The ‘top 10’ urological procedures: a study of hospital episodes statistics 1998–99
Objective To summarize the urological procedures recorded in Hospital Episodes Statistics (HES, 1998–99) and to examine the accuracy of coding and HES.
Materials and methods Data on the 10 commonest urological procedures from the Department of Health website ( www.doh.gov.uk/hes ) were extracted, summarized and presented.
Results Urethral catheterization, endoscopic procedures on the bladder, prostate, urethra and ureter (excluding ureteric stone extraction), minor open procedures on the foreskin and the vas, bladder instillation, extracorporeal shockwave lithotripsy and prostatic biopsy are the 10 commonest procedures, according to finished consultant episodes. There is published evidence that the data from coding and HES are not completely accurate.
Conclusion This study highlights shortfalls in the HES data (1998–99) which may directly affect the funding of urological services.
Hospital Episodes Statistics (HES) is a database containing medical and administrative details of all patients admitted and treated within NHS hospitals in England since 1989. It is constructed from data collected locally by the NHS via Patient Administration Systems or Hospital Information Systems. Diagnoses are coded using the International Classification of Diseases (10th Revision) (ICD-10) and operations are coded using the Office of Population Censuses and Surveys Surgical Operations and Procedures, Fourth Edition, Consolidated Version 1990 (OPCS-4). Information from HES is used by the Government, the NHS, and other organizations and individuals with an interest in health and healthcare information. Decisions on healthcare policies are based on HES data, and thus accuracy is vital. A major influence upon this data is the standard of clinical coding.
The first classification of surgical operations in Britain was published by the MRC in 1944 ; a three-character code was used with punch-card equipment. In 1950, the General Register Office developed a new system, incorporating the MRC system with the classification used by the USA Armed Forces . A second revision was published in 1969 with an optional fourth digit extension . The third revision was introduced in 1975 to provide comparability with the surgical operations section of the WHO International Classification of Procedures in Medicine .
With the rapid expansion and increased diversity of surgery, any coding system is at risk of becoming less than optimal. In 1982, The Steering Group on Health Services recommended that as a matter of urgency, OPCS should provide operation codes which reflected current clinical practice and develop procedures for the frequent updating of the classification . It was decided to issue a revised edition of the OPCS classification to provide the best possible basis for accommodating existing systems of coding, incorporating procedures not requiring the full operating-theatre environment and future developments in data, surgical innovations and operations. This process began in 1983 with the birth of the current revision (OPCS-4) in 1987.
OPCS-4 provides a definition of an operative procedure and outlines the concept of main operations during an episode of care. This system consists of 23 anatomically based chapters, related to body systems (part or whole) and is designated alphabetically. The alphabetic character for each chapter forms the leading digit of the 3- and 4-digit codes used. Within each organ section, the procedures listed are presented in descending order of complexity. For many purposes, it is acceptable to group codes at the 3-digit level. Where more precision is required, it is necessary to use the subdivision indicated by the final digit. As a result, in urology, ESWL was included. This classification is used in conjunction with other available classifications such as the ICD and methods of classification of urgency of admission (emergency or elective).
Several groups use the HES data; the Treasury uses it for the Public Expenditure Survey model for negotiating the Health Service budget; HES information also supports work on health monitoring, the impact of diseases and environmental conditions. In the NHS, it is used to provide measures of performance, i.e. (i) high-level performance indicators; (ii) direct outcome measures or clinical indicators of hospital care (e.g. readmission rates); (iii) National Data Quality Indicator; and (iv) Public Health Outcome Indicators. These examples form part of the Performance Assessment Framework. HES also allows an analysis of specific conditions, operations performed and bed-occupancy rates from the clinical and administrative information. Thus, the present study was undertaken to determine how useful HES data is for planning and monitoring urological services.
Materials and methods
To achieve the aim, the urologically related dataset was extracted from the HES tables (no. 5, main operations 3-digit OPCS codes; no. 19, main operations 4-digit OPCS codes) via the Department of Health website (www.doh.gov.uk/hes). We concentrated on the 10 commonest urological procedures performed (the ‘top 10’). These were presented under the headings of Finished Consultant Episodes (FCEs), number of admissions (emergency, waiting list and day case), the mean and median waiting time, mean length of stay and the number of bed days. The data reviewed at the time of this study had not been adjusted to account for shortfalls in the number of records submitted and for missing or invalid clinical information (i.e. ‘ungrossed’). Grossed data are available up to and including 1995–96.
