This is the first study comparing the effectiveness of a new form for investigating garbage codes for external causes in Brazil. Use of the new IDEC form obtained statistically significant percentage reductions in GC from external causes that were greater compared to the current standard procedure; only 7.5% of IDEC form deaths remained garbage with the new form compared to 34% in the matched sample. The performance of the IDEC form resulted in double the proportion of GC deaths not from undetermined intent being reclassified to non-garbage codes compared with deaths of undetermined intent. The IDEC form was particularly good at reclassifying GCs into falls, road injuries and interpersonal violence.
The characteristics of the matched sample, composed from the current standard procedure, was mostly similar with the 164 cases investigated using the IDEC form, with the exception of the IDEC form having a lower proportion of deaths of undetermined intent; however, we disaggregated our findings into this group and other GCs. The characteristics were compatible with previous studies, and showed a higher frequency of death in men, both younger and older, and deaths occurring in hospital for injury deaths than other deaths.(16, 20, 27)
The investigation with the IDEC form reduced the number of GCs by more than 90%. The results may vary due to the composition of localities, information sources and types of GC. GC reductions for external causes ranged from 39–83% in certain studies, which typically used data from forensic institutes as a central source.(28–30) Better results have been shown for unspecified accidents in some studies, (28, 29) and for undetermined intent in others(30). Recently a multi-source study reduced GC with undetermined intent by 84% and reclassified 11% of undetermined natural causes to external causes, pointing to greater contribution to the data from police, press, and forensic institutes. (16, 27) Interestingly, it was possible to qualify 67% of undetermined intent deaths relying on newspaper reports alone.(31)
We also verified a reclassification of cases between external causes and natural causes as well as the reclassification of unintentional injury to other valid codes. This was most evident when looking at reclassification of ill-defined causes to external causes, obtained from national and local data;(10) where forensic institutes accounted for 20% of the ill-defined causes in 2010.(12) The frequency of GC in hospitals was still high, although it was lower in specialized hospitals, particularly those treating trauma.(20)
In this study, most deaths of undetermined intent were reclassified into falls in both comparison groups, which may have been influenced to some extent by the demographic profile of the investigated GCs. There is evidence of more reclassification of undetermined intent for accidents, specifically falls;(28, 29) as well as more recent findings have shown more of these are classified to homicides.(16, 27, 30, 31) Evidence has indicated that the utilization of multiple sources of information improves the results of GC reclassification,(27, 30) although this still may not be sufficient for assigning a specific COD.(20) Thus, it was observed that certain sources contribute better to identifying a particular category of cause, e.g. police investigations identifying homicides, or newspaper reports of recent traffic accidents, etc.(10, 28, 31)
After more than 40 years of operation, SIM has its own challenges in modernization and in decentralizing in a relatively large country like Brazil; with some of its components being implemented partially inadequately.(32) The challenges may be greater in areas with poorer access to public services, especially small and medium-sized municipalities in countryside and rural areas.(10, 33) These challenges are compounded by the poor quality of death records for external causes, which has not been updated, highlighted by the poor agreement of the causes of death issued by forensic institutes compared with the Department of Health.(22) (34) Poor quality of death certificates may be partly due to forensic physicians' inclination to disregard the records related to the hospitalization period. A recent study found that there may be movements and discontinuities in the quality of the cause over time.(35) In other countries certain similarities related to some types of GC have been found in literature.(9)
The IDEC form generally collected enough data to improve the quality of external COD data in hospitals and forensic institutes. However, after the investigations, it was necessary to rearrange the logical sequence of the questions to make it more intuitive and possible to be used in different services. Causes that remained as a GC when there was no information on the circumstance of the death generally resulted from unavailability from police investigations. Most of these were due to poisoning (mainly by cocaine) and unspecified vehicle accidents, and where the locality of violence or accident was reported as having taken place in a private residence or public road without a witness. It was observed that events that occur at home are more likely to be classified with some type of GC.(20) The difficulty of defining a correct diagnosis for the external causes due to that they require additional information in order to determine the COD. If this is not part of the death registration, then misclassification may result.(36)
In addition to autopsy, the proper completion of the cause of death is, or should be, based on the availability of examinations and complementary data from the scene of the event, which remain the responsibility of specialized police and justice departments.(36, 37) Physicians from forensic institutes often document the nature of the injury, without specifying the circumstance of death, which results in a GC.(22, 28, 30) Findings from a preliminary police investigation into a violent death is sent along with the corpse to the forensic institute. The burial only occurs after release of the body with the death certificate and are linked to the same period of time, which should only be a short period of time following arrival at the forensic institute. However, when examinations and police investigation are continued for longer periods, possible new findings may be acquired which alter or clarify the cause and circumstance of the event.(12, 34) Police information is generally limited to a brief description of the event, often sufficient to characterize violence, but insufficient to characterize accidental events such as falls and poisoning.(28)
These earlier findings reinforce the hypothesis that the difficulty in properly diagnosing external causes extend beyond the contextual and structural issues of service and medical training reported by previous studies, which noted as key issues the non-use of instructional materials, low value attributed by the certifiers and lack of ability to describe the chain of events.(22, 38, 39) The set of evidence suggests that different needs and ways of working of the legal/police and the epidemiology of health are part of the problem.(16, 27) This dual role is not a particular feature of the Brazilian medicolegal death investigation system, as for example, in the United States, the death investigations also carry broad societal importance for criminal justice and public health. We must also consider the potential of the autonomy of federated entities as one of the factors of production of multiple realities, of highly varied state and local systems for investigating deaths.(40)
Although these different systems have in common the protection and guarantee of rights(41), each institution has different priorities and ways of working. For instance, public health has its central focus the victim and the associated risk factors, demanding a faster pace and shorter time in view of the need to implement timely actions.(42–45) The legal/police logic of the public security and justice sector, on the other hand, is a normative system that prioritizes the victim-perpetrator binomial as its core of attention, expressly with a slower pace and timeframe, especially in violent deaths involving crimes.(46)
Certain experiments have shown promise in the field of qualifying the COD, such as the use of an online death certification system to improve the cause of death record, especially when accompanied by a training program.(47) When errors are encountered on the COD in the death certificate, it is proposed to recapture the causes using the multiple COD, which can provide strong clues about the valid cause.(19, 48) A strategy implemented by other countries has been to postpone filling the cause in the death certificate until up to six days after death, facilitating the incorporation of test results and police investigations. This requires a prior death certificate to be issued, leaving out the circumstances of death until a full certificate can be issued following the conclusion of outstanding investigations.(27)
Some limitations were present in this study. In particular, the low number of cases, the geographical locations and the variety of codes investigated may have affected the results. Although testing was performed only in state capitals, the selection of deaths to review allowed us to investigate some cases from the interior of the states, which have distinctive epidemiological patterns.