Chapter 8
Auditing cause of death determinations

A national audit system

8.18Under a national audit system, experienced doctors would be employed to review the cause of death certificate with the aim of detecting and correcting error in the determination of the cause of death and in referrals to the coroner. However, unlike the current system of medical referees, a national audit system would have the additional features of:

Administration by central government

8.19In Chapter 4, we recommended that the Ministry of Health should have statutory responsibility for the death certification process. The Ministry of Health would therefore also have responsibility for funding and implementing the national audit system and for ensuring the quality of its outcomes. That would involve providing ongoing education to the practitioners employed to review the cause of death certificates (we have called them “cause of death reviewers”) and checking the accuracy of their assessments. The cause of death reviewers would be accountable to a minister for the outcomes of the audit system and should provide a publicly accessible annual report to that effect.

8.20We consider that centralisation brings an essential element of independence and integrity to the system. Medical referees have told us that their knowledge of local circumstances and local doctors enhances their ability to assess cremation certificates. While that may be so in some cases, it means that there is no scope to identify problems that occur nationwide and no ability to tailor training for common problems. It also risks a lack of independence if the medical referees generally know the certifying doctors. A structure that enables a more objective assessment is important.

8.21We have not formed a view on whether, although centrally accountable, cause of death reviewers should be located centrally or regionally. This may depend upon the availability of suitable staff to perform the role. We cannot see that locating cause of death reviewers regionally would be a problem so long as they are well connected to each other so that audits in one region can be informed by the lessons learned from other regions.

8.22One option is for the Ministry of Health to provide funding to the National Forensic Pathology Service to provide this service. Currently, that service is funded by the Ministry of Justice to provide forensic pathology services to coroners. Part of the core work of these pathologists is to analyse causes of death in light of medical notes and autopsy reports. While the purpose of the proposed audit function (detecting errors in cause of death certification) is slightly different from the purpose of their services for coroners, the skills required would be similar. It may be more efficient to incorporate this service into an existing service than to establish a new organisation.

How would the audit system work?Top

8.23The primary function of cause of death reviewers would be to detect and correct errors in cause of death determinations. In undertaking reviews, cause of death reviewers should have access to the deceased person’s medical notes and the ability to discuss the documentation with the certifying doctor so that a full analysis can be made to check whether the cause of death is accurate.

8.24It is likely that a two-level approach to auditing death certification, similar to the system developed in Scotland, would be the most efficient. Level one reviews would involve scrutiny of the entries on the online certification system and any other certification documentation and (if necessary) speaking to the certifying doctor. The purpose of this review is to identify omissions or errors in the way that the cause of death is stated (for example, confusing the primary cause of death with an antecedent cause and underlying conditions).

8.25If the death involves any one of a number of identified risk factors or “red flags”, a level two review would be carried out, which, in addition to reviewing the death certification entries and documents and speaking to the certifying doctor, would involve scrutiny of electronic or hard-copy medical records plus speaking to any other relevant people, such as the family, Police or other medical professionals involved in the care of the deceased. The purpose of a level two review is to detect deaths where the stated cause of death is not verified by the medical records or where circumstances of the death may give rise to questions as to the cause of death. This would include identifying deaths that should have been referred to the coroner but were not.

8.26When an error is detected, the cause of death reviewer should be required to discuss the error with the certifying doctor. That discussion will either provide further detail and context to satisfactorily address the reviewer’s concern, or an amendment to the cause of death will be agreed upon. If the cause of death reviewer cannot agree with the certifier, it will be necessary to refer the question to an independent adjudicator. That could be a coroner where the issue raised by the reviewer is that the death should have been referred to the coroner. In other cases, it should be a third doctor or other cause of death specialist.

8.27If the reviewer detects criminal behaviour, he or she should have a statutory duty to report that to the Police.

