Aims In England and Wales, doctors are charged with a responsibility either to report a death to the coroner or issue a medical certificate specifying cause of death. A lack of formal prescriptive or presumptive oversight has resulted in the promulgation by individual coroners of local reporting regimes. The study reported here identified overall and gendered variations in local reporting rates to coroners across the jurisdictions of England and Wales, consistent over time.
Methods Analysis was performed on Ministry of Justice (MOJ) data pertaining to the numbers and proportions of deaths reported to the coroner by jurisdiction over a 10-year period (2001–2010). Office of National Statistics (ONS) data provided the numbers of deaths registered in England and Wales over the same period to serve as a denominator for the calculation of proportions. Where coroner jurisdictions (and local authorities) had been amalgamated during this period, the combined reported and registered death figures have been included in line with the current jurisdiction areas.
Results While reporting rates for individual jurisdictions were found to be stable over the 10-year period, wide local variations in reporting deaths to coroners were found with no obvious demographic explanation. The gender of the deceased was identified as a major factor in local variation.
Conclusions The decision to report a death to the coroner varies across jurisdictions. Implications for coronial investigations are discussed and the need for wider research into coroners’ decision-making is proposed.
- ANALYTICAL METHODS
Statistics from Altmetric.com
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.
A coroner is an independent judicial officer responsible only to the Crown and must be a barrister, solicitor or medical practitioner of not less than 5 years standing.i Their legal and professional autonomy is unmatched in other judicial settings. There are currently 114 coroner jurisdictions in England and Wales served by 98 coroners.ii
There is no statutory duty placed upon a medical practitioner (eg, general practitioner) to report a death to the coroner. The legal duty resides with the coroner to hold an inquest when defined circumstances apply—for example, deaths which are violent or unnatural, or sudden and of unknown cause, or in prison.iii Limited advice as to which cases should be reported is to be found within blank medical certificate booklets, in Office of National Statistics (ONS) Advisory Group publications, in bespoke advice leaflets such as that issued by the Medical Protection Society and, tellingly, in locally produced guides issued by some coroners to medical practitioners working in their areas. In many locally produced guides coroners go beyond spelling out national provisions and seek to impose additional ‘local rules’, instructing doctors to report to the coroner all ‘deaths within 24 h of admission to hospital’, after ‘late diagnosis or treatment’, or ‘fractures or falls’ or ‘dementia’.
Start et al1 reported the findings of a study in which 196 clinicians and four coroners’ staff were invited to complete a postal questionnaire describing 12 fictitious case studies (10 of which contained a clear indication through national guidelines for referral to the coroner). 97% of participating general practitioners failed to recognise all those deaths which should have been reported for further investigations. In contrast, all participating coroners’ staff correctly identified all of the reportable cases in the study. Start and colleagues concluded that some doctors held disturbing misconceptions in relation to the coronial system, with implications for the evasion of medicolegal investigation resulting in consequences ranging from serious crime going undetected to loss of industrial pension or other appropriate compensation for relatives of the deceased (Start et al, 1993). The main confusion was found to lie between accidents and deaths associated with medical treatment as putative causes of death, with further confusion arising as a result of local variations in coroners’ practices thereby shaping doctors’ subsequent decisions to report. For example, there is persuasive evidence in the fundamental review of coroner services by Luce et al2 that the present system fails to identify some suicides, drug-related deaths and deaths in which adverse reactions to prescribed drugs may have contributed.
Berry and Heaton-Armstrong3 describe in their review of the coroner system a flawed process which requires comprehensive revision and is poorly understood to the extent of widespread ignorance by those within the medical profession who have to use it. This, they contend, is not helped by an inconsistent and unprofessional approach by coroners who cannot agree on what is a death from natural causes and what is not. Roberts et al4 demonstrated considerable variation in the way coroners approach borderline ‘natural cause’ cases. Sixteen clinical scenarios were circulated to coroners asking for a verdict and explanation; 64 were returned and there was near consensus as to verdict (>80% concordance) in only two of the 16 cases. The comments made for each case indicated that the differences reflected varying personal attitudes of each coroner. Thus coroners set their own ‘local tone’ as to what might be considered a natural death (and therefore potentially not reportable to the coroner) and what might be an unnatural death (and therefore reportable).
