The researcher adopted a cross-sectional study design to examine the effect of particular independent variables on the selected dependent variable. Each of the selected hospitals was considered as the main unit of analysis while the efficiency score was regarded as the dependent variable. The other independent variables considered were the TQM implementation and hospital attributes including the hospital ownership, hospital size and hospital location. The purpose was to compare the efficiency scores generated by the DEA to the independent variable levels (Abbott & McKinney, 2013; Harrison & Sexton, 2006; Wilkins, 2006).
According to Ray (2004), OEDC (2011) and Abbott and McKinney (2013), the DEA model has been applied to other healthcare related studies that were intended to examine the hospital efficiency based on the accepted variables validated by the earlier projects. Just like it has been done in the earlier projects, the variables considered for this study were consistent with other investigations on the hospital efficiency such as the total labor costs, beds in a hospital and the fixed and variable costs. The fixed costs considered were the total discharges and outpatient visits (outputs) and the supplies (inputs) (Huerta, Ford, Peterson & Brigham 2008; OEDC, 2011; Ray, 2004; Markham 2004). The quantitative study technique was applied to generate two types of data. The first set of data was acquired through the survey questionnaire and efficiency data retrieved from the American Hospital Association (AHA) database.
The current study is about quality improvement efforts and efficiency. The primary objective is to measure the impact of process improvement programs on the performance of hospitals in Mid-Atlantic region using data envelopment analysis. This is achieved by analyzing the impact of total quality management (TQM) on efficiency in the hospitals as evaluated using the Data Envelopment Analysis (DEA). According to Christensen, Coombes-Betz and Stein (2007), Sherman & Zhu (2006) and Bryman (2012), the aim of TQM is to continuously enhance the quality of service provided by eradicating waste so as to satisfy the client’s expectation. According to Labov, Ash and Boberg (2006) and Ozcan (2008), the TQM is a relatively new concept in the hospital sector compared to other industries, such as the manufacturing industry. Considering the guidelines presented by Hsu and Hu (2007) and Christensen et al. (2007), the level of TQM implementation was examined from the hospital’s top management’s perspective. The DEA is a technique that is designed to assess the relative efficiency of functionally comparable and independent units having a multiple incommensurate output and inputs (Huerta et al., 2008). In order to investigate the connection between the TQM implementation and efficiency, the study reviewed the general hospitals that operate in the East South Central region of the U.S, including the state of Kentucky, Alabama, Tennessee and Mississippi (Harrison, Coppola & Wakefield, 2004). During the study, the researcher targeted the specific respondents in each of the hospital’s top management positions. Each of the respondents consulted occupied at least one of the following positions: Chief Executive Officer (CEO), an administrator, a president, or a Director in each of the selected hospitals. Two types of data were considered. The first set of data was collected through the survey questionnaire, while the second set was retrieved from the American Hospital Association (AHA) database.
The questionnaire was formulated to evaluate the level of perceived implementation of total quality management (TQM) from the hospital’s top administration and to obtain the background information regarding each of the hospitals, including the expenses for supplies and materials, the population size, the service area and the personnel. A section of the data collected regarding each hospital’s efficiency was retrieved from the 1997/1998 American Hospital Association (AHA) guide to the Health Care Field Annual Survey of Hospitals. Each hospital was viewed as an interdependent decision making unit (DMU), following the DEA terminology. A total of total 97 DMUs were reviewed. According to Ancarani, Di Mauro and Giammanco (2009), the analysis of the total quality management implementation (TQM) for each DMU should be obtained by linking responses to the questionnaire and to the relative importance of the TQM constructs. Each hospital’s efficiency should be determined relative to other hospitals in the selected set. This should be obtained using the variable return to scale, with an input orientation (Dimitri, Vivian & Grosskopf, 2004). The nondiscretionary variable formulation should be applied to compute the relative efficiency by reflecting only the effects of controllable variables for each of the DMUs. Therefore, this was done considering the data on seven inputs and two outputs (Ancarani et al., 2009). The outputs should be the distinctive categories of patients, who consume the hospital services: the outpatients and the inpatients. The inputs should use the supply costs as material resources, the beds and facilities as capital resources, and the physicians, registered nurses, senior managers and other full-time equivalents as labor resources.
