Analysis of the completeness and consistency of records of violence against indigenous women in the health macro-region of Dourados, Mato Grosso do Sul state, Brazil, 2009-2020

Abstract Objective: To analyze the temporal trend of completeness and consistency of data on notifications of violence against indigenous women in the health macro-region of Dourados, state of Mato Grosso do Sul, Brazil, between 2009 and 2020. Methods: An ecological time series study was conducted using data from the Notifiable Health Conditions Information System; Prais-Winsten regression was used to analyze the trend of data completeness and consistency, as well as the proportion of completed and coherent fields. Results: A total of 2,630 cases were reported; completeness was found to be very poor in the variable “occupation” (48.9%) and poor in the variables “schooling” (68.3%) and “time of occurrence” (67.9%); in the analysis of temporal trends, only the variable “occupation” showed a decreasing trend (p = 0.045). Conclusion: The data analyzed demonstrated the need for improvement in the completeness of the variables “schooling”, “occupation” and “time of occurrence” of the violent act.


INTRODUCTION
Violence is a historical and social phenomenon present in humankind.Considered a health problem, it demands specific public policies, organization of practices and services, given its power to affect individual and collective health. 1 According to the World Health Organization's (WHO) World Report on Violence and Health, violence can be prevented and its impacts reduced through strategies such as promoting gender equality, restricting the availability and reduction of the harmful use of alcohol, as well as access to firearms, in addition to changing cultural and social norms that favor them. 2 In order to reduce and prevent cases of violence, it is necessary to implement a victim assistance network, with a multidisciplinary approach that intervenes in cases and prevents recurrences. 2cording to the WHO, one in three women worldwide has experienced some form of physical or sexual violence by an intimate partner or sexual violence by a non-partner. 3n Brazil, according to occurrence bulletins and femicide qualif ication by civil police stations, in the first half of 2022, nearly 700 women were victims of this fatal form of violence, constituting four murders every day, on average.Compared to data from the first half of 2019, released by the Brazilian Forum of Public Security, the increase in femicide in the country was 10.8% in 2022. 4 the context of the indigenous population, indigenous women are vulnerable to some form of violence due to increased contact with the urban environment, which has changed their traditional social and economic structure and contributed to the adoption of new habits and, consequently, tumultuous gender relations. 5A descriptive report, based on data from Public Safety Canada, assessing 516 homicides that occurred in 2014, showed that indigenous women in that country were 6.5 times more likely to be killed than nonindigenous women. 6 the state of Mato Grosso do Sul, located in the Midwest region of Brazil, high population density in some indigenous lands, conflicts over land ownership, contact with the nonindigenous population, the degradation of the original landscape, as well as the introduction of illicit drugs and alcohol, have contributed to the occurrence of violence against Guarani and Kaiowá women. 5 addressing violence against women, it is the responsibility of the health services to provide care and compulsory notif ication of these cases by health professionals, in a specif ic form designed for recording this violence.Once f illed out and entered into the Notifiable Health Condition Information System (Sistema de Informação de Agravos de Notificação -SINAN), this form generates data

Main results
There was a progressive increase in notifications over the years.Most of the variables showed regular or excellent completeness and consistency.In the analysis of temporal trend, only the "occupation" variable showed a decreasing trend.

Implications for services
Care for victims of violence is part of the daily routine of health services, and it is essential for health professionals to provide adequate compulsory notification for a comprehensive understanding of the victims' profile, thus assisting in addressing this issue.

Perspectives
Further studies are needed to understand the factors associated with violence against indigenous women, which could help the development of health promotion actions and violence prevention strategies targeting these women.
for planning public policies aimed at preventing and reducing these occurrences.However, the invisibility of violence against women in health services means that many cases go unreported, or when they are, the reports have several unfilled fields. 7ere is a scarcity of epidemiological studies that evaluate the SINAN database on violence against indigenous women.Thus, the completeness of a variable, characterized by the degree to which it is filled with a non-null value, and its consistency, defined by its coherence in relation to another variable, contributes to better surveillance of this health condition. 8,9e study aimed to analyze the temporal trend of completeness and consistency of data on notif ications of violence against indigenous women in the health macro-region of Dourados, state of Mato Grosso do Sul, Brazil, between 2009 and 2020.

