What happens to coroners’ recommendations for improving public health and safety? Organisational responses under a mandatory response regime in Victoria, Australia

Coroners in Victoria may issue recommendations as part of their written findings from a death investigation. The investigation may be desk-based, or involve an inquest, however most recommendations come from investigations that have gone to inquest [15]. Recommendations may be directed at any entity [16]. For simplicity, we refer to each instance of the death investigation-finding-recommendation continuum as "cases"; we refer to entities that are required to respond as "recipient organisations".

In any given case, a coroner may direct one or more recommendations to one or more recipient organisations. All combinations are possible: a recipient organisation may receive one or more recommendations, and recommendations may be directed at a single recipient organisation, or distributed across several. This heterogeneity necessitated the development of counting conventions and careful consideration of the appropriate level at which to collect and analyse data. A detailed description of our case ascertainment procedures and steps for determining the units of analysis are presented in Additional file 1.

We invited all organisations that received recommendations during the first 33 months (1 November 2009 to 31 July 2012) of Victoria’s new mandatory response regime to complete an online survey. Recipient organisations were eligible for inclusion in the study if they: a) had one or more recommendations directed to them; b) were mandated to respond (all "public statutory authorities and entities" are mandated, but Ministers are exempt); [13] and c) had provided a written response to the coroner. Some recipient organisations received recommendations arising from multiple cases during the study period. These "repeat players" were surveyed in relation to each case. To minimise respondent burden, however, if the most appropriate person within a recipient organisation had already been surveyed twice, they were not surveyed again.

Over the study period coroners issued recommendations in 146 cases to 228 recipient organisations. Several exclusions from this sample frame were necessary: 22 recipient organisations were not surveyed because the relevant person or unit had already completed two surveys; 13 recipient organisations had received recommendations by mistake; one case produced more than 50 recommendations directed at three recipient organisations, which were not surveyed because the recommendation-level questions in the survey would have been too onerous; two recipient organisations were not surveyed because initial contact revealed that another recipient organisation in the case had taken over responsibility for their response; and one recipient organisation from the private sector had gone into administration at the time of the survey. In addition, three organisations pre-emptively opted out of participating in the study; collectively, they received recommendations in 34 cases. These exclusions left a total of 153 recipient organisations (96 of which were unique), all of which were invited to participate in the study.

We invited recipient organisations to participate in the study a minimum of six months after the coroner had handed down findings and recommendations in the case. The average time from receipt of recommendations to survey date was 13 months (SD = 6.6). Initial contact was with either the signatory to the formal response letter or a contact person named in the letter. The goal of this initial approach was to identify the most senior person in the organisation who had direct knowledge of how the organisation had responded to the coroner’s recommendation(s). Once the target respondent was identified, we forwarded a study information sheet and a hyperlink to the online survey instrument. Non-respondents were re-contacted with three reminders spaced at two-week intervals.

We developed the survey instrument based on a review of relevant literature and input from a project Advisory Committee that included the chief coroners of Australia’s three most populous states (New South Wales, Queensland, Victoria). In addition, a draft instrument was pre-tested for content validity and comprehension with officials from two organisations that regularly receive coroners’ recommendations.

The survey asked participants, in relation to each recommendation received, to indicate if their organisation had: (i) taken action along the lines of what was recommended, (ii) taken the recommended action, but not in response to coroner’s recommendation; or (iii) not taken the recommended action. Participants then rated the importance of a range of factors in determining the nature of their response (on 5-point Likert scales ranging from "very important" to "not important"). Participants also rated their level of agreement with several statements about the relevance, appropriateness and effectiveness of both the coroners’ recommendations they received and the mandatory response regime in general (5-point Likert scale from "strongly agree" to "strongly disagree"). In addition, we queried participants about the costs of preparing their response and taking action (for those that did). Finally, for purposes of describing the sample, we extracted several variables (e.g. case dates, presiding coroner, type of death) from the written case findings.

Survey items measuring recipient organisations’ views of the recommendations and the mandatory response regime exhibited strong internal consistency (Cronbach’s alpha = 0.76 and 0.81, respectively), as did recommendation-level survey items measuring important factors in deciding whether to accept or reject recommendations (Cronbach’s alpha = 0.90).

Responses to questions about views of coroners’ recommendations and the mandatory response regime were analysed at the recipient organisation level, as were responses to questions about the costs of responding and taking action. All other analyses were conducted at the recommendation level.

For each recommendation, we calculated counts and proportions to describe recommendation uptake. Likert-scale responses rating importance were dichotomized (very important/important vs moderately important/of little importance/not important), as were Likert-scale responses rating level of agreement with specific aspects of the recommendation and response process (strongly agree/agree vs neither agree nor disagree/disagree/strongly disagree). "Don’t know" responses were excluded from all counts. Finally, we used chi-square analyses to test for relationships between recipient organisation’s uptake and views of recommendations. All analyses were conducted using Stata version 12 (Cary, North Carolina).

