A glance over the surveillance of fetal deaths of Jaboatão dos Guararapes in the Brazilian state of Pernambuco , Brazil , in 2014

This study aimed to achieve a complete evaluation of the records of investigation, and the principal indicators of fetal death surveillance of Jaboatão dos Guararapes, Pernambuco State, and their contributions Mortality Information System (Sistema de Informações sobre Mortalidade – SIM). The population of the study comprised all fetal deaths taking place in the year 2014, of people resident in the town. The data were obtained from death investigation records and the databank of the SIM. The completeness of the report forms of 68 fetal deaths, and 13 variables in the death certificates, before and after the investigation, was analyzed. In 2014 the rate of fetal mortality was 10.3‰. Of the 102 deaths, 86.3% (88) were investigated, and 67% (59) were investigated within a period of 120 days. Only nine (10.2% of the deaths investigated) were the subject of a final discussion to complete investigation. The Hospital Form was the most frequently filled in, and the Summary Form was the form most completely filled in. The Outpatient Form was the least well filled-in. In the death certificates, there were rectifications made in the 13 variables studied. The results showed deficiencies and operational difficulties in Fetal Death Surveillance in Jaboatão dos Guararapes. Conversely, the study revealed the contribution of the investigation process in qualifying of the SIM.


Introduction
In recent years fetal death has become an increasingly well-known subject through successive reports in publications about its scale, and its invisibility, creating pressures from many countries in the WHO for recognition and monitoring of these deaths 1 .
For there to be an effective monitoring of fetal deaths it is necessary to recognize the importance of timely and continuous production of information 2 and, also, to incentivize capture of data in households, in a way that is similar to what is already carried out with infant deaths in some countries, taking as a basis the audit of stillbirths recorded in health establishments 1 .
In Brazil, the investigation of fetal deaths was instituted by a basis in law put in place as from 2010 3 , and has been used as a tool of monitoring and oversight, so as to recognize the risk situations and foster adequate care for mothers at the prenatal phase and during birth 4 .Death oversight consists of a strategy for prevention of further deaths, because it involves a sequence of phases.These include a discussion of deaths by the Technical Group (Technical Group).This makes it possible to: detect corrections in the vital records, analyze the chain of determinant factors, and assess the healthcare network involved in the occurrences, with a focus on avoidability and proposing preventive and corrective measures.
To achieve this, it is important that there should be involvement of representatives of all the technical areas of healthcare in the discussion 5 .
Thus, Mortality Oversight makes it possible to continuously improve the experience and knowledge of the professionals involved, through analysis of the deaths 6 , and makes a complementary contribution in enhancing the quality of the vital statistics by bringing attention to the information and correcting it 7 .
To ensure effectiveness of the oversight and appropriateness of the information, Health Ministry Order 72 of January 11, 2010 establishes periods to be complied with for each of the steps that comprise the oversight of fetal deaths, with an obligation for the investigative process to be completed in up to 120 days from the date of the death 3 .
The information also needs to be of good quality, with significant data 8 .Thus, consistency and completeness of the variables are parameters that should be considered in the analysis of appropriateness and quality of the information, and in monitoring of it, which can contribute to enhancement of the records 9 .
Surveys on oversight of fetal deaths are scarce in Brazil, especially those that propose to analyze dimensions of the investigative process and their contribution to the Death Information System (SIM).The objective of this study was to evaluate the completeness of the investigation records, which are the principal indicators of fetal death oversight in Jaboatão dos Guararapes, in the Brazilian state of Pernambuco, in the municipality of Greater Recife, in 2014, and their contributions to the SIM.

