Clinical Coding Audit Report University Hospital Llandough Cardiff and Vale University Local Health Board

` Clinical Coding Audit Report University Hospital Llandough Cardiff and Vale University Local Health Board Mr Richard Burdon, ACC, NCS Approved Auditor, NWIS Mrs Helen Dennis, ACC, NWIS

CONTENTS University Hospital Llandough Clinical Coding Audit Report... 3 Executive Summary... 3 2 Introduction... 8 3 Aims... 9 4 Objectives... 9 5 Background... 10 6 Methodology... 13 7 Findings... 15 8 Conclusions... 26 9 Recommendations... 27 Page 2 of 48

University Hospital Llandough Clinical Coding Audit Report Executive Summary 1.1 Introduction 1.1.1 In seeking to undertake an audit of clinical coding services as part of their 2013/14 audit programme, the Wales Audit Office (WAO) has sought assistance from the NWIS Clinical Classifications Team to undertake audits of clinical coding accuracy across all Welsh Local Health Boards (LHBs) and Velindre NHS Trust. 1.1.2 This report outlines the findings and recommendations of the NHS Wales Informatics Service (NWIS) Clinical Classifications Team audit of clinical coding accuracy at Llandough Hospital. 1.2 Methodology 1.2.1 The sample audited was 99 Finished Consultant Episodes (FCEs), which were randomly generated from the activity data held within the Patient Episode Database for Wales (PEDW). Only FCEs from the specialties of General Medicine, General Surgery and Trauma & Orthopaedics were audited. The period audited covered episodes with an end date of 1st may 2013 31st August 2013 inclusive. Episodes were also limited to being no longer than 10 days in length. 1.2.2 The locally assigned classification codes were audited against national clinical coding standards using the information available in the patients case notes and relevant electronic systems (e.g. RADIS). 1.2.3 Attention was also paid to the patient case notes being used by the coders and auditors in order to assess their impact on the assignment of codes. 1.3 Findings 1.3.1 Below is a breakdown of the error rates: Code Type Primary Diagnosis Secondary Diagnosis Primary Procedure Secondary Procedure Total Number of Codes Reviewed Total Number of Correct Codes 98 77 78.57% 255 194 76.08% 96 87 90.63% 264 248 93.94% Percentage Correct 1.3.2 In addition to the percentages given above a number of specific findings were made by the auditors: The most prevalent type of errors uncovered during the audit are errors of omission. Errors due to codes being omitted account for 63.55% of all errors encountered during the audit (68 out of 103 errors), and in the case of Secondary Diagnosis errors account for 81.97% of all errors in that area (50 out of 61 errors). Page 3 of 48

Clinical coding staff do not appear to be consulting histology reports when assigning codes to an episode. Out of 21 primary diagnosis errors 5 (23.81%) were due to clinical coding staff not assigning a more specific code that was available from information contained within histology results that they had access to at the time of coding the episode. The clinical coding staff are consistently incorrectly assigning codes for haemorrhoids in the primary diagnosis position when they are an incidental finding or secondary diagnosis. This was the underlying reason for 7 of the 21 (33.33%) primary diagnosis errors. The coding of procedures by the department is of a good standard, in particular the assignment of secondary procedure codes. The clinical coding staff are not consistently applying the national standards below: o The correct code to apply for arthrosis with mention of more than one site is M15 Polyarthrosis 1 o The correct code to assign for a diagnostic statement of a PR Bleed is K92.2 Gastrointestinal haemorrhage, unspecified not K62.5 Haemorrhage of anus and rectum 2. o Following the introduction of ICD-10 4 th Edition in July 2012 the asterisk code of a dagger and asterisk set of paired codes can be assigned as the primary diagnosis code 3. 1.3.2 When looked at by specialty the percentage of correct codes are: Code Type Percent Correct in Percent Correct in Percent Correct in General Surgery General Medicine Trauma & Orthopaedics Primary Diagnosis 81.82% 77.14% 76.67% Secondary Diagnosis 74.63% 78.70% 73.75% Primary Procedure 87.50% 94.29% 89.66% Secondary Procedure 94.52% 91.55% 95.00% 1.3.3 The case notes used in the audit were generally of a poor standard. A more detailed account on the medical records is included in the Wales Audit Office report. The auditors encountered the following issues: The physical case notes are often messy and disorganised, with many temporary folders and loose documents. Documentation of diagnoses within the medical record is very poor. In particular the lack of clinical statements regarding any type of diagnosis is most problematic within the Trauma and Orthopaedic specialty. Many of the medical records within this specialty contain no clear diagnosis of a condition for which the patient is receiving the treatment they are undergoing. The clinical coder is forced to attempt to identify a primary diagnosis by interpreting the operation sheet itself. 1 NCCS ICD-10 4th Edition Reference Manual (2013), pg XIII-9 2 NCCS ICD-10 4th Edition Reference Manual (2013), pg XI-14 3 NCCS ICD-10 4th Edition Reference Manual (2013), pg 15 Dagger and asterisk system Page 4 of 48