The conclusion arrived at by Berry and Heaton-Armstrong is that the investigation of death continues to rely upon the application by medical practitioners of standards set variably by local coroners. Whether or not this is satisfactory practice depends upon the value attached to national consistency. Certainly the Crown Prosecution Service and the criminal courts are subject to extensive national guidance in an attempt to limit inconsistent or idiosyncratic decision-making, and there seems no reason why this should apply less to the process of death investigation.
The Government is presently reforming the process of death certification by appointing medical examiners to provide independent scrutiny of the cause of any death which is not reported to a coroner (expected implementation by 2014). The aim is to simplify and strengthen current certification arrangements, improve the quality and accuracy of data on cause of death and prevent multiple deaths going unnoticed and unchallenged.5 It is envisaged that in future medical examiners will have the discretion to report a death to the coroner according to an agreed national protocol which sets out the minimum level of scrutiny that must be applied. In an update produced for coroners,6 deaths reported to a coroner are said to be expected to decrease from the present national average of 46% to around 35%, although this is likely to differ in each area depending on its current baselines and local factors.
Analysis of reporting rates to coroners in England and Wales
In the 1950s, fewer than 20% of registered deaths were reported to the coroner. That figure is now 46%.iv
Tables 1 and 2 were developed by comparing data held by the ONS on registered deaths in local authority areas in England and Wales and data held by the Ministry of Justice (MOJ) on deaths reported to the current 114 coroner jurisdictionsv to show local reporting patterns for the 10 years (2001–2010).
Table 1 shows the 10 areas in England and Wales with the highest reporting rates for the period 2001–2010 and table 2 shows the 10 areas with the lowest reporting rates. Where coroner jurisdictions (and local authorities) have been amalgamated in the last 10 years, the registered and reported death figures have been included in the current jurisdiction area. This does mean that, on some occasions, while the results reflect the practices in an area over the decade, they may not accurately reflect current practices in so far as they are shaped by the incumbent coroner rather than his or her predecessors in jurisdictions swallowed up during the period.vi Reporting rates for other jurisdictions are available from the writer on request.
Reporting rates thus ranged from 12% of registered deaths in one jurisdiction to 87% in another, with no obvious explanation in features which distinguish high and low reporting areas. An obvious question is whether high rates are consistently so. This was checked by calculating a product moment correlation coefficient comparing reporting rates in the first and last year of the period under study. This proved to be extremely high, suggesting that the relative rates are stable over time.vii This was further checked by correlating rates for comparison of data 9 years apart and 8 years apart and so on, again supporting the conclusion that jurisdictions have stable relative rates of reporting. Reporting rate differences must therefore be attributable to some combination of local demography or local medicolegal practice which is consistent over time.
Figure 1 shows the total reporting rates to coroners for 2001–2010 for all jurisdictionsviii (plotted in 5% bands). The mean (SD) reporting rate is calculated as 45 (12)% of all registered deaths in a jurisdiction.
Eighty-three jurisdictions (75%) lie within reporting rates of 33% and 57% (1 SD above and below the mean), with 11 jurisdictions having reporting rates of <33% and 17 jurisdictions having reporting rates of ≥57%.
When the findings are placed on the geographical map of jurisdictions, there does appear to be a probable inverse relationship between larger jurisdiction areas and lower reporting rates to the coroner. Possible factors, which are only speculative, might be distance from a coroner's office, or necessary self-sufficiency in large rural communities or, indeed, the relative lack of supervisory or peer oversight in less densely populated areas.
Highly populated areas of course reported more deaths (a larger caseload), but no relationship was found between the reporting rate and size of the population.ix For example, the two coroner jurisdictions with the highest populations (Essex & Thurrock and North London) had reporting rates of 36% and 59%, respectively. There was, however, some relationship between higher levels of deprivation and higher reporting rates to the coroner.x This is likely to reflect a population less well known to general practitioners through health inequalities and therefore less capable of certification without referral to the coroner. Significant differences could still be found, though, in Blackburn and Manchester North, areas with similar levels of deprivation reporting 70% and 41% of all deaths, respectively.