Methods
This is a cross-sectional, descriptive, population-based study.The population comprised all the fetal deaths (weight ≥ 500g and/or gestation ≥ 22 weeks) that occurred in 2014, in women resident in Jaboatão dos Guararapes.
The municipality is part of Greater Recife, on the coast of the state of Pernambuco, with area of 256 km² and population estimated at 680,943 in 2014 -the second largest population of the State 10 .
The sources of the data were reports of investigation of fetal deaths in the municipality and the municipal databank of fetal deaths, by household, of the SIM.Investigation of fetal deaths in Jaboatão is carried out using a series of forms recommended by the Health Ministry [3][4] .These are: the Outpatient Form (F1); the Hospital Form (F2); the Household Form (F3); the Autopsy Report Data Collection Form (IF4); and the Infant and Fetal Death Investigation Summaries, Conclusions and Recommendations Form (IF5).
The completeness of the variables in the forms of the 68 deaths that had at least one of the investigation forms available -F1, F2, F3 and IF5 -was analyzed.The completeness of the IF4 forms was not analyzed since not all deaths were sent for autopsy, and even those which were sent did not have those forms duly filled in.
The Outpatient Form comprises 38 variables and is divided into two blocks: Identification and Prenatal care.The Hospital Form has 52 variables in three blocks: Identification, Birth care, and Annotations on the stillborn child.The Household Form has 96 variables in five blocks: Identification, Characteristics of the mother and the family; Gestation and prenatal care; Care given at birth; and Information on the child.IF5 (the Summary Form) is not divided into blocks, although it comprises 50 variables applicable to fetal deaths that allow for its subdivision, from identification information to recommendations to the health services.For each form a database in Microsoft Excel spreadsheet was created.
Frequency of rectification, and completeness, were analyzed, before and after the investigation of the following variables of the Death Certificate: gender, weight at birth, gestation period, mother's age, mother's level of schooling, district, location of the event, establishment where the birth took place, number of children born alive, number of children stillborn, type of birth, death in relation to birth, submission for autopsy.
The analysis for completeness was based on the score system proposed by Romero and Cunha 11 , adapted, as already used in national publications.This considers the proportion of fields ignored and/or not filled in, analyzing the incompleteness of the variables.This present work used the opposite -completeness of the information -for classification of the filling in of forms, assessing it into the following categories: Excellent (> 95.0%); Good (90.0 to 95.0%); Regular (80.0 to 89.9%); Bad (50.0 to 79.9%) and Very bad (< 50.0%).Relative frequencies and mean central trends of the completeness were analyzed, by block of variables, and for the form as a whole.
Also analyzed were the following indicators for fetal deaths: proportion of deaths investigated; proportion of deaths investigated within the period; proportion of deaths investigated late; proportion of deaths discussed by the Technical Group; proportion of deaths discussed within the period specified; and distribution of frequency of the investigation by type of form.
The research project was submitted to the Ethics Research Committee of the Medical Sciences Center of the Federal University of Pernambuco (CEP/CCS/UFPE) and approved.The consent of the Executive Secretariat for Health Promotion of Jaboatão dos Guararapes was also obtained.

