NLG (15) 401. DATE 29 September Trust Board of Directors Public REPORT FOR. Mr Lawrence Roberts, Medical Director REPORT FROM

Transcription

1 NLG (15) 401 DATE 29 September 2015 REPORT FOR Trust Board of Directors Public REPORT FROM Mr Lawrence Roberts, Medical Director CONTACT OFFICER Jeremy Daws, Head of Quality Assurance SUBJECT Monthly Mortality Report BACKGROUND DOCUMENT (IF ANY) Monthly Quality Report REPORT PREVIOUSLY CONSIDERED BY & DATE(S) Mortality Performance Committee 22 September 2015 EXECUTIVE COMMENT (INCLUDING KEY ISSUES OF NOTE OR, WHERE RELEVANT, CONCERN AND / OR NED CHALLENGE THAT THE BOARD NEED TO BE MADE AWARE OF) The Board s attention is drawn to the key points section of this report contained on page 4 of this report. HAVE THE STAFF SIDE BEEN CONSULTED ON THE PROPOSALS? N/A HAVE THE RELEVANT SERVICE USERS/CARERS BEEN CONSULTED ON THE PROPOSALS? N/A ARE THERE ANY FINANCIAL CONSEQUENCES ARISING FROM THE RECOMMENDATIONS? N/A IF YES, HAVE THESE BEEN AGREED WITH THE RELEVANT BUDGET HOLDER AND DIRECTOR OF FINANCE, AND HAVE ANY FUNDING ISSUES BEEN RESOLVED? N/A ARE THERE ANY LEGAL IMPLICATIONS ARISING FROM THIS PAPER THAT THE BOARD NEED TO BE MADE AWARE OF? N/A WHERE RELEVANT, HAS PROPER CONSIDERATION BEEN GIVEN TO THE NHS CONSTITUTION IN ANY DECISIONS OR ACTIONS PROPOSED? N/A WHERE RELEVANT, HAS PROPER CONSIDERATION BEEN GIVEN TO SUSTAINABILITY IMPLICATIONS (QUALITY & FINANCIAL) & CLIMATE CHANGE? N/A THE PROPOPSAL OR ARRANGEMENTS OUTLINED IN THIS PAPER SUPPORT THE ACHIEVEMENT OF THE TRUST OBJECTIVE(S) AND COMPLIANCE WITH THE REGULATORY STANDARDS LISTED N/A ACTION REQUIRED BY THE BOARD The Board is asked to note the contents of the Mortality Report

2 Directorate of Performance Assurance Monthly Mortality Report August 2015 Directorate of Performance Assurance, August 2015 Page 2 of 38

3 Board Report Mortality Summary August 2015 Contents 1.0 Introduction Board Action Key Points 4 At a Glance Mortality Dashboard Mortality Indicators Clinical Coding Indicators Update on Mortality Improvement Work Nurse Staffing Levels Glossary Understanding the different Standardised Mortality Ratios 38 Directorate of Performance Assurance, August 2015 Page 3 of 38

4 1.0 INTRODUCTION The monthly mortality report seeks to provide an update on the most recent information available to the Trust and the different work streams underway around this area to support the focus of reducing the Trust s current mortality ratio. 2.0 BOARD ACTION Following recent discussions with the local Health Scrutiny Committee and Healthwatch organisations, requests were made to simplify the monthly mortality report. As a result, the monthly report style has been amended. This newly revised report has been consulted on with Healthwatch and the Scrutiny Committee for their views. 3.0 KEY POINTS The Board s attention is drawn to the following key points: Dashboard at a glance page 5: - Crude mortality remains at 1.51% vs. 1.42% - Latest official SHMI, July 2015, has been published for the period Jan Dec 14, resulting in a numerical score of versus previous quarter s release of This is an improvement and results in the Trust being in the as expected banding once more - HED SHMI 112 (for the year to April 2015) vs. 109, (for the year to April 2014), currently the Trust places within the higher than expected banding according to the HED information. However this is only a provisional indication and the official SHMI will adjust this figure when all national data is included in 2 months time. - HSMR increase 111 vs. 105 reflecting an increase in crude mortality - RAMI slightly increased 104 vs Out of hospital SHMI increasing, particularly marked at Scunthorpe (128 vs. 104) - Goole SHMI, having increased is now on a downward path Page 10: Crude Mortality is above the national average but falling for the past 2 months and is below the peer average. Page 11: The non-elective crude rate in July 2015 was 2.28%, 0.31% lower than the rate in July 2014 and below the lower control limit. Page 12: Top 6 diagnosis groups with highest crude mortality remain unchanged (Respiratory, Cardiology, Infection/Sepsis, Gastroenterology, Cancer and Stroke) Page 12: July 2015 saw 10 less deaths than reported in June and 36 less than the previous 12 months average. Page 13: Monthly focus on stroke the group attending to present updates at the Mortality Performance & Assurance Committee this month. Directorate of Performance Assurance, August 2015 Page 4 of 38