METHODS
This was an ecological time series study that assessed the temporal trend of the degree of completeness and consistency of cases of violence against indigenous women aged 10 years and older, living in the health macroregion of Dourados, state of Mato Grosso do Sul.The data evaluated were reported on SINAN from 2009 to 2020.
The health macro-region of Dourados is located in the southern region of the state of Mato Grosso do Sul and borders Paraguay.In 2022, this macro-region comprised a population of nearly 1 million inhabitants, spread across 33 municipalities, 10 and approximately 60 thousand people living in indigenous lands, predominantly from the Guarani and Kaiowá ethnic groups, according to the report of the Special Indigenous Health District of Mato Grosso do Sul (Distrito Sanitário Especial Indígena do Mato Grosso do Sul -DSEI/MS) for 2023.Primary Health Care (PHC) for this population is provided by the DSEI/MS through health centers in their respective territories. 11e unidentified records from SINAN were provided by the state Center for Strategic Information in Health Surveillance (Centro de Informações Estratégicas de Vigilância em Saúde -CIEVS) in September 2022.SINAN is a national health information system, decentralized to municipalities, where data on compulsorily notifiable health conditions are entered.Notifications of violence are carried out by healthcare professionals and forwarded to municipal epidemiological surveillance services, which are responsible for entering the data into the system.
The analysis of the SINAN database on violence against indigenous women was evaluated for its completeness and consistency.Completeness is characterized by the degree of completion of the variable under analysis, measured by the proportion of notifications with categories other than those indicating the absence of data: 8 fields of variables classified as "unknown" or blank were considered unfilled.Consistency, on the other hand, can be defined as the proportion in which the relationship between two selected variables presents coherent values.An example of inconsistent data is the notification of violence against an eight-year-old child with complete high school reported in the notification.Consistency was analyzed only for the completed variables.The percentage was obtained by dividing the number of consistent relationships divided by the total number of relationships between the considered variables.
The notification years were grouped into biennia: 2009-2010; 2011-2012; 2013-2014; 2015-2016; 2017-2018; 2019-2020.In the analysis of completeness and consistency, the percentage of correctly filled variables (for each period) and the total percentage were calculated.For completeness, the following categories were considered: excellent, greater than or equal to 95.0%; good, from 90.0% to 94.9%; regular, from 70.0% to 89.9%; poor, from 50.0% to 69.9%; and very poor, less than or equal to 49.9%. 8or consistency, the considered categories were: excellent, greater than or equal to 90.0%; regular, from 70.0% to 89.9%; and poor, less than 70.0%. 12e annual percentage change (APC) was calculated using Prais-Winsten linear regression, with the biennium as the independent variable (y) and the percentage of completeness as the dependent variable (x). 13A significance level of 5% was used in statistical tests and a 95% confidence interval (95%CI) was calculated.A decreasing trend was considered when the 95%CI showed negative values, an increasing trend was considered when the 95%CI showed positive values, and stability when the 95%CI showed both positive and negative values. 14The analyses were performed using the R statistical software, version 4.1.2.