Results
In 2014, there were 9,921 births, 102 stillbirths and 9,819 live births in Jaboatão dos Guararapes.The rate of fetal mortality was 10.3 deaths per thousand births of resident mothers.Of the total of these stillbirths, 75.5% were born in Recife, compared to 20.6% born in Jaboatão and 3.9% in other municipalities.
Of the 102 fetal births, 86.3% (88) were investigated, and 14 were not investigated.Of the 88 investigated, none of the investigation forms of 9 of them (10.2%) had been located at the time of the study: 79 deaths were included in the analysis of the investigation.For 68 deaths, at least one of the investigation forms was available.
Of the deaths investigated, 67.0% (59) were within the period of 120 days.For 29 (33.0%)deaths investigated after expiry of the period, the mean number of days late was 63, with quartiles of 25% and 75% equivalent to 11 and 177 days, respectively.The minimum time of lateness was 2 days and the maximum was 393 days.Only 9 deaths (10.2% of the deaths investigated, and 8.8% of the total fetal deaths) were discussed by the Technical Group of the state; of these, 4 were considered within the maximum period of 120 days.
The main causes of delay reported by the death oversight team of the municipality were: rotation of the professionals in the technical team; difficulty in locating the address of the mother for household investigation; non-availability of information on medical records; and non-submission of the death certificate for deaths which took place in another municipality.The main reason relating to the low proportion of deaths concluded with a discussion was absence of a pediatric medical professional in the Technical Group of the municipality.
None of the 79 deaths studied presented more than one file per type of investigation (outpatient, in-hospital), even if the mother carried out her prenatal care in two different establishments or visited more than one hospital during the period of labor.
A higher frequency of filling in of the Hospital Form was observed, available for 65 (73.9%) of the 88 deaths investigated, corresponding to 82.3% of the deaths studied.In second place was the Household Form, with 26.1% (23) deaths investigated, or 29.1% of those studied.The outpatient form was filled in less frequently, for 11 (12.5%) of the deaths of the investigated or 13.9% of those studied.
For 31 of the stillbirths, the municipality had the cadaver transport record and the autopsy protocol.However, only for 9 of these (10.2% of the 88 deaths investigated) was the IF4 form (autopsy form) partially filled in, which hindered feasibility of study of completeness of that form.Further, for 11 deaths the investigation was terminated having only the information that comprises the IF4.
The IF5 form was only filled in in the discussions, and was present in eight cases.Although nine deaths were discussed, one case was not found in the files of the municipality.In the year under study, the discussion of the deaths was centralized in the State and the case not located may have been filed away by the Health Department of Pernambuco State.
The analysis of the completeness of the forms considered: the variables; the blocks; and the files.The file least well filled in was the Outpatient File (61.5%) and the best filled in was the Summary File (94.0%).The category of filling-in was 'regular' for the Hospital Form (80.9%) and for the Household Form (85.2%).
The Outpatient Form had a level of completeness that was considered 'bad' and the proportion of its Identification and Prenatal Care blocks with 'bad' filling-in were 61.62% and 61.44%, respectively.
The variables with the highest level of filling-in were: Number of the death certificate; name of mother; location of prenatal care; and information on mother's reproductive history (number of pregnancies, abortions, type of birth).These had 100% completeness, that is to say, excellent filling-in.By contrast, the variables relating to the characterization of the prenatal care received were categorized, predominantly as 'bad' (Table 1).Some variables of this file were not filled in (0% completeness) and classified as 'very bad': Number of the mother's SUS card; health/administrative district, Code in the National Health Establishments Register (CNES) of the prenatal care; general remarks; and remarks of the interviewer (Table 1).