What deaths should be reviewed?Top

8.28In designing a review system, we have considered whether every determination of cause of death should be reviewed or whether it is sufficient to review a random sample. The answer lies in the importance given to ensuring accuracy in each case versus improving levels of accuracy generally. Reviewing every cause of death would be important for many of the private purposes of cause of death determinations (such as the assessment of life insurance claims and knowledge of familial medical history). However, for public purposes such as the development of public health policies and programmes, it is more important that the accuracy of cause of death certification is improved overall.

8.29There are two significant negative consequences of reviewing every cause of death determination: cost and delay. Obviously, auditing every death would require significantly more resources than auditing a sample. It is not clear that the benefits to the public justify these costs. In relation to timing, for a review to be meaningful, it must be conducted prior to disposal of the body so that the body is available for further investigation if necessary. However, reviewing every death before disposal would produce significant delays in funerals, burials and cremations. We have received strong submissions that avoiding such delay is very important in New Zealand.85 On balance, we consider that these cost and timing considerations tip the balance in favour of auditing only a sample of deaths.

8.30Deaths that are referred to the coroner should be excluded from the national audit process because the purposes of the review will be satisfied by the coronial process.

8.31We have also considered whether deaths in hospitals should be excluded from the audit process. Some hospitals have established their own internal review committees to examine the quality of cause of death certification within those hospitals. For example, in 1993, Christchurch Hospital devised its own system of auditing and quality control of death certification following an inquiry into deaths of a number of patients of cardiothoracic surgeon Keith Ramstead. That audit system has resulted in significant improvements in the accuracy of MCCDs and in assessing whether a death should be reported to the coroner. Other hospitals, particularly larger hospitals, have similar review committees with differing processes.

8.32There are significant advantages in having one review process for all deaths, wherever they occur, where the lessons learned can be shared across all deaths. However, this is a more expensive approach. A cost-effective alternative is for hospital deaths to be excluded from the national audit system. If this approach was adopted, hospitals should be required to peer review their own cause of death determinations. We suggest that such peer-review systems must review a random sample of deaths and include a mechanism for providing feedback to the certifier when errors are identified.

8.33In addition, there should be some national oversight from the Ministry of Health of these hospital peer-review systems to ensure they produce quality outcomes and the trends and lessons learned from them are shared between hospitals and are used to train hospital doctors who certify the cause of death. A further feature could be for the central agency responsible for auditing death certification when death occurs in the community to have a role in providing oversight for hospital peer-review systems.

8.34In conclusion, we consider that the proposed national audit system should review a random sample of all deaths except hospital deaths and deaths that are referred to the coroner.

Referrals for auditsTop

8.35In addition to random sample reviews, cause of death reviewers should be able to receive referrals to review particular cases. This would be available if someone suspected there was an error in documentation related to a particular death (whether it be family members, funeral directors, the Police, the Health and Disability Commissioner or another party). This will mitigate the effect of auditing only a random selection of deaths. However, the cause of death reviewers should be entitled to dismiss a referral where they consider there are no circumstances that would justify review.

Targeted reviewsTop

8.36An important secondary function of cause of death reviewers should be to undertake targeted reviews of the cause of death certificates for deaths with particular characteristics. Decisions about the types of deaths targeted for review would be informed by a number of factors including:

8.37An example of a targeted audit would be a review of cause of death certificates in respect of deaths occurring in a particular aged care facility if there is cause for concern over a disproportionate prevalence of a particular cause of death. Such a review may detect inaccuracies in death certification or problems resulting from neglect at particular facilities (such as an unusually high number of deaths from falls). The evidence gathered from these reviews should be used to change practices, procedures and accountability mechanisms. It should also be used to educate doctors who certify the cause of death.

Support for and education of doctorsTop

8.38A corollary of centralising the review of cause of death certificates is the potential to make significant improvements to accuracy via formal systems of support and education for doctors. Systematic audits of a random sample of death certificates, together with the targeted audits, will quickly enable cause of death reviewers to develop a strong understanding of the risk factors for error across the whole country. The expertise that cause of death reviewers will develop in this role makes them ideally suited to support doctors to improve the accuracy of death certification. We consider that support should take three forms:

8.39In Issues Paper 23, we asked whether all doctors who are required to complete MCCDs should have access to independent advice. All except one of the 18 submitters who responded to this question agreed. Submitters thought that discussing cases with experts is very valuable and is likely to improve the accuracy of death certification, particularly for rural or sole-practice doctors. We agree and consider that cause of death reviewers would be well placed to offer this kind of service.

