House Of Quality Analysis In Health Care

House of Quality Analysis in Health Care Siamak Aghlmand Department of Health Management and Economics, School of Public Health, Tehran University of Medical Sciences, Iran, [email protected] & Rhonda Small Mother & Child Health Research, Faculty of Health Sciences, La Trobe University, Australia, [email protected]

Abstract Background: Adopting a formal and reliable method for linking patient requirements with the relevant performance measures of a care process is a top priority for high quality clinical care. Objective: To describe the concept and process employed in house of quality analysis, the heart of quality function deployment, by providing a step-by-step methodology and a case study from the maternity ward of Fayazbakhsh hospital in Iran. Methodology: We considered the house of quality analysis as a process with both input and output data. Major input data were patient requirements that were converted to key performance measures and targets, as principle output data, by two matrices including a relationship matrix and a correlation matrix. A case study: We illustrate the steps of translating the top 20 maternal requirements into six key performance measures throughout the house of quality analysis. Based on identified key performance measures, we also identified six necessary organizational functions to meet the 20 selected maternal requirements and increase maternal satisfaction. Discussion: The house of quality analysis provides a unique and rigorous method to translate patient information into relevant process performance measures. This is a key step in clinical process improvement. However, it is time-consuming and complex to adopt. Decreasing the amount of input data can simplify the house of quality analysis. Keywords: Quality, quality function deployment (QFD), house of quality (HoQ), maternity care, Iran

Introduction In view of the fact that meeting or exceeding ‘customer’/patient needs and requirements is essential to improve quality of care processes (Iacobucci et al., 1995), patient satisfaction has gained widespread recognition as a measure of quality in many health care organizations since the late 1980s. Nevertheless, care providers have given less attention to converting patient data into a set of useful decision-making information for quality improvement strategies to occur. The process for translating patient information into organizational terms for the improvement of care has become one of the key challenges in health care settings (Williams, 1994). Lack of action is partly attributable to entrenched attitudes, lack of interest, limited resources, restricted time, structural and cultural barriers, fear of negative experiences, and lack of experience in using quality tools and techniques (Bamforth et al., 2002 and Dodek et al., 2004). Quality function deployment (QFD) is a well-known product/service/process planning approach. It ensures that customer requirements are systematically taken into account throughout the product/service/process planning and design stages (Dodek et al., 2004 and Garon, 1992). QFD

-1-

emerged in Japan in the late 1960s and it was subsequently used by both manufacturing and service industries worldwide, yet its concept is still new in the health care arena (Akao et al., 2003). The house of quality (HoQ), as the heart of QFD, is a matrix that provides a conceptual map for process design and quality improvement. It is used as a construct for establishing priorities for process performance measures to satisfy customer requirements (Büyüközkan et al., 2004). Performance measures or quality attributes/characteristics specify what should be measured to predict customer satisfaction so that they are used to evaluate whether or not customer requirements are fulfilled (Madu, 2006). In summary, HoQ translates customer requirements into performance measures and their operational targets, in order to meet customer requirements and improve satisfaction with care (Terninko, 1997). Once customer requirements are combined with the process performance, the probability of experiencing real improvement is significantly increased (Lloyd, 2004). The result of HoQ analysis can derive the best combination of performance measures along with their target levels to design a process based on important customer requirements (Lin et al., 2006). HoQ also studies the relationship between the various elements of a system. According to general systems theory, external and internal environments interact with each other. In addition, the internal parts of a system interact with each other, and the interest of any part may conflict with the interests of other parts (Bertalanffy, 1950). HoQ quantitatively analyses the interaction between the outside of an organization (customer requirements) and the inside (performance measures). It also assesses the synergies and conflicts among the internal parts of an organization (Shin et al., 2000). The main objective of this paper is to convey the conceptual content and process of HoQ analysis and illustrate its application in a clinical area by providing a step-by-step methodology and a case study from the maternity ward of Fayazbakhsh hospital, an Iranian Social Security affiliated hospital in Tehran.