Within the Trauma and Orthopaedics specialty the discharge summaries contain no diagnosis of the condition being treated. The Bluespier sheets (used for capturing information in relation to operations within the Trauma and Orthopaedics specialty) already have OPCS-4 codes printed on them when they are received for coding. These codes are selected by clinical staff and are frequently incorrect. Each sheet also has a diagnosis printed on it; however the accuracy of these diagnostic statements is questionable as they often conflict with the body of the medical record. Where typed operation sheets were found in the case notes, they were clear and easy to follow. Histology results can often take some time to be reported and are therefore not available to clinical coding staff at the time of coding. 1.3.4 The above issues were represented across the entire sample of case notes examined, and appeared to be representative of the general condition of the case notes. 1.4 Conclusions 1.4.1 The clinical coding of procedures at Llandough Hospital is generally of a high standard of accuracy. 1.4.2 The clinical coding staff at Llandough Hospital are up to date with their required training and generally demonstrate a sound grasp of national clinical coding rules and standards. However, there are problems with the application of certain specific national standards (see section 1.3.3 above). 1.4.3 The number of errors of omission identified during the audit points to a lack of indepth analysis of the medical record prior to assigning codes. Staff appear to be rushing and demonstrate a lack of care, particularly regarding the assignment of secondary diagnoses. Errors of omission have also been caused by the incorrect assignment of codes recording haemorrhoids as the primary diagnosis when they were incidental findings, and therefore the condition that should have been assigned in the primary position has not been recorded. 1.4.4 The failure to utilise available histology results in order to assign the most specific codes to episodes of care has negatively impacted the accuracy of clinical coding within the department. 1.4.5 The poor standard of documentation across both electronic and physical patient records has been the direct cause of a number of coding errors. In addition, it significantly increases the difficulty of assigning accurate codes to patient episodes for all clinical coding staff. 1.4.6 The presence of OPCS-4 codes, and the lack of certainty surrounding the accuracy of these codes (as well as the uncertainty regarding the accuracy of any diagnosis codes selected by the consultants) on the Bluespier operation sheets causes problems for clinical coding staff when assigning both OPCS-4 and ICD-10 codes. Page 5 of 48

1.4.7 Clinical coding staff are not clarifying issues (whether created by poor documentation or complex clinical issues) with clinicians on a regular basis. 1.4.8 Currently all queries for research regarding anatomy and other clinical queries must be submitted via the Clinical Coding Manager; none of the clinical coding staff have routine access to the internet to undertake such research. This is not regarded as best practice. 1.4.9 Whilst all Llandough Hospital clinical coding staff are fully up to date with the necessary level of core clinical coding training, the lack of staff with the ACC qualification prevents the organisation from being assured that its coding staff are coding to a recognised national standard and makes it impossible to ascribe a base line level of expertise to the clinical coders within the department. 1.4.10 The lack of a regular programme of audits of the work of clinical coding staff makes it impossible for the Clinical Coding Manager to be sure of the level of accuracy of coding being assigned by the department. Coupled with the lack of a PDR process for staff it is extremely difficult for the department to measure and improve the quality of its coded data. 1.4.11 The current structure of the clinical coding department is not supportive of the provision of high quality clinical coded data. The lack of any band 5 supervisory positions or audit trained staff within the department as a whole, and no management staff based at Llandough Hospital; coupled with the lack of an ongoing programme of regular audits does not allow the clinical coding manager to sufficiently review the quality of the clinically coded data being created by the clinical coding staff at Llandough Hospital. 1.5 Recommendations 1.5.1 The Llandough Hospital clinical coding department should endeavour to maintain the good standard of procedure coding accuracy. 1.5.2 All clinical coding staff should continue to maintain their attendance on required training sessions. In addition all clinical coding staff should ensure that they are familiar with all current national standards and that their OPCS-4 and ICD-10 books are annotated appropriately. 1.5.3 Clinical coding staff at Llandough Hospital must ensure that they take adequate time with each episode to fully extract relevant data and assign codes using the full 4-step coding process. In addition, local training sessions reinforcing the importance of accurately capturing all relevant information from the medical record should be arranged as soon as possible. 1.5.4 The Clinical Coding Manager should remind all clinical coding staff of the importance of using the full medical record as a source of information for clinical coding. Histology results in particular often provide detailed diagnostic information that can be used by the clinical coder to assign an accurate code. 1.5.5 An immediate effort should be made to ensure that staff within Llandough Hospital who have responsibility for clinical case notes and the wider medical record are aware of the need for good practice regarding their use. In particular, attention should be drawn the Royal College of Surgeons Standards for Clinical Records. Significant issues with individual case notes should be highlighted using the Page 6 of 48

relevant local incident reporting procedures in order to ensure that attention is drawn to this issue and that possible clinical risks are being highlighted. 1.5.6 It is inappropriate for clinicians without the relevant clinical classifications expertise and understanding of national (UK and Wales) clinical coding standards to identify clinical classification codes within operational, clinical IT systems. The Clinical Coding Manager should liaise with the parties responsible for the creation of the Bluespier sheets regarding the issues of concern they create for the department. 1.5.7 The Clinical Coding Manager should immediately reinforce to all clinical coding staff the importance of clarifying any issues caused by a lack of clarity in the documentation with the responsible consultant, as per national standards. 1.5.8 The Clinical Coding Manager should seek to ensure all members of the clinical coding department staff are given access to the internet as soon as possible, as a basic resource for research regarding anatomy and other clinical queries. 1.5.9 All clinical coding staff should be encouraged and supported to gain ACC status as soon as possible. 1.5.10 The Clinical Coding Manager should implement a programme of regular audits of clinically coded data created by the department as soon as possible. The results of these should be fed back to clinical coding staff as necessary, and as part of a yearly PDR process. 1.5.11 The current departmental management team should immediately investigate the possibility of re-structuring the department to allow the creation of a minimum of 2 supervisory positions. Page 7 of 48