Even though in the existing literature regarding the analysis of performance of organizations the key terms such as “efficiency” and “productivity” are used interchangeably, in the current study the terms were regarded as separate but related concepts (Ozcan, 2008). The term “productivity” was regarded as the ratio of outputs to inputs, while “efficiency” was regarded as the relationship between the observed and optimal values of inputs and outputs and of an organization (Mujis, 2004). In line with Mujis (2004)’s explanations, efficiency was regarded as a component of productivity. The main hypothesis of the study was that successful performance and implementation of the TQM results in increased efficiency of the hospital in relation to three key control variables, including the hospital location, hospital ownership and hospital size. This hypothesis should be true if the average efficiency score within a group of DMUs is significantly different from the average efficiency scores of other groups.
Analysis of variance (ANOVA)
In statistics, the analysis of variance (ANOVA) is described as a collection of models and the related procedures that are used to test the means (averages) by dividing the overall observed variance into different units (Blondeel, Morris, Hallock and Neliga, 2014). According to Blondeel et al. (2014), the ANOVA method was formulated by Ronald A. Fisher in the 1920s. This explains why in many articles, the technique is identified as the Fisher’s Analysis of variance or Fisher’s ANOVA. Based on Hu & Huang (2004)’s and Mujis (2004)’s descriptions, there are many types of ANAOVA depending on the number of treatments and the manner in which they are applied to the subjects in the test/measurement. For instance, the one-way ANOVA method is applied to test for the difference between 3 or more independent groups. The one-way ANOVA for the repeated tests is applied, which means that the same elements are used for each treatment (Hu & Huang, 2004; Mujis, 2004). According to Hu and Huang (2004), this method can be dependent on carryover effects. The factorial ANOVA is appropriate when the researcher/experimenter intends to examine the effects of two or more variables. The most frequently used factorial ANOVA type has been the 2 by 2 design, in which two independent variables, with two levels are evaluated. The factorial ANOVA can also use multiple levels, such as the 3 by 3, or the higher order such as 2 by 2 by 2 (Hu & Huang, 2004; Mujis, 2004). Based on the Blondeel et al. (2014)’s recommendations, the analyses with higher number of factors often make the calculations lengthy and make the outcomes /results difficult to understand and interpret. Such tests/experiments are often costly and time consuming. The multivariate analysis of Variance (MANOVA) is considered appropriate when more than one dependent variable is involved in the text.
In this study, the key variable of interest is the TQM implementation. According to Griffin and Museus (2011), the hospitals with the highest level of TQM implementation are viewed as more efficient. In the current study, it is considered that the various hospital characteristics have potential effects and determine the efficiency of the hospital. Every hospital is treated independently together with its operations and administration. Therefore, this experiment is a factorial design with two distinct independent variables. The analysis of variance (ANOVA) is applied to explore the distinction in efficiency in terms of total quality management (TQM) implementation. Griffin and Museus (2011) described ANOVA as a parametric test, which is useful to compare and analyze the disparity among k sample means, in which k is equivalent to or bigger than two. According to Griffin and Museus (2011), ANOVA presumes that inaccuracies are normally and independently distributed (NID) with a mean of zero and the variance
The mathematical model presented by Griffin and Museus (2011), supports two factors for the hospital ownership and can be written as follows:
- Yh=μ+aI + βj+ Yy+ Eh
- Yh is equal to the hth observation efficiency in each i, j treatment combination and the population is equal to N(μr ,σ’)
- μ Denotes the grand population mean (average) irrespective of the treatment
The μr denotes the effect of the treatment, the total quality management (TQM) implementation, while i denotes, the “higher”, “highest medium” and “low hospital levels”. βj represents the effect of treatment , TQM implementation, in which i stands for the not-for-profit (NFP), for- profit, non-federal government (NFG) hospitals and the federal government (FG) (Harrison et al., 2004).
Yy denotes the effect of the relations between the treatment i and j.
The Eh stands for the error term and is equivalent to N(0, σ’)
The mathematical representation of the design experiment with the two hospital location factors can be expressed as follows:
- Yh=μ+aI + βj+ Yy+ Eh
Whereby the Yh stands for the observation in each i, j treatment arrangement and is presented as N(μr ,σ’)
The μ denotes the grand population mean irrespective of the treatment.
ai stands for the effect of the treatment, the total quality implementation; i represents the high, highest medium and low hospital levels.
