Introduction to sus pbR and hes




Дата канвертавання27.04.2016
Памер65.08 Kb.






Introduction to SUS PbR and HES

1Contents





1 Contents 1

2 Introduction 1

3 Differences between the data sets 1

4 Appendix 4



2Introduction


Hospital Episode Statistics (HES) is the source of official statistics on hospital activity. The information that feeds it comes from Secondary Uses Service (SUS). Another function of SUS is to calculate the tariff, which is how much commissioners should pay providers under payment by results. This calculation is undertaken within a separate component of SUS known as PbR SUS.
The Better Care Better Value (BCBV) tool uses data from the PbR SUS data set, as information on the tariff is central to the indicators. As HES data are also fed from SUS, the activity values will be broadly consistent. However, there are some differences between the processing of the two datasets, which this paper goes on to outline, which could result in differences in numbers recorded against the datasets. In such instances, the HES data provide the official activity figures. However, the PbR SUS data provide useful management information on the activity for which a commissioner paid a provider under PbR.
The period covered by the indicators on the BCBV tool will be those for which the corresponding HES data have already been published.

3Differences between the data sets


Figure 1. Illustrative data flows

Both HES and SUS PbR are populated from the SUS database, which in turn is populated from data input by users in hospital trusts, using patient administration systems (PAS).
The tariff of PbR qualifying activity has been populated onto the PbR SUS data. Given that the BCBV tool is based on calculating potential saving opportunities this financial data is central to each of the BCBV indicators. The PbR tariff information is not available on the HES data.
In PbR SUS data, the in-patient hospital episodes have been converted to hospital spells. Again, this is convenient for BCBV since indicators based on length of stay or readmissions require the hospital spell to be considered rather than the episode of care within a spell.
There are a number of limitations when considering SUS PbR data in comparison to HES. Each of these limitations has been analysed and considered acceptable for the implementation of the BCBV tool given project timescales and data availability. These limitations are considered in turn below.


  1. Cumulative vs. snapshot

A key difference between the way in which HES and SUS PbR are populated is the cumulative nature of HES.

Users in hospital trusts have the opportunity to re-state prior months if errors, exclusions or duplications are identified. This is reflected in HES, as on a monthly basis, the year to date position is extracted. This means that HES is the ‘gold standard’ in terms of quality and completeness as issues can be rectified throughout the year. The SUS PbR data is extracted from the SUS database on a monthly basis and no such restating of prior months occurs. Therefore, where a trust has restated a prior month at any point during the year there will be divergence when comparing HES and SUS PbR.


  1. Spell Construction

The SUS PbR data is spelled; that is, the episodes of care under a consultant within a hospital are combined per patient to produce a single hospital stay or spell. Where the patient data on the episode record is incomplete it is not possible to combine episodes to produce spells. In these examples the record is excluded from the SUS PbR data, but, given that HES is episode based will be included in the HES data.


3% of elective hospital episodes belong to spells with more than one episode, 22% of the total number of episodes belong to spells with more than one episode. Approximately 97% of finished consultant episodes have a valid NHS number1. It is expected that a small proportion of records would be excluded for this reason.


  1. Cleaning and de-duplication of data

HES data is processed in order to clean a number of fields. These are typically cleaned so that only particular values are allowed, invalid values being set to default values. Cross checking also occurs, so that for example, dates are compared to ensure that illogical combinations are excluded. Similarly an effort is made to ensure that duplicated episodes are removed from HES. The SUS PbR extract in contrast contains all records present in the data set at the time of the extract, and while some fields are cleaned or set to default values when invalid input is received there are no cross checks performed.




  1. Index of Multiple Deprivation (IMD)2

One of the derivations performed in the generation of the HES database is that the IMD ranking of a record is derived from the residential postcode of the patient.


The IMD index is based on the concept that distinct dimensions of deprivation such as income, employment, education and health can be identified and measured separately. These dimensions are then aggregated to provide an overall measure of multiple deprivation and each individual lower super output area (LSOA) is allocated a deprivation rank and score. An LSOA will comprise of one or more postcode; so, given the postcode of a patient it is possible to determine the IMD ranking of the patient’s residence.
The BCBV indicators use the IMD ranking of the patient as some group patients together by age band, gender, deprivation band and treatment / condition type in order to compare like with like when considering relative performance. It is also important as financial opportunities are based on the demographics of the population.
The SUS PbR data does not contain the full postcode of the patient since this is considered to be a sensitive field. Instead, the postcode sector is available – this is illustrated below.