2 Introduction 2.1 The Admitted Patient Care data set (APC ds), and the clinically coded data contained within, is arguably the single most important source of management information in use within NHS Wales. The availability of timely, complete, accurate-coded APC data are an essential pre-requisite for numerous current and emerging decision support processes. 2.2 Welsh LHBs and Velindre NHS Trust are mandated to clinically code the finished consultant episodes (FCEs) for every patient admitted to a Welsh NHS hospital. Organisations are required to accurately code information relating to all diagnoses and procedures relevant to each individual episode of care experienced by a patient. 2.3 Welsh LHBs and Velindre Trust are currently monitored against two national performance measures of clinical coding completeness. These are: 95% of all FCEs are clinically coded within 3 months of the episode end date; 98% of all FCEs are clinically coded for any given rolling 12 month period. 2.4 There are currently no national performance indicators or measures for clinical coding accuracy. 2.5 Clinical coded data are used for a variety of uses and it impacts on a number of areas including: Healthcare planning (including service reconfiguration); Performance management (notably the production of Tier 1 and other Welsh Government performance indicators and measures); Providing the basis of the Risk Adjusted Mortality Index (RAMI), a current WG priority area. Health needs assessment; Evaluation of treatment and outcome analysis; Benchmarking; Chronic disease management (and the linkage of datasets); Provision of information for research; The production of official statistics and ad-hoc requests; Financial costing and resource utilisation mapping; Ad hoc requests (be they Ministerial, AQs, media/public and so on); Identification of at risk populations; Identification of frequency and occurrence of disease; The monitoring of (often high cost) services provided by the Welsh Health Specialised Services Committee (WHSSC); Clinical coding data is central to a range of national information initiatives, such as the annual financial costing process and patient-level costing It is current WG policy for healthcare data to be made more readily available to the general public, media etc. under its transparency agenda. Where clinical coding information is being shared, this will further raise the importance of that data being accurate and the need for the Service to be assured that this is the case. 2.6 It is a therefore a requirement that clinical coded data are accurate, consistent, complete and coded in a timely fashion. Page 8 of 48

2.7 Clinical coding audit is currently the only means by which it is possible to assure the accuracy of clinical coded data. 2.8 As part of its 2013/14 audit programme, the Wales Audit Office (WAO) has decided to audit the quality of clinical coding services across NHS Wales. This programme work will see the WAO reviewing the processes, procedures, resources and executive support associated with Welsh clinical coding service, whilst the NHS Wales Informatics Service (NWIS) Clinical Classifications Team will oversee an audit of clinical coding accuracy (i.e. the assignment of ICD-10 and OPCS-4 classifications codes by Welsh clinical coding staff) across Wales. 2.9 Support from NWIS was requested by WAO, as it is recognised that audits of clinical coding accuracy requires specific clinical coding expertise that is not present within WAO. 2.10 This report outlines the findings and recommendations of the NHS Wales Informatics Service (NWIS) Clinical Classifications Team audit of clinical coding accuracy at Llandough Hospital. The audit was carried out between the 17 th and 21st of February 2014 and was undertaken by two Accredited Clinical Coders from NWIS. 3 Aims 3.1 The aim of this audit was to assess the accuracy of the clinically coded data produced by Llandough hospital by comparing the codes assigned by the clinical coding department against national clinical coding standards. 3.2 This report aims to provide a benchmark that can be used by the clinical coding department within Llandough Hospital and Cardiff and Vale University Local Health Board, to identify areas for improvement within the organisation and aid in the identification and planning of future training needs. Conclusions and recommendations based on areas of both good and poor practice found are provided to achieve this. 3.3 It also aims to evaluate the quality of the source documentation used by the coders and the local policies and procedures used at Llandough Hospital. 4 Objectives 4.1 The objectives for the audit were: To assess the clinical coding data against national clinical coding standards; To identify and report areas of good and bad practice; To review and assess the accuracy of the source documentation used for clinical coding; To assess the level of clinical involvement with the coding department and to what degree this impacts on the coding process and coding accuracy; To make recommendations designed to support future improve in the accuracy of clinically coded data within the hospital; Highlight training issues within the department. Page 9 of 48

5 Background 5.1 Llandough Hospital is one of two hospitals within the Cardiff and Vale ULHB at which clinical coding staff are based, the other being the University Hospital of Wales (UHW) 5.2 Demographics & Staffing 5.2.1 Cardiff and Vale ULHB generated a total of 153,368 Finished Consultant Episodes (FCEs) in the 2012/13 financial year. Of these, Llandough Hospital generated 46,263 FCEs. 5.2.2 Clinical coding staff at Llandough Hospital assign codes to episodes that take place in both acute and community hospital sites within the Health Board. 5.2.3 Cardiff and Vale ULHB achieved 88.2% completeness for clinical coding as of the submission date at the end of March 2014, and 90.1% completeness for the rolling 12 months to March 2014. This is below the target amounts of 95% and 98%. 5.2.4 The Clinical Coding Department is part of the Information Management and Technology Directorate. The management team within the department currently consists of an acting band 6 Clinical Coding Manager based at UHW, with responsibility for all clinical coders in the ULHB. There is currently a vacant post for a band 5 Assistant Clinical Coding Manager. The department has 31 clinical coders 23 based at UHW and 8 based in Llandough. The table below gives a breakdown of the whole Time Equivalents (WTE) by site: Whole Time Equivalents (WTE) by Site Band Llandough UHW Total 3 1.00 9.31 10.31 4 6.22 11.26 17.48 5 - - - 6-1.00 1.00 5.2.5 During the period being audited there was 1 WTE band 5 vacancy within the UHW coding department. 5.2.6 Coders range in experience in the coding department from 22 years to 1 year and 9 months as coders. 5.2.7 The Llandough Hospital has clinical coders based centrally within 2 adjacent open plan offices located inside the main hospital building. 5.2.8 During the period of time examined by this audit the coding department had a backlog of approximately 19,000 FCEs. 5.3 Workloads 5.3.1 Cardiff and Vale ULHB clinical coding staff do not have an expected amount of FCEs that they are expected to code per day or per year as individuals. Instead, the department as a whole is expected to code 600 FCEs per day. This is not covered in the department s Clinical Coding Policy document, nor is the clinical coding staff made aware of this target as part of any Personal Development Review (PDR) process. Page 10 of 48

