cited data, the hospital prepared to respond to 20.4 medically complicated births, for some months the hospital would have been prepared and for other months would have been underprepared. The occurrence of medically complicated births, as described in the time series analysis, varies by month. It is not a steady state.

Too often, it is assumed that any month that had more than 20.4 medically complicated births was a “bad” month, and any month in which there were fewer than 20.4 medically complicated births was a “good” month. However, these assumptions do not recognize the natural variation that characterizes the number of medically complicated births per month. The run and control charts systematically describe this month-to-month variation and provide the ability to deduce specific conclusions based on the pattern of variation around a mean value.

To respond efficiently and effectively, the quality analyst must be able to respond to individual occurrences and the pattern in which they are encountered. As such, variation is very important to managers. As stated, period-to-period variation is a natural characteristic of most occurrences presented in a time series analysis. Although traditional statistics do provide a measure of this variation—the variance and standard deviation—these statistics provide no way to describe the actual month-to-month variation in the data. They mask it. As also stated, TQM is based upon the premise that variation is a natural phenomenon and should be used, not masked, as we strive to improve the systems we manage. From this perspective, variation is an expected characteristic of any process, and we need an approach that will help us analyze the variation inherent in any process. A certain amount of the expected variation of occurrences is random; it happens by chance. There is nothing we can or should do to try to control this type of variation except to build into our management and clinical systems the ability to respond to it. As managers, we should expect this natural amount of variation. Another amount of this variation, however, may represent a signal that a fundamental change or shift is occurring, requiring a change in the process. The critical term here is “signal.”

The primary challenge is to differentiate validly between the natural or random variation expected in all occurrences examined over time (e.g., time series analyses) and variation that constitutes a signal of a fundamental positive or negative change requiring management action. TQM tells us that variation is natural and we need to learn from the variation about the systems or processes that produce these occurrences. We must redesign systems only when the variation signals us that fundamental negative change is occurring. TQM is also telling us that random variations cannot be described or analyzed using traditional statistics because the (random) variation does not adhere to the requirements of the Central Limit Theorem and the Standard Normal or Beta or Gaussian Probability distributions. In other words, traditional averages or means, standard deviations, and 95% confidence intervals are irrelevant. These statistical tools cannot be used to describe and analyze variation. More importantly, they cannot help to discriminate between natural variation and variation that signals a fundamental change.

The Run Chart

The run chart is a time series analysis. Its horizontal axis is time and its vertical axis is the data axis. The chart is a data plot; it usually includes a minimum of 12 observations. The data points are connected.

An Example of a Run Chart

The center line of the run chart distinguishes this chart. The center line is the median value of the data. In 

Figure 14-9

 the median is 14. The median is used to represent the middle of the data distribution. The mean is not used because a mean can be overly influenced by a small number of very low or high data values.

A run is defined as one or more consecutive data points on the same side of the median. Data points on the center line are ignored. The run chart allows the quality analysts to draw specific conclusions about the data based on the number of runs presented by the data. Once a run chart is prepared, the analyst:

· 1. Counts the number of appropriate data points (ADP). This equals the total number of data observations minus the number of observation on the median line on the run chart. In 
Figure 14-9
, the ADP is 17. Seventeen of the data points are not on the median line.

· 2. Estimates the lower level (LL) and upper level (UL) number of runs using the following formula:

 

 

 

 

=

=

 

=

UL = (0.59 × ADP) + 2.70
LL = (0.41 × ADP) − 1.78
In  Figure 14-9 , ADP = 17,
then the UL				=	(0.59 × 17) + 2.70
12.73 or 13
then the LL	(0.41 × 17) − 1.78
5.19 or 5

·

· Figure 14-9An Example of a Run Chart

· 3. Counts the number of runs (R), defined as the number of one or more consecutive data points on the same side of the median. 
Figure 14-9
 demonstrates eight runs.

· 4. If the number of Rs is lower than the LL or greater than the UL, the variation is based upon a change in the process and warrants additional and detailed analysis. If the R is between the UL and LL, the variation is considered natural, and further analysis is not needed.

· 5. The LL for 
Figure 14-9
 has been calculated as 5, and UL as 13. If the number of runs is less than 5 or greater than 13, then a fundamental change has occurred that warrants further detailed analysis and potential system change. Because the data in 
Figure 14-9
 indicates eight runs, the analyst would conclude the variation in the number of falls is within natural limits and no systematic change is needed.

· 6. Other points to consider signals of process variations that warrant additional analysis include:

· a. If any one run presented by the data has 7 (when ADP is less than 20) or 8 (when ADP is 20 or greater) data points.

· b. If any one run has 14 points that consecutively zigzag the median.

· c. If any run, including the data points on the median, have 6 or more consecutive increasing or decreasing values.

