«Zero Day Stay ‘Emergency’ Admissions in Thames Valley Higher volumes at particular acute sites after adjusting for population characteristics Dr ...»
Supporting your commitment to excellence
Zero Day Stay
in Thames Valley
Higher volumes at particular acute sites after
adjusting for population characteristics
Dr Rod Jones
Healthcare Analysis & Forecasting
Table of Contents
Table of Contents
Effect of the Healthcare System
Implications to PbR
Effect of Population Characteristics
Method of Analysis
Population Factors Influencing ‘Admission’
Effect of Distance on Zero Day Emergency Admissions
Effect of Acute Thresholds
Specific Comments at HRG Chapter Level
Chapter B (Ophthalmology)
Chapter C (ENT, Oral & Maxillofacial Surgery)
Chapter G (General Surgery – Hepatobiliary and Pancreatic)
Chapter K (Endocrinology & General Medicine)
Chapters M (Gynaecology) and N (Obstetrics)
Volume of ‘Excess’ Zero Day Stays
Role of Assessment Units
Implications to the National Tariff
Benchmarks for Zero Day Stay Emergency Admissions
Appendix One: Population characteristics influencing the volume of zero day stay emergency ‘admissions’
Appendix Two: High volume of ‘emergency’ admission to Ophthalmology at the ORH
Appendix Three: High volume of ‘emergency’ admission to Oral & Maxillo-facial Surgery at the ORH
Appendix Four: Effect of assessment units and other changes at MKGH upon the trends in total emergency admissions
Appendix Five: National average percentage zero day emergency stays at HRG level.
Final Draft – October 2006 Aims To demonstrate that zero day stay emergency admissions are largely a by-product of ‘assessment’ activities.
To provide PCT commissioning and PBC leads with an insight into the PBR implications of zero day stay emergency admissions.
To calculate the volume of zero day stay emergency admissions in particular locations that should arise due to population charactistics.
To assess if zero day stay emergency admissions represent a valid and unique activity which could justify a separate PbR tariff.
This analysis covers any activity reported as an ‘emergency’ admission with a zero day length of stay. As such it will include admissions to observation wards, medical and surgical assessment units, clinical decision units and A&E assessment units. It is also possible that it includes zero day admissions to avoid breaching the four hour A&E target and may also include activities that may otherwise be regarded as an A&E attendance. There is no easy way of determining the exact nature of each type of zero day activity except by detailed audit of the activities at each acute site.
3 of 33Dr Rod Jones (Statistical Advisor) Mobile: 07890 640399
Final Draft – October 2006 Executive Summary This work analyses the results from 2.13 million head of population with144, 000 zero day stay ‘emergency’ admissions per annum. Analysis is at lower super output area level (LSOA)1 covering all extremes of age profile, deprivation, ethnic composition (Asian & Black) and distance to the nearest acute site2 using data for the three years 2003/04, 2004/05 and 2005/06 with volumes normalised to 2005/06 out-turn. Data is analysed at Health Resource Group (HRG) chapter level where each chapter corresponds to a body system, i.e. Nervous System, Vascular System, etc.
A unique relationship between deprivation and increased zero day stay emergency admission is confirmed for each individual HRG Chapter. Ethnicity has a variable effect depending on the specific HRG chapter and ethnic type.
In general, zero day stay emergency admissions increase with decreasing distance to the nearest acute site. They are especially high for the population living within six kilometres (km) of the acute site. However this relationship is unique to each acute site and for some sites such as the Oxford Radcliff and Royal Berkshire Hospital there is no increase in zero day stay emergency admissions for patients living close to the hospital. The highest distance related ‘excess’ is seen in Milton Keynes.
The key finding of this work is that zero day stay ‘emergency’ admission are mainly a by-product of Assessment Units. High volumes of zero day stays arise when ‘assessment’ activities are administratively separated from A&E activities. This division is justified for particular conditions. However, distance specific relationships and sitespecific thresholds drive the overall volume of zero day stay emergency admissions more so than the characteristics of the population such that the PbR cost born by some locations is disproportionatly high.