Methodology HoQ is a process with both input and output data (Figure 1). The input data are: 1. Important customer requirements along with their weight 2. Important performance measures 3. Benchmarking data (benchmarks) The output data are: 1. The weight and correlation values of performance measures 2. Key performance measures (with high-weight and high-correlation) 3. Target level for each key performance measure (Chaplin et al., 2000) Some primary activities should be performed before starting data collection. First, a cross-functional and multidisciplinary team must be assembled based on the process to be studied. Team membership must include people who know the process best. Just-in-time training needs to be provided for the team members. Topics for inclusion are the concepts and background of quality improvement, teamwork, HoQ, and the overview of statistical process control (SPC). The team members then review the process by creating a detailed flowchart (Brown et al., 2005). Subsequently, the team identifies the most important customer segment of the process (e.g., patient, physician, nurse or insurer). At this time, the team members can commence to gather the necessary input data of HoQ process in the following order: Important customer requirements along with their weight This step is the most important but also the most difficult and time-consuming stage of HoQ analysis because all the following steps depend on this stage. This step is called the voice of customer (VoC) analysis. Here, our aim is not to explain VoC analysis in detail and we only highlight some basic methodological issues. At the first stage, the team members interview a small sample of customers (15 to 20 people) and ask ‘why’ and ‘how’ they use the services. Customer requirements or demanded qualities (DQs) are then extracted by re-framing customer responses as brief positive statements. Subsequently, the team organizes DQs on a tree diagram to rank them by ‘analytical hierarchy process (AHP)’ in which the data are compared pairwise using a 1-9 scale (Terninko, 1997 and Chaplin et al., 2000).

-2-

The most highly ranked DQs, a maximum of 20, are used to conduct a baseline survey with a larger group of customers (representative sample) to assess: (1) their preferences tied to selected DQs, (2) their satisfaction level with given services in the study organization and its competitor(s), (3) the Kano levels of requirements, including assumed (basic), expected (revealed), and unexpected (exciting) requirements. The results of the customer survey are then entered into the quality-planning table (QPT), in which the weight of DQs is determined. To this end, the team identifies three variables including, target, improvement ratio, and sales point by assessing the survey results. A target value for a DQ is found by comparing customer preference, customer satisfaction, and the Kano’s level of the DQ, which is determined by a 1-5 scale. The improvement ratio is the ratio of the target value to the current customer satisfaction with the target organization’s services. Finally, sales point represents organizational ability to meet a DQ. The rating scales of 1, 1.2, and 1.5 are used to express no, medium, and strong ‘sales points’, respectively. The weight of a DQ is calculated by multiplying the DQ’s importance by improvement ratio, and sales point. The weight of DQs is also expressed as a percentage (DQs’ relative weight) (Chaplin et al., 2000 and Duhovnik et al., 2006). Finally, selected DQs along with their relative weight are entered into HoQ matrix (Figure 1) Important performance measures The team uses brainstorming to generate performance measures for each selected DQ. Brainstorming should include a review related to categories such as people, methods, equipments, materials, and the process environments. An affinity diagram and a fishbone diagram (a DQ as effect and relevant performance measures as causes) can be used to organize the generated ideas into like categories. Identified performance measures are subsequently ranked for each DQ by a set of appropriate criteria (e.g. effectiveness, cost, feasibility) using a 1-5 scale. Lastly, a few highly ranked performance measures, as important performance measures, are selected for HoQ analysis (Figure 1), while at least one important performance measure is kept for each of the selected DQs (Chaplin et al., 2000 and Brown et al., 2005). Benchmarking data The last input data of the HoQ matrix are benchmarking data (benchmarks). A benchmark is a measure of the best practice or performance standard against which an organization’s performance is compared. The first step in benchmarking is to identify a competitor organization(s). Then, another team must be organized at the competitor organization. This team assesses ‘key performance measures’ on their own process (this will be more fully described later) (Lloyd, 2004). As shown in Figure 1, selected DQs along with their weights, important performance measures, and benchmarks are entered into the HoQ matrix as the input data. The output data of the HoQ process are provided by two matrices (Figure 1): 1. Relationship matrix The relationship matrix, as the central part of HoQ analysis, evaluates the strength of linkages between DQs and performance measures. Selected DQs and important performance measures are entered into the rows and columns of the matrix, respectively. The strength of relationships is assessed by an asymmetrical four-point scale, which uses zero, 1, 3, and 9 to represent no, weak, medium, and strong relationships, respectively. Then, the absolute and relative weights of performance measures are calculated as follows: The absolute weight of the jth performance measure equals the sum of the values obtained by multiplying the relationship with the ith DQ by the corresponding relative weight of that DQ. The absolute weight is given by Equation (1): n

Pj = ∑ ( Di × Rij )

(1)

i =1

Where n represents the total number of DQs, Pj represents the absolute weight of the jth performance measure, Di represents the relative weight of the ith DQ, and Rij represents the relationship between the ith DQ and the jth performance measure (Lin et al., 2006).