The terms used were defined as follows: (a) FCEs; the number of HES records submitted on behalf of NHS hospital providers that relate to episodes of inpatient care ended during the 1998–99 financial year. Each period of care under the same consultant is known as an episode. If the care of an inpatient is transferred to another consultant, a separate HES record is constructed. There will be two or more HES records relating to the patients' stay in hospital. The total for the FCE is therefore higher than that for admissions. (b) Admissions; the number of episodes that were the first in the spell of inpatient treatment (record with an episode order of 1). This includes patients who were admitted in the previous years. For some HES admission analyses, such episodes are omitted, and this can produce a slightly lower value that is more relevant to analyses that concentrate on hospital activity falling wholly within a particular year. (c) Emergency admissions; the number of first episodes with an admission method indicating the admission was an emergency. (d) Waiting list admissions; the number of first episodes with an elective admission method indicating that the admission was from a waiting list. (e) Mean waiting time; the mean waiting time in days for admissions from the waiting list. Waiting time in HES is the period between the date of the decision to admit and the date of eventual admission. Days of deferment and suspension are not taken into account. The waiting time statistics from HES are not compatible with the official waiting-list values because the latter provide an indication of the numbers waiting to be admitted on a particular date, and how long they have been waiting up to that date. The mean waiting time does not necessarily give a true picture of the service provided for most patients because there are often very few patients with very long waiting times. The median waiting time is therefore also quoted. (f) Mean length of stay; the mean number of days for a continuous period of treatment within a single hospital provider. (g) Day case; the number of episodes relating to patients who are admitted electively from a waiting list, or as planned admissions, and are treated during the course of that day. (h) Bed days; the sum of all the days that patients' in the group occupied hospital beds during the HES year. To obtain a better estimate of the true total, the occupied bed-days for FCEs during 1998–99 has been increased by a factor to compensate for the days contributed by episodes that had not finished on the last day of the HES year.
The ‘top 10’ urological procedures according to FCEs, number of admissions, emergency admissions, waiting list admissions and day-case admissions, mean and median waiting time, mean length of stay and the number of bed days are summarized in Table 1.
Table 1. A summary of the HES on urological procedures in England (1998–99); the superscript numbers refer to further information in the footnotes. The number in brackets corresponds to the number of episodes, except for waiting time, length of stay and bed days, when it denotes the number of days. The procedures for each code are listed after the table
FCE and admissions
Diagnostic cystoscopy was the most frequent procedure, with more than half of the episodes as day cases. Unfortunately, the data available did not allow further analyses of the type of cystoscope (rigid/flexible) or anaesthesia (general/sedation/local) used. The other most frequent procedures are urethral catheterization, endoscopic extirpation of a bladder lesion, endoscopic resection of male bladder outlet, vasectomy, foreskin procedures, bladder instillation, endoscopic urethral procedures, ESWL, other endoscopic operations on the ureter (e.g. endoscopic removal of ureteric calculus, which has a different procedure code) and prostatic biopsy. The number of FCEs for therapeutic nephroscopic operations on the ureter, therapeutic ureteroscopic operations on the ureter and other endoscopic removal of calculus from ureters were 282, 2740, 1157, respectively, compared with 12 144 FCEs of ‘other’ therapeutic endoscopic operations on the ureter. No further information was available from the tables or Department of Health website as to the nature of these procedures.
Emergency admission episodes
Urethral catheterization was the commonest procedure involved in emergency admissions in urology. It is a distinct entity as it is not an operation. However, it appeared repeatedly in the ‘top 10’ lists in FCEs, admissions, emergency admissions, waiting list admissions, and the number of bed days. Next came diagnostic endoscopy of the bladder, endoscopic resection of male bladder outlet and percutaneous needle nephrostomy. Within this last category 57% (1959 of 3418 episodes) were needle biopsy of the kidney. Drainage of the kidney only accounted for 40% of cases. The rest of the list consisted of ‘other’ ureteroscopic procedures, open drainage of bladder, testicular operations (most were coded as fixation of testis, reduction of torsion and exploration), endoscopic extirpation of bladder lesion, scrotal operation (mainly coded as ‘exploration’) and other operations on bladder (mainly suprapubic catheterization).