5 Mortality Dashboard In-Hospital Crude Rate Deaths Discharges National SHMI HED SHMI HSMR RAMI 1.51% (year to Jul-2015) 1, , (year to Dec-2014) 112 (year to Apr-2015) 111 (year to May-2015) 104 (year to Jun-2015) 1.42% (year to Jul-2014) 1, , (year to Sep-2014) 109 (year to Apr-2014) 105 (year to May-2014) 103 (year to Jun-2014) In and Out of Hospital HED SHMI - moving annual totals Trust in/out SHMI: 107 v 122 (year to Apr-2015) Grimsby 111 v 116 Scunthorpe 104 v 128 Goole in/out 75 v v 117 (year to Apr-2014) in/out SHMI: 106 v 115 in/out SHMI: 105 v 121 SHMI: 15 v 70 Key: Full SHMI In Hospital SHMI Out Hospital National Average HED SHMI outlier status: "Higher Than Expected" Top Six Diagnosis Groups: Crude Mortality Trending last 24 months (Aug-13 to Jul-15) Key: Monthly SHMI Linear Monthly SHMI Mean UCL LCL HED SHMI (year to Apr-2015) (year to Apr-2014) Non Elective In Hospital Crude Mortality by day of admission/discharge Produced by Information Services, August 2015 HED SHMI (year to Apr-2015) (year to Apr-2014) Directorate of Performance Assurance, August 2015 Page 5 of 38

6 What are we doing about it? Cardiology The cardiology work stream, led by Dr Morgan, has recently reported on its work to the mortality performance and assurance committee. Themes emerging from case reviews include: Lack of consistency regarding the use of oxygen Last offices not being recorded in notes Atrial fibrillation not always treated in the community Lack of availability of palliative care nurses Actions now being taken include: Oxygen standard prescription going to NMAF/governance Nursing taking up last offices issue Dr Ali presenting to GP training session End of life group reviewing single number for access to palliative care services Some of these issues have also been identified in other mortality work streams. Respiratory The respiratory work stream continues to seek out opportunities for improvement in quality of care and Dr Yasso is currently delivering a number of training sessions to consultant colleagues and trainees about the themes identified and the lessons learned to date. Highlights of the group s recent work include: Bottlenecks in A&E/ CDU triage have been resolved through adding the COPD pneumonia pathway alert to Symphony in A&E Trust-wide and updating the admission/handover sheet in CDU DPOW smoking cessation: a senior HCA has been appointed in a stop smoking service liaison role A recommendation has been made that the term LRTI (lower respiratory tract infection) for the purpose of coding should not be used alone without specifically mentioning whether it is for eg community acquired pneumonia, hospital acquired pneumonia aspiration, TB, bronchiectasis; and LRTI should not be used in death certificates A full-time fourth consultant has been appointed at DPOW to support respiratory in-reach to reduce time to specialist opinion for acutely ill respiratory patients Directorate of Performance Assurance, August 2015 Page 6 of 38

7 Liaison with bed managers to ensure that COPD patients at SGH are transferred to ward 22 Oxygen prescription chart has been approved and will be implemented Trust-wide Alert for patients at risk of Co2 retention has been added to A&E triage and will be added to WebV Respiratory in-reach in DPOW and SGH respiratory patients to be reviewed within 24 hours of admission. Mortality work steam update cross-discipline projects This work stream, led by Professor Carrock Sewell, recently reported to the Mortality Performance and Assurance Committee. Updates on the group s eight strands of work included: Reducing patient transfers data analysis completed, process to be redesigned following clinician analysis Structured ward rounds junior doctors have created an aide memoir to use as a template which will be rolled out throughout the Trust Handovers continuing to promote WebV as preferred system of electronic handover of all patients Using SBAR considering using SBAR on structured ward round card and all ward phones Reducing misidentification ward boards are being updated to display information on misidentification events Improving health records quality audits are being run each month on departmental notes and a plan is being worked up to reintroduce notes splitting teams Improving quality of discharge letters all electronic discharges now have feedback buttons on letters so GPs can like or dislike a letter and any comments Developing systems for tracking use of clinical pathways on WebV WebV board has approved the principle, design team are now developing the system; pathways will be linked to the WebV menu screen. Directorate of Performance Assurance, August 2015 Page 7 of 38

8 1.0 Introduction 2.0 Board Action 3.0 Key Points This section 4.0 Mortality Indicators 4.1 Crude Mortality Overview 4.2 Crude Mortality by Diagnosis Groupings 4.3 NLAG Non Elective Crude Mortality by Day of Admission / Discharge 4.4 Standardised Mortality Indicators 4.0 Clinical Coding Indicators 5.0 Update on Mortality Improvement Work 6.0 Nurse Staffing Levels 7.0 Glossary Directorate of Performance Assurance, August 2015 Page 8 of 38

9 4.0 MORTALITY INDICATORS The following section of the Trust s Mortality Report is compiled by Information Services. It contains high level analysis of NLAG s crude mortality, Summary Hospital Level Mortality Indicator (SHMI), Hospital Standardised Mortality Ratio (HSMR) and Risk Adjusted Mortality Index (RAMI) 4.1 Crude Mortality Overview Crude Mortality Indicators Dashboard The following dashboard looks at our crude mortality indicators at a Trust and site level. Indicator Source: Information Services / CHKS Aug-2014 to Jul-2015 Prev 12 mths Annual Change NLAG Moving Annual Total (MAT) Crude Mortality Rate v Peer The first graph in this crude mortality section shows the Moving Annual Totals (MAT) for the NLAG crude mortality rate against peer. This includes all deaths. A MAT is the sum of the individual twelve monthly figures up to and including the reporting month e.g. twelve months to July This methodology helps to obtain a trend with less variance. A crude mortality rate is simply the number of deaths divided by the number of discharges expressed as a percentage. The discharges in the methodology exclude well babies. Statistical Process Control (SPC), showing 95% upper (UCL) and lower (LCL) control limits, has been applied to NLAG crude mortality in the graphs below to highlight excessive variation. Peer Compared to Peer CRUDE MORTALITY Trust 1.51% 1.42% 0.09% -0.02% DPOW 1.49% 1.46% 0.03% -0.04% M1 Crude Mortality Rate 1.53% SGH 1.66% 1.49% 0.18% 0.13% GDH 0.39% 0.47% -0.08% -1.14% M2 Non Elective Crude Mortality Rate Trust 3.38% 3.15% 0.23% 0.26% DPOW 3.27% 3.18% 0.09% 0.15% 3.12% SGH 3.44% 3.07% 0.37% 0.32% GDH 5.66% 5.64% 0.02% 2.54% Trust 1,617 1, DPOW M3 Number of Deaths n/a n/a SGH GDH Source: Information Services / CHKS Directorate of Performance Assurance, August 2015 Page 9 of 38