-3-

Each relative weight of a performance measure is the corresponding absolute weight, which is rescaled as a percentage. The number of DQs predicted and how effectively they are predicted influences the weight of the performance measure. The higher-ranking performance measures highlight key points on the process where important DQs can be significantly met (Chaplin et al., 2000). 2. Correlation matrix A correlation matrix is constructed on the roof of the HoQ based on the relationships among performance measures. In fact, the synergies and conflicts among performance measures are studied by this matrix (Chaplin et al., 2000). A five-point asymmetrical scale is used to measure the strength of relationship between two performance measures. The rating scale 9, 3, 0, -1, and -3, represents a strong positive relationship, a positive relationship, the lack of any relationship, a negative relationship, and a strong negative relationship, respectively. The mean of values related to each performance measure (average correlation rate), are placed at the base of the corresponding column (Figure 1). The average correlation values close to 9, 3, 0, -1, -3 represent strong positive, positive, no, negative, and strong negative correlation, respectively (Lin et al., 2006). Key performance measures After determining the weight and correlation values of important performance measures, the team uses these criteria (i.e., weight and correlation) to select a few performance measures as ‘key performance measures’ from the list of important performance measures. The combinations of key performance measures must cover the whole range of selected DQs (Chaplin et al., 2000). For accurately calculating the extent of key performance measures, the current process performance must be evaluated in the target organization and its competitor(s) at the same time, using appropriate control charts. There are many different kinds of control charts based on the type of assembled data (e.g., measurable or countable data with Poisson or binomial distributions). Control charts show how a process is performing over time and whether it is varying within an expected range or not. A control chart has three basic components: (1) a centreline, usually the mathematical average of all the samples plotted; (2) upper and lower statistical control limits (= Centreline ± 3δ ) ; and (3) performance data plotted. Control charts distinguish between two types of variation: one caused by common causes and the other caused by special causes. When variation comes only from common causes, the data points fall within the control limits and the process is said to be ‘stable’. In contrast to common causes, special causes of variation are out-of-the-ordinary events that disrupt the usual flow. Points outside the control limits indicate special causes, which require immediate attention. After removing special causes and stabilizing process performance in the target organization and its competitor(s), the averages of data are considered as the values of key performance measures and benchmarks (Kelley, 1999). Finally, these values are entered into the base of the HoQ columns for analysis (Figure 1). The target values of key performance measures In the final step of HoQ analysis, the team begins to plan the actual services by setting target values for key performance measures. Benchmarking data are incorporated at this step. Target values for each key performance measure can be determined based on the following criteria (Chaplin et al., 2000): 1. The importance of the key performance measure 2. The gap between the levels of the performance measures in the target organization and the competitor organization (benchmark) 3. The technical capacity of the target organization to improve the key performance measure 4. The availability of the necessary resources At the conclusion of this step, key performance measures and their target values are identified as the output data of HoQ analysis. Lastly, the team can clearly define a set of organizational functions and tasks to reach the target levels of key performance measures for meeting customer needs and requirements.

