big data analytics to step change in pathology services

1
Authors: R LITTLEWOOD, S DOUGLAS 1 , N ANTONIOU 2 M LAFFAN 3 (1)applied strategic, London, UK (2)Business Mathematics, London, UK (3)Imperial College Healthcare NHS Trust, London, UK INTRODUCTION The Healthcare Trust A major Healthcare NHS Trust provides pathology service at 3 main sites for medicine, surgery and other specialties. A budget of £25M provides pathology services for 200,000 people amounting to more than 2M tests per year Strategic reviews highlight the resources allocated to pathology in UK NHS 1 and the potential to achieve operational improvements 2 The Pathology Test service Analytical approach OBJECTIVES Understand terrain of demand and drivers leading to map of capacity utilization in a pathology service Use an extensive set of data from key points along the pathway to understand performance within the pathology test process Address how one of the laboratories in the Trust network could optimize performance ANALYSIS RESULTS METHODS Applying big data analytics and queue theory to identify opportunity for step change in pathology services activity Insert your Logos Activity review database Full blood count (FBC) Coagulation (COAG) •1.2M case records •21.6M data points •204,000 tests •1M tests FBC Hospital 1 100% Hospital 1 FBC Hospital 2 Primary Care Others (n=10) Lab 4 1M 0.6M Laboratory 1 Laboratory 2 Laboratory 3 •80% of FBC tests were ordered from hospitals; they are the primary driver of demand 1 Hospital 1 Hospital 2 Primary Care Lab 1 Lab 2 Lab 3 •Considering all 1M FBC tests, 0.6M tests were ordered in hospital 1. Of this 0.6M test volume, 239,000 (41%) processed at Lab 1 2 Analysis of activity in pathology testing shows reasons for delays in achieving results and more importantly, why full capacity cannot be achieved today There are limits to productivity (blood test rate) in blood testing inherent in the current process: simple solutions exist for some limiting factors Hospital 1 Hospital 2 Primary Care Laboratory 1 Laboratory 2 Laboratory 3 Average FBC turn around time increases as day progresses; peaking at 4pm 5 Hospital 1 Hospital 2 Primary Care Laboratory 1 Laboratory 2 Laboratory 3 As FBC volume received at Lab 1 per hour falls turn around time increases; delayed negative impact of FBC arrival on turn around time 6 Descriptive statistics: Analysis of 1M FBC records was completed to understand process delays 9 10 11 12 13 14 15 16 17 18 19 20 0 4000 8000 12000 16000 20000 Hour of day (for Mon- Thur inc.) Analyses of test by location was recorded to map the relationship between performance and demand •Volume of FBC samples arriving at Lab 1 peaks at 10am, declines as day progresses •Demand for pathology test service is greatest in the morning 3 Hospital 1 Hospital 2 Primary Care Laboratory 1 Laboratory 2 Laboratory 3 •Activity in Lab 1 peaks at 11am; there is a 1 hour delay between peak arrival and peak reporting of tests •Reporting rate slows as the day progresses 4 8 9 1011121314151617181920 0 40 80 120 160 Hour of day (for Mon- Thur inc.) Demand: Assessment of FBC tests arrival at Lab 1 per hour, Mon-Thurs 8 9 10 11 12 13 14 15 16 17 18 19 20 0.7 0.9 1.1 1.3 1.5 Average FT/ hrs Test receive hour of day (for Mon-Thur inc.) Activity: Assessment of FBC tests reported at Lab 1 per hour, Mon-Thurs 8 9 1011121314151617181920 1 6 11 16 21 26 2 2.4 2.8 3.2 3.6 FBC number arriving Average turn around time / hr Test receive hour of day (for Mon-Thur inc.) Performance: Assessment of FBC test turn around time/ hour received at Lab 1 Tests FBC COAG Installed capacity Max # tests per machine/ hour 140 28 No. of machines 3 2 Laboratory potential Max machine capacity/ hour 420 60 Observed performance Mean tests reported/ hour 60 13 Max tests reported/ hour 95 20 Performance analysis: Assessment of slowest FBC quartile in Lab 1 - turn around time v number received per hour Process map: pathology test to result Load to machine, run test Collect, ship, drop off Sample & local label Receive & label enter data Validate results Release results Process flow limit at step 3: Front of house sample processing limits to 80 samples/ hour Potential block in process flow at step 5: Clinical validation limits capacity Installed capacity in Lab 1 is high Activity potential is defined by rate of test completion by analyzer machines and test turn around time Measurement of true activity highlights gap between performance and maximum potential Understanding reasons for noted performance can facilitate a step change in performance ACKNOWLEDGEMENTS/ REFERENCES 1 Carter, P. Report of the Second Phase Review of NHS Pathology Services in England. London: DH, 2008. 2 Carter, P. Operational productivity and performance in English NHS acute hospitals: Unwarranted variations. London: DH, 2016 T. Lumley assisted in the preparation of this poster; the team at HH Pathology Service provided essential input. Contact information: [email protected] FBC pathology tests Primary Care Trusts Hospitals Others 1M tests • 0.8M tests • 2 major hospitals , 3 laborator ies • 0.2M tests • 43 PCTs • 6,900 tests • 10 organisation s served Step-by-step process review defines capacity and utilization for FBC pathology tests Clinical results and lab management systems recorded activity and outcomes of tests The data described incidence of exemplar tests (FBC and coagulation) over time, with parameters to show healthcare professional ordering the test, location of the patient having the test and site of the laboratory in which the test was analysed Records did not contain any information that Test time • Time from patient collection site to pathology lab receiving Laboratory Transit • Time from lab receiving sample to result reported to doctor Analysis based on turn around time includes processing and analysis of samples in laboratory as well as time in transit Big data activity analytics, queue modeling and process mapping were applied CONCLUSIONS/ RECOMMENDATIONS Big data analytics, queue modeling and process mapping shows reasons for under utilization of installed capacity and options for optimization in an NHS pathology service Service advised to deal with blocks in flow by addressing the registration of samples and introducing smart process for sample validation Demand for testing should be controlled and phased by programming standard times for inpatient routine tests and using off schedule overnight time for primary care service NHS UK can make huge operational gains by analyzing big data operational activity •1.2M tests ordered per year by 13 different organisations • 80% FBC; 20% COAG ordered Process described as order to action loop to facilitate analysis Service provided by a number of in- house laboratories operating advanced analyzers capable of high throughput, automated assessment of samples The service tracked performance including volume of activity and turnaround time to complete a test, following receipt of samples Process loop Blood to test to report Data influen ces action Service users order product 21 st CONGRESS JUNE 9-12 2016 European Hematology Association 160 120 80 40 0