10 Comment: For the twelve months to July 2015, the crude mortality rate for the Trust was 1.51%, an increase of 0.09% compared to the rate of 1.42% for the twelve months to July NLAG s performance continues to be better than peer. The gap to the national average is still evident. For the twelve months to July 2015 there were 1617 deaths in hospital NLAG Monthly Crude Mortality Rate v Peer The following graph shows the monthly trend of Trustwide crude mortality rates against peer. Source: Information Services / CHKS Comment: The crude rate in July 2015 was 1.05%, 0.08% lower than the rate in July 2014 and below the lower control limit (outlier). Note the winter peaks across the years in the graph and where the NLAG rate is higher than the upper control limit (outliers); higher mortality rates are expected in winter months. (The peer rate for the latest month and the last two months national average rates have been estimated in the above graphs). Non Elective Crude Mortality As the majority of deaths occur within non elective patient admissions, the following section looks at non elective crude mortality NLAG Moving Annual Total (MAT) Non Elective Crude Mortality Rate v Peer The following graph shows the Moving Annual Totals (MAT) for the NLAG non elective crude mortality rate against peer. Source: Information Services / CHKS Directorate of Performance Assurance, August 2015 Page 10 of 38

11 Comment: Note that for all months NLAG s non elective crude mortality rate has been above peer. The crude non elective mortality rate for the twelve months to July 2015 was 3.38%, an increase of 0.23% from the rate of 3.15% for the twelve months to July For the twelve months to July 2015 there were 1585 non elective deaths NLAG Monthly Non Elective Crude Mortality Rate v Peer The following graph shows the monthly trend of Trustwide non elective crude mortality rates against peer. Source: Information Services / CHKS Comment: The non elective crude rate in July 2015 was 2.28%, 0.31% lower than the rate in July 2014 and below the lower control limit (outlier). Note the winter peaks across the years in the graph and where the NLAG rate is higher than the upper control limit (outliers); higher mortality rates are expected in winter months. (The peer rate for the latest month and the last two months national average rates have been estimated in the above graphs). Directorate of Performance Assurance, August 2015 Page 11 of 38

12 4.2 Crude Mortality by Diagnosis Groupings Patients Died In hospital by Primary Diagnosis Summary Group The following table shows the number of deaths in each diagnosis group. The diagnosis group is based on the primary diagnosis at the time of death Grand Latest Month Latest Month Total Vs. Previous Vs. 12 Primary Diag Summary Group Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Month Months Avg Respiratory Cardiology Infection Gastroenterology Cancer Stroke Renal Trauma and Orthopaedics Vascular Diabetes and Endocrine Neurological Rheumatoid DVT/PE Neonatal Urinary Tract Miscellaneous Psychological Haematology Grand Total , Source: Information Services 12 Months' Trend Comment: The latest month of July 2015 had 10 fewer deaths than the previous month and 36 fewer deaths than the previous twelve months average. Respiratory, Cardiology, Infection, Gastroenterology, Cancer and Stroke are the six diagnosis areas with the higher number of deaths. Directorate of Performance Assurance, August 2015 Page 12 of 38

13 4.2.2 Crude Mortality SPC Trend Charts for Diagnosis Summary Group Work Streams The following graphs show monthly crude mortality over the last twenty four months for one of the top six primary diagnosis groups. Accompanying each graph is a table showing the crude mortality rates split by site. Goole is excluded from the crude mortality rates table as due to the low number of deaths at this site, the rates fluctuate. Goole is included in the secondary table showing the number of deaths. Stroke Group - SPC Chart Stroke Group - Crude Mortality Rates Primary Diag Summary Group Jun Jul Stroke Stroke Group - Number of Deaths Source: Information Services Latest Month Vs. Previous Month 12 Month Avg DPOW 2.9% 2.6% -0.2% 8.0% -5.3% SGH 7.5% 0.0% -7.5% 7.8% -7.8% Grand Total 6.1% 0.8% -5.3% 7.9% -7.1% Primary Diag Summary Group Jun Jul Stroke Latest Month Vs. Previous Month 12 Month Avg Latest Month Vs. 12 Months Avg Latest Month Vs. 12 Months Avg DPOW SGH GDH Grand Total Comment: There is one special cause variation on the graph, in the most recent month when the crude mortality fell outside the lower control limit. When considering site performance, Grimsby has a slightly higher twelve month average mortality rate than Scunthorpe (8.0% v 7.8%). Directorate of Performance Assurance, August 2015 Page 13 of 38