5.3.2 The clinical coding department aims to achieve 100% completeness for a given month by the 10th working day of the following month. 5.3.3 In order to reduce the outstanding backlog of uncoded episodes within the department, extensive use was made during the period of the audit of third party contracted clinical coding staff. These contractors worked only on historical backlog episodes, with current episodes of coding being undertaken by regular staff within the department. 5.3.4 In addition to their general clinical coding role, members of the department are also responsible for changing any errors in the information held on the Patient Administration System regarding the patient episode that they identify whilst assigning codes. 5.3.5 The table below shows the amount of FCEs each coder produced in the period 1st of April to 31st of December 2013, against the expected workload for individual coders (this has been averaged to 4836 FCEs per year based on the department target of 600 FCEs per day, and then pro-rata reduced to the amount expected in the 9 month period for which productivity figures were available): Hours worked per week Expected FCE per year Apr-Dec 2013 FCE coded: Apr-Dec 2013 30.0 3627 2448 37.5 3627 4506 28.9 3627 3723 33.0 3627 8927 37.5 3627 3916 28.9 3627 2734 37.5 3627 1999 37.5 3627 3657 Total 29016 31910 Notes 5.4 Training 5.4.1 None of the clinical coding staff at Llandough Hospital hold the ACC qualification. 5.4.2 All of the clinical coders meet the minimum training requirements of having completed the Clinical Coding Foundation Training Course and a Clinical Coding Refresher Training Course within the last 3 years. 5.4.3 There is currently no NCS approved Clinical Coding trainer or auditor on site. All the department s training needs are currently met by D&A Consulting; a commercial company supplying clinical coding training who provide all training services to NHS Wales via a national training contract agreed with NWIS. 5.4.4 The department has a detailed induction process for all new staff which is set-out in the departments Clinical Coding Policy document. The policy document itself has not been updated since February 2012. Page 11 of 48

5.4.5 New staff begin in the department as band 3 Trainee Clinical Coders. After 18 months they are able to sit an internal assessment in order to progress to the role of a band 4 Clinical Coder. This internal assessment is created and marked by the clinical coding manager. 5.4.6 All band 4 clinical coders are expected to mentor band 3 trainee clinical coders as necessary. 5.4.7 As well as assigning classifications codes to inpatient episodes, the Clinical Coding Department also assigns codes to outpatient activity (approximately 3020 in the 2013 calendar year). 5.4.8 Workloads at Llandough Hospital are divided primarily by means of clinical specialties, with each member of the clinical coding staff being assigned a specialty they are responsible for. Due to the case-mix of the hospital and the numbers of FCEs involved, some specialties require multiple members of the clinical coding staff assigned to them, whilst other clinical coders may be responsible for several smaller specialties. There is no rota for alternating staff across specialties, and many coders do not rotate to different specialties at all. In addition to this, some staff are also allocated work by other methods as necessary, such as by ward or by consultant. 5.4.9 Clinical coding staff at Llandough Hospital do not currently undergo any PDR reviews. 5.5 Assignment of codes 5.5.1 Coding is carried out using the Medicode clinical coding encoder system from 3M. This interfaces with the Cardiff and Vale UHB Patient Administration System (PAS), PMS. Medicode is installed individually on the computers of clinical coding staff, rather than using a central networked incidence of the system, which limits some of the functionality of the system (such as not being able to centrally amend issues or make changes which then automatically apply across the Medicode of all clinical coding staff, or use Medicode's inbuilt coding validation reports to identify basic errors in assigned codes). 5.5.2 Codes are assigned to episodes using both the ICD-10 4th Edition and OPCS 4.6 classifications. 5.5.3 The main source documentation used at Llandough Hospital is the patient s physical case notes. In addition to this, the clinical coding staff have access to electronic discharge summaries for all three specialties being audited. They also have access to GP letters, test results and scan results through the clinical portal system. 5.5.4 Only some of the clinical coding staff within Cardiff and Vale ULHB have access to the internet, though all have access to the intranet. 5.5.5 Histology results are available to clinical coding staff. However, the time taken by the pathology department to create the histology report often means that the report is not available at the time of the coder assigning codes to the episode. In this circumstance the clinical coding staff are required to keep a record of the patient details and periodically check if the report is available (though there is no formal Page 12 of 48