The study was approved by the human research ethics committees at the University of Melbourne and the Victorian Department of Justice.

Participants from 90 of the 153 recipient organisations surveyed completed the online survey, a 59% response rate. Forty-three of those 90 recipient organisations responded in relation to recommendations issued in one case only; the other 47 received and responded to recommendations in multiple cases.

The 90 recipient organisations received a total of 164 recommendations (mean = 1.9; range, 1–7). These recommendations were issued in 74 coronial cases relating to 86 deaths. The recommendations varied widely in form, content and specificity. Recurrent themes included requests for changes relating to education programs, information handling practices, organisational policies and procedures, documentation, governance, public awareness activities, regulations, the built environment, and the use of preventive devices.

Characteristics of the recipient organisations and deaths are detailed in Table 1. Statutory authorities or regulators (28%), government departments (24%), and health care institutions (16%) were the most common types of recipient organisations; together they accounted for 68% of all recipient organisations. Transport (24%), complications of health care (17%), and drowning (10%) were the leading causes of death. There were no significant differences between participants and non-participants across any of the characteristics reported in Table 1.

Cases were presided over by 19 different coroners working in metropolitan (n = 12) and regional (n = 7) areas of Victoria. An inquest was held in 69% (51/74) of cases and three-quarters (123/164) of all recommendations arose from cases that went to inquest. The average time elapsed between the death date and the date the recommendations were handed down was 34.8 months (SD = 17.7).

Fifty-nine percent (53/90) of recipient organisations reported having had an opportunity to provide information to the coroner during the investigation. A majority of this group (40/53) felt their views had been taken into account in formulation of the recommendations. For 93% (153/164) of recommendations, recipient organisations indicated that they were clear about what they were being asked to do.

In response to 37% (60/164) of recommendations, recipient organisations implemented the recommended action (referred to hereafter as "accepted"). In 36% (59/164) of recommendations, the recommended action was taken, but not in response to the coroner’s recommendation ("supplanted"). Recipient organisations did not take the recommended action in the remaining 27% (45/164) of recommendations ("rejected").

Prevention of future injury or death was cited as an important reason for taking action in relation to nearly all (56/60) of the recommendations that were accepted (Table 2, upper half). Other reasons cited as important in accepting the recommendations were the reputational implications of not acting (37/60) and awareness that a written response was mandatory and would be posted online (31/60).

Nearly half (18/45) of the recipient organisations that rejected recommendations indicated that an important reason for doing so was that the recommended action was not logistically viable (Table 2, lower half). Other reasons cited as important in the decision to reject were that the recommendation was not relevant to the organisation’s business or its operations (16/45) and the recommended action was not economically viable (11/45).

Among accepted recommendations, the most common types of action taken were implementation of education programs or training (36/60) and adoption or modification of policies, procedures or standards (28/60) (Table 3). For nine accepted recommendations, the recipient organisation had resolved to implement the coroner’s recommendation, but the precise form of action was still under consideration.

In nearly half (28/59) of supplanted recommendations, an investigation into the death by the recipient organisation itself was cited as the reason for having taken the action prior to the coroner’s recommendation (Table 4). The next most commonly-cited prompts were the death itself (24/59) and the coroner’s investigation (12/59).

The estimated financial costs to recipient organisations of preparing their response ranged from zero (20/90) to over $100,000 (1/90), with 79% of recipient organisations (71/90) estimating the costs at less than $20,000. Among the recipient organisations that implemented all or some of the recommendations directed to them, 57% (42/74) estimated the financial cost of implementing the recommendation at less than $50,000 and 15% (11/74) estimated costs of between $50,000 and $500,000. One organisation estimated the cost of implementing the coroner’s recommendation at more than $1,000,000.

Approximately 60% of recipient organisations agreed that the recommendations they had received were useful to their organisation (54/90) or to their industry as a whole (56/90; Table 5). However, only half (45/90) believed the recommendations would be effective in preventing death and injury in the future. There was substantial support for making findings, recommendations and responses available to the public, with the vast majority of recipient organisations agreeing with the statement that it is a good idea to publish coroners’ findings and recommendations (83/90) and organisations’ responses (78/90) on the Internet. In line with their experience with recommendations, however, only about half the recipient organisations (43/90) thought the new regime would result in long-term reductions in preventable death and injury in the community.

Views varied systematically according to how the recipient organisation had responded to the recommendations directed to them. For example, compared to all other recipient organisations, organisations that implemented the recommendations they received were more likely to perceive them as useful (43/57 vs 11/33, p < 0.001), regard the mandatory response regime as a good idea (47/57 vs 20/33, p = 0.02), and agree that it would help to decrease death and injury (32/57 vs 11/33, p = 0.04).