1. Kevin Hudson Programme Manager, Clinical Commissioning Information, NHS Somerset Detailed Analysis of Emergency Admissions in Somerset FULL RESULTS 0 2 nd December 2011

2. Note to readers For summary presentation, review this document in slide show. For full results, review without slide show. 1

3. Content of this Presentation Background, objectives, datasets & tools Summary of overall findings Some examples of detailed observations – Observations relating to demographics of the patient population – Observations relating to non-Acute settings and processes – Observations relating to Acute Trust settings and processes Ongoing questions and next steps 2

4. OBJECTIVES, SOURCES & TOOLS 3

5. Background Project initiated by the QIPP board to fully understand why emergency admissions have increased in Somerset and quantify the causes. Steering group of senior directors, chaired by David Slack: – met weekly across October through conference call – attendees from PCT, Acute Trusts, Somerset Partnership and South West Ambulance. – Analysis led by Kevin Hudson and mediated through a Data Analyst Group across all organisations. – Data Analyst Group subjected observations to peer review prior to discussion at the steering group. Sought to achieve common story for Emergency Admissions in Somerset. Shared at visit of Intensive Support Team on 2 nd November 2011 Initiated further work and further communication. 4

6. Objectives To quantify and stratify the current trend in rates of emergency admission in Somerset, and particularly in relation to Taunton & Somerset NHS Foundation Trust. To investigate potential correlating factors to the trend, and to attempt to quantify the relative influences of: – Changes in the demographics of the local population. – Changes in processes taking place in non-Acute settings – GP Practices, Out of hours and ambulance trusts. – Changes in processes taking place in Acute settings. To further understand causes of A&E attendances and their link to emergency admissions (not part of these results, this work will follow). 5

7. Data Sources used in this report (Listed as this analysis expands) SUS data from April 2008 re-costed at 2011/12 tariff. (Master Data Set). Received and uploaded for analysis. Timed admission data from Trusts Ambulance Activity data by patent and Practice Somerset Primary Link Activity (main emergency admission avoidance scheme) RISC system data on patients’ likelihood of a future unscheduled admission. Out of Hours activity data by Practice Practice performance data as used in QOF 6

8. * ‘Admission Components’ (subsets that have been used to analyse the data) SUS data through the NERRT2 tool allows interrogation to Practice and Federation level of a number of components of Emergency Admission: Patient Demographics (“Admissions of”): Analysis for under 16s, 16-64, over 65s, 75s & 85s. Providing Trust (“Admissions to”): Analysis by each individual Trust providing to Somerset patients Method of Admission (“Admissions through”): A&E admissions, GP admissions, other admissions. Clinical Condition (“Admissions for”): Analysis by HRG chapter plus analysis of codes for ‘major / minor’ procedures and ‘with’ or ‘without complications Length of Stay (“Admission Stay”): Zero days, 1 day, 2-3 days, 4+ days. Time of Admission: including “in hours” and “out of hours” 7

9. * Tools - NERRT2 (Overall Trend) (main analytical tool developed to review datasets – now on Dashboard) The front page of NERRT2 reports monthly recorded activity and reports variance for Financial Year to Date, 12 month growth, and growth from 2008/9: 8

10. Admission Component Summaries One page report can be printed from the NERRT2 tool. Outlines admissions growth in all particular components. Can be set for Somerset overall and for any Federation or Practice. There is also a similar one page report to outline each Practice’s & Federation’s admission performance. Can be set for all admissions or for any of the individual admission components show opposite. Also shown graphically 9

11. NERRT2 - Analysis by ‘Admission’ Component NERRT2 analytical tool reports growth in admission rate by each admission component – both graphically and in tabular form. 10

12. NERRT2 - Analysis by Practice & Federation NERRT2 analytical tool reports overall admission rate by Practice and Federation (the Federations are highlighted in the chart below) 11

13. NERRT 2 Analysis by Practice & Federation (cont) NERRT2 analytical tool also reports growth in admission rate by Practice and Federation (the Federations are highlighted in the chart below) 12

14. * Emergency Admission Trends (Raw Results – based on original / unadjusted SUS data) Analysis of SUS data has allowed the following ‘headline figures’ to be calculated Somerset Trusts have agreed these reported growth statistics. 13

15. Summary of admission rates and growth by Federation The Federations with the highest ‘raw’ admission rate are Taunton & West Somerset The Federations with the highest admission growth are Bridgwater and Taunton 14

16. * Performance against Contract (source – SLAM) Figures have been compared to planned (contractual) activity and variances are reported below: 15

17. * Which reinforces the key objective of this analysis: Why have Somerset emergency admissions grown by 5.6% in the past year? What contributory factors can be identified in relation to this growth? What is the contributory amount of each of those factors? And… Why have emergency admissions to Taunton and Somerset NHS Foundation Trusts grown by almost double this amount? (9.7% growth compared to 5.6% for Somerset) And again, what contributory factors can be identified in relation to this growth? What is the contributory amount of each of those factors? 16

