The Impact of Lean Implementation in Healthcare: Evidence from US Hospitals.

Transcription

1 The Impact of Lean Implementation in Healthcare: Evidence from US Hospitals. Yong Taek Min Assistant Professor & Program Director of MS in Health Science Department of Health Sciences Marieb College of Health & Human Services Florida Gulf Coast University Joseph D. Restuccia Professor of Health Care and Operations Management Boston University Questrom School of Business Co authors: Jay Kim & Michael Shwartz Operations and Technology Management Department Boston University Questrom School of Business Quality Challenges Health care in the United States is not as safe as it should be At least 44,000 people, and perhaps as many as 98,000 people, die in hospitals each year as a result of medical errors that could have been prevented.. (Institute of Medicine, 2000) The US health care delivery system does not provide consistent, high quality medical care to all people Indeed, between the health care that we now have and the health care that we could have lies not just a gap, but a chasm. (Institute of Medicine, 2001) 1

2 Cost Challenges 17.8% of GDP; over $3.2 Trillion (CMS, 2015) Lean implementation in Healthcare 2

3 Lack of clear understanding of Lean in Healthcare My response to this question (how many hospitals are doing or implementing lean?) is usually something along these lines. When I think about what lean and lean thinking really means, I think it s about transformation of management (Deming s term). Most people who have researched lean for all industries (like Bob Emiliani) cite very low percentages of real lean adoption (1 3%). I think this is about right for healthcare 1% maximum. Mike Stoecklein in 2013 from Mark Graban s Lean Blog Lack of rigorous research on the effectiveness of Lean in Healthcare We have not studied this question adequately.. Helen Zak and Alice Lee in personal communications (2017, April 15) 3

4 Research question Does Lean work in Healthcare? Theoretical Framework Multilevel Structure (Shah et al., 2008) Lean philosophy Lean principles Lean practices 4

5 Theoretical Framework Multilevel Structure (Shah et al., 2008) Lean philosophy Lean principles Highest abstraction level Philosophy perspective Customer value centered waste elimination (Ohno, 1988; Womack et al., 1990) Difficult to observe and measure Lean practices Theoretical Framework Multilevel Structure (Shah et al., 2008) Lean philosophy Lean principles Lean practices Lowest abstraction level Physical manifestations of Lean philosophy Operationalization of Lean principles Easy to observe and measure Mostly used in lean implementation in healthcare and empirical studies (Shah and Ward, 2003) 5

6 Theoretical Framework Multilevel Structure (Shah et al., 2008) Lean philosophy Lean principles Intermediate level Guiding goals to implement Lean philosophy Link philosophical view and practices view Lean practices Theoretical Framework Multilevel Structure (Shah et al., 2008) Lean philosophy Lean principles Right abstraction level to apply and study lean in healthcare Lean practices 6

7 Health care delivery operations Service operations (Nie and Kellogg, 1999) Intangibility, simultaneous production and consumption, heterogeneity, perishability, and labor intensity Direct customer participation in the service process adds complexity (Chase and Tansik, 1983). Hospital: Complex service organization (Tucker et al., 2006) Interdependent work units whose work must be coordinated to provide customer service but often have conflicting priorities. Inapplicability of some lean manufacturing practices/tools to healthcare (Mazzocato et al., 2010; Poksinska, 2010; Radnor et al., 2012) Lean manufacturing principles Womack & Jones (1996) 5 lean principles Identify value from a customer perspective Map the value stream Create flow Establish pull Seek perfection Spear & Bowen (1999) 4 rules of TPS Standardization of work Seamless work flow Direct links between suppliers and customers Continuous improvement based on scientific methods, under the guidance of a teacher, at the lowest possible level 7

8 Patient focus (centeredness) Any activity that does not create value for the end customer is waste and should be eliminated (Ohno, 1988) Identify value from a customer perspective (Womack & Jones, 1996) Ambiguous understanding of customer in healthcare (Bushell et al., 2002; Endsley et al., 2006; Young & McClean, 2008) Work process in health care organizations is mostly organized around health care providers without a clear focus on the convenience of the patient (Bahensky et al., 2005; Dickson et al., 2009; Fillingham, 2007; Kim et al., 2007) Patient centered care (Institute of Medicine, 2001) Standardization of care Standardize work (Spear & Bowen, 1999) All work should be highly specified to its content, sequence, timing, and outcome. Map the value stream (Womack & Jones, 1996) Eliminate non value adding steps. Standardizing care process based on evidence based best practice Use of simple checklist can significantly improve quality of care (Gawande, 2007, 2010) Sustained reduction in catheter related bloodstream infection (Pronovost et al., 2006) 8