Ghauri and Gronhaug (2010) recommended that the appropriate instrument should be adopted for the intended subjects. Since the study is mainly intended to examine the perceptions of the top management regarding the TQM implementation and to measure the hospitals’ efficiency, the top management in each hospital is the subject. In this case the top management is the director, the president, the administrator and/or the chief executive officer (CEO). The top leaders are the members of the management team in the hospital business. In order to ensure uniformity in the study sample, two simple rules were followed during the selection of the hospitals. The first rule was to use the study of AHA regional classification to organize the environmental differences in each of the hospitals. Based on National Center for Health Statistics (2010)’s descriptions, AHA categorizes the United States into 10 classifications based on the hospitals’ geographical locations. That meant that The East South Central region of the United States was chosen in this study and the following states were considered: Alabama, Mississippi, Tennessee and Kentucky. The second rule was to use the operational factors to manage the operational differences in the patients’ attributes of a hospital. The operational factors considered were the average length of stay and services offered. Therefore, the hospitals considered were those that were found to offer the general surgical and medical services and that were in the short-term stay category. Following these rules, a total of 440 hospitals were selected as the target sample for the study.
A survey questionnaire was designed to comprise of two sections. The first section contained thirty-three questions to reflect the seven proposed TQM constructs to examine the top manager’s perception regarding the TQM implementation in the respective hospital. The questionnaire items were randomly arranged and were based on the recommendations of Denscombe (2007) and Collins & Hussey (2009). The panel comprised the faculty members of Marketing and Management Department at the Mississippi University and a Total Quality Management (TQM) professional of the North Mississippi Medical Center in Tupelo, Mississippi. Based on the observations and suggestions obtained from the panel, some changes were made in order to enhance the wording and to eliminate the chances of misinterpretation and to reflect the hospital environment. The respondents were asked to demonstrate the level of agreement with four statements describing their hospital. In response to each of the statements listed, the respondents were given a chance to choose a 1 for “Strongly Disagree”, 2 for “Disagree”, 3 for “neither agree nor disagree, 4 for “agree” or 5 for “strongly disagree”. In case the respondent felt that he/she did not have sufficient information to respond to the question, they were asked to indicate “N/A”. The second part presented four questions about the hospital’s service area, population size, location, supplies and material expenses and the structure of personnel employed.
The data considered in this study were retrieved from the 1997/1998 American Hospital Association (AHA) Guide to the Health Care Field Annual Survey of Hospitals and a questionnaire survey. The data in the analysis of the TQM implementation for each of the hospitals were drawn from the questionnaire survey, while the efficiency measure data from the AHA. According to Bryman (2012), the annual survey presents the hospital specific information on approval, facility utilization, service mix, personnel, expenses and health care system. The survey also contains information on hospital ownership and the number of beds, which are employed as control variables in this study. Each of the 440 general medical and surgical hospitals operating in the East South Central region of the United States was e-mailed and invited to participate in the study. A letter outlining the purpose of the study and requesting the people to take part in the study was attached to the email. A few weeks after the first mailing, the researcher sent a follow-up mail to all the hospitals that had not provided any reply to the first letter. The mailing and follow-up resulted in a total of 105 questionnaires being returned. This was 23.9 percent of a response rate. 4 out of the 105 hospitals admitted to have implemented at least some quality programs. However, four hospitals did not complete the questionnaires correctly and therefore were excluded from the study. Lastly, 97 hospitals were reviewed. A total 39 not-for-profit, 18 for-profit, 32 non-federal government and 8 federal government hospitals were examined.
The study used the Data Envelopment Analysis (DEA) to determine the accuracy and efficiency of each hospital within the selected set. Harrison & Sexton (2006) explained that the Data Envelopment Analysis (DEA) utilizes various inputs and outputs concurrently and the modeling constituent in a Data Envelopment Analysis assessment is the identifications/ recognition and measurement of inputs and outputs. The identification of the outputs and inputs should be based on the understanding of the resources that the management utilizes to dispatch the services offered. In general, the hospital as a health care provider has an organized structure for the medical staffs, patients’ beds, supplies, facilities and equipment. Based on the information presented by Harrison & Sexton (2006), the conceptual output of better service provided is not easy to measure. This implies that the intermediary levels of health services need to be used. In the current study, the choice of variables depends on the available data as retrieved from the public reports issued by the hospitals. To compute the DEA efficiency score, the study utilizes two outputs and seven inputs. The outputs are the distinctive categories of patients for the hospital services.