-4-

A Case Study In October 2005, a team consisting of two physicians and five midwives was formed at the maternity ward of Fayazbakhsh hospital to set improvement targets for maternity care by HoQ analysis. This team mapped the process of maternity care and identified ‘women in labour’ as the key customer segment of the process. Between 31 January and 4 February 2006, the midwives conducted in-depth structured interviews with women following birth (n=18) to identify their needs and requirements. Through these interviews, 54 maternal requirements (DQs) were identified and numbered. Identified DQs were organized on a tree diagram to rank by analytical hierarchy process (AHP). From the rank order, the twenty highest-ranked DQs were selected to design a three-part questionnaire for collecting data from a large sample of women. To this end, a self-completed questionnaire, using Likert scale type responses was designed based on the previously identified top twenty DQs. The questionnaire was then piloted with 15 women. From piloting, the team could estimate the variance related to maternal preferences (=1.92) and calculate the required sample size (n=82 with α=0.05, S2=1.92, and d=0.3). After modifying three questions, the final questionnaire provided scale scores ranging from 0–21 and was found to be reliable, with a Cronbach’s α of 0.90 indicating very high internal consistency (DeVellis, 2003). Between 23 July and 19 September 2006, the questionnaires were voluntarily completed by a random sample of women (n=89), who had given birth at Fayazbakhsh hospital within the previous year. For weighting the 20 selected DQs, the median of the answers to the questionnaires, as the results of the survey, were entered into the quality-planning table (QPT). The team calculated the DQs’ weights by determining three variables i.e., targets, improvement ratios, and ‘sales points’ (Table 1). In the next stage, the team generated 160 performance measures using brainstorming and 20 causeand-effect diagram. The performance measures were then ranked by the consensus of the team members using the three criteria of ‘feasibility’, ‘sustainability’, and ‘intervention possibility’. By doing this, the twenty-seven performance measures which acquired the highest ranks, were numbered as ‘important performance measures’ for HoQ analysis. For weighting the first-rated performance measures, the 20 selected DQs along with their weights, and the 27 performance measures were entered into the rows and columns of the HoQ matrix, respectively. After determining the strength of relationship cell-by-cell between each row and column using the asymmetrical 4-point scale, the absolute and relative weights of the performance measure were calculated by the ‘QFD Designer V4’ software (Figure 2). For checking the synergies and conflicts among the 27 performance measures, each cell of the roof, which was the intersection of two performance measures, was assessed using the asymmetrical 5point scale. Next, the mean of scores related to each performance measure were considered as the average correlation rate of the performance measure (Figure 2). Subsequently, the team selected six performance measures as ‘key performance measures’ to evaluate the 20 selected DQs, based on the highest weight and correlation values, including: 1. % of clinicians using agreed clinical guideline 2. % of obstetricians present 24 hours a day in the maternity ward 3. % of women receiving labour and childbirth education at the antenatal unit 4. % of women with a companion in the postnatal unit 5. % of doctor visits where a folding screen was used in the labour unit 6. % of women who have access to warm water during their hospital stay In the next stage, data connected to key performance measures were collected in Fayazbakhsh and Babak (competitor) hospitals for a period of 25 days by a constant sample size per day (n=4). The team then studied the current process performance of maternity care at the two hospitals by developing ‘np charts’ for each key performance measure because the data related to all key performance measures had ‘binomial distributions’. The control charts showed the out of control points and instability in the process at Babak hospital (Figure 3). After removing special causes, the average of all the samples plotted, as the level of the key performance measures in Fayazbakhsh hospital and benchmarks from Babak hospital were entered into the base of the HoQ matrix. In the final stage, the team identified the target values for the key performance measures based on the resources and capacities of the study hospital, as well as the benchmarks (Figure 2).

-5-

By completing the HoQ analysis, the team established six necessary functions for maternity care at Fayazbakhsh hospital that would better meet maternal requirements and increase maternal satisfaction (Table 2).