9 Seamless coordination Create direct and unambiguous linkages for every customersupplier connection (Spear & Bowen, 1999) Create simple and direct pathways for every product and service (Spear & Bowen, 1999) Create flow (Womack & Jones, 1996) Complex system with interdependent functional departments and professional work groups (Tucker et al., 2006) Overly complex and uncoordinated system requiring unnecessary steps and patient handoffs (Institute of Medicine, 2001) Continuous improvement Seek perfection (Womack & Jones, 1996) Continue improvement processes until a state of perfection is reached. Improvement must be made in accordance with the scientific method, under the guidance of a teacher, at the lowest possible level (Spear & Bowen, 1999) Frontline clinician empowerment (Aherne, 2007; Bahensky et al., 2005; Dickson et al., 2009) Top and mid level management commitment (Fillingham, 2007; Jimmerson et al., 2005; Massey & Williams, 2005) 9

10 Lean healthcare principles Womack & Jones (1996) Spear & Bowen (1999) Lean Healthcare Principles Identify value from a customer perspective Standardization of work Patient (customer) focus Map the value stream Seamless work flow Standardization of care (work) Create flow Establish pull Seek perfection Direct links between suppliers and customers Continuous improvement based on scientific methods, under the guidance of a teacher, at the lowest possible level Seamless coordination Continuous improvement Lean healthcare principles Lean healthcare principles Patient (Customer) Focus Identify value from patient (customer) perspective Improve care (work) process with a clear focus on needs and flow of a patient Standardization of Care (Work) Specify care process with regard to content, sequence, timing, and outcome; Based on evidence based best practice Seamless Coordination Create seamless connections and pathways between multiple departments and workgroups Improve patient flow and reduce errors Continuous Improvement Continue improvement process until a state of perfection is reached; From frontline clinician (worker) level; Based on scientific methods; With management support 10

11 Hypotheses H1: Each lean healthcare principle implementation is significantly associated with hospital performance. Individual lean healthcare principle implementation H2: High lean healthcare organizations show better hospital performance. Overall lean implementation Measures for lean healthcare principles 1. Lean measures development Select hospital activities/programs that operationalize Lean healthcare principles from Quality Improvement Activities Survey: 30 measures Face validity test: A panel of researchers reviewed the lean measures 2. Lean measures modification: Exploratory data analysis Principal Component Analysis (Factor loading > 0.40): Construct validity Corrected Item to Total Correlation (> 0.30): Reliability Eliminated 8 measures 3. Lean measures validation: Confirmatory data analysis Confirmatory Factor Analysis (Significance of factor loading): Construct validity Cronbach s alpha: Reliability Validate 22 lean measures for 4 lean healthcare principles 11

12 Data Quality Improvement Activities Survey Developed to assess the nature and extent of hospitals quality improvement activities 173 questions: endorsed by IHI, Leapfrog group Commomwealth Foundation, Health Research & Education Trust, American Hospital Association Entire 4,237 nationwide short term, nonfederal, general service hospitals ( 25 beds) in 2006 Five point Likert scale Respondents: CQO Completed surveys: 470 hospitals Lean principle Item name Hospital activities Patient Focus Pat_1 * Clinicians involve patients and families in efforts to improve patient care (PAT) Pat_2 * The hospital regularly communicates achievement of hospital-wide quality goals to the general public Pat_3 Patient advisory groups Pat_4 Specific strategies to reduce the number of patients assigned to each nurse Pat_5 Patient flow improvement strategies Standardized Std_1 * Patient care processes are standardized, where and when appropriate Care (STD) Std_2 Evidence-based practice guidelines/clinical pathways Std_3 Disease- or condition-specific QI projects Std_4 Chronic disease registries Std_5 Planned care for chronic illness (Wagner s Chronic Disease Model) Std_6 * Standing orders Seamless Seam_1 * People and processes are in place to identify, analyze, and act upon all adverse events to prevent future occurrences Coordination Seam_2 * People and processes are in place to identify, analyze, and act upon near misses to prevent future occurrences (SEAM) Seam_3 * The hospital s structure and work processes impede coordination across departments and workgroups (reverse coded) Seam_4 * There is little coordination of QI efforts across departments and workgroups (reverse coded) Seam_5 Shared clinical governance by nurses and physicians Seam_6 Use of advanced practice nurses (APNs) to coordinate or manage patient care Seam_7 Pharmacists place in patient care units Seam_8 Multidisciplinary rounds Continuous Cont_1 Progress toward achieving hospital-wide quality goals is tracked and communicated to clinical staff Improvement Cont_2 QI project results are regularly communicated to clinical staff (CONT) Cont_3 Senior managers regularly celebrate successful QI projects and give recognition to project team members Cont_4 Management walk-arounds to identify quality problems or issues Cont_5 Corrective action is taken if progress toward achieving hospital-wide quality goals is not adequate Cont_6 QI activities to improve workforce recruitment, retention, and development Cont_7 Profiling of individual provider performance Cont_8 Work process redesigning or re-engineering Cont_9 Benchmarking within the hospital Cont_10 Benchmarking with other hospitals Cont_11 Learning best practices from other industries * Items eliminated after exploratory data analysis 12

