Intended for healthcare professionals


Using death certificates to identify malpractice might be difficult

BMJ 2001; 322 doi: (Published 03 February 2001) Cite this as: BMJ 2001;322:303
  1. Adam Coldwells, development and research manager (adam.coldwells{at},
  2. Fiona Fraser, clinical audit assistant,
  3. Pam Tavendale, trust-wide clinical governance coordinator,
  4. George Crooks, associate medical director,
  5. Gordon Peterkin, medical director
  1. Grampian Primary Care NHS Trust, Aberdeen AB25 2ZP
  2. Grampian Primary Care NHS Trust, Royal Cornhill Hospital, Aberdeen AB25 2ZH

    EDITOR—In January 2000 Dr Harold Shipman, a general practitioner in Greater Manchester, was convicted of murdering 15 of his patients. 1 2 This conviction has fostered considerable comment and a need to prevent any recurrence. Although a quantitative technique to identify malpractice has limitations,3 a description of basic data should act as a starting point.

    We performed some simple investigations on death certificate records for 1998, after agreement from the general practice subcommittee. Our comments highlight some difficulties associated with the investigation of death certificates from an electronic record.

    The first problem concerned technical issues with the dataset. These included identification of a general practitioner, as opposed to a hospital consultant; electronic sorting was difficult because many fields (including general practitioners' names, initials, and addresses) were inconsistent. In 30 of our 6000 records the address of the patient rather than the general practitioner was recorded, and general practitioners have satellite surgeries, which legitimately adds a second address permutation.

    The second problem was that difficulties in identifying general practitioners in the Grampian register were confounded by out of hours cover, trainees, and locum cover from outside Grampian. One particular problem arose when we seemed to identify a general practitioner who signed 30 death certificates; it transpired that two general practitioners have identical surnames and initials and are both based in the same area. Including the doctor's registration number on the death certificate would yield a useful additional field.

    The average (SD) number of certificates signed by general practitioners grouped into practices was 27.65 (19.87) (range 1-92). The median was 22. The average number for an individual general practitioner was 5.85 (4.47) (range 1-24; median 11).

    Grampian has some 93 general practices; 88 appeared in this analysis. Inspection of the data raised many questions—for example, where do individuals die (city versus rural); do individuals get appropriate choice?—and issues concerning workloads and staffing levels.

    The trust cannot escape its corporate responsibility and, specifically, its need to show clinical governance. To this end a multidisciplinary group has been set up to establish a mechanism for judging the quality of an individual's death.


    1. 1.
    2. 2.
    3. 3.
    View Abstract