In this study the 12 acute hospital sites (both within and outside of TV) providing care to the residents of TV is used to define 12 hospital emergency catchment areas3. Each output area was allocated to an acute site catchment using straight line distance4.
Each acute site at the centre of a catchment area does not provide a full range of services, i.e. spinal surgery, burns care, etc; however, it is illustrative to see how relative rates of zero day stay emergency admission vary between different catchment areas. The implications to PbR are discussed. HRG chapter benchmarks and estimates of excess activity have been calculated for each Local Authority, PCT and Acute site.
1 Each LSOA contains around 1,000 to 3,000 head of population. LSOA nest together into electoral wards and can be further nested into PCT or Local Authority boundaries.
2 Straight line distance is measured in km.
3 The 12 acute sites are as follows: Basingstoke, Frimley Park, Heatherwood, Hemel Hempstead, Hillingdon, Horton, Milton Keynes, Oxford Radcliff, Royal Berkshire, Stoke Mandeville, Swindon, Wexham Park, Wycombe.
4 This method assumes that the bulk of the population would normally go to the nearest acute site for emergency care. Around 5% of emergency admissions are to out-of-area hospitals; however for the purpose of establishing good correlations the approximation is fit for purpose.
4 of 33 Dr Rod Jones (Statistical Advisor) Mobile: 07890 640399 Final Draft – October 2006 Key Points Effect of the Healthcare System
Effect of Population Characteristics Rates increase with the Index of Multiple Deprivation (IMD)7, and some HRG • chapters show increased levels of admission due to ethnic populations.
Attempts to analyse Chapter N (Maternity & Neonatal) were frustrated by what • appears to be widespread inconsistency in how events are counted and coded.
5 Surgical procedures only account for 8% of all zero day stay emergency admissions and these are concentrated in what may be called surgical emergency ‘day case’ procedures.
6 Assumes an average ‘real’ cost of £300 per zero day stay ‘emergency admission’
Final Draft – October 2006 Introduction In recent years Thames Valley has shown the highest apparent growth in the volume of emergency admissions in England, however, analysis reveals that this is exclusively related to emergency admissions with a zero day stay, i.e. there has been almost no growth in the volume of non-zero LOS emergency admissions over the past three years. These zero day stay emergency admissions appear to arise when an acute trust shifts the interface from A&E to an Assessment Unit, i.e. activities which would previously have been reported as an A&E attendance are now counted as an ‘emergency admission’ or are counted twice as an A&E attendance and then as a zero day stay emergency ‘admission’.
While part of this shift may represent best practice it acts to confound the analysis and creates a specific PbR problem for two reasons. Firstly around one-third of nonsurgical8 HRGs still do not have a short stay tariff, i.e. a zero day stay is paid for at the same price as a full length stay. Secondly the current short stay tariff includes 0 and 1 day stays and appears to over-remunerate the vast majority of zero day stays. For this reason all zero day LOS emergency admissions have been analysed to determine if there is the potential for material differences across Thames Valley.
Method of Analysis Refer to the companion report covering non-zero day LOS emergency admissions for a full description of the analytical methods.
The only modification was to simplify the effects of distance into just two groups, namely, 0 to 6 km and 6 km. This simplification was required due to the smaller volumes of 0 day stays, i.e. the number of variables in the model was reduced to a level appropriate to the data.
During the process of analysis it was noted that the sum of residuals was higher than expected9. This is interpreted as evidence for the fact that the so-called zero day emergency ‘admissions’ do not have the characteristics of a true ‘emergency’ admission, i.e. the real age profile is most probably closer to that applicable to A&E attendance than to an ‘emergency’ admission. In addition there is huge variation between sites in the relative volumes of admissions, i.e. the activities reported as a zero day stay ‘emergency’ admission are more characteristic of A&E, intermediate or primary care unscheduled care than an ‘admission’.