Based on Harrison & Sexton (2006) ’s descriptions, the inpatient stands for a patient that is admitted to the hospital to obtain inpatient service, while the outpatients describe those who are not hospitalized, but only visit the hospital for regular medical services and use the outpatient facilities (Coelli, Rao, O’Donnell and Battese, 2005). The input variables signify the numbers of equipped hospital beds and existing facilities as proxies for capital resources (Patton, 2002; Grosskopf, 2004; Coelli et al., 2005). They also describe the supplies and materials expenses and the number of active and associate physicians, senior managers, registered nurses and the full-time equivalents (FTEs) as labor resources. Coelli et al. (2005) describe the selected input and output with references. The input variables, facilities and beds need licenses from the authority in order to remove or add them, and the hospital administration do not have the power to raise or lower their levels. The input variables are said to be exogenously fixed and the final decisions regarding the inputs are beyond the control of the hospital’s top administration. The inputs are also described as nondiscretionary variables (Harrison & Sexton, 2006).
Since the DEA is an external strategy, Grosskopf (2004) pointed out that it is subject to bias or errors. The decision making units (DMUs) that apply the extreme inputs or produce extreme outputs are said to be efficient. This implies that one may commit mistakes with the extreme points. It also implies that the DEA is so sensitive to errors in the data (Patton, 2002). More distinctively, if an efficient Decision Making Unit has data errors, the shape of the true efficient frontier of the site might be altered. This may affect many DMUs causing them to become less efficient. In order to avoid coding and data entry errors, the data used in the study need to be carefully scrutinized (Patton, 2002). In line with Patton (2002)’s recommendations, the researcher carefully inspected the data to avoid possibilities of inconsistencies and coding errors. Additionally, the collected data were systematically scanned and compared with those retrieved from the earlier years and a summary statistics regarding each of the variables. This was intended to detect the presence of errors or any inaccurate information (Patton, 2002).
Since it is hard to establish the way in which inputs of concern react to the adjustments in the inputs, the researcher decided to apply the variable returns-to-scale as a basis of choosing an envelopment surface to determine the relative efficiency. If the evaluation established that hospital is inefficient, then the manager is expected to seek ways to maximize the efforts to reduce the excessive inputs while maintaining the level of outputs (Huerta et al., 2008). This implies that the input orientation was appropriate. In line with the Huerta et al. (2008)’s descriptions, this study employs the BCC input orientation theory/model together with the nondiscretionary variables to determine the relative efficiency of each of the DMUS in the particular set.
In this study, correlation and reliability were effectively addressed. According to Huerta et al. (2008), correlation describes the association between two variables. Reliability describes the ability or degree to yield the same result from the same experiment/measure. The correlation should be achieved through the development of scales and use of multiple items to measure a single construct. The items used to measure one construct should be related among themselves and distinct from the other variables (Al-Balushi et al, 2014). The correlation matrix is therefore appropriate for the study based on Huerta et al. (2008)’s research, where it was affirmed that it helps to assess and evaluate the degree, direction and worth of relationship that exists between the two variables without imagining that one variable is independent and the other is dependent. The association that was discovered among the variables in this study will be displayed in the appendix section. The correlations/associations that are significantly different at the 1 percent level are represented by “**” while those correlations/associations that are deviating from zero at the 5 percent level are represented by “*”.
The objects within the construct are believed to be highly interconnected except one in the consumer focus. Some associations/correlations between the items in the various constructs also exist (Al-Balushi et al, 2014). In the nature of each of the searches, it is expected that every item that is assigned a particular construct is likely to affect the measurement of other factors (Coelli et al., 2005). Considering Coelli et al. (2005)‘s verifications, it is assumed that some correlation therefore exists between the constructs.