14 4.2.3 NLAG Non Elective Crude Mortality Rate by Day of Admission / Discharge Twelve Months to July 2015 In Hospital Rates: The following radar graphs show the NLAG non elective in hospital crude mortality rate by day of admission and by day of discharge. Source: Information Services Comment: The admission day with the highest in hospital non-elective crude mortality rate (3.9%) is Sunday. Wednesday admissions have the lowest in hospital non-elective crude mortality rate (3.0%). The in hospital non-elective crude mortality rates for weekend discharges are higher than the rates for weekday discharges. Saturday and Sunday discharges have the highest in hospital non-elective crude mortality rate (4.9% and 5.0% respectively) and Friday discharges have the lowest rate (2.7%). Out of Hospital Rates (within 30 days of discharge): The following radar graphs show the NLAG non elective out of hospital crude mortality rate by day of admission and by day of discharge. Source: Information Services Comment: The admission day with the highest out of hospital non-elective crude mortality rate is Thursday (2.0%), closely followed by Wednesday and Saturday (1.8%). Sunday admissions have the lowest out of hospital non-elective crude mortality rate (1.3%). The out of hospital non-elective crude mortality rate is highest for Friday discharges (2.5%) and lowest for Saturday and Sunday discharges (0.7% and 0.5% respectively). Directorate of Performance Assurance, August 2015 Page 14 of 38

15 4.2.4 NLAG Non Elective Out of Hospital (within 30 days of discharge) Crude Mortality Rate by Day of Death Twelve Months to July 2015 The following radar graph shows the NLAG non elective out of hospital deaths as a percentage by day of death. Source: Information Services Comment: The highest percentage of deaths within 30 days of discharge occurs on Fridays (15.8%), closely followed by Thursdays and Saturdays (15.3% and 15.1% respectively). The lowest percentage of deaths within 30 days of discharge occurs on Sundays (12.6%) and Mondays (12.9%) NLAG Non Elective Out of Hospital Deaths (within 30 days of discharge) by No of Days from Hospital Discharge to Date of Death in Community Twelve Months to July 2015 The following pie chart shows the NLAG non elective out of hospital deaths as a proportion by the number of days from hospital discharge to date of death in the community. Source: Information Services Comment: The majority of deaths within 30 days of discharge occur after a week of discharge (61%). 8% of deaths within 30 days of discharge occur between 0-1 days of discharge. Directorate of Performance Assurance, August 2015 Page 15 of 38

16 4.3 Standardised Mortality Indicators The following section provides high level analysis of NLAG s performance in three of the key mortality indicators - Summary Hospital Level Mortality Indicator (SHMI), Hospital Standardised Mortality Ratio (HSMR), and Risk Adjusted Mortality Index (RAMI) Summary Dashboard Indicator Nationally Published SHMI Jan14-Dec14 SHMI National Position Oct13-Sep14 SHMI National Position Change in National Position M5 Summary Hospital Level Mortality Indicator (SHMI) Trust / / place Indicator Provisional SHMI (HED sourced) May14 - Apr15 Prev 12 mths Annual Change National Peer Compared to Peer Trust DPOW M6a Provisional SHMI 100 SGH GDH M6b M6c In Hospital Provisional SHMI Out of Hospital Provisional SHMI Trust DPOW SGH GDH Trust DPOW SGH GDH Indicator HSMR Jun14 - May15 Prev 12 mths Annual Change National Peer Compared to Peer M7 Hospital Standardised Mortality Ratio (HSMR) Trust Indicator RAMI M8 Risk Adjusted Mortality Index (RAMI) - All Conditions Jul14 - Jun15 Source: Information Services / CHKS / HED / HSCIC. Prev 12 mths Trust Annual Change National Peer Compared to Peer Directorate of Performance Assurance, August 2015 Page 16 of 38

17 A section on each of the mortality indicators follows: Nationally Published SHMI The Summary Hospital-level Mortality Indicator (SHMI) is the nationally agreed mortality indicator. It is a ratio of the observed deaths in a trust over a period of time divided by the expected number given the characteristics of patients treated by that trust. It includes deaths occuring up to 30 days post discharge and excludes day cases. The national average SHMI is 100. A more detailed description of the SHMI methodology and its comparison to other mortality indices can be seen in the glossary. Latest Nationally Published SHMI mortality position The most recent Summary Hospital Level Mortality Indicator (SHMI) was published in July 2015 and covers the January 2014 December 2014 time period. The Trust s SHMI score was 110 ranking 121 out of the 137 NHS provider organisations included in data set this continues to be officially within the expected range. In the previous quarter s SHMI release NLAG was ranked 122 out of 137, also with a score of 110. The SHMI includes all deaths in hospital and those deaths that occurred within thirty days of discharge. The indicator uses data that is normally around six months out of date, for example the July 2015 release covered the period January 2014 December NLAG s SHMI in National Context The following chart illustrates the Trust s most recent SHMI score in relation to those of all Trusts nationally. Source: Information Services / Information Centre In and Out of Hospital Split One of the SHMI contextual indicators that are published is the rate of SHMI deaths that occurred in and out of hospital. NLAG had 69.1% of SHMI deaths occurring in hospital the national rate was 71.5%. The SHMI indicator is not solely a hospital based mortality indicator, but is influenced by wider community-based healthcare also. Directorate of Performance Assurance, August 2015 Page 17 of 38