Waiting list admissions
In this group, cystoscopy again came first, followed by vasectomy, endoscopic male bladder outlet resection, foreskin procedures, urethral catheterization, endoscopic extirpation of bladder lesion, other operations on bladder, e.g. closure of cystostomy, change/removal of suprapubic catheter, introduction of therapeutic agents (most), ESWL, endoscopic urethral operations and other procedures on the bladder outlet, mainly prostatic biopsy.
Epididymal procedures did not feature in the ‘top 10’ list when the median waiting time was used. However, procedures involving penile prosthesis, female bladder outlet support, reversal of vasectomy, urethral repair, reconstructive penile and bladder operations persisted. The maximum mean waiting time is 217 days for penile prosthesis while the maximum median waiting time is 169 days for abdominal operations to support the female bladder outlet. None of these is indicated for life-threatening conditions or cancer. However, all of them are used to improve the patients' quality of life.
The procedures in the day-case list included cystoscopy, vasectomy, urethral catheterization, foreskin procedures, bladder instillation, ESWL, prostatic biopsy, endoscopic extirpation of bladder lesion, endoscopic urethral procedures and ‘other’ endoscopic operations on ureter.
The 10 commonest urological procedures are thus urethral catheterization, endoscopic procedures on the bladder, prostate, urethra and ureter (excluding ureteroscopic extraction of calculus), minor open procedures on the foreskin and the vas, bladder instillation, ESWL and prostatic biopsy. Except for urethral catheterization, endoscopic extirpation of bladder lesions and endoscopic procedures on the prostate, urethra and ureter, most were day-case episodes. Among these, flexible cystoscopy, vasectomy, ESWL and bladder instillation can be undertaken as ‘office’ procedures.
Urethral catheterization featured frequently in the ‘top 10’ lists, being the procedure consuming the most bed days. This may be because 43% of the patients admitted with problems involving urethral catheters were >75 years old (29 623 of 69 600 episodes), and under urological care. Those who were not fit for operative intervention were managed by long-term catheterization. They then underwent prolonged hospital stay because of unfavourable social circumstances and/or multiple coexisting medical problems, over which urologists have little influence or control.
The present study assessed HES data from NHS trusts in England and not Scotland, Wales and Northern Ireland. Data for Scotland are available at http://www.show.scot.nhs.uk/isd/, but statistics for Northern Ireland and Wales are only available on special request (contact details via www.doh.gov.uk/hes). The present study sought to uncover the inaccuracies present within the HES dataset for 1998–99 (England), and thus highlight a major problem with the data. A comparative study assessing all available years in England and all home nations would be interesting as a future study, which we are planning.
When using the HES dataset, several limitations exist which must be acknowledged, as discussed below.
Difficulties with comparing yearly trends
Changes in NHS organization, e.g. alterations in the number and boundaries of Health Authorities, and methods used to record inpatient information, do not allow the easy evaluation of yearly trends. Comparing diagnoses before and after April 1995 (when ICD-10 was introduced) is not straightforward because ICD-10 has an entirely different coding structure compared with ICD-9.
Although operations and diagnosis are recorded in HES, there is little information on the effectiveness of treatment. Furthermore, HES does not provide information on drugs used in hospitals or outpatient activities (e.g. prostatic biopsy and urodynamics). However, in return for this potential enhancement in data, the additional costs and administrative burden involved in collecting an extra 80 million outpatient records, in addition to the 12 million inpatient records already collected annually, is enormous.
Accuracy of data
The ultimate reliability of any medical database depends on how accurately the data are collected, recorded and coded. Accurate coding is the only means of ensuring that statistical information drawn from clinical data truly reflects the patterns and activities of the individual clinician, and his/her unit and the hospital, upon which the workload and performance will be measured and future budgets allocated. This applies to both NHS and the private healthcare system.