Validity and Reliability
Al-Balushi, et al. (2014) defined the term “reliability” as the ability or degree to generate the same outcome from a given measurement. In Christensen et al. (2007)’s argument, if a particular instrument such as a survey questionnaire obtains the same response/feedback from a given respondent multiple times, then the questionnaire should be considered as a highly reliable instrument. If a survey collected the same feedback from one participant to the next under identical conditions or positions, that would show a high dependability over the respondents (Christensen et al., 2007). In this study, a number of methods were considered, that could be used to estimate the reliability. In line with the information presented by Christensen et al. (2007), the methods were the alternative form method, the test-retest technique and the internal consistency procedure. According to Huerta et al. (2008), the test-retest method and the alternative form method have severe drawbacks which are experienced while administering two forms of the instruments for determining reliability. Unlike the test-retest method and the alternative form method, the internal consistency reliability procedure is based on a single administration of a single instrument form. The internal consistency measuring instrument may be described as the level to which the items in the instrument are homogeneous or consistent (Huerta et al., 2008). Huerta et al. (2008) claim that whenever the scores of a particular instrument are considered as being composed of random and true error components, then the instrument’s reliability will be increased when an item is added to measure the same phenomenon. Considering the and Hu & Huang (2004)’s descriptions, the most frequently used techniques for estimating the internal consistency include the Cronbach’s coefficient and the test split-half test methods (Mujis, 2004). In the current study, the Cronbach’s test technique was applied in order to assess the internal consistency of the TQM implementation in which each construct was comprised of multiple items. According to Mujis (2004), the higher the coefficient, the more reliable the construct becomes.
In line with these Huerta et al. (2008)’s explanations, an internal consistency test was conducted separately for each of the constructs. The excel tables were used to display the results of Cronbach’s coefficient test and the alpha coefficient ranges. Following the Mujis (2004)’s guidelines, there are various considerations in the determinations regarding the levels of acceptance and reliability coefficients. Huerta et al. (2008) also argued that the researcher should determine the amount of error that he/she is ready to accept, while considering the particular situation and context of the study. Huerta et al. (2008)’s suggestions are similar to those provided by Mujis (2004), who pointed out that “relatively low” reliability coefficients of.50 or.60 can be tolerated in the early stages of research or could be the hypothesized measures of a particular construct. According to Mujis (2004), the reliabilities of.70 and above are needed when the measure is expected to be used for determining the differences among groups. The least reliability value of.90 is important when the scores are intended to be used for making crucial decisions regarding individuals. Based on Huerta et al. (2008)’s and Mujis (2004)’s guidelines, the values of coefficient alpha in this study will be used to show that each construct is adequately reliable.
Schwandt (2001), Fouka & Mantzorou (2011) and Lewis & Thornhill (1997) explained that in every study that involves human participants, the researcher is expected to ensure that the rights and dignity of the respondents are protected. In line with this recommendation, every participant in this study was issued with a consent form to allow them confirm their acceptance and willingness to take part in the study. The consent forms were distributed to the selected respondents, who took part in the study. The forms also presented essential notification information to explain to the participants the purpose and objective of the study and to emphasize to them that they had the right to withdraw. This action was in line with Fouka and Mantzorou (2011)’s and Blaxter, Hughes and Tight (2010)’s recommendations that it is always important to ensure that participants are not under pressure when participating in the study. The ethical and consent forms clearly outlined the key issues regarding the safeguarding the respondents’ confidentiality and privacy protection. It was also intended to assure the respondents that the collected data would be for research purposes only.
In brief, the study examines the impact of process improvement programs on the performance of hospitals in Mid-Atlantic Region using Data Envelopment Analysis (DEA). In order to achieve the objective, the study evaluates the effect of TQM implementation on hospital’s efficiency in applying the factorial ANOVA design. The hospitals considered are those that operate in the Mid-Atlantic Region, and particularly those that provide the general surgical and medical services with the short-term stay. The level of total quality management (TQM) implementation is determined through the survey questionnaire from the perspective of the top management officials in each of the selected hospitals. The BCC input orientation model is applied together with the nondiscretionary variables in order to determine the relative efficiency.
Abbott, M.L., and McKinney, J. (2013). Understanding and applying research design. Hoboken, NJ: John Wiley & Son, Inc.
Al-Balushi, S.S. et al. (2014). Readiness Factors for Lean Implementation in Healthcare Settings–A Literature Review. Journal of Health Organization and Management, 28 (2), 2-2.
Ancarani, A., Di Mauro, C. & Giammanco, M.D. (2009). The impact of managerial and organizational aspects on hospital wards efficiency: Evidence from a case study. European Journal of Operational Research, 194 (1), 280-293.