written process for this). Once it is available, the coder then makes any necessary amendments to the assigned clinical codes to ensure they are accurate. 5.5.6 The clinical coding department at Llandough Hospital employs a single 0.47 WTE clinical coding support staff whose role is to retrieve and return case notes for the clinical coding staff in order to maximise the amount of time they can spend assigning codes. In addition they also identify and locate missing uncoded episodes for the department. Due to the volume of patient episodes however, some clinical coding staff are still required to collect and return case notes. 5.5.7 The retrieval and coding of deceased patient records is prioritised within the department, followed by the retrieval and coding of the Trauma & Orthopaedic and Cardiothoracic specialties. 5.5.8 They do not make use of Read Codes, Clinical Terms or SNOMED-CT. 5.6 Previous Audits and Recommendations 5.6.1 The last external audit was carried out in 2008 by the National Leadership and Innovation Agency for Healthcare (NLIAH). Although a draft report was created, no formal report of the audit was ever presented to the ULHB. 5.6.2 There is no process of regular audits in place at Llandough Hospital although coding checks are conducted in an ad hoc fashion as required. Results from these are fed back to the individual clinical coding staff involved, and any issues identified are e-mailed to all members of the clinical coding staff. 5.6.3 The Clinical Coding Manager does not regularly run any validation reports to identify basic errors in the coded data (see 5.5.1). 5.6.4 A number of clinicians are involved in the validation of their clinically coded data. 6 Methodology 6.1 A pre-audit questionnaire regarding details of the organisation of clinical coding services in the LHB was completed by the Clinical Coding Manager. 6.2 A list of 270 FCEs, drawn from three specialties, was randomly generated from the Patient Episode Database for Wales (PEDW) the national database of APC ds activity data. PEDW is managed and maintained by NWIS. 6.3 The planned number of episodes audited was 30 from each of the 3 specialties below: General Medicine General Surgery Trauma and Orthopaedic. 6.4 The episodes audited were limited to those with an episode end date of 1st may 2013 31st August 2013 inclusive. Episodes were also limited to being no longer than 10 days in length. 6.5 Staff at Llandough Hospital were required to provide the auditors with access to the written case note records associated with the requested FCEs. Page 13 of 48

6.6 The clinical coding record for each episode was generated from Medicode and a copy attached to the relevant set of case notes. 6.7 The auditors then assessed the locally coded data against the National Clinical Coding Standards (see Appendix 1) and the Welsh Clinical Coding Standards (see Appendix 2) using ICD-10 and OPCS 4.6 classifications. 6.8 Codes were audited as one of 4 types: Primary Diagnosis codes (i.e. the main condition treated); Secondary Diagnosis codes (including External Cause Codes and Morphology Codes); Primary Procedure codes; Secondary Procedure codes (including Chapter Z site codes). 6.9 Any errors were assigned to an Error Type (see Appendix 3), which specified the exact nature of the error. This information was then tabulated to calculate the statistical information required (see Appendix 2) 6.10 The errors are of two general types non-coder errors and coder errors. Non-coder errors are those errors identified as being due to a factor external to the individual coder, such as an encoder system which automatically re-sequences codes, or a local coding policy which instructs the coder to assign codes in a way which contravenes national standards. Coder errors are errors in the coding made by the coder themselves. 6.11 For statistical reasons and due to the judgemental nature of a code being relevant to an episode, those error types where coding staff have assigned more codes than the auditor deems relevant (i.e. overcoding ) are not counted as errors when calculating the error percentages. However, the numbers of these errors are reported and examples given for information and training purposes. 6.12 An analysis of the errors is given in Appendix 1. 6.13 The recommended minimum percentage of correct codes are: 90% for Primary Diagnosis and Primary Procedure 80% for Secondary Diagnosis and Secondary Procedures The Accredited Clinical Coding (ACC) exam also stipulates a minimum requirement of 90% accuracy for all clinical coding staff sitting the National Clinical Coding Qualification (NCCQ) exam. Furthermore, the above targets are consistent with the requirements set out in the NHS England Information Governance Toolkit requirement 505 (attainment level 2) and audits of coded data carried out by NCS auditors on English Coders. 6.14 Case notes which did not contain the episode to be audited were marked as Unsafe To Audit (UTA) and removed from the sample and replaced. 6.15 A total of 98 episodes were examined. Page 14 of 48

7 Findings 7.1 The percentages of correctly assigned codes are given below: Code Type Total Number of Total Number of Percentage Correct Codes correct codes Primary Diagnosis 98 77 78.57% Secondary Diagnosis 255 194 76.08% Primary Procedure 96 87 90.63% Secondary Procedure 264 248 93.94% The percentage of codes that were correct was below the recommended level in both primary and secondary diagnosis coding. However, the percentage was above the recommended level in both primary and secondary procedure coding. 7.2 When looked at by specialty, the percentage of correct codes is: Code Type Percent Correct in Percent Correct in Percent Correct in General Surgery General Medicine Trauma & Orthopaedics Primary Diagnosis 81.82% 77.14% 76.67% Secondary Diagnosis 74.63% 78.70% 73.75% Primary Procedure 87.50% 94.29% 89.66% Secondary Procedure 94.52% 91.55% 95.00% It should be noted that of the 98 episodes examined 44 (44.90%) contained no errors in any position. A breakdown of the error types assigned is given below at section 7.4 onwards. 7.3 In addition to the percentages given above a number of specific findings were made by the auditors: The most prevalent type of errors uncovered during the audit are errors of omission. Errors due to codes being omitted account for 63.55% of all errors encountered during the audit (68 out of 103 errors), and in the case of Secondary Diagnosis errors account for 81.97% of all errors in that area (50 out of 61 errors). Clinical coding staff do not appear to be consulting histology reports when assigning codes to an episode. Out of 21 primary diagnosis errors 5 (23.81%) were due to clinical coding staff not assigning a more specific code that was available from information contained within histology results that they had access to at the time of coding the episode. The clinical coding staff are consistently incorrectly assigning codes for haemorrhoids in the primary diagnosis position when they are an incidental finding or secondary diagnosis. This was the underlying reason for 7 of the 21 (33.33%) primary diagnosis errors. The coding of procedures by the department is of a good standard, in particular the assignment of secondary procedure codes. The clinical coding staff are not consistently applying the national standards below: Page 15 of 48