18. OVERALL FINDINGS This has been a very detailed analysis (over 200 slides of results). However the key points that would be offered to readers are as follows: 17

19. * Key Observations Certain Components show ‘irregular’ or ‘step-change’ growth: – Obstetrics, female reproductive system, Somerset PCT zero length of stay – 4% of admissions account for 21% of reported growth. – More work is required to understand these observations. Populations and age of patients have grown: – 31% of current year emergency admission growth (& 56% of growth since April 2008) can be accounted for by changes in population and changes in the age of that population. A change of ‘Provider Mix’ is observable. – Patients in certain parts of Somerset that might previously have been admitted elsewhere appear more recently to be admitted to Taunton & Somerset NHS Trust. – Of approx 1000 ‘additional admissions’ observed last year at T&S, 66% due to change in mix and originate from patients attending Practices beyond the Trust’s ‘core catchment’ – ‘Effect of Mix’ varies by HRG chapter: Significant for Cardiac Conditions & Musculo- skeletal System. Less significant for Digestive System complaints. 18

20. ** Summary of Headline Figures From original “observed growth” of 5.6% in Somerset, 3.0% remains as ‘underlying’ or ‘unexpected’. For Taunton & Somerset, from original observed growth of 9.7%, 4.1% remains unexpected. For Yeovil, observed growth has increased. 19

21. Summary of admission rates and growth by Federation The Federations with the highest admission rate are Taunton & W.Somerset The Federations with the highest ‘underlying’ growth are Bridgwater and Taunton 20

22. * ‘Underlying Growth’ The ‘underlying’ (unexpected) trend of emergency admissions has been calculated and investigated further. Key observations include:  A less than average growth of admissions for patients over 55 years. o Particularly focussed in certain Federations. o Is this evidence of success of QIPP Schemes (or due to other factors)?  A higher than average growth of admission for patients under 55 years.  Growth of admissions of ‘under 55s’ particularly concentrated at T&S: o through both GP and A&E admission routes, particularly short lengths of stay. o focussed in ‘Digestive System’, ‘Male Reproductive’ & ‘Immunology...contacts with Health Services’.  High growth of admissions for children at Yeovil District Hospital (YDH)  Higher growth for patients ‘with complications’ compared to those ‘without’ 21

23. * Analytical Links to other sources Analytical links have also been made to other data sources:  A observed decline in the use and efficiency of Somerset Primary Link.  No observable correlation between growth of admissions & out of hours service.  No observable correlation between growth of admissions & patient’s abilities to access GP Practices.  Ambulance conveyance trends appear mirror overall observations including observed demographic differences and change in mix.  Nursing home analysis show differences in admission rates per home.  Timed data received from T&S show potentially useful information regarding growth of admissions at particular times of day. 22

24. DETAILED OBSERVATIONS The following detailed observations are perhaps worthy of further investigation and clinical debate... Given the time constraints in any meeting, we can only just briefly touch on some of the observations made – but it is recommended that this evidence presented is understood in detail by any of those involved in the planning and execution of QIPP objectives for emergency care. 23

25. CHANGES IN DEMOGRAPHICS OF LOCAL POPULATION - Analysis of population and patient age changes in Somerset - Adjusting admissions for ‘natural’ changes in population and age - Detailed analysis of adjusted admissions for ‘unexpected changes’ - Analysis of admissions of patients of different ages in different Federations Effect on Emergency Admission of... 24

26. Practice Population (Federation) Growth Practice Population statistics have been uploaded to NERRT2 as below. Evidence of data cleansing in 2009 through Summary Care Record. Some federation populations have grown faster than the average…. 25

27. The effects of ’data cleansing’ on practice population files In Somerset, some age bands have show a greater change across 2009 data cleanse period than others – Adults of working age demonstrated below: 26

28. The effects of ’data cleansing’ on practice population files (cont) But the elderly age bands do not seem sot show this change. (e.g. Over 65s below). How will the reported growth in emergency admissions due to population changes change if we took out this “data cleanse” effect? 27

29. Adjusted Demographic Growth: Based on ‘re-estimated’ populations… Patients aged 65 or over has grown by 2.8% in the last 12 months. Adults under 65 has grown by 1%. Children has grown by 0.4%. 28

30. ‘Discounted Admissions’ Method to adjust for changes in population & age. Age band ratios have been set. (<16, 16-64, 65+, 75+, 85+) Practice populations are reported obtained for each age band in each Practice, on a quarterly basis. From April 2008, monthly practice populations by age have been estimated (linear extrapolation between each quarter’s reported actual figures). A ‘population increment factor’ is calculated for each month for each Practice for each Age band (increment factor = popn now / popn @ Apr08). Emergency Admissions for each month for each Practice for each age band are ‘discounted’ by the population increment factor. ‘Discounted’ (adjusted) admissions per Practice per month are calculated and reloaded into NERRT2 for analysis and comparison. 29