Blaxter, L., Hughes, C., and Tight, M. (2010). How to research, (4th ed.). Berkshire, England: Open University Press.
Blondeel, P.N., Morris, S.F.,. Hallock, G.G., & Neligan, P.C. (eds.). (2014). Perforator Flaps: Anatomy, Technique, & Clinical Applications, Second Edition Balkema-proceedings and monographs in engineering, water, and earth sciences (2nd ed.). New York: CRC Press.
Bryman, A. (2012). Social research methods. Oxford, UK: Oxford University Press.
Christensen, E.H., Coombes-Betz, K.M., & Stein, M.S. (2007). The Certified Quality Process Analyst Handbook. New York: ASQ Quality Press.
Coelli, T. J., Rao, P., O’Donnell, D.S., & Battese, G. E. (2005). Data Envelopment Analysis. An Introduction to Efficiency and Productivity Analysis, 161-181.
Collins, J., and Hussey, R. (2009). Business research: A practical guide for undergraduate & postgraduate students, (3rd ed.). Basingstoke, UK: SAGE.
Denscombe, M., (2007). The good research guide, (3rd ed.). Berkshire, England: Open University Press.
Fouka, G., and Mantzorou, M. (2011). What are the major ethical issues in conducting research? Is there a conflict between the research ethics and the nature of nursing?. Health Science Journal, 5(1), 3-14.
Ghauri, P., and Gronhaug, K. (2010). Research methods in business studies. (3rd ed.). Harlow, England: Prentice Hall.
Griffin, K.A., and Museus, S.D. (Eds.). (2011). Using mixed methods to study intersectionality in higher education: New directions in institutional. San Francisco, CA: Jossey-Bass.
Grosskopf, S., Dimitri M., and Vivian V. (2004). Competitive effects on teaching hospitals. European Journal of Operational Research 154 (2), 515-525.
Harrison, J. P., Coppola, M. N., & Wakefield, M. (2004). Efficiency of federal hospitals in the United States. Journal of Medical Systems, 28 (5), 411-422.
Harrison, J. P. & Sexton, C. (2006). Improving efficiency frontier of religious not-for-profit hospitals. Hospital Topics, 84 (1), 2-10.
Hsu, P.F. & Hu, H.C. (2007). The development and application of a modified data envelopment analysis for assessing the efficiency of different kinds of hospitals. International Journal of Management, 24 (2), 318-330.
Hu, J.L. & Huang, Y.F. (2004). Technical efficiencies in large hospitals: A managerial perspective. International Journal of Management, 21 (4), 506-513.
Huerta, T.R., Ford, E.W., Peterson, L.T. & Brigham, K.H. (2008). Testing the hospital value proposition: An empirical analysis of efficiency and quality. Health Care Management Review, 33 (4), 341-349.
Labov, W., Ash, S., & Boberg, S. (eds.) (2006).The Atlas of North American English: Phonetics, Phonology, and Sound Change: a Multimedia Reference Tool, Volume 1. New York, Walter de Gruyter.
Markham, A. N. (2004). Internet communication as a tool for qualitative research. In D. Silverman (Ed.), Qualitative research: Theory, method and practice (pp. 95–124). Thousand Oaks: Sage.
Mujis, D. (2004). Doing Quantitative Research in Education with SPSS. Thousand Oaks: Sage.
National Center for Health Statisitics. (2010). Health United States, 2010: With special feature on death and dying. Hyattsville, MD: U.S. Government Printing Office.
OEDC. (2011). HealthData2011, Retrieved from <http://stats.oedc.org/Index.aspx?DataSetCode=HEALTH_STAT>.
Ozcan, Y.A. (2008). Health Care Benchmarking and Performance Evaluation. New York: Springer.
Patton, M. (2002). Qualitative research & evaluation methods (3rd ed). Thousand Oaks, CA: Sage.
Ray, S. C. (2004). Data envelopment analysis. Cambridge: Cambridge University Press.
Sherman, H. D., & Zhu, J. (2006). Service productivity management: Improving service performance using data envelopment analysis (DEA). New York: Springer.
Sherman, H. D., & Zhu, J. (2013). Analyzing performance in service organizations. Sloan Management Review. 12(1), 30-49
Wilkins, C. (2006). A qualitative study exploring the support needs of first-time mothers on their journey towards intuitive parenting. Midwifery, 22 (1), 169-180.