Sir,

Clinical coding is a process by which descriptions of diseases, injuries or procedures are assigned a numeric or alphanumeric designation. Coding provides a mechanism for standardizing the recording of information and therefore, if accurate, is a valuable tool for audit, epidemiological studies, healthcare planning, and resource allocation.

In recent years clinical coding has become increasingly important and there are at least two main reasons for this. First, coded data is now used to calculate surgical complication statistics, which may be used to compare the performance of individual units or surgeons. If there are errors in clinical coding the reported complication rates will be inaccurate. Jain et al1 examined 80 consecutive cataract extractions that had been coded by clinical coders as having had a surgical complication. Fifty percent of the patients were found to not actually have experienced a complication at all. Of the 40 patients who had experienced a surgical complication only 15 were coded accurately.1

The second reason for the increased focus on accurate clinical coding is the Payment by Results reimbursement scheme. In the past National Health Service (NHS) Trusts were paid according to block contracts or negotiated cost and volume contracts. Under Payment by Results, NHS Trust reimbursement is dependent on the actual patient-level clinical activity undertaken and accurate reimbursement relies on an accurate record of each patient's hospital episode.

In most institutions clinical coding is carried out by non-medically trained clinical-coding staff. Several earlier studies have indicated poor reproducibility of clinical coding.2, 3, 4 We decided to investigate the coding of surgical procedures in the ophthalmology theatre at our institution with the hope that we might identify ways to improve the accuracy of clinical coding.

For a two-week period surgeons in the ophthalmology theatre at Lincoln County Hospital were asked to assign codes to each surgical procedure they carried out. A coding audit sheet was compiled that contained the 190 most common ophthalmology codes based on the Office of Population Census and Surveys (OPCS) 4.4 coding system (Figure 1). Surgeons were asked to tick as many boxes as they felt were relevant to fully describe the surgical procedure undertaken. Coding was also performed, as normal in our institution, by clinical coders who were not informed that an audit was taking place. The total number of procedures carried out during this period was ascertained from the theatre logbook and iPath Operating Room Management Information System (ORMIS). When there was disagreement between the codes assigned by the coders and surgeons the case notes were retrieved and analysed.

Figure 1
figure 1

Coding Audit Sheet for use in theatres by the surgeon performing the procedure.

Full size image

During the two-week period, 120 patients had surgical procedures carried out, including 87 cataract extractions, six strabismus procedures, four blepharoplasties, two trabeculectomies, two penetrating keratoplasties, and 19 other procedures ranging from an intraocular lens exchange to an eyelid laceration repair. Overall, the quality of coding provided by the coding department was excellent. The clinical coders identified 117 (97.5%) of these patients compared with the surgeons, who recorded details for 110 (91.8%) of the patients (Figure 2). The clinical coders made 2 (2/117, 1.7%) major coding errors (code for wrong procedure). The errors included coding a trabeculectomy as a trabeculotomy and coding a planned extracapsular cataract extraction as a phacoemulsification procedure. In addition, the clinical coders made nine minor coding errors that mainly related to miscoding the type of anaesthetic used. The surgeons made no major or minor coding errors. The codes provided by the surgeons were also more detailed than those from the clinical coders. The coders used a median of four codes for each procedure (range 2–4 codes) compared with the surgeons, who used a median of eight codes for each procedure (range 3–12 codes). The surgeons commented that there were no adequate codes for some procedures; for example, there was no code for trypan blue ophthalmic solution or capsular tension rings.

Figure 2
figure 2

Venn diagram showing the number of patients recorded from each source.

Full size image

We have found there is less uncoded clinical activity when coding is carried out in the clinical coding department than when the operating surgeon codes procedures directly in the theatre; however, input from surgeons can improve the accuracy of coded information. Uncoded clinical activity is not reimbursed and therefore carries large financial implications. For example, the elective tariff for a phacoemulsification cataract extraction and lens implant was £720 in 2006. In our audit, the estimated cost of the ten procedures missed by the surgeons would be £6898 for just this 2-week period.

For the patients who were coded, the surgeons provided more accurate and detailed codes. A possible explanation for this is the different methodology used for the two study groups. We purposefully did not inform the coders that the audit was taking place as we felt that this would not provide an accurate reflection of the coding practice in our institution. By contrast, the surgeons were aware that an audit was taking place. Despite the extra detail provided by the surgeons, in most cases this would not alter the reimbursement. The primary reason for this is that the tariffs used in Payment by Results do not always reflect the complexity of the procedures; however, detailed coding is likely to become more important as the tariffs become further individualized. The Department of Health has stated that ‘the more detail that is captured about the patient's treatment the greater potential for Payment by Results to differentiate between routine and more complex cases and achieve fairer reimbursement’.5

The quality of clinical coding may be improved by increasing the awareness of clinical coding among medical staff. Simple steps such as ensuring that entries in the notes are legible and without abbreviation will make coding easier. Other measures might include increasing the collaboration between medical staff and coders to develop more comprehensive coding systems and by holding regular clinical coding audit. The Audit Commission has set out its intention to regularly audit clinical coding. A pilot audit of 17 organizations in 2006 showed high levels of coding errors. In one trust, over 25% of the primary procedure codes were incorrect. The Audit Commission has recommended that clinical coding be given a high priority within the NHS and has called for more investment in training clinical coders.

In conclusion, clinical coding is a valuable tool, but to be so it needs to be reproducible and accurate. Although clinical coders usually carry out coding, all healthcare professionals have a responsibility to ensure that coding is as accurate as possible.