Postcode - LS1 6AE

Postcode Sector - LS1 6
This means that it is not possible to determine the IMD ranking of the patient from SUS PbR.
It was decided to use the postcode sector to generate an approximation of the IMD. The IMD ranking of the centroid (geometric centre) of the postcode sector is used, and IMD rankings are grouped to create broad bandings of deprivation for use in indicator generation.


  1. Secondary procedure and diagnosis codes

A number of BCBV indicators require analysis of secondary procedure and / or diagnosis codes. PAS recording allows users to record many procedure and diagnosis codes for each hospital episode. HES contains the first 20 diagnosis and 24 procedure codes, SUS PbR is limited to the first 13 of each. This means that records which have a relevant code in position 14 to 20 or 14 to 24 will not be treated correctly in those indicators which consider these secondary positions. Analysis has shown that less than 0.2% of records have a non NULL value in any of the 14 to 20 secondary diagnosis positions3, while less than 0.005% have a record present in the 14 to 24 secondary procedure positions. See appendix for details.




  1. Baby Ages

A single BCBV indicator (Managing variation in emergency admissions) requires the age of the patient in months for babies under the age of one year. This is because the definition of the ambulatory care sensitive (ACS) condition Influenza and pneumonia states that patients under the age of 2 months should not be considered preventable emergency admissions whereas patients over 2 months should.


SUS PbR does not distinguish between patients under the age of one year so this distinction is not possible.

Again, an analysis of HES was performed which showed that of the emergency admissions for this condition only 4 records during 2008/09 were in relation to a baby of 2 months or less. This was therefore considered to be an acceptable limitation.





  1. Data Quality

The HES publication is accompanied by a data quality note describing the known issues with the data4. The SUS PBR data is likely to present these same issues, and potentially further issues, at any point in time. These data quality notes should be read in conjunction with the detailed indicator specifications in order to determine the likely impact on the indicators shown in the tool.



4Appendix

Count of 2008-09 HES episodes with non-NULL values in diagnosis and procedure codes shown.




 

Episodes with non-NULL values

% of total

 

 

Episodes with non-NULL values

% of total

Main Diagnosis

16,239,503

99.87%

 

Main Procedure

16,233,417

99.83%

Diagnosis 1

11,263,561

69.27%

 

Procedure 1

6,769,776

41.63%

Diagnosis 2

7,597,889

46.72%

 

Procedure 2

3,242,807

19.94%

Diagnosis 3

5,277,690

32.46%

 

Procedure 3

1,569,529

9.65%

Diagnosis 4

3,649,284

22.44%

 

Procedure 4

748,038

4.60%

Diagnosis 5

2,473,356

15.21%

 

Procedure 5

414,166

2.55%

Diagnosis 6

1,618,533

9.95%

 

Procedure 6

240,860

1.48%

Diagnosis 7

1,042,862

6.41%

 

Procedure 7

151,039

0.93%

Diagnosis 8

676,589

4.16%

 

Procedure 8

99,863

0.61%

Diagnosis 9

430,421

2.65%

 

Procedure 9

68,052

0.42%

Diagnosis 10

277,534

1.71%

 

Procedure 10

46,909

0.29%

Diagnosis 11

179,330

1.10%

 

Procedure 11

33438

0.21%

Diagnosis 12

102,975

0.63%

 

Procedure 12

1,490

0.01%

Diagnosis 13

30,428

0.19%

 

Procedure 13

704

0.00%

Diagnosis 14

4,001

0.02%

 

Procedure 14

523

0.00%

Diagnosis 15

2,630

0.02%

 

Procedure 15

412

0.00%

Diagnosis 16

1,648

0.01%

 

Procedure 16

291

0.00%

Diagnosis 17

1,095

0.01%

 

Procedure 17

226

0.00%

Diagnosis 18

734

0.00%

 

Procedure 18

191

0.00%

Diagnosis 19

502

0.00%

 

Procedure 19

158

0.00%

 

 

 

 

Procedure 20

118

0.00%

 

 

 

 

Procedure 21

100

0.00%

 

 

 

 

Procedure 22

87

0.00%

 

 

 

 

Procedure 23

75

0.00%

Total Records

16,261,332

 

 

 

16,261,332

 




1 HES 2009/10 in-patients

2 See http://www.communities.gov.uk/publications/communities/indicesdeprivation07?view=Standard for details on IMD

3 HES 2008/09 in-patients

4 http://www.hesonline.nhs.uk/Ease/servlet/ContentServer?siteID=1937&categoryID=97


Page of


База данных защищена авторским правом ©shkola.of.by 2016
звярнуцца да адміністрацыі

    Галоўная старонка