The Hospital File comprises three blocks.The level of filling-in of the first block (Identification) was 'excellent' for all the variables except Number of mother's SUS card (10.8%, or 'very bad').The Birth Care block had completeness of 82.9% ('regular'), since the filling-in of its components varied from 100% (excellent) to 33.9% (very bad).The variables with worst filling-in of the block were Establishment CNES Code, date of previous birth and Medication during pregnancy, with 33.9%.38.5% and 40.0%, respectively.In this block, filling-in was at 'excellent' level for only 13 out of 36 of the variables, and further, there was 'bad' filling-in for variables relating to pregnancy risk and the quality of care provided at the birth: Time of membrane rupture, Aspect of the amniotic liquid, Maternal complications during labor, Risk factors, Largest interval without evaluation of the fetus, and Medication used (Table 2).
Completeness of the last block of the Hospital File (Annotations on the stillborn child) was 68.7% ('bad').Of the 9 variables, 5 are semi-structured and their filling-in was 'bad' (70.8%) to 'very bad' (16.9%).The other presented 'excellent' to 'regular' filling-in, the most complete being Conditions of birth (95.4%) (Table 2).
In the Home File, filling-in of the blocks varied from good to bad.Those most completely filled in were: Birth care received (90.7%) and Characteristics of the mother and the family (90.4%); the worst filled in was Identification (74.4%).In the latter, although filling-in was 'excellent' in 5 of the 10 variables, it was 'very bad' in two: Mothers' SUS card number (17.4%) and Health/administrative District (8.7%).In the block 'Characteristics of the mother and the family' filling-in was 'excellent' for 13 of the 23 (56.5%) variables, one was 'bad' (Paid work) and one 'very bad (Date of last prior birth) -(Table 3).
The group named 'Pregnancy and prenatal' comprised 14 variables with 'excellent' filling-in.Of these 4 had 100% completeness (Whether mother had prenatal care, Reason for not having prenatal care, Month when prenatal care began, and Number of prenatal consultations).At the same time, important information on access to the Health Center, High risk prenatal care (HRP-NC) examinations and Drugs had 'bad' filling-in.In the information on Care at birth, 12 of the 20 variables (60.0%) had 'excellent' filling-in and 4, 'bad' .And in the last block, only the variable Observation of the family was classified as 'excellent' , with completeness at 100%.However, the variable Repercussions in the family had 'very bad' filling-in (34.8%) (Table 3).
Of the 50 variables of the Summary File that applied to fetal deaths, 41 (82.0%) had 'excellent' filling-in, with 100% completeness in all of them.Only two variables dealing with classification of avoidability of deaths had 'very bad' filling-in (SEADE Foundation and Expanded Wigglesworth), both with 0% completeness.Only the classification of avoidability through intervention by the SUS was filled in.
Mother's level of schooling, which was filled in with excellence on the Home File, had a 'bad' level of filling-in (75.0%) on the Summary File (Table 4).
When comparing the completeness of 13 variables of the Death Certificate, before and after investigation, the contribution of death oversight in redemption of the information from the SIM was observed.The completeness of the information on Mother's level of schooling and Delivery of the body to autopsy was classified as 'bad' before resumption of the investigative  process (79.4% and 62.8%, respectively) and its level of completeness became 'regular' after the process of investigation of the files began (85.3% and 86.3%, respectively).The degree of filling-in of the group of 13 variables before and after their being considered in the investigation was 88.9% and 94.9%, respectively, with the level of completeness going from 'regular' to 'good' .There were rectifications in 13 variables studied in the death certificate (Table 5).Errors of record were found in 11 variables analyzed in the death certificate.The variable Number of stillborn children showed a considerable change in distribution of frequency after the investigation.Before the investigation 64.7% of the mothers had experienced a loss of fetus previously, and after the investigation, 60.8% of the mothers had not.The completeness of this variable changed from 'regular' to 'good' .There was alteration in 49 of the 88 deaths investigated (55.7%) in the information on the number of children who had previously died.Some inconsistencies were also found in the filling-in of the variables of the death certificate, such as: Gestation age equal to zero, two or seven weeks; Number of live births equal to 16 for a mother aged 16; and Caesarean birth for a death that took place in public, in the streets.