13 Lean principles implementation Lean Principles # Mean Std. 25% 50% 75% Dev. (Med) Patient Focus Standardization of Care Seamless Coordination Continuous Improvement Lack of clear implementation of lean in healthcare in spite of hard work in improvement Hospital quality indicators Health care quality assessment model (Donabedian, 1980; 1988) Quality is multidimensional concept Health care quality needs to be measured under three categories: Structure: Attributes of the setting in which patient care occurs Material resources, human resources, organizational structure Process: What is actually done in giving and receiving care Outcome: Effects of care on patient Improvement in the health status Degree of satisfaction of the patient 13

14 Hospital quality indicators Structure: Structural attributes of hospitals as control variables Process: Process quality: Adherence to evidence based process of care Hospital Compare data 15 process of care measures for AMI, HF, PN Outcome: Clinical outcome: Risk adjusted O/E mortality ratio Medicare Provider Analysis and Review (MEDPAR) data APR DRGs risk adjustment system Perceived quality: Patient experience Hospital Consumer Assessment of Healthcare Provider and Systems (HCAHPS) data CMS Data sets Medicare Provider Analysis and Review (MedPAR) Patient level data to track patient care history and outcome Risk adjustment Clinical outcome quality Hospital Compare Hospital level data 15 process of care measures for AMI, Heart failure, and Pneumonia Process quality Hospital Consumer Assessment of Healthcare Provider and Systems (HCAHPS) Hospital level data Perceived quality (Patient experience) 14

15 Hospital quality indicators Process quality: Adherence to evidence based process of care N ph denotes the number of patient cases hospital h handled for process p s ph denotes the associated quality score (adherence rate) for hospital h and process p N h denotes the total number of patient cases that hospital h treated for the three conditions Hospital quality indicators Clinical outcome quality: Risk adjusted in hospital mortality (O/E ratio) D sdh denotes the number of deaths in severity class s in DRG d in hospital h EMR sd denotes the expected in hospital mortality rate for a patient in severity class s in DRG d N sdh denotes the number of patients in severity class s in DRG d in hospital h 15

16 Hospital quality indicators Perceived quality: Patient experience HCAHPS data Average percentage of the respondents who rated highest scores (9 or 10 out of 10, 5 out of 5 point scale) for the questions: How do you rate the hospital overall? Would you recommend the hospital to friends and family? Hospital quality indicators Quality measures # Mean Std. Dev. Min Max Process quality (Adherence to process of care) Clinical outcome quality (Risk-adjusted in-hospital mortality O/E ratio) Perceived quality (Patient experience)

17 H1: Impact of individual lean principle implementation on hospital performance Multivariate Regression Analysis Dependent variables: Hospital quality performance indicators Process quality, Clinical outcome, Perceived quality Independent variables: Lean healthcare principles Patient focus, Standardization of care, Seamless coordination, Continuous improvement Control variables: Hospital structural attributes Bed size: Small (26 99 beds), Medium ( ), Large (>400) Ownership: For profit, Not for profit, Government owned Location: Metropolitan, non metropolitan Correlation coefficients of control and independent variables Variance Inflation Factors (VIFs) of the variables were far below 3 with most of VIFs below 2: No multicollinearity H1: Impact of individual lean principle implementation on hospital performance Variables Process quality Clinical outcome Perceived quality Patient focus (p=0.144) Standardization of care Seamless coordination Continuous improvement (p=0.052) (p=0.569) ** (p=0.001) (p=0.083) (p=0.089) (p=0.238) (p=0.806) *** (p=0.000) (p=0.379) * (p=0.010) (p=0.340) (p=0.591) ** (p=0.003) F value *** (p=0.000) R square p < 0.1; *p < 0.05; **p < 0.01; ***p <