RESULTS
Between 2009 and 2020, a total of 2,630 notifications of violence against indigenous women were recorded in the health macroregion of Dourados, state of Mato Grosso do Sul, all of which were analyzed in this study.There was an increase in the number of notifications in the period from 2009 to 2020, ranging from 54 in the first biennium (2009-2010) to 833 in the last biennium (2019-2020); as well as an increase in the number of notifying institutions, from 5 in the first biennium to 37 in the last biennium (Figure 1).Women aged 10 to 19 years accounted for the majority of registrations for the entire period (33.0%), as shown in Table 1.
The lowest completeness was recorded for the variable "occupation" (48.9%), classified as very poor.The variables "schooling" (68.3%) and "time of occurrence" (67.6%) showed poor completeness.The highest degrees of completeness, classif ied as good, were observed for the variables "physical violence" (93.5%), "number of individuals involved" (93.6%), "sex of the perpetrator" (90.5%), "prophylaxis for STDs" (90.0%) and "blood collection" (90.0%).Completeness was considered regular for 42 variables.As for the analysis of temporal trend, only the variable "occupation" showed a decreasing trend in the period 2009-2020 (p = 0.045); for 31 variables, the temporal trend was increasing; and for 18 variables, the trend was stable (Table 2).
Regarding consistency, the relationships between (i) reproductive age (10 to 49 years) and pregnancy status (not applicable), (ii) psychological violence (yes) and the types of aggression (excluding threats) and (iii) the sex of the perpetrator (male) and the relationship (mother) showed regular consistency.Excellent consistency was observed in the relationship between (i) disability/disorder (no) and type of disability/disorder (yes), and between (ii) sexual violence (yes) and type of sexual violence (unknown) (Table 3).

DISCUSSION
The results of this study showed an increase in notifications over the 12 years selected, from 2009 to 2020.The percentage of completeness was classified as regular for the majority of the variables considered.In the analysis of the temporal trend of completeness according to biennia, only the variable "occupation" showed a decreasing trend.Consistencies remained from regular to excellent quality.During the

ORIGINAL ARTICLE
Analysis of the SINAN database quality       The increase in the number of notifications may be related to the fact that, as of 2010, violence was included in the compulsory notification list of diseases in sentinel health units, in accordance with the Ministry of Health Ordinance MS/GM No. 2,472, dated August 31, 2010; and from the following year, 2011, with the publication of Ordinance MS/ GM No. 104, dated January 25, 2011, violence became part of the list of National Compulsory Notification List of Diseases. 15,16With the creation of the Special Secretariat for Indigenous Health (Secretaria Especial de Saúde Indígena -SESAI), the organization of the Indigenous Special Health Districts (Distritos Sanitários Especiais Indígenas -DSEIs) and the inclusion of multi-professional healthcare teams, two factors would have contributed to the increase in notifications: (i) increased efforts for violence notifications within DSEIs, since 2013, and (ii) the publication of Ordinance No. 1,271, dated June 6, 2014, which was responsible for the implementation, in the same year, of the SINAN notification units in the indigenous territory. 17,18e increase in the number of notifications may be related to the increase in the number of notifying units registered with the National Health Establishment Registry (Cadastro Nacional de Estabelecimentos de Saúde -CNES).A study conducted in the state of Rio de Janeiro, between 2009 and 2016, aimed at analyzing the consistency of 103,841 notifications of interpersonal and self-inflicted violence, found a 284% increase in notifications over the period, an increase associated with better structuring of notifying units. 19In Belém, capital city of the state of Pará, there was also an increase in notifications of violence alongside the increase in the number of notifying units between 2009 and 2011. 20Therefore, the increase in the number of notifying units, associated with a better structure, staffed with trained professionals who are sensitized to the issue of violence, may contribute to the increase in notifications.
Taking into consideration the target population group of this study, the highest number of notif ications occurred among women aged 10 to 19 years.In the general population, one of the factors that can contribute to the increase in notif ications of violence against this age group is the awareness of professionals for their reporting; among indigenous women, specifically within the context of the ethnic groups studied, the first menstrual period (menarche) marks a rite of passage where the child becomes a woman, allowing for intimate-sexual relationship and childbearing. 21As a result of this new status, indigenous women aged 10 to 15 years are subject to potential violence perpetrated by intimate partners: boyf riends, spouses, ex-boyf riends, ex-spouses.In this present study, 409 indigenous pregnant women were reported as victims of violence in this age group, with physical violence being the most frequent, with 226 cases, followed by sexual violence with 141 cases.Of the total number of pregnant women, 258 indigenous women (63.1%) were between 11 and 15 years of age (data not shown).
With regard to the occurrence data block, the variable "physical violence" showed excellent completeness, the time of the occurrence showed poor completeness, and the other variables showed regular completeness.Collecting information on notif ications of violence, such as the place and time of occurrence, type of violence and whether there was recurrence, is important for mapping cases and providing security and protection actions for victims. 22According to a study conducted in the city of Amambaí, also in the state of Mato Grosso do Sul, between 2007 and 2013, among the Guarani and Kaiowá ethnic groups, it was common for reports to be made to institutions by close relatives or village leaders, not by the victim himself. 5However, this trend of reports being based on secondary sources