31. * Admissions of patients of different ages Emergency admission growth adjusted for rises in population (at each Practice) and rises in the age of the population. 30% of last 12 months growth and 56% from April 2008 ‘removed’ during this adjustment – accounts for natural population growth. Adjusted Admission Growth represents that not explained by underlying changes in the population or age – this is then analysed further. Adjusted Emergency Admission Growth does not appear to be even across all patient ages: – The growth of different ages also differs by Federation. – Highest admission growth for adults aged under 55. – For Elderly patients, in some Federations, adjusted admissions have reduced. Examples of these observations are shown in the following slides.... 30

32. Components of ‘Age Discounted’ growth (source NERRT2 (Pop’n growth adjusted) / Component Summary Table) Underlying admissions by Age Band (after allowing for population growth): Emergency admissions of Children (16 & under) has risen by 6.1% Cohort of patients aged 16 to 54 have grown by on average 7.0% Adjusted Admissions of patients over 55 has risen by only 0.7% 31

33. * Admissions of Children (source NERRT2 (pop’n growth adjusted / Practice Benchmarking Table) Overall adjusted Admissions of under 16s & under has risen by 6.1% Significant difference between Federations: Highest growth in, South Somerset (23%), Central Mendip (17%) & Bridgwater (12%) 32

34. * Admissions of Adults under 55 (source pop’n growth adjusted NERRT2 / Practice Benchmarking Table) ‘Population growth adjusted’ admissions of adults under 55 has risen by 6.8% Significant difference between Federations: Highest growth in, Taunton (17%), CLIC (16%) & Bridgwater (14.2%) 33

35. * Admissions of Adults over 55 (source pop’n growth adjusted NERRT2 / Practice Benchmarking Table) ‘Adjusted’ admissions of adults 55 and over has risen by only 0.6% Some difference between Federations: Drop in admissions in Central Mendip (-2.9%), North Sedgemoor (-1.7%), Bridgwater (-0.8%), East Mendip (-0.3%). 34

36. * Admissions of Elderly – over 75 (source NERRT2 pop’n growth adjusted / Practice Benchmarking Table) Overall Discounted Admissions patients aged over 75 has risen by only 0.2% Falls in admissions of over 75s again in North Sedgemoor (-2.8), Bridgwater (-3.1%) Highest Growth again in CLIC (6.4%), Taunton (1.7%), South Somerset (2.0%) What can be inferred from these statistics concerning effect of QIPP scheme pilots? 35

37. * Bridgwater admissions by age band (as an example of different rates of admission for different ages) Why is there significant differences in rate of growth of admissions before & after 55? Is this evidence of success of Long Term Condition work? – or something else? But if so, what is causing the increase in the under 55s which is masking LTC results? 36 Note Weighted Effect

38. * Key Growth components - adults under 55 Strong ‘differential growth’ between T&S and YDH. Equal growth through A&E and by GP Admission. Highest growing HRG Chapters: Digestive system, Immunology... & contacts with Health Services, Urinary Tract & Male Reproductive, Cardiac Surgery, Respiratory. 37

39. * Key Growth components - adults over 55 Potentially less ‘differential growth’ between T&S and YDH? Growth through A&E but a decline by GP Admission. Highest growing HRG Chapters: Cardiac Surgery, Immunology... & contacts with Health Services, Respiratory, (Digestive system up on 1.9%). 38

40. * Under v’s Over 55 – Length of Stay Significant differences in growth length of stay. Under 55s demonstrate strong admission growth for zero days, 1 day and 2-3 day stays. Over 55s demonstrate growth for 1 day stays but declines on 2-3 day and 4+ day stays. 39

41. Ongoing Questions Can we state that 30% of the growth in admissions in the past year is due to natural rises in the population and the age of that population? – 56% of admission growth from April 2008 caused by demographic changes? – Is our adjustment for effects of “data cleansing” valid? What is causing higher growth in emergency admissions of: – Children? – Adults under the age of 55? Why have emergency admissions of the elderly grown less? What can be inferred from the difference in growth in elderly emergency admissions between Federations: – Evidence of effectiveness of QIPP schemes? – Evidence of different practices at Providers? – Influenced by other factors? 40

42. Adults under 55 – what actions? What can be inferred from the component analysis for adults under 55? Why is the differential growth rates to T&S / YDH more marked? Why is growth equally from both GP & A&E when there are marked differences in these channels for the over 55s? What can be inferred from the highest growing HRGs? – Digestive System – Immunology...& contacts with Health Services – what are these admissions. – Urinary Tract & Male Reproductive System What can be inferred from the difference in growths in length of stay between the under 55s and the overs. 41