Upload: richard-littlewood

Post on 23-Feb-2017

181 views

Category:

Health & Medicine


3 download

TRANSCRIPT

Page 1: Big data analytics to step change in pathology services

Authors: R LITTLEWOOD, S DOUGLAS1, N ANTONIOU2 M LAFFAN3 (1) applied strategic, London, UK(2) Business Mathematics, London, UK(3) Imperial College Healthcare NHS Trust, London, UK

INTRODUCTIONThe Healthcare Trust • A major Healthcare NHS Trust provides pathology service at 3 main sites for medicine,

surgery and other specialties. A budget of £25M provides pathology services for 200,000 people amounting to more than 2M tests per year

• Strategic reviews highlight the resources allocated to pathology in UK NHS1 and the potential to achieve operational improvements2

The Pathology Test service

Analytical approach

OBJECTIVES• Understand terrain of demand and drivers leading to map of capacity utilization in a

pathology service• Use an extensive set of data from key points along the pathway to understand

performance within the pathology test process• Address how one of the laboratories in the Trust network could optimize performance

ANALYSISRESULTS

METHODS

Applying big data analytics and queue theory to identify opportunity for step change in pathology services activity Insert your

Logos

Activity review

database

Full blood count (FBC)

Coagulation (COAG)

• 1.2M case records• 21.6M data points

• 204,000 tests• 1M tests

FBC

Hospital 1

100%

Hospital 1 FBC

Hospital 2

Primary Care

Others (n=10) Lab 41M 0.6M

Laboratory 1

Laboratory 2

Laboratory 3

• 80% of FBC tests were ordered from hospitals; they are the primary driver of demand

1

Hospital 1

Hospital 2

Primary Care

Lab 1

Lab 2

Lab 3

• Considering all 1M FBC tests, 0.6M tests were ordered in hospital 1. Of this 0.6M test volume, 239,000 (41%) processed at Lab 1

2

• Analysis of activity in pathology testing shows reasons for delays in achieving results and more importantly, why full capacity cannot be achieved today

• There are limits to productivity (blood test rate) in blood testing inherent in the current process: simple solutions exist for some limiting factors

Hospital 1

Hospital 2

Primary Care

Laboratory 1

Laboratory 2

Laboratory 3

• Average FBC turn around time increases as day progresses; peaking at 4pm

5

Hospital 1

Hospital 2

Primary Care

Laboratory 1

Laboratory 2

Laboratory 3

• As FBC volume received at Lab 1 per hour falls turn around time increases; delayed negative impact of FBC arrival on turn around time

6

Descriptive statistics: Analysis of 1M FBC records was completed to understand process delays

9 10 11 12 13 14 15 16 17 18 19 200

4000

8000

12000

16000

20000

Hour of day (for Mon-Thur inc.)