o The correct code to apply for arthrosis with mention of more than one site is M15 Polyarthrosis 4 o The correct code to assign for a diagnostic statement of a PR Bleed is K92.2 Gastrointestinal haemorrhage, unspecified not K62.5 Haemorrhage of anus and rectum 5. o Following the introduction of ICD-10 4 th edition in July 2012 the asterisk code of a dagger and asterisk set of paired codes can be assigned as the primary diagnosis code 6. 7.4 Unsafe to Audit (UTA) There was one episode which was marked as UTA (1.01% of the total number of sets of case notes looked at). As per the methodology described above, they were removed from the audit and replaced. Example: There was no information in the medical notes pertaining to the episode to be audited. 7.5 Primary Diagnosis Codes The primary diagnosis was correct in 78.57% of the episodes audited (77 of the 98 primary diagnoses). These were broken by specialty as follows:- General Surgery primary diagnosis correct - 81.82 % (27 out of the total of 33) General Medicine primary diagnosis correct - 77.14% (27 out of the total of 35) Trauma and Orthopaedic primary diagnosis correct - 76.67% (23 out of the total of 30) A breakdown of the errors in primary diagnoses by their associated error types is given below (see Appendix 3 for a detailed explanation of the error keys): Error Type Number of Errors Percentage of FCEs with Error PD3 7 7.14% PD4 2 2.04% PDIS 3 3.06% PDO 7 7.14% PDD 1 1.02% PDI 1 1.02% 7.5.1 Primary Diagnosis Incorrect at Third Character Level (PD3) There were seven primary diagnosis errors incorrect at third character level (7.14%). Example: ULHB Coding Auditor Coding Z03.8 Observation for other suspected Z08.0 Follow up examination after surgery for diseases and conditions malignant neoplasm Z85.0 Personal history of malignant neoplasm of digestive organs 4 NCCS ICD-10 4th Edition Reference Manual (2013), pg XIII-9 5 NCCS ICD-10 4th Edition Reference Manual (2013), pg XI-14 6 NCCS ICD-10 4th Edition Reference Manual (2013), pg 15 Dagger and asterisk system Page 16 of 48

The information in this patient s medical notes stated that they were attending for follow up examination after previous surgery for a malignant neoplasm. The code assigned by the clinical coder is therefore incorrect as it does not record this. 7.5.2 Primary Diagnosis Incorrect at Fourth Character Level (PD4) There were two primary diagnosis errors incorrect at fourth character level (2.04%). Example: ULHB Coding K80.2 Calculus of gallbladder without cholecystitis Auditor Coding K80.1 Calculus of gallbladder with other cholecystitis The information in the medical record for this episode of care stated that the patient had calculus of the gallbladder with chronic cholecystitis. The clinical coder has chosen the wrong code from the tabular list to assign 7. 7.5.3 Primary Diagnosis Incorrectly Sequenced (PDIS) There were three primary diagnosis errors which were incorrectly sequenced (3.06%). Example: ULHB Coding Auditor Coding M51.1 Lumber and other intervertebral disc G55.1* Nerve root and plexus compression in disorders with radiculopathy intervertebral disc disorders G55.1* Nerve root and plexus compression in M51.1 Lumber and other intervertebral disc intervertebral disc disorders disorders with radiculopathy Z88.0 Personal history of allergy to penicillin Z51.2 Other chemotherapy Z88.0 Personal history of allergy to penicillin The information in this episode of care showed that it was the nerve root compression that was being treated. Therefore it is appropriate to assign the asterisk code for nerve root compression in the primary position, as per the standard introduced in ICD-10 4 th Edition 8. 7.5.4 Primary Diagnosis Omitted (PDO) There were seven primary diagnosis omission errors (7.14%). Example: ULHB Coding I84.9 Unspecified haemorrhoids without complication K57.3 Diverticular disease of large intestine without perforation or abscess Z80.0 Family history of malignant neoplasm of digestive organs Auditor Coding R19.4 Change in bowel habit Z80.0 Family history of malignant neoplasm of digestive organs I84.9 Unspecified haemorrhoids without complication K57.3 Diverticular disease of large intestine without perforation or abscess 7 NCCS ICD-10 4th Edition Reference Manual (2013), pg 5 Individual codes 8 NCCS ICD-10 4th Edition Reference Manual (2013), pg 15 Dagger and asterisk system Page 17 of 48

