• patient safety, case record review
• adverse events

In their paper in this issue of QSHC in which they redesign current case (or medical) record review methods for the purpose of detecting adverse events and teasing out opportunities for preventing recurrence of this patient harm, Woloshynowych et al¹ state that “in our view the full potential of retrospective record review has yet to be explored”. Their view on the so far unrealised potential of the medical record is true, despite current efforts in some countries.

The medical record has a time honoured place as a chronological account of the clinical state and care of a particular patient. This place has been reinforced by the routine use of the medical record as part of any legal or insurance inquiry about care, as well as in clinical or peer review processes that are designed to improve the delivery of health care. But, unlike an incident report or computer alert, the medical record has a passive nature that requires and rewards systematic examination after an alerting process identifies the records that need review. That alert may be an adverse clinical outcome—for example, death of the patient—or of a procedural form such as a complaint or subpoena. For effective use of the medical record to improve health care, two elements are therefore needed—an alerting system to indicate records worthy of detailed examination and a systematic approach to that examination. The paper by Woloshynowych et al provides new insights into such a systematic approach to medical record review. It builds on the previous work that started with the Californian Insurance Feasibility Study² to estimate the incidence of adverse events and adds questions from experience with clinical investigation of adverse events to better understand potential causes and contributory factors. This is a major step forward in using the medical record for improvement of health services, by understanding causation then generating recommendations for actions to prevent recurrence or alleviate effects of adverse events.

But what about the commonly cited shortcomings of the medical record?³ Availability and legibility of the record and reliability of the judgements made by reviewers are key areas for comment. The Quality in Australian Health Care Study⁴ reported that in only 24 out of 15 000 records from 28 Australian hospitals was there insufficient information in the record to confirm a suspected adverse event, and that there was 80% agreement (χ = 0.55) between two senior medical reviewers as to the presence or absence of an adverse event in the remaining records. From similar work in New Zealand, the reliability of determination of adverse events was 87.5% agreement (χ = 0.47).⁵ These findings indicate that case record review performs better than any other currently available modality for detecting the frequency of adverse events to patients. But it is also clear that performance could be better.

Given the fundamental importance of successful case record review and its progression towards automation with the electronic health record, the patchy success of case record policy is both surprising and disappointing. When this is combined with relative lack of investment in the content and accessibility of these records, it is not surprising that records do not enjoy the reputation that this unique data source should have. The following list of essential elements comes from the reviewers in the Quality in Australian Health Care Study⁶:

• Medical admission.

• Medical discharge.

• Discharge summary.

• Medical continuation notes.

• Management or treatment plans.

• Consultations.

• Results of investigations.

• Short stay documentation.

• Referral/follow up letters from GPs or specialists.

• All volumes of the medical record simultaneously accessible.

They further recommended:

• Standardise the layout of the medical record

• Use admission dividers as a standard requirement for medical record filing.

• Devise a policy for loose sheet filing.

• Use an integrated medical record for all healthcare professionals.

• Keep all records relating to a patient in a single medical record rather than divided between clinics or sites of care.

• Devise standardised systems for linking mother and baby records after discharge.

If we could add the suggestions from Woloshynowych et al on detecting and preventing adverse events, then we could further improve the quality and usefulness of medical records. Even allowing for the fact that these latter recommendations need further formal evaluation and probable refinement, what is really lacking is the leadership to improve the usefulness of medical records for improvement. The incentives for many of the contributors to the medical record to improve their contribution consistent with particular standards are few and the barriers may seem large. If we then automate a current “faulty” system of medical records into an “electronic health record” we can only expensively perpetuate the problem.

We are now at a crucial time where more public discussion on the role and content of the medical record is required, and it is not clear where the leadership for this work is located. Once agreed, then appropriate standards can be internationally agreed and local incentives for implementation can be elucidated. There are current standards in this area, but it is not clear whether they take into account all the necessary functions of the record, especially the role of the record in quality improvement. Even when standards are in place—for example, through accreditation in the US or Australia—it is not obvious that these standards are being followed. Perhaps the readers of this journal could help this discussion gain momentum by using the rapid response function http://qhc.bmjjournals.com/cgi/eletter-submit/12/6/402 to answer three questions:

1. What are the key functions of the medical record?

2. What are the key elements that would need to be included in a medical record in order to achieve these functions?

3. How do we implement the recommendations from 2 above?

You, as the readers of QSHC, can make a difference here.