43. CHANGES IN PROCESSES AT NON ACUTE SETTINGS - Somerset Primary Link - Out of Hours Service - Primary Care Access - Ambulance Service Effect on Emergency Admission of... 42

44. * Non Acute Factors Somerset Primary Link data analysed using NERRT2 – There appears some Practice & Age coding errors which may require review. – Currently showing reduced use of SPL & increase in proportion of patients to DGH. Out of Hours usage analysed using NERRT2 and correlated to Admissions – Shows variability in use of Out of Hours Service by Practice & Federation. – However no correlation with decline of Out of Hours and Growth of Admissions. Practice Access performance correlated to Admissions – No correlation with poor Practice Access and increased A&E Admissions. – A small correlation with poor Practice Access and increased A&E Attendances Residential & Nursing Homes – RISC analysis – Analysed for each Federation – RISC scores, admissions and ‘cost’ of patients. – Significant variation in admissions per resident observed between homes. The following slides illustrate the observations made... 43

45. Somerset Primary Link (SPL) Analysis Analysis of SPL provided dataset shows a 19.2% decline in use of SPL. Also shows proportionately more patients going to DGHs. If a key SPL indicator is the proportion of patients admitted to DGHs, then a small diminution in service can potentially be observed (i.e. proportionally more patients admitted to DGHs): – For the 12 months to July 2010, 73.6% of SPL referrals were admitted to DGHs – For the 12 months to July 2011, 81.6% of SPL referrals were admitted to DGHs Somerset Partnership information team may wish to review this dataset. 44

46. NERRT2 – Somerset Primary Link source NERRT2 (SPL) / Trend Dashboard) SPL data from Jan 09 uploaded to NERRT2 to be analysed. Overall Use of SPL down by 19.2% compared to last year – although this is due to reported ‘spike’ around July ‘10. Includes all reported SPL episodes (including 6000 episodes with no Practice code) 45

47. * NERRT2 – SPL Analysis by Outcome (source NERRT2 (SPL) / Trend Dashboard) SPL code 7 ‘outcome types’ (see below). Whilst with this dataset use of SPL has dropped by 19.2%, admissions to DGH have decreased by only 10.4%. This may warrant further investigation – indication of lessening of SPL effectiveness? 46

48. * NERRT2 – SPL Analysis by Provider (source NERRT2 (SPL) / Trend Dashboard) For DGH destinations, although overall DGH admissions from SPL have fallen by 10.4%, to T&S it has fallen by 7.8% whilst to YDH and Weston it has fallen further (-14.8% & -26.5%). Could SPL be contributing in a small part to the observations in Change of Mix? 47

49. NERRT2 – Out of Hours GP source NERRT2 (out of hours) / Trend Dashboard) Out of Hours GP data from April 2009 uploaded to NERRT2 to be analysed In Somerset, Out of Hours usage level (up by 0.2%) as at July 2011. 48

50. Out of Hours GP v’s Federation Use source NERRT2 (ambulance) / Benchmark Table) Growth of use of Out of Hours GP analysed by Federation and Practice Different Federations report different trends (N.Sedge up by 3.3%, S.Som down by 3.4%) What can be inferred from this information 49

51. * Out of Hours GP Correlation Analysis Growth of OOH GP per Practice compared to growth of Emergency Admission. Correlation Coefficient calculated. No significant correlation found to growth (or decline) of use of OOH GP service and Practice growth of emergency admissions – either overall or for admissions of “adults of working age”. 50

52. * Practice Access Performance Practice access performance (via the quarterly GP survey) as of March 2011 and Jun 2009 compared to emergency admission growth reported for that Practice. No significant correlations have been observed for either growth of admissions against poor practice access or overall levels of admissions. But A small / moderate correlation (R=0.34) has been observed when comparing Practice access performance against a measured rate of A&E Attendance for patients of that practice. 51

53. * Residential Homes – RISC analysis (source RISC Nursing Home Analysis / CCG Information Dashboard) Reported generated following Federation requests. Now published on CCG Dashboard Why is there disparity between the number of admissions / cost per resident? Why does it seem that residential homes costing the NHS more than nursing homes? Shared with Federations for further clinical analysis and local reflection. 52

54. Analysis of South West Ambulance Data – Conveyances.... Data has been restated for all conveyances “from Somerset” (rather than “to Somerset”) as was previously reported. 53

55. * Ambulance Data Somerset Conveyance Data received from SW Ambulance Trust – Analysed through NERRT2 to calculate latest trends – Admission Components also analysed and compared to SUS trends A reasonable correlation to Admission trends has been observed – Similar overall growth in conveyances to growth in admissions. – Similar increased growth for adults of working age. – Increased conveyances to T&S, fewer to YDH support notion of a ‘change in mix’ Data also provides some new fresh insights to emergency admissions – Conveyances by Urgency & clinical condition. – Commissioning data for ‘see and treat’ and ‘hear and treat’ still to be analysed. The following slides illustrate the observations made... 54