8.40There have been a number of formal studies into the effect of educating doctors on their accuracy in determining the cause of death:

8.41Although these studies are few in number and diverse in their methods, they provide some evidence that education of doctors can produce significant improvements in accuracy. Based on these studies, best-practice training should:

8.42In addition, the 2007 United Kingdom study provided evidence that the “audit, educate, audit” method can be very effective. We envisage that, through targeted reviews, cause of death reviewers may identify particular groups of doctors who have a high rate of error. Education would then be targeted based on the most common errors being made in practice.

Funding the audit systemTop

8.43In Chapter 3 we described options for sources of funding for doctors who certify the cause of death. We have conducted a similar analysis of the proposed new audit process and we consider that the audit process should be publicly funded for the following reasons:


R15The statute should create a statutory role of “cause of death reviewer” to be appointed by the Minister of Health.

R16A function of cause of death reviewers should be to undertake a review of a random sample of all deaths (except deaths that occurred in hospital and deaths that have been referred to the coroner) for the purpose of:
  • detecting error in the determination of the cause of death;
  • detecting deaths that should have been referred to the coroner; and
  • providing education and support to doctors who certify the cause of death.
R17Additional functions of cause of death reviewers should be to:
  • review deaths referred to them;
  • undertake targeted reviews of deaths; and
  • provide support and education for doctors who certify cause of death.
R18The statute should provide that, when a cause of death reviewer detects an error in the determination of the cause of death, the reviewer must:
  • discuss the error with the certifying doctor with a view to reaching agreement (if necessary) about amending the certification of the cause of death; and
  • if agreement cannot be reached, refer the death to the coroner or to another authorised doctor for adjudication.
R19If the reviewer detects evidence of criminal activity, the reviewer must report the death to the Police.

85For example, Dr Martin Sage contrasted New Zealand to England and Wales: “[…] there are practical cultural differences between England and Wales and New Zealand in this regard: in England and Wales the whole after-death process (that is, certification of death, release of the body to funeral directors, with or without autopsy) usually progresses at what New Zealanders would regard as an infuriatingly lackadaisical pace, certainly over a period of many days (often 5 -7 days or more) which are apparently accepted by families in the UK but which would be entirely intolerable to many sectors of our society.”
86T Weeramanthri, W Beresford and V Sathianathan “An evaluation of an educational intervention to improve death certification practice” (1992) 13 Aust Clin Rev 185.
87Major errors included mechanisms of death without a legitimate cause of death, improper temporal sequencing of diseases and competing causes of death. Minor errors included the omission of time intervals for the presence of diseases, the use of abbreviations and the inclusion of the mechanism of death (but with a legitimate cause of death).
88KA Myers and DR Farquhar “Improving the accuracy of death certification” (1998) 158 CMAJ 1317.
89Eindra Aung, Chalapati Rao and Sue Walker “Teaching cause-of-death certification: lessons from international experience” (2010) 86 Postgraduate Medical Journal 143.
90Dhanunjaya R Lakkireddy and others “Improving Death Certificate Completion: A Trial of Two Training Interventions” (2007) 22 J Gen Intern Med 544.
91Christian P Selinger, Robert A Ellis and Mary G Harrington “A good death certificate: improved performance by simple educational measures” (2007) 83 Postgrad Med J 285.
92Submissions indicated that, in hospitals, it is usually the junior doctors who are asked to certify death. Given the natural movement of junior doctors through their training system, regular training is important to capture new doctors.
93See Chapter 5 for further discussion on the timing of the obligation to notify the death to the Registrar-General.