Analyses of test by location was recorded to map the relationship between performance and demand

• Volume of FBC samples arriving at Lab 1 peaks at 10am, declines as day progresses

• Demand for pathology test service is greatest in the morning

3

Hospital 1

Hospital 2

Primary Care

Laboratory 1

Laboratory 2

Laboratory 3

• Activity in Lab 1 peaks at 11am; there is a 1 hour delay between peak arrival and peak reporting of tests

• Reporting rate slows as the day progresses

4

8 9 10 11 12 13 14 15 16 17 18 19 200

40

80

120

160

Hour of day (for Mon-Thur inc.)

Demand: Assessment of FBC tests arrival at Lab 1 per hour, Mon-Thurs

8 9 10 11 12 13 14 15 16 17 18 19 200.7

0.9

1.1

1.3

1.5

Ave

rage

FT/

hrs

Test receive hour of day (for Mon-Thur inc.)

Activity: Assessment of FBC tests reported at Lab 1 per hour, Mon-Thurs

8 9 10 11 12 13 14 15 16 17 18 19 201

6

11

16

21

26

2

2.4

2.8

3.2

3.6FBC number arriving

Average turn around time / hr

Test receive hour of day (for Mon-Thur inc.)

Performance: Assessment of FBC test turn around time/ hour received at Lab 1

Tests FBC COAGInstalled capacityMax # tests per machine/ hour

140 28

No. of machines 3 2Laboratory potential Max machine capacity/ hour

420 60

Observed performanceMean tests reported/ hour

60 13

Max tests reported/ hour 95 20

Performance analysis: Assessment of slowest FBC quartile in Lab 1 - turn around time v number received per hour

Process map: pathology test to resultLoad to

machine, run test

Collect, ship, drop off

Sample & local label

Receive & label enter

dataValidate results Release

results

Process flow limit at step 3: Front of house sample processing limits to 80

samples/ hour

Potential block in process flow at step 5: Clinical validation limits

capacity

• Installed capacity in Lab 1 is high• Activity potential is defined by rate of test

completion by analyzer machines and test turn around time

• Measurement of true activity highlights gap between performance and maximum potential

• Understanding reasons for noted performance can facilitate a step change in performance

ACKNOWLEDGEMENTS/ REFERENCES1 Carter, P. Report of the Second Phase Review of NHS Pathology Services in England.

London: DH, 2008. 2 Carter, P. Operational productivity and performance in English NHS acute hospitals:

Unwarranted variations. London: DH, 2016 • T. Lumley assisted in the preparation of this poster; the team at HH Pathology Service

provided essential input.• Contact information: [email protected]

FBC pathology tests

Primary Care Trusts Hospitals Others

1M tests

• 0.8M tests• 2 major

hospitals, 3 laboratories

• 0.2M tests• 43 PCTs

• 6,900 tests• 10 organisations

served

Step-by-step process review defines capacity and utilization for FBC pathology tests

• Clinical results and lab management systems recorded activity and outcomes of tests

• The data described incidence of exemplar tests (FBC and coagulation) over time, with parameters to show healthcare professional ordering the test, location of the patient having the test and site of the laboratory in which the test was analysed

• Records did not contain any information that might identify the test recipient nor clinical results

Test time

• Time from patient collection site to pathology lab receiving

LaboratoryTransit• Time from lab

receiving sample to result reported to doctor

• Analysis based on turn around time includes processing and analysis of samples in laboratory as well as time in transit

• Big data activity analytics, queue modeling and process mapping were applied

CONCLUSIONS/ RECOMMENDATIONS• Big data analytics, queue modeling and process mapping shows reasons for under

utilization of installed capacity and options for optimization in an NHS pathology service• Service advised to deal with blocks in flow by addressing the registration of samples and

introducing smart process for sample validation• Demand for testing should be controlled and phased by programming standard times for

inpatient routine tests and using off schedule overnight time for primary care service• NHS UK can make huge operational gains by analyzing big data operational activity 

• 1.2M tests ordered per year by 13 different organisations• 80% FBC; 20% COAG ordered

• Process described as order to action loop to facilitate analysis

• Service provided by a number of in-house laboratories operating advanced analyzers capable of high throughput, automated assessment of samples

• The service tracked performance including volume of activity and turnaround time to complete a test, following receipt of samples

Process loopBlood to test to reportData

influences action

Service users order product

21st CONGRESSJUNE 9-12 2016European Hematology Association

160

120

80

40

0