Discussion The identification of customer requirements is only a starting point for quality improvement; the next important step is how to link the customer requirements with process performance (Lloyd, 2004). In health care, much patient data are gathered, but these are not usually connected with clinical processes and organizational functions. The applied method confirmed that HoQ provides a logical and systematic means to achieve this goal. The results of the case study also showed that we could meet the top 20 maternal requirements and significantly increase maternal satisfaction through just simple 6 functions with defined valid targets. We are not aware of any alternative methods or techniques to achieve these results (Ramaswamy, 1993). However, there is a clear need to test the impact of implementing the identified functions on maternal satisfaction in future studies. In spite of many successful applications of HoQ worldwide, there are some notable impediments to its adoption. The large size of the HoQ matrix and the time-consuming nature for the method make it challenging and inefficient to manage a large project, or to apply it in everyday clinical settings without considerable resources and organizational commitment (Shin et al., 2000 and Logan, 1997). Using a well-trained team with a good background in quality improvement as well as applying appropriate software can help to solve this dilemma. Since the number of the matrix cells is equal to the product of the rows and columns, so the number of customer requirements (DQs) and performance measures selected for HoQ analysis will determine the size and complexity of the matrix. Therefore, proper ranking of input data and selecting only a few highly ranked items can reduce the level of detail in HoQ analysis (Chaplin et al., 2000). It is also useful to anticipate the barriers of HoQ analysis before attempting to use this method (Dodek et al., 2004). Comparing our work with what has appeared in the literature, we found few health care related QFD articles in Medline and just a few papers published in industry journals. Most of the papers were not about clinical processes nor were patients the key customers. The review of the literature also revealed some methodological shortcomings that we have attempted to overcome in this study: Evaluation of current process performance For the purpose of process improvement through process redesign using QFD strategy, studying current performance of the process is mandatory. Otherwise, the process improvement will not be demonstrable. Control charts can be used to document that the process is stable, and QFD process can be applied confidently. Furthermore, we cannot acquire an accurate quantity of performance measures as well as benchmarks, without removing special causes. This important step is ignored in most published studies (Chaplin et al., 1999, Einspruch et al., 1996, Moores, 2006, Lim, 2000, Lim et al., 1999, Dijkstra et al., 2002, Radharamanan, 1996 and King, 1994). Benchmarking Although, identifying relevant benchmarks is required for objective goal setting, a technical benchmarking process has not been applied in most of the published studies. As a result, the target levels of key performance measures have been subjectively determined without using a set of valid criteria (Chaplin et al., 1999, Einspruch et al., 1996, Moores, 2006, Lim, 2000, Lim et al., 1999, Dijkstra et al., 2002, Radharamanan, 1996 and King, 1994). In conclusion, relating patient data to care processes is a key part of any quality improvement initiative. HoQ analysis, as the heart of the quality function deployment (QFD) strategy, is a wellestablished method for this purpose, although there is as yet little experience of it in the health care arena. Here, we have presented a step-by-step methodology, which proved feasible and useful in the context of improving maternity care at our study hospital. Whether the methods of our case study have value in other settings and with other clinical processes will only be seen with further work using the methods described.

-6-

Acknowledgements Our thanks are due to Tehran University of Medical Sciences (TUMS) and Iranian Social Security Organization (SSO) for financial support of this research. The authors are most grateful to Soheila Mohammadi, Faeze Bahrami, Tahmineh Farkhani, Parisa Ghafari, Zahra Pourkalhor, Hila H. Vaziri, and Shahnaz T. Zahrani for their invaluable contributions as members of the study team. We would also like to thank Fayazbakhsh Hospital and La Trobe University for their sincere collaboration in development of this research.