Thus, there appears to be no obvious demographic reason for such wide variations in local reporting rates.
The mean post mortem rate for 2010 was found to be 46%, ranging from 20% to 66%. As expected, more reported deaths brought more post mortems, but a moderate negative correlation was found between higher reporting rates and post mortem ratesxi—that is, areas in which a greater proportion of registered deaths were reported tended to conduct proportionately fewer post mortems on those deaths.
All inquest verdicts across all jurisdictions (England and Wales) from 1995 to 2011 inclusive were analysed. Initial analysesxii showed that individual jurisdictions are consistent over time in their proportionate use of inquest verdict types. However, jurisdictions vary widely across the country in verdict use, suggesting that verdict patterns may be more a product of an individual coroner's decision-making style than a reliable indicator of local patterns of death. Even jurisdictions with similar overall caseloads report very different verdict profiles. This ‘consistent variance’ is a key focus of this research.
Table 3 shows the range of proportionate verdict use across jurisdictions for 2011 only,7 and demonstrates wide variation in the use of verdicts across England and Wales. There is, of course, a link back to reporting patterns as, for example, a coroner with low use of the natural causes inquest verdict may be one whose guidance to doctors yields fewer cases which suggest a natural death.
Reporting of death according to the sex of the deceased
When death reporting rates in England and Wales (2001–2010) are looked at according to the sex of the deceased, 49% of male deaths (range 81–13%) were reported to the coroner compared with 39% of female deaths (range 70–11%). Jurisdictions with high (or low) reporting rates for men were found to have high (or low) reporting rates for women.xiv Again, jurisdictional reporting rates appeared quite consistent over time, with the relationship dropping from very strong to strong for male deaths and to moderate for female deaths when the first and last years of the dataset (2001–2010) are correlated.xv Therefore, in this study deaths of men were found on average to be 26% more likely to be reported to the coroner than deaths of women. This held across all jurisdictions in England and Wales (and, in some areas, was up to 48%xvi more likely to be reported). Since women die later than men, it would be crucial to examine age-specific reporting rates for which data are not readily available.
Further research: coronial decision-making
One might consider the reporting of a death to the coroner as the first stage in a three-stage decision-making process which describes our system of death investigation. Once a death has been reported, it is subject to two further fundamental coronial decisions: (1) whether to proceed to inquest; and (2) what the resulting inquest verdict should be.
There are proportionately fewer inquests for women than for men. For men, 16% of all deaths reported to coroners (2001–2010) proceeded to inquest while, for women, the figure was 8%. Once at inquest, verdicts of natural causes were recorded more often for women than for men (28% vs 22%). So, fewer women were reported to the coroner, fewer women proceeded to inquest and fewer women at inquest were considered to have died unnaturally. For unnatural deaths, men were over-represented in industrial disease and suicide verdicts while women were over-represented in narrative verdicts and accidents.
There is therefore evidence of different verdicts being reached differentially according to the sex of the deceased. Analyses of the difference in the proportionate use of verdicts by sex of the deceased suggest that some coroners are ‘gendered’ in their approach to inquest verdicts—that is, they are consistently more likely to favour a particular verdict when dealing with a death, according to the sex of the deceased.
The degree to which decisions are gendered in different jurisdictions appears to be consistent over time for all verdicts except industrial disease where the association between year and degree of ‘genderedness’ diminishes over time (r=0.82, p<0.001). The degree to which different verdicts are gendered varies across jurisdictions, with the verdicts ‘unlawfully killed’ and ‘drug non-dependent’ being the most highly gendered—that is, the proportions of verdicts being particularly different according to gender.
The proportions in which deaths are reported to the coroner in England and Wales vary widely across coronial jurisdictions. Local rates are consistent over time.
Coroners in England and Wales vary widely in their use of verdicts. Local profiles of verdicts are consistent over time.