The run chart provides analysts with a systematic approach to assess variation. This assessment of variation is the basis for additional analysis using TQM tools and techniques.

The Control Chart

Run charts and time series analyses can be used to describe the variation in occurrences. More important, control charts can be used to analyze this variation. A control chart tracks variation to determine whether it occurs within predetermined boundaries or limits (natural limits). These charts also provide the ability to draw certain analytic conclusions based upon whether the variation crosses the threshold boundaries used on a control chart. When the variation crosses the control chart’s boundaries, fundamental change in the system’s outcomes has occurred. Variation across or near a boundary is a signal to analyze the underlying system and potentially redesign it.

The concept of control charts is simple. It involves taking a time series analysis and adding appropriate boundaries, called control limits. The control limits represent the band of natural variation inherent in the time series analysis. When these natural variation boundaries are crossed, the analyst should look for fundamental change in the underlying process that creates the occurrences 
and (potentially) change the system. Given this approach, the challenge is to find an approach that can be used to establish the appropriate boundaries of the control limits. A control chart is created when these limits are added to a time series analysis.

Control charts provide analysts the ability to assess the variation in occurrences and use the variation as a signal that an intervention is needed. They are based on the fundamental premise that variation is a natural, not abnormal, characteristic examined over a period of time. Control charts accept this natural variation inherent in processes and systems and tell us when we need to act and when the variation being reported is natural and beyond the control of the organization and the manager.

There are many types of control charts based on whether the data are continuous or discrete. The following two provide the analyst with a starting point to use control charts.

· 1. Time series for moving ranges, also known as a range chart and a moving range chart

· 2. Time series for individual values, also known as an X chart

Control Limits for Control Charts

Control limits are the threshold boundaries that are added to a time series analysis to create a control chart. These limits, expressed as an upper limit and lower limit, are intended to provide managers an interpretative context. As stated, when the variation remains within the control limits, the variation is natural. No organizational response is appropriate. The underlying process or system is acting naturally and producing a natural level of variation. Natural variation occurs within the control limits. We always find natural variation. When the variation exceeds control limits and/or presents related characteristics involving these limits and the center line, then the quality analyst must act. The variation is no longer a natural property. It represents a dramatic signal that a fundamental change has occurred in the process or systems underlying the occurrences being examined. Again, notice the use of the term “signal.” The preceding two statements are the fundamental concepts underlying the appropriate generation and use of control charts.

When time series analyses present individual values, not averages or subgroup averages, the average moving range can be calculated and used to establish threshold values. The moving range is calculated by determining the absolute between a preceding and succeeding month (

Table 14-4

). For example, the number of medically complicated births in

January

was 19, and in February was 27. The moving range calculation is the difference between January and February, or 8. In

March

the number of medically complicated births was 20. The moving range, expressed as an absolute number, for February–March is the absolute difference or 7, not −7. A chart can be used to examine the moving ranges, the month-to-month variation that may be occurring.

Table 14-4 Number of Late Surgeries

Month

January

February

March

April

May

June

July

August

September

  89

October

November

 156

December

 

 

 

Total

Number of Surgeries	Number of Late Surgeries
43 5	112
40 1	12 9
572	186
409	103
577		89
329	67
467		156
301	9 4
23 5
325	127
378
444	124
4873		1432

Calculate the moving range by time interval (e.g., month), sum the differences, and then divide this sum by the number of range calculation. In the example, divide by 11 to determine the average moving range. For the example, the average moving range is 5.5 births.

The upper control limit for a moving range chart (of individual values) is the average moving range × 3.27—3.27 is a constant, just as 1.96 is the constant used to calculate a 95% confidence interval when we use the central limit theorem.

In our example, therefore, the upper control limit for the moving range chart is:

Average Moving Range = 5.5 × (the constant) 3.27 = 17.985

For this control chart, 17.985 is the upper threshold limit used to evaluate the variation presented in the time series analysis we have been using involving medically complicated births. It is the upper boundary or control limit. Given the nature of this chart there is no lower limit.

The center line for a moving range chart (of individual values) is the actual average moving range. When the change from one month to another is more than 17.985, a value outside the boundary of values established by upper control limits, then interpret this change as sufficient to justify some type of additional attention involving more detailed system analysis, design, and/or implementation.

When we use control charts, we also must use a time series for individual values, known as an X chart. To prepare an X chart:

· 1. Present a time series analysis that presents individual values over time.

· 2. Add to this time series analysis chart a center line based upon the average of all values. For our example, the average of all values is 20.42 medically complicated births.