56. NERRT2 – Ambulance Conveyances (source NERRT2 (ambulance) rev’d / Trend Dashboard) Ambulance Conveyance Data uploaded to NERRT2 to be analysed In Somerset, conveyances up by 6.5% for year to Jun 2011 and 11.4% up since April 08 Do our observations on emergency admissions match ambulance information? 55

57. * Ambulance Conveyances v’s Demographics (source NERRT2 (ambulance) rev’d / Component Table) Conveyances of Children up by 13.7% and adults of working age up by 8.5% Conveyances of the over 65s up by only 4.1%. Shows the same pattern of growth as overall emergency admissions. 56

58. Ambulance Conveyances v’s Destination (source NERRT2 (ambulance) rev’d / Component Table) Only data on conveyances to Somerset destinations received. Although conveyances up by 6.5%, conveyances to T&S up by 9.4%, to YDH up by 5.6%. Further evidence in a “shift in mix” towards T&S? 57

59. Ambulance Conveyances v’s Urgency (source NERRT2 (ambulance) / Component Table) Red status conveyances up by 15.2%, Amber status up by 6.8% Green status conveyances down by 0.8% What can be inferred from this information? 58

60. * Ambulance Conveyances v’s Clinical Condition (source NERRT2 (ambulance) / Component Table) Data only to March 2011. Clinical coding changes from April 2011 prevents comparison. Highest rising conveyances for: Health Care Professional Emergency (22%), Chest Pain (10%), Sick person (21%)... Also Overdose (10%), Stroke (12%), Abdo Pains (18%), ‘Falls’ represents 14% of all conveyances by has only risen by 4%. What can be inferred from this information? Is the coding sufficiently robust? 59

61. Questions Arising from Ambulance Analysis Can the Ambulance data be refined further to support more detailed analysis? Is the geographical area analysis supportive of the concept of a change in mix? What can be inferred from the observed trend in: – Increase in conveyances by Demographics? – Increased admissions to T&S over the average? – Increased admissions of red and amber status? – Variations in conveyances by clinical condition? Activity that does not result in a conveyance also needs to be analysed (see & treat / hear & treat). This data has been received and is in the process of analysis 60

62. CHANGES IN PROCESSES ACUTE PROVIDERS -Service Changes - HRG chapters with higher growth rates - Analysis of Length of Stay - “Major” v’s “Minor” procedures -Changes in Provider Mix -Analysis of Time of Emergency Admissions Effect on Emergency Admission of... 61

63. * ‘ Service Change’ Adjustments The following adjustments are made to the data to excluded identified areas of growth occurring during the analysed time period and so calculate the underlying ‘like for like’ growth of emergency admissions: – Zero Length of Stay, Somerset PCT (632 Somerset Admissions in 12mths to Jul11, 46% growth). – Female Reproduction (1094 Somerset Admissions in 12mths to Jul11, 33% growth) – Obstetrics (491 Somerset Admissions in 12 mths to Jul11, 91% growth. All other components reported as higher than average growth have growth consistently and steadily across reported time periods. The observations in Female Reproduction & Obstetrics are not yet understood (and they need to be), but are excluding from the ongoing analysis in the attempt to understanding ‘underlying trends’ 62

64. Zero LOS for HRG - Female Reproductive System Using Zero LOS data subset viewed through NERRT2 Female Reproductive HRG shows start of growth from April 2010. Is this the start of a new service. Should it be excluded for ‘like for like’ comparison. T&S to investigate further & provide explanation. 63

65. Zero LOS for HRG - Obstetrics Using Zero LOS data subset viewed through NERRT2 Obstetrics HRG shows start of growth from Oct 2010. Is this the start of a new service. Should it be excluded for ‘like for like’ comparison. T&S to investigate further & provide explanation. 64

66. * Growth of Admissions by HRG Chapter (source NERRT2 (pop’n adjusted) / Component Table) Following HRG Chapters account for the highest rise in Emergency Admissions: What is accounted for by Immunology....& other contacts with Health Services? Digestive System, Cardiac Surgery and Respiratory account for next highest elements of emergency admission growth. 65

67. * Length of Stay (Under v’s Over 55) Significant differences in growth length of stay between over & under 55s. Under 55s demonstrate strong admission growth for zero days, 1 day and 2-3 day stays. Over 55s demonstrate growth for 1 day stays but declines on 2-3 day and 4+ day stays. 66