18 4.3.3 HED Provisional SHMI Following the acquisition of the University of Birmingham Hospitals Healthcare Evaluation Data (HED) reporting product, we can now report on more up to date SHMI data. You will note that the July 2015 nationally published SHMI contained data up to December 2014; the HED data currently shows data to the end of April Data in this analysis should be treated as provisional. Note that HED has not yet been updated with the latest SHMI model from HSCIC. Current SHMI mortality position The Trust s latest SHMI position (using the HED system for the period May 2014 to April 2015) is a score of This is in the higher than expected range, and shows a decrease from the April 2014 to March 2015 position (113.2) which was also in the higher than expected range. The following table shows the provisional SHMI for the year to April 2015 split by hospital site plus the split between the in hospital and out of hospital elements of SHMI: Site Patient Spells Deaths Expected Deaths Variation from Expected SHMI (full) In Hosp SHMI Out of Hosp SHMI Grimsby Scunthorpe Goole Grand Total Source: Information Services / HED. Comment: Grimsby has a higher SHMI score (113) than Scunthorpe (111) and Goole (55). There is a higher out of hospital SHMI for the Trust and the two larger sites a difference of 15 points between the two elements of SHMI at trust level. NLAG - relative SHMI mortality performance The Trust is ranked as 120 of the 137 NHS provider organisations included within the mortality data set. This is some way from where the Trust wishes to be. The Trust has moved out of the as expected range banding. The following funnel plot graphically represents this. Source: HED Directorate of Performance Assurance, August 2015 Page 18 of 38

19 Trustwide Provisional SHMI Trending to April 2015 The following graph shows the moving annual total (MAT) for our SHMI score. Source: Information Services You can see that the Trust, and the Grimsby and Scunthorpe provisional SHMI scores have consistently been higher than the national average score of 100. Goole s SHMI has dropped off dramatically this is in part due to the removal of emergency admissions from summer The majority of provisional SHMI scores (blue line on graph) match the quarterly official SHMI scores (orange points). Where there is a slight difference this can be explained by two factors. Firstly, the statistical modelling data is more up to date on the provisional SHMI than it is for some of the more historic official SHMI statistics. This can have the effect of minimally changing the more historic provisional SHMI scores. Secondly, the provisional SHMI uses more up to date data, most notably out of hospital deaths from Office of National Statistics (ONS). Mortality performance Analysis of In and Out of Hospital SHMI Trending Graph Practice in our hospitals is not the only factor influencing the SHMI score. The split of the SHMI into the in and out of hospital components gives us an indication of the SHMI performance for these two elements of the health community. The methodology used for the split of the full SHMI is the same as the one used in the Boston Consulting Group SHMI analysis undertaken on behalf of our commissioners in early The following graph shows the moving annual total scores for the full SHMI, the in hospital SHMI and the out of hospital SHMI for the Trust. Directorate of Performance Assurance, August 2015 Page 19 of 38

20 Provisional SHMI: In/Out of Hospital SHMI Score Trending Graphs The following graph shows the moving annual total scores for the full SHMI, the in hospital SHMI and the out of hospital SHMI for the Trust. Source: Information Services/HED. Comment: The trend of the in hospital element of the SHMI mirrors the trend of the full SHMI itself over the periods shown. Note the gap on the graph between the in and out of hospital SHMI. The following graphs show the moving annual total scores for the full SHMI, the in hospital SHMI and the out of hospital SHMI for each of the two main hospital sites. Source: Information Services / HED. Comment: Note the overall reduction of out of hospital SHMI over the periods shown. The in hospital SHMI has remained largely static over the last nine months. Directorate of Performance Assurance, August 2015 Page 20 of 38

21 Source: Information Services / HED. Comment: Note the overall increase of out of hospital SHMI and the wide gap between the in and out of hospital SHMI Hospital Standardised Mortality Ratio (HSMR) The Dr Foster Hospital Guide is released each November/December. The December 2013 release showed NLAG with an overall HSMR score of 109 for the twelve months to March This was on the borderline of the higher than expected banding (the upper limit is ) and showed a reduction of 9 points from the score of 118 for the previous year (twelve months to March 2012). The HSMR was also provided for emergency weekend and weekday admissions. These figures showed a higher HSMR (113) for weekend emergency admissions than for those on a weekday (109). Both HSMRs were in the as expected banding. In terms of more recent data, the following, sourced from the Healthcare Evaluation Data (HED) toolkit, shows the Trust s most recent 12 months HSMR score in comparison with other national Trusts. Source: Information Services/HED Directorate of Performance Assurance, August 2015 Page 21 of 38

22 Comment: NLAG has a HSMR of 111 for the twelve months to May 2015, the national average being 100. A section on the Trust s RAMI position follows Risk Adjusted Mortality Index (RAMI) NLAG RAMI Score in National Context The Trust now uses the SHMI as its key standardised mortality ratio (SMR), however it is prudent to monitor performance on other indicators if we have them available to us. The Trust has access to the CHKS Signpost product to monitor other areas of performance such as emergency readmission rates, outpatient did not attend (DNA) rates and new to review ratios. Signpost also gives us access to Risk Adjusted Mortality Index (RAMI). The RAMI is a standardised mortality ratio that is casemix adjusted. It uses a different methodology from SHMI. Where possible, a RAMI score should be compared to a peer value as the older the statistical model is, a score of 100 is not necessarily the norm. CHKS update, or rebase, the model once a year. The graph below shows the Trust s most recent 12 months RAMI score in comparison with other national Trusts. Source: CHKS Comment: NLAG has a RAMI score of 104. The national average RAMI score is 97 and the local peer of similar Trusts is 102. Note that CHKS have re-based the RAMI model since the previous report; the usual outcome of a rebase is that index scores rise by a number of points. Directorate of Performance Assurance, August 2015 Page 22 of 38