Before being incorporated into the main HES database, all data submissions are verified, validated and inappropriate entries overwritten, to ensure its accuracy. In addition, comparison with the Korner aggregate return (KP70) checks the extent to which HES fulfils its aim of capturing all FCEs. This is completely independent of HES and provides annual baseline totals of FCEs, which can be assessed against HES. The grossing process involves a comparison of the KP70 and HES figures by consultant speciality. An adjustment is then applied selectively to compensate for missing or invalid diagnoses, to ensure consistency and comparability across the whole database.
Data collection inaccuracy
As HES data ultimately comes from coding, we assessed the accuracy of coding. We identified several studies which addressed this issue; 10% of discharge summaries in a vascular unit over an 8-month period contained errors, with 22% of these involving the diagnosis . Mistakes were found in nearly a third of all the diagnostic codes in general surgery over the same period . In plastic surgery, there was an overall error of 22% in the procedure codes over a 2-month period . A comparison of codes ascribed by internal and external coders for main diagnosis and procedure over a period of 2 years in two NHS hospitals revealed an exact agreement in only 43% of cases . In urology, Ballaro et al. showed errors in over a third of urological FCEs from one unit and concluded that the primary cause was inaccurate coding. When the completeness of HES data on private patients treated in NHS hospitals was compared against information obtained directly from private patient unit managers, there was a difference of 13% for inpatient admissions and 21% for day-case admissions, despite the existing data verification and grossing process . We were unable to identify any study which concluded that coding or HES accuracy was satisfactory, thus fuelling the concern over the reliability of these data.
Improving the HES data
The accurate assignment of diagnosis and procedure codes relies on an understanding of medical terminology, adherence to the principles of OPCS-4 (procedures) and ICD-10 (diagnoses), and the development of coding skills through training and experience. The Australian health department in 1996 introduced a medical coding accreditation examination. Those who were most likely to pass were: those who worked in hospitals with >110 beds; spent 25 h each week on coding or related activities; and had been coding for 3 years . The introduction of accreditation in the UK may provide an incentive and the means to improve the quality of coding and the professional pride of the staff, provided that proper training, support and pay are integrated.
Importantly, encoding software now exists which is reported to improve data quality  and improve accuracy . As expected, limited training is required to use the software.
The health professional team has a collective responsibility to ensure the accuracy of case notes and discharge summaries. Involvement of senior members of the clinical team in assigning procedure codes, e.g. in the operating notes in operating theatres, and completing the diagnoses and codes in discharge summaries, will improve the accuracy of the data. This was confirmed by a prospective audit of discharge summary error, where the seniority of the doctor was inversely proportional to the errors made (2% of consultant discharge summaries contained errors, compared with 7% for the senior registrars, 10% for registrars and 17% for senior house officers) .
There is currently no HES data ‘cleaning’ or quality control. Despite the resources required, regional or national audit of HES examining the accuracy of its data is clearly essential, given that inferences are drawn from such data with important implications in healthcare planning and resource allocation.
To maintain and monitor a database like HES is not easy; it requires resources in manpower and financial capital. For countries with a less mature healthcare monitoring system, involvement of experienced clinicians in data collecting and coding will be an important first step towards a cost-effective system. Important data to be collected should include diagnoses, nature of admissions (emergency or elective), procedures and operations performed, complications, and length of stay. There should be active participation in auditing the accuracy of such data, at both local and national levels.
In conclusion, the range and depth of HES data is vast. Endoscopic procedures, minor operations on the male external genitalia, bladder instillation and ESWL are the commonest procedures undertaken by urologists. The HES data are important, especially under the current climate of clinical governance and ‘performance-related’ budget allocation. However, decision-makers must be made aware that the reliability and the conclusions drawn from such data are far from perfect. Without the correct data input, the financial outcome for urology and other specialities will struggle to match the reality of current needs. As urologists, it is to our professional and patients' advantage to take steps to co-ordinate, improve and monitor the accuracy of coding and HES data.
H.R.H. Patel, Institute of Urology and Nephrology, University College London Medical School, 48 Riding House Street, London, W1W 7EY, UK.