68. * ‘Major’ v’s ‘Minor’ HRG procedures (source NERRT2 (pop’n adjusted) / Component Table) HRG codes have been analysed depending on whether for ‘major’ or ‘minor’ procedures (as identified in the descriptive text) and ‘with complications’ or ‘without complications’ Why does there appear to be more growth of admissions for ‘major procedures’ and ‘with complications’? Evidence of change in clinical Practice or changes in coding interpretation? 67

69. Those were the measured variances for Somerset, but what about the individual Trusts.... Analysis of Population Growth Adjusted Emergency Admissions to - Taunton & Somerset NHS Foundation Trust and - Yeovil District Hospital NHS Foundation Trust... 68

70. * T&S Growth by HRG Chapter (source NERRT2 (pop’n adjusted / T&S extract) / Component Table) Most significant growth in Cardiac Surgery (13.1%), Digestive System (15.3%). 69

71. * T&S Growth by demographic band (source NERRT2 (pop’n adjusted / T&S extract) / Component Table) Paediatric admission growth – 2.1% Significant growth of admissions for adults under 55 – 18.8% growth. Growth of admissions for adults over 55 – 3.3%. 70

72. * T&S Growth components - adults under 55 27% growth through A&E, 10% growth through GP Digestive System – 23% of all admissions, up by 28%. 71

73. * YDH Growth by HRG Chapter (source NERRT2 (pop’n adjusted / YDH extract) / Component Table) Excl ‘undefined groups’, most significant growth in Respiratory (7.1%), Mouth (23.4%). Highest fall in admissions for Musculoskeletal (-18.1%) & Cardiac Surgery (-4.8%). 72

74. * YDH Growth by demographic band (source NERRT2 (pop’n adjusted / YDH extract) / Component Table) Significant Paediatric admission growth – 13.6% Unlike T&S, emergency admissions of adults under 55 has declined – 5.3% down. For adults over 55, decline of admissions is less (-2.7%). Growth of admissions for adults over 55 – 3.3%. 73

75. * YDH Growth components - children 28% growth through A&E, 5% growth through GP Childhood & neonates – 77% of all admits, up by 7% / Undefined groups 11% of admits Longer length of stays up by 22% (2-3 days) and 38% (4+days) 74

76. * ‘Major’ v’s ‘Minor’ HRG procedures at T&S & YDH (source NERRT2 (pop’n adjusted) T&S & YDH extracts / Component Table) This coding analysis shows different results for each Trust. What can be inferred? 75

77. * Length of Stay Analysis at T&S & YDH (source NERRT2 (pop’n adjusted) T&S & YDH extracts / Component Table) Again, this shows different growth profiles at different Trusts. What can be inferred? 76

78. Analysis at Trust Level – Ongoing questions After adjusting for movements in population and age, the previous slides represent the “reality” in terms of activity movements at each of those individual Trusts. Each Trust has a very different variance profile, particularly when compared against one another. But what is the true picture for Somerset? – What variances are due to real difference in process and performance at Trusts? – What variances might be due to changes in mix (i.e. Patients who would have been previously admitted at one Trust, now being admitted at another. The following slides illustrate the work that has been undertaken to quantify any ‘change in mix’ that might have taken place in Somerset.... 77

79. Evidence supporting a “change in Provider mix” towards Taunton & Somerset. 78

80. * Change in Provider Mix The greatest growth in admissions to T&S appears to be from Federations ‘further away’. Even more stark at Practice level – Practices on the border between T&S and another Trust show the biggest growth in admissions to T&S. Endeavour to quantify the effect of this observation: – Overall – Analysis of Practices beyond the ‘inner catchement area’ – In what components is this observation occuring. Which HRGs? – Can the effect be observed for Practices at HRG level? Is it equal for all? Need to ensure through this observation, other factors are not overlooked. The following slides illustrate the observations in relation to change in mix... 79

81. Usage of T&S - Practice Scatter Chart (Source: Jacq Clarkson, NHS Somerset Statistician) Plotting T&S use across 2 periods shows more Practices of lower %age T&S use using T&S more in the second time period (see red-dashed area) Practice distribution is not spread equably around the x=y (expected) line (red line). 80

82. Usage of T&S – Age Standardisation Analysis (Source: Jacq Clarkson, NHS Somerset Statistician) Plotting T&S use against ONS population statistics and standardising for age, sex and deprivation identifies particular areas in Somerset which have highest growth of admissions to T&S. Generally they are concentrated on the boundaries between Trust catchment areas. 81

83. * Quantify Movement in ‘Provider Mix’ Taking age / population adjusted data, calculate the additional admissions observed at Taunton from Practices in Federations not necessarily expected to refer to Taunton. Whilst a ‘diminishing list’ the top 20 Practice in this category suggest 66% of the observed overall variance can be attributed to these Practices on the boundary of catchment areas. 82