23 1.0 Introduction 2.0 Board Action 3.0 Key Points 4.0 Mortality Indicators This section 5.0 CLINICAL CODING INDICATORS 5.1 Depth of Coding 5.2 Recording of Co-morbidity Codes 5.3 Signs and Symptoms R Codes 5.4 Palliative Care Coding 6.0 Update on Mortality Improvement Work 7.0 Nurse Staffing Levels 8.0 Glossary Directorate of Performance Assurance, August 2015 Page 23 of 38

24 Clinical Coding Indicators 5.0 CLINICAL CODING INDICATORS The following report outlining coding indicators is compiled by Information Services. It contains analysis of NLAG s performance in relation to those coding indicators that have an impact on mortality indicators such as SHMI, RAMI and HSMR. Executive Summary: The data contained within this section illustrates: During the last year, the average depth of coding has remained static in the trust and continues to be better than the peer average. In July 2015 there were 5.0 diagnoses per coded episode compared to a peer of 4.6. Over the last year, the number of co-morbidities recorded and coded has remained static. In July 2015 there were 8,502 co-morbidity codes. The continued full collection and recording of co-morbidities should be reasserted to clinicians. Over the last year, the percentage of admissions with an R signs and symptoms code has risen 8.7% in July 2015 compared to 7.1% in July The percentage of first episodes with an R code as a primary diagnosis has been better than peer since October In June 2015, the trust had 8.7% of first episodes with a primary diagnosis R code the peer average was 10.4%. Continued Clinical Coding input with clinicians on recording appropriate, specific diagnoses should continue. With regards to coded palliative care (Z515 code), the trust performance is slightly below the peer in the latest month for the percentage of episodes with a palliative care code (Trust 0.76% v peer 0.80%). A lot of work has been invested into improving the recording in, and subsequent coding of, clinical casenotes. Pro-active work continues to take place in the communication between clinical staff delivering care and the coding team to ensure what is recorded on the data systems within the Trust is as accurate and in-depth as possible. Directorate of Performance Assurance, August 2015 Page 24 of 38

25 5.1 Depth of Coding Depth of coding is the average number of diagnosis codes per episode of care. A high depth of coding reflects a wide source of clinical information captured in the casenotes that then goes on to be coded. It is widely recognised that a high depth of coding may be an advantage in relation to mortality indicators such as SHMI, RAMI and HSMR as it helps to accurately reflect the total number of expected deaths. This said, the quality of the source diagnoses in the casenotes is also important having many non-specific diagnoses will not benefit the Trust in relation to mortality indicators. Depth of Coding April 2012 to present The following graph shows the depth of coding for all episodes of care from April 2012 to present. Source: Information Services/CHKS Comment: The average depth of coding for the Trust has remained largely static over the last year. You will see that since April 2012, Grimsby has performed above the peer average. Scunthorpe has been improving and since January 2013 had mostly matched or beaten the peer average however in the latest few months the depth of coding dropped below the peer. The depth of coding for the Trust was 5.0 diagnoses per coded episode of care in July 2015 a decline from 5.1 in July 2014 but performance continues to be better than peer (4.6). In the most recent month there were 52,113 diagnoses coded across the trust. An analysis of the recording of co-morbidity codes, which can positively affect the expected number of deaths in mortality ratios, such as SHMI, follows. Directorate of Performance Assurance, August 2015 Page 25 of 38

26 5.2 Recording of Co-Morbidity Codes The recording and coding of co-morbidities is important to monitor as they affect the risk given to the patient in the SHMI statistical model. If co-morbidities are not recorded this could be reducing the expected number of deaths and therefore potentially raising our SHMI score. Fully recording co-morbidities also benefits RAMI and HSMR. Number of co-morbidity codes coded The following graph shows the number of co-morbidity codes coded from April 2012 to present. Source: Information Services Comment: The amount of co-morbidities coded remained largely static over the past year. The continued collection of co-morbidities, which should be supported by clinical management, should be stressed to help to continue to improve the amount of co-morbidities collected and coded. We have no benchmarking data to derive a peer comparison. There were 8,502 co-morbidity codes collected in July 2015 across the Trust. An analysis of the recording of signs and symptoms codes, which can adversely affect the expected number of deaths in mortality ratios, such as SHMI, follows. Directorate of Performance Assurance, August 2015 Page 26 of 38

27 5.3 Signs and Symptoms R Codes The recording and coding of primary diagnoses is important as this is one of the data items that affects the risk attached to the patient in the SHMI statistical model. If a diagnosis is recorded as a query or is not specific, then this is coded as an R signs and symptoms code. These R codes hold a lower risk, this is turn reduces the expected number of deaths having the outcome of a higher SHMI score. A reduction in R codes will also benefit RAMI and HSMR. Percentage of patient admissions with an R code as a primary diagnosis The following graph shows a site level trend of the percentage of patient admissions with an R code as a primary diagnosis. * where multi episode spell has a primary diagnosis of an R Code in the first two episodes or where a single episode spell has a primary diagnosis of an R code in that single episode. Source: Information Services Comment: The percentage rate of admissions with an R code signs and symptoms diagnosis has showed an increase over the past year. A reduction will be achieved by clinicians recording appropriate, specific diagnoses upon admission. From October 2012, there was a substantial drop in the percentage of admissions with an R code signs and symptoms diagnosis. This reflects the work that was carried out with clinicians by Clinical Coding. However, the Trust level percentage has increased to 8.7% in July 2015 from 7.1% in July Numerically, there were 816 admissions with an R code in July 2015 this has increased from 680 in July Use of Signs and Symptoms R Codes Benchmarked Position Using the CHKS benchmarking system, we can benchmark our use of R codes against a peer average. The following graph shows the percentage of first episodes with a primary diagnosis of an R code. Directorate of Performance Assurance, August 2015 Page 27 of 38