The clinical information in this episode of care stated that the patient was being admitted for investigations due to a change in bowel habit and a family history of bowel cancer. The haemorrhoids were an incidental finding of the colonoscopy 9. 7.5.5 Information available at the time of audit not available at the time of coding (PDI) There was one primary diagnosis errors due to information being available to the auditors that was not available at the time of coding (1.02%). Example: ULHB Coding R22.4 Localised swelling, mass and lump, lower limb Auditor Coding D17.2 Benign lipomatous neoplasm of skin and subcutaneous tissue of limbs M8850/0 Lipoma NOS The clinical information in this episode of care stated that the patient had a lump on their calf, but during the episode the lump was excised and a diagnosis of a lipoma was made on the histopathology report. The clinical coder has not checked the histology result when it became available in order to amend the assigned codes to reflect the correct diagnosis 10. 7.5.6 Primary Diagnosis Documentation Issue (PDD) There was one primary diagnosis error due to documentation issues (1.02%). Example: ULHB Coding Auditor Coding M89.35 Hypertrophy of bone pelvic region M24.85 Other specified joint derangements, not and thigh elsewhere classified pelvic region and thigh M76.15 Psoas tendinitis, pelvic region and M89.35 Hypertrophy of bone Pelvic region and thigh thigh J45.9 Asthma, unspecified M76.15 Psoas tendinitis, pelvic region and thigh F17.1 Mental and behavioural disorders due J45.9 Asthma, unspecified to use of tobacco Z88.0 Personal history of allergy to penicillin F17.1 Mental and behavioural disorders due to use of tobacco Z88.6 Personal history of allergy to analgesic Z88.0 Personal history of allergy to penicillin agent Z88.6 Personal history of allergy to analgesic agent The information in this patient s medical notes made it was extremely difficult to ascertain the primary diagnosis due to lack of accurate documentation. The primary diagnosis was not overtly stated within the medical record, and the Bluespier operation sheet was organised by anatomical area of the hip, relating separately the parts of the procedure performed on each area. This made identifying the procedure and reason for it particularly difficult. As the patient was referred to as having a Snapping Hip, and it was this that was mainly treated by the procedure undertaken, the primary diagnosis code assigned by the auditor is to record this condition. 9 NCCS ICD-10 4th Edition Reference Manual (2013), pg 9 Primary Diagnosis 10 NCCS ICD-10 4th Edition Reference Manual (2013), pg 9 Specificity Page 18 of 48

7.6 Secondary Diagnosis Codes Including External Cause and Morphology Codes The secondary diagnoses codes were 76.08% correct (194 out of the total 255 secondary diagnoses). These were broken down by specialty as follows: General Surgery secondary diagnosis correct 74.63% (50 out of the total of 67) General Medicine secondary diagnosis correct 78.70% (85 out of the total of 108) Trauma and Orthopaedic secondary diagnosis correct 73.75% (59 out of the total of 80) A breakdown of the errors by their associated error types is given below (see Appendix 3 for detailed explanation of error keys): Error Key Number of Errors Percentage of Secondary Diag with Error SD3 3 1.18% SD4 5 1.96% SDO 47 18.43% SDNR 7 - MCI 3 1.18% MCO 3 1.18% 7.6.1 Secondary Diagnosis Incorrect at Third Character Level (SD3) There were three secondary diagnoses incorrect at third character level (1.18%). Example: ULHB Coding Auditor Coding D37.4 Neoplasm of uncertain or unknown behaviour of oral cavity and digestive organs D37.4 Neoplasm of uncertain or unknown behaviour of oral cavity and digestive organs M8261/1 Villous adenoma NOS M8261/1 Villous adenoma NOS K62.5 Haemorrhage of anus and rectum K92.2 Gastrointestinal haemorrhage, unspecified The information in this patient s notes stated that they had haematochezia which is bleeding per rectum (PR) not a haemorrhage of the rectum itself 11. 7.6.2 Secondary Diagnosis Incorrect at Fourth Character Level (SD4) There were five secondary diagnoses incorrect at fourth character level (1.96%). 11 NCCS ICD-10 4th Edition Reference Manual (2013), pg XI-14 Page 19 of 48

Example: ULHB Coding Auditor Coding K80.2 Calculus of gallbladder without K80.2 Calculus of gallbladder without cholecystitis cholecystitis K83.4 Spasm of sphincter of oddi K82.4 Cholesterolosis of gallbladder The information in this patient s notes stated that they were suffering from Cholesterolosis of the gallbladder not a spasm of the sphincter of oddi. Therefore the correct code to assign would be K82.4. 7.6.3 Secondary Diagnosis Not Relevant (SDNR) There were seven secondary diagnoses assigned which were not relevant. Example: ULHB Coding Auditor Coding K56.2 Volvulus K56.2 Volvulus T88.4 Failed or difficult intubation Z88.8 Personal history of allergy to other drugs, medicaments and biological substances There was no indication of failed or difficult intubation in the medical record for this episode of care. The information in the record stated that there was difficulty introducing the colonoscope during the procedure undertaken, however the scope is not a tube and it is not therefore appropriate to use the code T88.4. 7.6.4 Secondary Diagnosis Omitted (SDO) There were forty seven secondary diagnoses omitted (18.43%). Example: ULHB Coding Auditor Coding M51.1 Lumber and other intervertebral disc G55.1* Nerve root and plexus compressions in disorders with radiculopathy intervertebral disc disorders (M50-M51 dagger) G55.1* Nerve root and plexus compressions M51.1 Lumber and other intervertebral disc in intervertebral disc disorders (M50-M51 disorders with radiculopathy dagger) Z88.0 Personal history of allergy to penicillin Z51.2 Other chemotherapy Z88.0 Personal history of allergy to penicillin The information in this episode of care stated that the patient was being admitted for a sacral epidural injection it is therefore required to assign the code Z51.2 in a secondary position 12. 12 NCCS ICD-10 4th Edition Reference Manual (2013), pg XXI-16 Page 20 of 48