84. * Demographic component of ‘Additional T&S Admissions’ These statistics (and those that follow) are after adjustment for population and age growth. Most of the additional admissions to T&S appear to be patients aged between 16 and 65. Admissions of adults of working age has risen by 13.2% at T&S compared to 3.8% overall. This component growth potentially accounts for 93% of the total observed difference. 83

85. * HRG Chapter component of ‘T&S additional admissions’ 80% of the additional admission to T&S can be observed from 3 HRG Chapters: Cardiac Surgery (30%), Digestive System (26%), MusculoSkeletal (23%) 67% of Cardiac & Musculo variance explained by change in mix but only 51% of Digestive. 84

86. Cardiac Surgery – ‘Change in Mix’ Repeat Practice analysis for admissions for cardiac surgery admissions only. Top 20 Practices outside inner T&S catchment (Bridgwater, Taunton & West Somerset) account for 67.7% of the total additional observed difference in this HRG Chapter. 85

87. Musculoskeletal System – ‘Change in Mix’ For Musculoskeletal system, Top 20 Practices outside inner T&S catchment account for 67.3% of total additional difference in the HRG Chapter. Similar observations as for admissions for Cardiac Surgery. 86

88. * Digestive System – ‘Change in Mix’ For Digestive system, Top 20 Practices outside inner T&S catchment only account for 51.3% of total additional difference in the HRG Chapter. Change in provider mix has had an effect here but not as significant as for Cardiac Surgery Reinforces that this observation cannot be used as the explanation in all cases. 87

89. Analysis of T&S data of time of Emergency Admission... 88

90. Time Profile of Admissions ‘Timed data’ received from T&S and analysed in details Originally ‘in hours’ v’s ‘out of hours’ was analysed assuming 08:30 to 17:30 as ‘in hours’ Data then re-analysed using 11.30 to 20.30 in hours – matching time profile below. These analyses demonstrated markedly different results…. Work taken to analyse time periods in details – particularly 17:30 to 20:30 window... 89 In Hours

91. * NERRT2 Variant ‘Admission Time Profile’ Analysis Tool Importing timed data into NERRT allows presentation of admission profile in units of ‘admissions per half hour’ – weekdays by half hour periods, weekends by 3 hr time periods. Further Summarisation by 3 hour weekday intervals and ‘in hours’ and ‘out of hours’ 90

92. * NERRT2 Variant ‘Admission Time Profile’ Analysis Tool Can also analyse movement in admission rate for same time segments. This shows overall admission growth happens most between 14:00 & 20:30. Little growth taking place in mornings, later evenings and weekends. 91

93. Some examples of during what times of the day particular components of Emergency Admissions demonstrate highest levels of growth. 92

94. Time Profile Observations Most growth of Admissions of Adults of Working Age – 12:00 to 18:00 Most growth of Admissions of Elderly happening later - 18:00 to 21:00 Growth of A&E Admissions across a wide time band - 10:30 to 18:30 GP Admissions overall flat growth, but time of admission delayed by approx 4 hours. Growth of cardiac admissions – even across all time bands Growth of respiratory admissions – narrow time band – 15:00 to 21:00 Zero Length of stay – most admissions earlier in the day I day Length of stay – most admissions later in the day The following slides illustrate the observations made.... 93

95. 94 Most growth in admissions of adults of working age is taking place earlier in the working day – between 12:00 and 18:00 ‘Time Profile’ growth of Admissions of patients of working age (17 – 64)

96. 95 Although growth is less overall, most growth of elderly admissions is taking place later in the working day – 15:00 to 21:00 Decline of admissions in late morning suggest later admissions. * ‘Time Profile’ growth of Admissions of patients aged over 65

97. 96 ‘Time Profile’ growth of Admissions through A&E. Wide period of growth between 10:30 and 20:30. Less admission between 20:30 and 23:00.

98. 97 * ‘Time Profile’ growth of Admissions through GP. Overall referrals down by 5%. But evidence that GP admissions are also now happening later in the working day.

99. 98 ‘Time Profile’ growth of Admissions for Cardiac Surgery & Primary Cardiac Condition Growth in admissions spread more evenly across the working day. Reflective of the ‘immediate’ urgency of this clinical condition?

100. 99 * ‘Time Profile’ growth of Admissions for Respiratory System Growth in admissions concentrated in later parts of the working day (15:00 to 21:00). Reflective of the ‘long term’ characteristics of this condition?

101. 100 ‘Time Profile’ of Admissions that result in zero length of stay Not unsurprisingly – profile in zero length of stay admissions show most admissions between 09:00 and 18:00

102. 101 ‘Time Profile’ of Admissions that result in one day length of stay Not unsurprisingly, most 1 day admissions take place with admissions later in the day. Also at weekends.