18 H2: Impact of overall lean principle implementation on hospital performance Cluster Analysis Classify hospitals into groups with similar lean principles implementation. Maximize homogeneity of objects within clusters, while maximizing heterogeneity between clusters. Clustering variables: 22 measures for lean healthcare principles. Two stage procedure (Hair et al., 2009) Hierarchical clustering: to determine the most appropriate number of clusters. Non hierarchical clustering (K means cluster analysis): to produce the final clusters. Decision rule: Largest increase in agglomeration coefficient (heterogeneity) H2: Impact of overall lean principle implementation on hospital performance 3 Cluster solution (hospital groups) (n = 416) Cluster 1 (n = 122): Highest level of implementation across 4 lean principles. Cluster 2 (n = 172): Intermediate level of implementation. Cluster 3 (n = 122): Lowest level of implementation PAT STD SEAM CONT 18

19 H2: Impact of overall lean principle implementation on hospital performance 3 Cluster solution (hospital groups) (n = 416) Cluster 1 (n = 122): High lean group Cluster 2 (n = 172): Medium lean group Cluster 3 (n = 122): Low lean group 3 Clusters are significantly differentiated in overall lean implementation (p=0.000). Lean principles High lean (n=122) Cluster 1 Medium lean (n=172) Cluster 2 Low lean (n=122) Cluster 3 Patient Focus 3.42 (0.05) 2.78 (0.04) 1.92 (0.05) Standardization of Care (0.05) (0.04) (0.05) Seamless Coordination (0.06) (0.04) (0.05) Continuous Improvement (0.04) (0.03) (0.04) Note. Values represent cluster mean and standard error for each principle. F value (p value) F = (p = 0.000) F = (p = 0.000) F = (p = 0.000) F = (p = 0.000) H2: Impact of overall lean principle implementation on hospital performance 3 Clusters showed significant differences in quality performance Process quality (p=0.000), Clinical outcome (p=0.000), Perceived quality (0.020) High lean (n = 122) Cluster 1 Medium lean (n = 172) Cluster 2 Low lean (n = 122) Cluster 3 F value (p value) Process quality 0.85 (0.006) 0.82 (0.006) 0.80 (0.009) F = (p = 0.000) Clinical outcome 1.00 (0.024) 1.08 (0.029) 1.26 (0.053) F = (p = 0.000) Perceived quality 0.67 (0.008) 0.65 (0.008) 0.64 (0.014) F = 3.96 (p = 0.020) Note. Values represent cluster mean and standard error for each performance indicator. 19

20 Findings & Contribution Individual lean healthcare principle implementation (H1): Patient focus: Clinical outcome (p=0.083) Standardization of care: Process quality (p=0.052), Clinical outcome (p=0.089), Perceived quality (p=0.010) Continuous improvement: Process quality (p=0.001) Overall lean healthcare principles implementation (H2): Higher lean group showed better quality performance. Process quality (p=0.000) Clinical outcome (p=0.000) Perceived quality (p=0.020) Empirical evidence for Lean works in healthcare Large sample of hospitals: 470 nationwide hospitals Objective quality indicators: CMS data sets Findings & Contribution Limitation Secondary data: Quality Improvement Activities Survey Dated data Cross sectional study not longitudinal study Illustrates a rigorous approach for developing healthcare lean principles and extent of lean implementation Evaluates the relationship between lean implementation and hospital quality performance Sets a baseline for comparison in current studies of how lean implementation and impact have changed over time 20