7.6.5 Morphology Code Incorrect (MCI) There were three incorrect morphology codes assigned (1.18%). Example: ULHB Coding Auditor Coding D12.5 Benign neoplasm of colon, rectum, anus and anal canal Sigmoid colon D12.5 Benign neoplasm of colon, rectum, anus and anal canal Sigmoid colon M8140/0 Adenoma NOS M8210/0 Adenomatous polyp D12.8 Benign neoplasm of colon, rectum, anus and anal canal D12.8 Benign neoplasm of colon, rectum, anus and anal canal M8140/0 Adenoma NOS M8210/0 Adenomatous polyp The clinical information in the medical record for this episode of care stated that the lesions were adenomatous polyps; however the clinical coder has assigned a code for an adenoma of a type Not Otherwise Stated. 7.6.6 Morphology Code Omitted (MCO) There were three morphology codes omitted (1.18%). Example: ULHB Coding R22.4 Localised swelling, mass and lump, lower limb Auditor Coding D17.2 Benign lipomatous neoplasm of skin and subcutaneous tissue of limbs M8850/0 Lipoma NOS The clinical information in this episode of care stated that the patient had a lump on their calf, but during the episode the lump was excised and a diagnosis of lipoma was made on the subsequent histopathology report. The clinical coder has not checked the histology result when it became available in order to amend the assigned codes to reflect the correct diagnosis 13. 7.7 Primary Procedure Codes There were 96 primary procedure codes assigned. The primary procedure was correct in 90.63% of the episodes audited (87 of the 96 primary procedures). These were broken down by specialty as follows: General Surgery primary procedures correct 87.50% (28 out of the total of 32) General Medicine primary procedures correct 94.29% (33 out of the total of 35) Trauma and Orthopaedic primary procedures correct 89.66% (26 out of the total of 29) A breakdown of the errors by their associated error types are shown below (see Appendix 3 for detailed explanation of the error keys): 13 Welsh Standard: Icd-10 Classification Neoplasm Morphology Coding - http://nww.classificationstandards.wales.nhs.uk/ Page 21 of 48

Error Key Number of Errors Percentage of Primary Procedures with Error PP3 2 2.08% PP4 4 4.17% PPO 2 2.08% PPD 1 1.04% 7.7.1 Primary Procedure Incorrect at Third Character Level (PP3) There were two primary procedure codes incorrect at third character level (2.08%). Example: ULHB Coding Auditor Coding W20.1 Primary open reduction of fracture of W23.1 Secondary open reduction of fracture of long bone and extramedullary fixation using bone and intramedullary fixation HFQ plate NEC W31.4 Cancellous chip autograft of bone Y66.3 Harvest of bone from iliac crest W31.9 Other autograft of bone unspecified Z71.8 Specified ulna NEC Z71.8 Specified ulna NEC Z94.2 Right sided operation Z94.2 Right sided operation Y80.4 Intravenous anaesthetic NEC Y66.3 Harvest of bone from iliac crest Z94.2 Right sided operation Y71.3 Revisional operations NOC Y80.4 Intravenous anaesthetic NEC The information in the medical record for this episode of care stated that this was revisional / secondary reduction of the fracture and intramedullary fixation together with a bone graft. Secondary open reductions of fractures are assigned codes from W23. The codes W23.1 and W23.2 have the abbreviation HFQ (However Further Qualified) in the descriptor, meaning that no additional detail in the operation documentation will require code assignment. Due to this the code for a graft of bone to the ulna is not required in addition 14. 7.7.2 Primary Procedure Incorrect at Fourth Character Level (PP4) There were four primary procedure codes incorrect at fourth character level (4.17%). Example: LHB Coding Auditor Coding H22.9 Diagnostic endoscopic examination of H22.1 Diagnostic fibreoptic examination of colon colon unspecified and biopsy of lesion of colon Y84.8 Other anaesthetic, other specified Z28.2 Caecum The clinical information in this episode of care stated that the patient had a biopsy of the caecum in addition to the examination. The clinical coder has not assigned the correct 4 th character to record this. 14 Clinical Coding Instruction Manual OPCS-4.6, pg 30 Page 22 of 48

7.7.3 Primary Procedure Omitted (PPO) There were two primary procedures omitted (2.08%). Example: ULHB Coding Auditor Coding H51.3 Stapled haemorrhoidectomy P23.7 Posterior colporrhaphy with mesh reinforcement Y84.2 Sedation NEC H51.3 Stapled haemorrhoidectomy Y84.2 Sedation NEC The information in this episode of care stated that the patient had both a rectocele and haemorrhoids so a repair of the rectocele was carried out as well as the haemorrhoidectomy. The repair of the rectocele is a more significant procedure than the haemorrhoidectomy; therefore it should be assigned as the primary procedure. 7.7.4 Primary Procedure Documentation Issue (PPD) There was one primary procedure code error due to documentation issues (1.04%). Example: ULHB Coding Auditor Coding W89.1 Endoscopic chondroplasty T83.2 Division of muscle Z90.2 Hip NEC Z57.8 Specified muscle of hip or thigh W83.3 Endoscopic shaving of auricular W89.11 Endoscopic chondroplasty cartilage Z90.2 Hip NEC Z84.3 Hip joint Z94.2 Right sided operation W08.5 Partial excision of bone Y80.3 Inhalation anaesthetic NEC Z76.9 Femur NEC W83.8 Therapeutic endoscopic operations on other articular cartilage Y05.5 Debridement of organ NOC Z84.3 Hip joint Z94.2 Right sided operation Y80.3 Inhalation anaesthetic NEC The Bluespier operation sheet for this episode was organised by anatomical area of the hip, relating separately the parts of the procedure performed on each area. This made identifying the overall procedure and reason for it particularly difficult. As the patient was referred to as having a Snapping Hip, and it was this that was mainly treated by the procedure undertaken, the primary diagnosis code assigned by the auditor is to record this condition. Therefore the code assigned as the primary procedure by the auditor was the one which represented the treatment of the Snapping Hip in this case the division of the patient s muscle. 7.8 Secondary Procedure Codes There were 264 secondary procedures codes assigned. These secondary procedure codes were 93.94% correct (248 out of the 264 secondary procedures). They were broken down by specialty as follows: General Surgery secondary procedure correct 94.52% (69 out of the total of 73) Page 23 of 48