103. Time Profile Analysis – next steps Is this analysis meaningful? How do you overcome the potentially large “so what!” factor? Can identified “short windows” of admission growth be targeted? Does it mean “growth” or “delay” (i.e. “bunched activity”)? Can we use intelligence about the time of growth of admissions to target investment of emergency admission avoidance schemes? Does evidence of “delay” or “bunching” of admissions (i.e. Activity happening later in the day) indicate potential for efficiency review. Going forward: Further reflection on this analysis is required. 102

104. RECAP OF KEY OBSERVATIONS & ONGOING QUESTIONS In Summary 103

105. * Key Observations Certain Components show ‘irregular’ or ‘step-change’ growth: – Obstetrics, female reproductive system, Somerset PCT zero length of stay – 4% of admissions account for 21% of reported growth. – More work is required to understand these observations. Populations and age of patients have grown: – 31% of current year emergency admission growth (& 56% of growth since April 2008) can be accounted for by changes in population and changes in the age of that population. A change of ‘Provider Mix’ is observable. – Patients in certain parts of Somerset that might previously have been admitted elsewhere appear more recently to be admitted to Taunton & Somerset NHS Trust. – Of approx 1000 ‘additional admissions’ observed last year at T&S, 66% due to change in mix and originate from patients attending Practices beyond the Trust’s ‘core catchment’ – ‘Effect of Mix’ varies by HRG chapter: Significant for Cardiac Conditions & Musculo- skeletal System. Less significant for Digestive System complaints. 104

106. * ‘Underlying Growth’ The ‘underlying’ (unexpected) trend of emergency admissions has been calculated and investigated further. Key observations include:  A less than average growth of admissions for patients over 55 years. o Particularly focussed in certain Federations. o Evidence of success of QIPP Schemes (or due to other factors)?  A higher than average growth of admission for patients under 55 years.  Growth of admissions of ‘under 55s’ particularly concentrated at T&S: o through both GP and A&E admission routes, particularly short lengths of stay. o focussed in ‘Digestive System’, ‘Male Reproductive’ & ‘Immunology...contacts with Health Services’.  High growth of admissions for children at Yeovil District Hospital (YDH)  Higher growth for patients ‘with complications’ compared to those ‘without’ 105

107. * ‘Other data sources’ Trends and correlations to other data sources have also revealed:  A observed decline in the use and efficiency of Somerset Primary Link.  No observable correlation between growth of admissions & out of hours service.  No observable correlation between growth of admissions & patient’s abilities to access GP Practices.  Ambulance conveyance trends appear mirror overall observations including observed demographic differences and change in mix.  Nursing home analysis show differences in admission rates per home.  Timed data received from T&S show potentially useful information regarding growth of admissions at particular times of day. 106

108. Recap of Key Results Growth adjusted for ‘Service Changes’, population rises & change in mix. Differences between patients aged over & under 55, also children. Certain key clinical conditions demonstrating growth in those age bands. 107

109. Federation Priorities From NERRT 2 it is possible to review Admission Component summaries for each Federation – which indicate components with highest growth. Different areas of priority appear for different areas of Somerset. 108

110. And in Summary it is possible to frame some key ongoing questions for those involved with Emergency Admissions in Somerset... 109

111. Next Steps: Key Ongoing Questions In addition to the quantification of components of emergency admission growth, this analysis has identified some particular areas for further consideration and research: 1.What is causing observed ‘Service / Step Changes’ in Female Reproduction & Obstetrics? 2.Why is there a measured decline in the use of the SPL? Why do more SPL patients appear to go to DGHs? 3.Observations on Ambulance Analysis: What can be inferred from growth in conveyances and differential growth to different Trusts? 4.How much can the hypothesis of ‘change in mix’ explain observed differences between Trusts? o Can this hide other factors that may be having a particular effect in Taunton? 1.What can be inferred from the lower than average growth in admissions of the elderly and the localities where this is happening the most? o Is this evidence of LTC QIPP effectiveness? 2.What has caused higher than average growth in patients aged under 16 and those adults aged under 55? o Why do we observe growth particularly at T&S, through both A&E & GP admissions, and for Digestive System, Male Reproductive & ‘other contacts with health services’? 110

112. Key Ongoing Questions (cont) 5.What can be inferred from the lower than average growth in admissions of the elderly and the localities where this is happening the most? o Is this evidence of LTC QIPP effectiveness? 6.What has caused higher than average growth in patients aged under 16 and those adults aged under 55? o Why do we observe growth particularly at T&S, through both A&E & GP admissions, and for Digestive System, Male Reproductive & ‘other contacts with health services’? 7.What can be inferred from detail study of growth of ‘Admission-Time profiles’ to improve the outcomes for admitted patients? 8.And Overall: How can we ensure that analytical observations are kept up to date and, most importantly, converted into clinical explanations? 111

113. End of results For further queries contact: Kevin.Hudson@somerset.nhs.uk 01278 727 401 / 07717 530 220 112