28 The percentage of first episodes with a primary diagnosis of an R code Source: Information Services/CHKS Comment: The graph shows that since October 2012, our rate against peer average for the percentage of first episodes of care with an R code as a primary diagnosis has improved (dropped). However, the trend has been increasing since December Our rate in June 2015 was 8.7% compared to a peer value of 10.4%. Number of Multiple Consultant Episodes within the Same Spell on the Same Ward split by Operational Group The following graph shows the number of Consultant Episodes where the previous consultant episode in the same spell of care started on the same ward as the current episode s starting ward. *All multiple episodes in the graph above are for consecutive episodes, e.g. if a patient has multiple episodes on the same ward, the episodes must be 1 then 2, etc. If there is a break in the pattern, e.g. episode 1 then 3 - these are not counted. Source: Information Services Comment: The number of multiple consultant episodes within the same spell on the same ward at Trust level has reduced to 98 in July 2015 from 342 in July Directorate of Performance Assurance, August 2015 Page 28 of 38

29 5.4 Palliative Care Coding The recording and coding of palliative care (Z515 code) for appropriate patients will exclude these patients from the RAMI indicator. The code is also used to adjust the Dr Foster HSMR statistic. Presently, the SHMI indicator makes no adjustment for palliative care. Percentage of episodes with a Z515 palliative care code Benchmarked Position The following graph shows the percentage of episodes of care which were coded with a Z515 palliative care code against the peer average. Source: Information Services/CHKS Comment: Trust performance in the coding of palliative care had dipped below peer during the past year but has moved closer to peer in the last few months and beat the peer in May In June 2015, the trust coded 0.76% of episodes with the code v a peer average of 0.80%. Directorate of Performance Assurance, August 2015 Page 29 of 38

30 1.0 Introduction 2.0 Board Action 3.0 Key Points 4.0 Mortality Indicators 5.0 Clinical Coding Indicators This section 6.0 UPDATE ON MORTALITY IMPROVEMENT WORK 6.1 Action already taken 6.2 Action now being taken 7.0 Nurse Staffing Levels 8.0 Glossary Directorate of Performance Assurance, August 2015 Page 30 of 38

31 Update on Mortality Improvement Work 6.0 UPDATE ON MORTALITY IMPROVEMENT WORK 6.1 Action Already Taken During 2013/14 and 2014/15 significant improvements were noted in the Trust s mortality position. These were as a result of a number of improvement projects that have been undertaken assessing both data quality and clinical care. These improvement projects have been focussed and guided by the monthly provision of the latest data contained within this monthly mortality report, presented and then scrutinised by the Mortality Performance and Assurance Committee (MPAC). This is then in turn provided to the Quality, Patient Experience Committee (QPEC) for their assurance of MPAC s actions, before finally being presented to the Trust Board and then becoming publically accessible. Another source of valuable information regarding the clinical care and quality thereof was the use of the mortality trigger tool review process which screened out all deaths and ensured clinical review of cases with triggers, both from a nursing and medical standpoint. This provided recent quantitative and qualitative data on clinical practice. It provided themes requiring further focus. From these two sources of information (1) the monthly mortality report and (2) the mortality trigger tool review process, the Trust developed a focussed and targeted quality evaluation and improvement programme. As part of this, the following areas where prioritised for improvement projects: Clinical areas where the data illustrated highest levels of mortality: o Stroke services o Respiratory medicine o Gastroenterology o Sepsis o Haematology/oncology o Diabetes and endocrine o Acute kidney injury/renal failure Themes identified as areas relating to poor quality: o Hospital acquired pneumonia o Fluid management o Cardiac arrests o Venous thromboembolism (VTE) o Safe staffing. Each of these quality improvement projects were asked to scope out the problem, using anecdotal observations of the teams working in these areas, the feedback from the mortality trigger tool review process and the monthly mortality report. Once they had identified the main issues, they begun to develop ways of targeting these issues with a view to improvement. Each project was monitored by the centrally held mortality action plan. As a result of these projects the following improvements were made: o o Centralisation of the hyper-acute stroke service on the Scunthorpe site to ensure that the specialist and finite resource, was fully able to deliver hyperacute stroke care, as a result mortality performance in stroke, looking at the service as a whole has seen significant improvements. Development and roll-out of respiratory pathways to enable admitting teams to prescribe evidenced based treatment, reducing delays to crucial medications and investigations. Directorate of Performance Assurance, August 2015 Page 31 of 38