Deaths of women in England and Wales are less likely to be reported to the coroner than deaths of men. Female deaths reported are less likely to proceed to inquest than male deaths, and female deaths proceeding to inquest are less likely to result in a verdict of unnatural death than male deaths.
Some coroners seem especially ‘gendered’ in their decision-making in that they are consistently more likely to favour a particular verdict according to the sex of the deceased.
Contributors MM designed the data collection tools, collected data for the whole analysis, wrote the statistical analysis plan, cleaned and analysed the data and drafted and revised the paper. He is guarantor. JR monitored the data collection, analysed the data and revised the draft paper. RA monitored the data collection and revised the draft paper.
Competing interests None.
Provenance and peer review Not commissioned; internally peer reviewed.
Data sharing statement Analysis of the reporting rates of deaths to coroners for all jurisdictions in England and Wales 2001–2010 is available from the corresponding author on request.
↵i Section 2(1) Coroners Act 1988.
↵ii Coroners Statistics 2011 England and Wales Ministry of Justice Statistics Bulletin.
↵iii Section 8(1) Coroners Act 1988.
↵v (n=111). With the exception of the Queen’s Household and Isles of Scilly for very low numbers and the City of London for which deaths registered could not be obtained separately (included in Inner North London). Absolute accuracy in registered deaths could not be achieved for 13 coroner jurisdictions due to MOJ composition being based on parts of parishes. A standard approach was taken whereby data were allocated to the jurisdiction which MOJ composition suggested had much the greater part of a parish. No jurisdiction was composed in this way alone, the vast majority being a combination of unitary authorities or metropolitan boroughs and thereby matching with ONS data.
↵vi Notes for completion on the MOJ coroner collection form (2010) state that “Section (i) (deaths reported to the coroner) should include all cases in which coroner investigated the circumstances of a death personally or by his officer, by letter or telephone….”. We are therefore satisfied that all relevant cases should be recorded, making our analysis valid in terms of which deaths are reported. There should be no case, for example, for coroners’ officers rejecting reports of death without recording the fact of the report. We contend that coroners have vicarious responsibility for all investigations under their jurisdiction and that comment on their decision-making, by reference to their areas overall output, is valid.
↵vii The relationship was investigated using Pearson product moment correlation coefficient. There was a very strong positive correlation between the two variables of reporting rates for 2010 and 2001 (r=0.900, n=111, p<0.0005).
↵viii With the exception of the Queen’s Household and the Isles of Scilly for very low numbers and the City of London for which deaths registered could not be obtained separately (included in Inner North London).
↵ix Investigated using Pearson product moment correlation coefficient using the variables reporting rate (for 10 years) and 2010 population figures calculated by summing local authority area populations according to the current jurisdiction composition provided by the MOJ, r=0.022, n=111, not significant.
↵x Using variables reporting rate (for 10 years) and 2010 IMD data for local authorities E&W only (using the highest ranked deprivation score within each jurisdiction published by Dept for Communities and Local Government 24 March 2011), r=−0.399, n=99, p< 0.0005.
↵xii The relationship between the 6 most common verdicts (which account for 98% of all 2011 verdict outcomes) between 2011 and 2009 was investigated as a test analysis using Pearson product-moment correlation coefficient. There was a consistently strong positive correlation for all years between the variables of ‘other verdicts’ and ‘natural causes’, (r between 0.946 and 0.905); and a similarly strong and consistent across years positive correlation for the verdicts of industrial disease, suicide, accidents and open verdicts, r between 0.894 and 0.834; n=112, p< 0.0005.
↵xiii Ten year male and female reporting rates were investigated using Pearson product-moment correlation coefficient. There was a very strong positive correlation between the two variables, r=0.975, n=111, p< 0.0005.
↵xiv The relationship between male and female reporting rates 2001 and 2010 was investigated using Pearson product-moment correlation coefficient. There was a strong correlation between the two male variables, r=0.739, n=111, p< 0.0005 and a moderate correlation between the two female variables, r=0.691, n=111, p=<0.0005.
↵xvi Bridgend and Glamorgan Valleys reporting rates of registered deaths 2001–2010: men 68% and women 46%.