21 Questions and comments are welcome! Yong Taek Min: Joseph Restuccia: 21

22 Correlation coefficients of variables BEDSIZE GOVERN FORPROFIT URBAN PAT STD SEAM CONT BEDSIZE a 1 GOVERN b ** 1 FORPROFIT c * 1 URBAN d ** ** PAT e ** * ** 1 STD f ** ** ** ** 1 SEAM g ** ** ** ** ** 1 CONT h ** ** ** ** ** ** 1 ** Correlation is significant at the 0.01 level (2 tailed). * Correlation is significant at the 0.05 level (2 tailed). a. BEDSIZE: Bed size; b. GOVERN: Government owned hospital; c. FORPROFIT: For profit hospital; d. URBAN: Urban hospital; e. PAT: Patient focus; d. STD: Standardization of care; e. SEAM: Seamless coordination; f. CONT: Continuous improvement. b. Variance Inflation Factors (VIFs) of the variables were far below Evidence based Process of Care Measures for AMI, HF and PN Acute myocardial infarction (AMI, Heart attack) 1. Aspirin at arrival 2. Aspirin prescribed at discharge 3. Angiotensin converting inhibitor (ACE) inhibitor or angiotensin receptor blocker (ARB) for left ventricular systolic dysfunction (LVSD) 4. Beta blocker prescribed at discharge 5. Beta blocker at arrival 6. Adult smoking cessation advice/counsel Congestive heart failure (HF) 1. Left ventricular function assessment 2. ACE inhibitor or ARB for LVSD 3. Discharge instructions 4. Adult smoking cessation advice/counseling Pneumonia (PN) 1. Oxygenation assessment 2. Pneumococcal vaccination status assessment 3. Initial antibiotic received within 4 hours of hospital arrival 4. Blood culture performed in emergency department before first antibiotic received in hospital 5. Adult smoking cessation advice/counseling 22

23 Characteristics of Responding vs. Population Hospitals Hospital Characteristics Categories Responding hospitals (n = 470) n (%) Population hospitals (N = 4,222) n (%) Bed size beds 145 (30.9) 1,810 (42.9) beds 237 (50.4) 2,000 (47.4) 400 or more beds 88 (18.7) 412 (9.7) Region Midwest 161 (34.3) 1,223 (28.9) Northeast 87 (18.5) 593 (14.1) South 160 (34.0) 1,638 (38.8) West 62 (13.2) 768 (18.2) Ownership For profit 21 (4.5) 662 (15.7) Government, nonfederal 96 (20.4) 951 (22.5) Nonprofit 353 (75.1) 2,609 (61.8) Teaching status Yes 67 (14.3) 275 (6.5) No 403 (85.7) 3,947 (93.5) Network affiliation Yes 164 (34.9) 1,278 (30.3) No 261 (55.5) 2,187 (51.8) Missing values 45 (9.6) 757 (17.9) System affiliation Yes 214 (45.5) 1,900 (45.0) No 253 (53.8) 2,322 (55.0) Missing values 3 (0.6) Medicare DSH Yes 289 (61.5) 2,428 (57.5) No 116 (24.7) 889 (21.1) Missing values 65 (13.8) 905 (21.4) Metropolitan county Yes 302 (64.3) 2,486 (58.9) No 168 (35.7) 1,735 (41.1) Missing values 1 (0) Confirmatory Factor Analysis & Cronbach s Alpha Lean principles (Code) Patient focus (PAT) Standardization of care (STD) Seamless coordination (SEAM) Continuous improvement (CONT) Lean components (Code) Evidence-based practice (STD_EVD) Chronic care management (STD_CHR) Scientific methods (CONT_MTHD) Frontline clinical staff commitment (CONT_FRONT) Management commitment (CONT_MGMT) Item name Factor loading t value Variance explained (R 2 ) Cronbach s alpha Pat_ *** Pat_ *** Pat_ *** Std_ *** Std_ *** Std_ *** Std_ *** Flow_ *** Flow _ *** Flow _ *** Flow_ *** Cont_ *** Cont_ *** Cont_ *** Cont_ *** Cont_ *** Cont_ *** Cont_ *** Cont_ *** Cont_ *** Cont_ *** Cont_ *** ***. Path loading is significant at the level (2 tailed). Factor loadings: Significant at level Verifying the posited relationships Variance explained (R 2 ): Ranged Cronbach s alpha: Ranged Acceptable for exploratory study High level of reliability and validity of the measurement instrument 23

24 Overall measures of fit Measures of fit Absolute Statistic measures Results Recommended value for close fit χ 2 -Test statistic (188) NA (d.f.) RMSEA, point estimate RMSEA, 90% CI (0.064; 0.076) (0.00; 0.08) Standardized RMR Incremental NNFI or TLI CFI Parsimonious Normed χ Absolute fit measures Meet suggested cutoff: RMSEA ( 0.08) 90% CI of RMSEA (0.00; 0.08) RMR ( 0.10) Incremental fit measures Meet moderate cutoff NNFI ( 0.80) CFI ( 0.80) Overall, the developed measures show reasonable model to data fit Reliable and valid measures 24