32 o o o o o o In collaboration with the respiratory pathways, a sepsis pathway, a more generic series of protocols, was designed and implemented to ensure that patients on admission with sepsis receive potentially lifesaving antibiotics sooner and more consistently. Work has been underway to redesign access for emergency patients requiring urgent endoscopic assessment with the drafting of a joint rota between medics and surgeons. An AKI guideline, protocol and improved guidance on the Web V system, based on abnormal blood work, have been developed and are currently being approved for use. Increased joint working is the objective behind new guidelines to help feed those patients unable to eat to reduce patient deterioration but also to lessen the risk of patients developing hospital acquired pneumonia. Improved guidance, in line with recently published NICE guidance, and a bespoke fluid prescription sheet have been drafted and are currently being approved for use. An improved tool to help get to the root cause behind in-hospital cardiac arrests has been developed and is currently being piloted. It is hoped this will provide effective real time information demonstrating the aspects of care that could be improved in this area to reduce the number of cardiac arrests or to work to improve end of life care planning still further. 6.1 Action Now Being Taken The work going forward into 2015/16 Whilst these improvement projects have made good progress, further improvement is still possible. As such, recent changes in the assurance mechanisms around mortality, leading to the renaming and repositioning of the Mortality Performance Committee (MPC) to the Mortality Performance and Assurance Committee (MPAC), has led to a stocktake in the plans around mortality improvement and a refreshed approach. The fundamental use of the monthly mortality report has not changed. What has changed however within the report is the increased use and reliance of the crude mortality indicator, and less use and reporting from the statistical standardised mortality ratios (SMRs). The reason for this in part was the usefulness front line clinicians found in using these indicators, and most importantly how much access they had using these to access the individual patient records in order to scrutinise the level of quality provided to them from their services. Using the previously relied upon data, access to patient level information to facilitate case note review was a protracted process leading to delays and resulted in their efforts to review and improve care being a project, rather than a day to day part of their management process arrangements. The new refreshed mortality report is designed to be more concise and focussed primarily on crude mortality. An additional benefit of this approach is the timeliness of the data, being only a month behind the present day. To further simplify both the process in reporting this material, but also the data available to front line teams, the information each improvement team will receive is the refreshed mortality report with embedded links to enable them to access the patient specific information that makes up the area s crude mortality performance for any given month. This strategic change moves the ownership of the data from corporate support team members to the front line clinical teams who own the service, thus enabling them to access information when it suits them and their work plans. Directorate of Performance Assurance, August 2015 Page 32 of 38

33 Using this refreshed mortality report, and the focus on crude mortality, reveals six key clinical groupings which have the greatest levels of mortality. This is not to imply these six areas have the highest levels of excess death, rather, simply these areas are those that have the highest numbers, so using the Pareto 80:20 principle, these are the areas where greatest levels of improvement, to the maximum benefit of local service users, can be gleaned. The six areas are unsurprisingly areas where mortality would likely be expected and are as follows: o o o o o o Cardiology Gastroenterology Stroke Cancer/End of life care Infection/Sepsis Respiratory The fundamental approach to take now in these areas again is nothing new. Each have been asked to ensure they have a medical lead, nursing lead, therapy lead and any other multidisciplinary team involvement, relevant to their specific area. Using the refreshed mortality report and embedded links to the patient specific detail, each of these groups is asked to reflect on the previous actions taken (for those groups who have operated previously i.e. respiratory) ensure that any outstanding actions are factored into their new plans and using the also refreshed mortality trigger tool review process, now renamed as the Quality of Care Outcomes Tool to illustrate the process is all about quality not specifically mortality, review cases of mortality and determine what areas of quality could be focussed on and improved. Each of the six groups will be supported to undertake these projects and to further support them in unblocking any obstacles they face during the course of their improvement work, each will be invited on a regular basis to attend the Mortality Performance and Assurance Committee (MPAC) for feeding back on their progress. Regular updates on each of these groups progress will be reported in this monthly document, on page 6, following the at a glance mortality dashboard. Directorate of Performance Assurance, August 2015 Page 33 of 38

34 1.0 Introduction 2.0 Board Action 3.0 Key Points 4.0 Mortality Indicators 5.0 Clinical Coding Indicators 6.0 Update on Mortality Improvement Work This section 7.0 NURSE STAFFING LEVELS 7.1 Transparent Monthly Reporting of Ward Staffing Levels 8.0 Glossary Directorate of Performance Assurance, August 2015 Page 34 of 38

35 Nurse Staffing Levels 7.0 NURSE STAFFING LEVELS 7.1 Transparent Monthly Reporting of Ward Staffing Levels The Trust are committed to ensure that patients and their relatives receive high quality, compassionate care. In order to do this, it is vital that the Trust has the right number of staff, with the right skills available on each ward. Whilst nurse staffing levels has been a constant theme within the Trust s monthly mortality report, alongside work on-going to monitor and report, the Government have recently published the Hard Truths document, following the Francis report into Mid Staffordshire. This report called for more transparent monthly reporting of ward by ward staffing levels. In response to this, the Trust, from June 2014, began publishing nurse staffing levels on its internet site and also on NHS Choices. This greater transparency reports how many staff, both registered nurses and health care assistants, are on shift over the month versus how many were planned. Alongside this, ensuring board to ward oversight, a detailed report on this subject will also be reported to the Trust Board. This has become a regular standing agenda item. Included within this will be an explanation of this information in greater detail alongside explaining what actions are being taken if ward areas fall short. This more detailed board paper is also be made available on the Trust s internet site. This greater transparency has also been rolled out on ward areas throughout the Trust, allowing patients and their relatives to see at a glance prominent boards displaying staffing numbers (see picture to the left). These display how many nurses are actually on duty compared to how many were planned, alongside what action is being taken if levels fall short. This approach has recently been commended during a recent Patient Safety conference. As a result of this increased reporting, all of which is in the public domain, to prevent duplication of reporting, this information will no longer feature within this monthly report. For access to this more detailed information simply follow the link below: Directorate of Performance Assurance, August 2015 Page 35 of 38