Finally, there is the suggestion that there is more ambiguity in the HRG codes than may otherwise be expected. Considerable overlap is noted between Chapter N (Female Reproductive) and Chapter M (Pregnancy, Childbirth & Neonates), i.e. it is possible to code the same event in different ways such that it is allocated to different HRG chapters. In particular HRGs M09, M14, M15 and M18 are likely to overlap with N12 if record keeping and coding is ambiguous. Such coding ambiguity may be expected when unscheduled care activities are given a diagnosis simply for the purpose that one is recorded.
Final Draft – October 2006 Population Factors Influencing ‘Admission’ Refer to the companion report for specific comments regarding the role of the Index of Multiple Deprivation (IMD) and ethnicity on the relative volume of admissions.
Coefficients in the model covering these fundamental population characteristics are given in Appendix One. The level of ‘excess’ zero day stays is calculated for each HRG Chapter after adjusting for the fundamental population characteristics of age profile, IMD and ethnicity (Asian or black).
Effect of Distance on Zero Day Emergency Admissions The effect of distance on the volume of emergency admissions has been recognised for many years. The distance effect is usually modelled with some form of decay function such as a power function.
In this study the distance decay was initially simplified into two parts, namely, 0 to 6 km and 6 km. Model testing showed that the inclusion of the factor covering 0 to 6 km was sufficient to give adequate model specificity. Table One gives the proportion of the TV catchment population living within 6 km of various acute sites. As can be seen this proportion ranges between 35% and 75% and thus there is ample scope for a large excess of unscheduled care events arising from the nearby population.
10 Table One: Proportion of total catchment population living within 6 km of an acute site.
The additional admissions arising from the population living within 6 km of an acute site are given in Table Two. All other acute sites do not appear to have any additional admissions from this portion of the population, i.e. it is the system behaviour and not the population characteristics which influence the volume of zero day stays.
In Table Two a figure of 66% implies that there are 66% more ‘admissions’ for people living within 6 km compared to people living six km after adjusting for the effects of age, deprivation and ethnicity.
Note that in Milton Keynes where 71% of the population lives within six km of the acute site the overall ‘excess’ of ‘admissions’ is compounded by very high levels of additional ‘admissions’ arising from this population, i.e. the acute site appears to be functioning (for whatever reasons) as an alternative to primary care rather than an ‘acute’ site.
Table Three: Site thresholds for zero day stay ‘admissions’. Data at HRG Chapter level is averaged over three years and adjusted to 05/06 out-turn. This acts to adjust for the progressive increase in volumes of zero day stays due to assessment units opening over the passage of time.
Important: Explanation of how to interpret a site threshold The site threshold is that portion of the total excess after stripping out anydistance related effects. Hence for the ORH and RBBH the site threshold explains any total excess of zero day stays, however, for Milton Keynes the site threshold of say 146% (as in Table Three) implies that at MKGH all the excess of persons arriving at the hospital with the potential to become a zero day stay emergency admissions have a 46% higher chance of becoming a zero day admission than elsewhere. So if 66% (as in Table Two) more people arrive at MKGH (living within 6 km) than may otherwise arrive elsewhere then the total percentage converting to a zero day emergency will be 46% of the baseline 100% plus 46% of the additional 66% giving 46% + 30% = 76% more than the TV average.
All PCTs using the Swindon & Marlborough Acute Trust should note that in 2004/05 this trust had the 8th highest % zero day emergency stay in England. PCTs may incur additional costs for ambulance and A&E journeys to this site. The MKGH has the 2nd highest percentage of zero day stays.
Table Two: Additional zero day emergency ‘admissions’ arising from the population living within 6 km of the acute site.
The simple fact that there is such a great disparity between sites implies that there are system specific effects. It is suggested that the ambulance service may play an important role in these system specific effects and the Oxfordshire system is worthy of specific comment.