References Akao Y, Mazur GH. The leading edge in QFD: past, present, and future, Int J Qual Reliab Manag 2003;20:20-35. Bamforth S, Brookes NJ. Incorporating the voice of multiple customers into product design, P I Mech Eng B-J Eng 2002;216:809-813. Bertalanffy LV. An outline of general system theory, Br J Philos Sci 1950;1:134-165. Brown DS, Bonacum D, Vonderheide-Liem D. The Kaiser Permanente FMEA model- simplified for healthcare personnel. J Healthc Qual 2005;27:48-55. Büyüközkan G, Ertay T, Kahraman C, Ruan D. Determining the importance weights for the design requirements in the house of quality using the fuzzy analytic network approach. Int J Intell Syst 2004;19:443-461. Chang CL, Cheng BW. To establish a continuing care system of discharge planning by QFD. Total Qual Manag Bus Excel 2003;14:903-918. Chaplin E, Terninko J. Customer Driven Healthcare: QFD for Process Improvement and Cost Reduction. Milwaukee, WI, USA: ASQ Quality Press, 2000. ChaplinE, Bailey M, Crosby R et al. Using quality function deployment to capture the voice of the customer and translate it into the voice of the provider. Jt Comm J Qual Improv 1999; 25: 300-315. DeVellis RF. Scale Development: Theory and Applications. 2nd ed., London, UK: Sage Publications; 2003. Dijkstra L, van der Bij H. Quality function deployment in healthcare: methods for meeting customer requirements in redesign and renewal. Int J Qual Reliab Manag 2002;19:67-89. Dodek PM, Heyland DK, Rocker GM, Cook DJ. Translating family satisfaction data into quality improvement. Crit Care Med 2004;32:1922-1927. Duhovnik J, Kusar J, Tomazevic R, Starbek M. Development process with regard to customer requirements. Concurrent Eng-Res A 2006;14:67-82. Einspruch EM, Omachonu VK, Einspruch NG. Quality function deployment: application to rehabilitation services. Int J Health Care Qual Assur 1996;9:42-47. Garon G. Is 'the House of Quality' a useful tool for capacity planners and managers? Proceedings of the 18th International CMG Conference; 1992 Dec 7-11; Reno, NV, USA: ProQuest-CSA; 1992. p. 865-870. Iacobucci D, Ostrom A, Grayson K. Distinguishing service quality and customer satisfaction. J Cons Psyc 1995;4:277-303. Kelley DL. How to Use Control Charts for Healthcare. Milwaukee, WI, USA: ASQ Quality Press; 1999. King B. Techniques for understanding the customer. Qual Manag Health Care 1994;2:61-67. Lim PC, Tang NKH, Jackson PM. An innovative framework for health care performance measurement. Manag Serv Qual 1999;9:423-433. Lim PC, Tang NKH. The development of a model for total quality healthcare. Manag Serv Qual 2000;10:103-111. Lin MC, Wang CC, Chen TC. Strategy for managing customer-oriented product design. Concurrent Eng-Res A 2006;14:231-244. Lloyd RC. Quality Health Care: A Guide to Developing and Using Indicators. 4th ed., Sudbury, MA, USA: Jones and Bartlett Publications, 2004. Logan GD, Radcliffe DF. Potential for use of a house of quality matrix technique in rehabilitation engineering. IEEE Trans Rehabil Eng 1997;5:106-115.

-7-

Madu CN. House of Quality (QFD) in a Minute. 2nd ed., Fairfield, CT, USA: Chi Publications, 2006. Moores BM. Radiation safety management in health care: the application of quality function deployment. Radiography 2006;12:291-304. Radharamanan R, Godoy LP. Quality function deployment as applied to a health care system. Comput Ind Eng 1996;31:443-446. Ramaswamy R, Ulrich K. Augmenting the house of quality with engineering models. Res Eng Des 1993;5:70-79. Shin JS, Kim KJ. Complexity reduction of a design problem in QFD using decomposition, J Intell Manuf 2000;11:339-354. Terninko J. Step-by-Step QFD: Customer-Driven Product Design. 2nd ed., Boca Raton, FL, USA: St. Lucie Press, 1997. Williams B. Patient satisfaction: a valid concept? Soc Sci Med 1994;38:509-516.

Table 1: The quality-planning table

5.0 4.5 5.0 3.0 5.0 5.0 5.0 5.0 4.5 4.0 5.0 5.0 5.0 3.5 4.5 5.0 4.0 5.0 5.0 5.0

E E U U E E E U E E E E E A E A A E E E

5.0 5.0 5.0 4.0 5.0 5.0 5.0 5.0 4.5 4.0 5.0 5.0 5.0 4.0 4.5 5.0 4.0 5.0 5.0 5.0

1.25 1.25 1.00 1.33 1.67 1.25 1.25 1.00 1.13 1.00 1.43 1.00 1.43 1.00 1.13 1.00 1.33 1.25 1.00 1.25

1.0 1.5 1.0 1.2 1.0 1.2 1.0 1.5 1.0 1.2 1.0 1.0 1.2 1.2 1.0 1.2 1.0 1.2 1.2 1.2

DQ Absolute Importance DQ Relative Importance (%)

Sales Point

4.0 4.0 5.0 3.0 3.0 4.0 4.0 5.0 4.0 4.0 3.5 5.0 3.5 4.0 4.0 5.0 3.0 4.0 5.0 4.0

Improvement Ratio

4.0 5.0 3.0 4.0 3.0 4.0 4.0 4.0 4.0 3.0 5.0 4.0 5.0 4.0 4.0 5.0 4.0 4.0 4.0 4.0

Target

1. Provision of comfort 2. Well-being of mother 3. Painless vaginal examination 4. Normal vaginal delivery 5. Companionship after delivery 6. Listening to the fetal heartbeat 7. Immediate opportunity to see the newborn 8. Low-pain Labour 9. Quick response to requests 10. Helping mother with breastfeeding 11. Caring and sensitive staff 12. Labour and childbirth education 13. Well-being of baby 14. Bed linen changed frequently 15. Privacy during delivery and vaginal examination 16. Clean maternity ward 17. Improved hospital facilities 18. Quick admission 19. Short labour 20. Frequent monitoring