Discussion
The proportion of fetal deaths investigated in Jaboatão dos Guararapes was high (86.3%),higher than the percentage observed in Brazil in 2013 -72% for infant and fetal deaths 5 .However, when analyzing the total completed with discussion or even the proportion of deaths investigated within the period of 120 days, one can infer that there were difficulties in the investigative process.
Lack of financial and human resources, lack of knowledge of their duties by the investigating team, and incomplete filling-in of the hospital data have been shown as factors that hinder the investigative process 12 .Similarly, there are evidences that barriers to access to the patient medical records and to the death certificate contributed to the delay in the conclusion of the case within the established period 12 .Also, it is possible that obstacles arise for carrying out of the home interview, due to refusal by the family, change of address or non-existence of the address, as stated in studies on child deaths carried out in the Northeast 13,14 .The time of investigation higher than that specified works against appropriateness of the information for decision on actions for intervention and improvement of care, as well as contributing to abandonment of the investigative process, since at every moment new cases arise that need to be given priority.
The Hospital Form had a higher frequency of filling-in, different from the reality found in another municipality of the Northeast with child deaths in 2009 and 2010 15 .This can be explained by the fact that the majority of stillbirths of residents in Jaboatão in 2014 were born in Recife and in establishments that had Hospital Epidemiology Group which, given the mandatory character  imposed by the Brazilian legislation 3,16 , carried out the investigation and filling-in of the form soon after the occurrence of the death.The Household Form showed a higher degree of filling-in than the Hospital Form.Questions related to inappropriate filling-in or even to incomplete and illegible records on the medical record sheet could have contributed to the worse filling-in of the Hospital Form, compared to the Household Form 13,17,18 .Further, the better completeness of the Home Form is in line with what is found by other authors 15 who refer to the contribution of the family health team in the investigation 13,14 , facilitating the location of addresses and making families sensitive to the importance of the interview.Appropriate filling-in of this form contributed to collection of important information from the Outpatient and Hospital Forms, since their blocks deal with data on the care given in those establishments.
The information on the outpatient data had a lower number of forms filled in (1:8) and also the worst degree of filling-in (61.5%) of the forms analyzed.Caetano, Vanderlei and Frias (2013) also found lower completeness in the Outpatient Form for infant deaths in Arapiraca, in the State of Alagoas 15 .Surveys on quality of medical records in Basic Healthcare Units indicated that information of implications in continuity of care were precarious 19 .
Additionally, the low number of forms for collection of autopsy data filled in indicates that although there is access to the autopsy services the information is not being used to its full potential.However, the data provided through this procedure could contribute to a better understanding of the causes of death.
The summary form was the one that had greatest completeness, but it was found to have been filled in only for eight of the deaths, which  were discussed and, thus, concluded.The low percentage of filling-in of this form points to the difficulties in the consolidation of cases 12,15 .The lack of training of the team and the lack of support from the medical professional for the due corrections in this stage of the investigation are some of the reasons already reported by other investigators 12 .
At present, the municipality has a pediatric professional within the Technical Group for discussion of the deaths, which facilitates the conclusion of the investigations and increases the proportion of deaths discussed.The closing of the cases with discussion is essential for consolidating the analysis of the case 15 , and also for disclosure of the results and submission of the  recommendations to the managers of the various sectors involved 4,12 .Further, it is necessary to incentivate the correction of the information even in the cases in which the deaths were not concluded with investigation.
When analyzing the completeness of the variables of the death certificate before and after the investigation of the deaths, it was observed that the oversight of fetal deaths made a contribution to enhancing the quality of the SIM, both in redemption of information that had been ignored and also in correction of information recorded in the death certificate, or indeed typing mistakes.
The high percentage of alteration in the number of prior stillborn births may be attributed to mistaken inclusion, in the number of losses of fetus reported, of the death that is the subject of the report, itself.According to the instruction manual for filling in of the death certificate, this variable should not include the death to which the document refers 20 .
The variable 'Body sent for autopsy' in the death certificate also showed a high percentage of alteration, and this is similar to what has been found by other authors 7 , and may reflect the small importance given by the medical profession to the filling-in of this field.The other inconsistencies reported may be the result of errors in the typing of the death certificate into the sys-tem or, of doubts on the correct filling in of the death certificate.
These findings are in line with the statement by investigators on the importance of validation of the information of the death certificate by the health oversight 7 and show that analysis of the database before the investigation can result in erroneous interpretation in the distribution of some variables.
Quality of data in the death certificate depends not only on access to health technologies but also on the doctor's comprehension of the dynamic of events that surround the chain of causes of the death and of his commitment to the production of reliable statistics 7 .
The results showed that there were deficiencies and operational difficulties to be overcome in oversight of fetal deaths in Jaboatão dos Guararapes, in 2014.The small number of deaths that were completed with a discussion indicates that the oversight on fetal deaths does not comply with the purpose for which it was proposedsince it is only in the discussion that the problems are identified, and the recommendations for improvements of the care network are made, and avoidability of the deaths analyzed.At the same time, it revealed the contribution of the investigative process in improving the quality of the SIM.
Improvement in quality of the information on fetal deaths will require permanent training of doctors in filling in of the death certificate, and of the teams involved in the operationalization of the SIM and in all the dimensions of the investigative process 7,12,18 .
For this, it is essential that critical thinking should be developed on the role of oversight of death and the importance of complete, reliable and correct and appropriate information for the planning of actions to improve the quality of integrated healthcare for pregnant mothers and births, so as to prevent further deaths and reduce the percentage of stillborn children.

Collaborations
LFB Santa Maria and TVB Araújo state that they contributed to the conception of the project, analysis and interpretation of the data, drafting of the article and final approval of the version to be published.

Table 2 .
Degree of filling-in and completeness of hospital investigation form for fetal death (F2 Form).Jaboatão dos Guararapes, 2014.

Table 4 .
Degree of filling-in and completeness of Investigation form -Summary, Recommendations and Conclusions -of fetal death (IF5 Form).Jaboatão dos Guararapes, 2014.

Table 5 .
Degree of filling-in, completeness and rectification of variables of the death certificate before and after investigation of fetal deaths.Jaboatão dos Guararapes, 2014.