· 3. Add to the chart lines indicating the upper and lower natural process limit. These limits equal the average of all values ± 2.66 times the average moving range. For our example, the upper natural process limit is:

Average of all values: 20.42 medically complicated births + 2.66 × (average moving range of 5.5) = 35.05

For our example, the lower natural process limit is 5.79. Rounded, the upper and lower natural process control limits for the X chart are 35 and 6.

Interpreting Control Charts

Use the following rules to interpret X charts and moving range charts. It is imperative that both types of control charts are prepared from the same data. Each type of chart presents different characteristics associated with the variation. An intervention is warranted when:

· 1. A single monthly value falls beyond a control limit. This constitutes a signal that a fundamental change may be occurring.

· 2. At least three out of four consecutive values are closer to one of limits than they are to the center line. This also constitutes a signal that a fundamental change may be occurring.

These two preceding signals constitute warning signals that a change may be occurring.

· 1. Whenever eight or more successive values fall on the same side of the central line, fundamental change in the underlying process or system has occurred.

· 2. Whenever three or more successive values fall outside of the control limits, a fundamental change has occurred in the underlying process or system.

The occurrences being analyzed and the limit being violated (i.e., upper or lower) indicate whether the change is positive or negative.

Analysts need the ability to identify and quantify measures of quality, establish systems to monitor quality, and analyze quality use using scatter, run, and control charts. When signals indicate that variation is not natural, the analyst must employ diagnostic tools such as flow charts, fish bone diagrams, force field analysis, and Pareto charts. To develop system revisions and/or to enhance the quality of outcomes, analysts also use brainstorming and consensus building techniques as well as general systems flow charting.

EXERCISES

· 14-1 Do a force field analysis (FFA) on the driving and restraining forces that influence your ability to do well in a specific academic course. One driving force may be “your desire to learn.” One restraining force may be “your need to devote time to other work.”

Table 14-5Number of Student

Complaint

s by Complaint Type

September

16

October

 7

November

 34

December

 40

12

 34

  9

January

 13

10

 45

February

 20

 4

 13

  9

March

 23

18

 23

1432

April

 31

 3

 14

 13

May

  4

 12

Total

Complaint Type	1	2	3	4	5	6	Total Visits
		34	38		45				13	103	824
		23	24		40			9	56	956
	14	50	11		20	26	1167
67	17	1034
	31		12	15	645
60	6	1645
97		4	3
43	0	745
25	0	16	1	456
241	396	81	196	139	257	8904

Complaint

· 1. The quality of service received

· 2. Waiting time was too long

· 3. Follow-up care was not available

· 4. Clinic was hard to find in the building

· 5. The medical care/treatment took too long

· 6. They could not find my medical record

· 14-2 Using the following data, prepare a run chart, scatter chart, and control chart. A late surgery is defined as any surgical operation that was started more than 30 minutes after its scheduled time (
Table 14-4
).

· 14-3 The ambulatory health service at a university is experiencing an increased number of student complaints concerning the services it offers in its walkin urgent care clinic.

The Basis of Complaint

· Type 1 The quality of service received

· Type 2 Waiting time was too long

· Type 3 Follow-up care was not available

· Type 4 Clinic was hard to find in the building

· Type 5 The medical care/treatment took too long

· Type 6 They could not find my medical record

Using these data, select a complaint for analysis. Your analysis must include a fish bone chart as well as other types of charts and techniques you deem necessary to appropriately analyze this data (

Table 14-5

). What do you recommend and why?

Case Study 2: Quality Analysis

Due Week 9 and worth 90 points

The ambulatory health service at a university is experiencing an increased number of student complaints concerning the services it offers in its walk-in urgent care clinic. Using the data in Table 14-5 on page 296 of the textbook, select a complaint for analysis. Your analysis must include a fish-bone chart, other appropriate charts (run and / or control), and any other techniques you deem necessary to analyze the data appropriately.

Write a two to three (2-3) page paper in which you:

1. Construct a fish-bone chart using Word or MS Paint.

2. Construct a run and / or control chart using Excel.

3. Recommend to the ambulatory health service on how it can improve the services it offers in its walk-in urgent care clinic, based on your analysis. Provide a rationale for your recommendation.

Your assignment must follow these formatting requirements:

· Be typed, double spaced, using Times New Roman font (size 12), with one-inch margins on all sides; citations and references must follow APA or school-specific format. Check with your professor for any additional instructions.

· Include a cover page containing the title of the assignment, the student’s name, the professor’s name, the course title, and the date. The cover page and the reference page are not included in the required assignment page length.

The specific course learning outcomes associated with this assignment are:

· Analyze projects using the tools of quantitative methods.

· Write clearly and concisely about quantitative methods for health services using proper writing mechanics.

Click 

here

 to view the grading rubric.

Due Week 9 and worth 30 points

· Homework

. Chapter 14: Exercise 14-2 (page 297 of the text)