D

C DQ Kano Level*

Demanded Qualities

B Customer Importance Rating Fayazbakhsh Hospital Rating Babak Hospital (competitor) Rating

A

5.0 9.4 3.0 6.4 5.0 6.0 5.0 6.0 4.5 3.6 7.1 4.0 8.6 4.8 4.5 6.0 5.3 6.0 4.8 6.0

4.5 8.5 2.7 5.8 4.5 5.4 4.5 5.4 4.1 3.2 6.4 3.6 7.7 4.3 4.1 5.4 4.8 5.4 4.3 5.4

*A: Assumed, E: Expected, U: Unexpected The table has listed high-ranked maternal requirements (column A) and the results of maternal survey (column B). Targets, improvement ratios, and ‘sales points’ (column C) have been identified by assessing the survey results. The weight of the DQs has been calculated as the output data of QPT (column D).

-8-

Current Performance (%)

Target Performance (%)

Table 2: The key elements for improving the process performance of maternity care based on the maternal requirements at Fayazbakhsh hospital

1

% of clinicians using agreed clinical guideline

0

70

Using agreed evidence-based clinical guideline by clinicians

2

% of obstetricians present 24 hours a day in the maternity ward

60

80

Full-time residing obstetrician in the maternity ward

3

% of women receiving labour and childbirth education at the antenatal unit

17

80

Labour and childbirth education in the prenatal and admission units

4

% of women with a companion in the postnatal unit

36

60

Having a companion after normal vaginal delivery

5

% of doctor visits where a folding screen was used in the labour unit

0

80

Using a folding screen in the labour and delivery units

6

% of women who have access to warm water during their hospital stay

57

70

Access to warm water during the hospital stay

No

Key Performance Measure

-9-

Necessary Organizational Function

Assessing correlations among performance measures by the asymmetrical 5-point scale:

Correlation Matrix

Strong Positive = 9 Positive = 3 No Relation = 0 Negative = -1 Strong Negative = -3

(3) Important Performance Measures

Strong = 9 Medium = 3 Weak = 1 No Relation = 0

2

Quality-planning Table

1

Assessing relationships between DQs and performance measures by the asymmetrical 4-point scale:

Identified by

(1) Important DQs

Identified by

Field Interview and AHP

Relationship Matrix:

(2) Weight of Important DQs

3

1) Weight of Performance Measures 2) Average Correlation Rate of Performance Measures

3) Selected (Key) Performance Measures (4) Benchmarks

4) Targets of key Performance Measures

DQs: Demanded Qualities AHP: Analytical Hierarchy Process

Figure 1: The data of house of quality (HoQ) HoQ is constructed in six parts: 1) Important DQs 2) DQs’ weight derived from quality-planning table 3) Important performance measures, 4) Relationship matrix that identifies relationships between DQs and performance measures, 5) Correlation matrix (top roof) that identifies correlation rates among performance measures, and 6) Benchmarks The output data of HoQ are (the down rows): 1) Performance measures’ weight identified by the relationship matrix, 2) Average correlation rate of the performance measures identified by the correlation matrix, 3) Selected (key) performance measures for improvement (based on their weight and correlation rates), 4) Target levels of the selected performance measure (based on the organizational recourses and comparison of the selected performance measures values with the benchmarks)

- 10 -

- 11 -

Fayazbakhsh Hospital

Babak Hospital

np Chart: % of women receiving labour and childbirth education at the antenatal unit, September 2006

np Chart: % of women with a companion in the postnatal unit, September 2006

np Chart: % of women who have access to warm water during their hospital stay, September 2006

Figure 3: The ‘np charts’ of three key performance measures of maternity care at Fayazbakhsh hospital and Babak hospital. The charts show that the process performance at Fayazbakhsh hospital was stable, but unstable in Babak hospital

- 12 -