ORIGINAL ARTICLE
Analysis of the SINAN database quality may contribute to the reduced completeness of information about some variables.On the other hand, the high completeness of the variable "physical violence" may be associated with the fact that it is easily recognizable, due to its visible marks on the victims.
Health services are essential for monitoring v i o l e n c e a n d s h o u l d b e p a r t o f t h e comprehensive care and protection network for people experiencing violence, involving Social Assistance, Education, Public Security, the Judiciary, the Guardianship Council and civil society.Among the indigenous population, for the referral of victims to social protection agencies in order to resolve cases of violence, in addition to the aforementioned agencies, it is important to involve the sector responsible for indigenous policy, social control, and indigenous associations and collectives. 23reover, concerning indigenous people, the low completeness identif ied in the evolution and referral block of cases may be related to professionals' unpreparedness to handle violence cases in this population, either due to language barrier or due to the lack of professional training for intercultural encounters; or even for the difficulty indigenous women face in accessing other services in the network.such as the Judiciary or Public Security, given the challenging access to some indigenous lands and reserves.According to the Violence Mapping, conducted by the Guarani and Kaiowá indigenous women's collective in the southern region of Mato Grosso do Sul, many indigenous women experience domestic violence but not all of them receive support and/or shelter. 24Women do not report such violence because they have nowhere to go, do not know how to start over or have no prospects other than their ethnic-social situation limited horizon.Those who receive the necessary support and shelter still depend on the transportation provided by SESAI, the National Foundation for Indigenous Peoples (Fundação Nacional dos Povos Indígenas -FUNAI) and the Guardianship Council, among others. 24ferring victims of violence to other services in the protection network is part of the process of comprehensive care, contributing to the humanization of care. 25When providing care, it is the responsibility of healthcare professionals to detect possible violence, provide support, care, notification and referral of the victim to other services in the protection network. 25The incompleteness of these data makes it difficult to analyze the care flow in the protection network: its absence can either be the result of not completing the specific field -even if the woman has been referred to the protection network -or can be related to the victim not being referred to other services.
Currently, indigenous peoples have limited access to comprehensive and specialized health care.Most of the institutions are located in cities, far f rom indigenous lands and reserves, making referral and continuity of care difficult. 26,27In addition to the distance from the services of the women's protection network, this population faces challenges in complying with the legal requirements related to violence: police officers, when requested, are not always able to be present in the indigenous territory; and in certain situations, it is women who cannot communicate with the police force because the territory lacks telephone signal. 24e consistency analysis demonstrates the coherence of information in the notifications of violence against indigenous women, allowing for data collection through the association between selected variables.A similar result was observed in a study conducted in Recife, the capital city of the state of Pernambuco, between 2009 and 2012, evaluating the completeness, consistency and duplicity of violence notification records on SINAN. 28factor that can compromise the notification of violence against women is the perception among professionals that domestic violence is exclusively related to Public Security and the Judiciary, with no need for involvement from Public Health. 29Another relevant point to take into consideration is the feedback to the notifiers by the epidemiological surveillance team.Feedback on the profile of reported cases, data quality and its implications can facilitate the improvement of information quality, as professionals responsible for reporting violence, upon realizing its effectiveness and results, may feel more motivated. 28Furthermore, in the care of indigenous peoples, sociocultural differences between healthcare professionals and indigenous victims of violence may hinder the process of reporting cases.
In order to improve the quality of completion of the notif ication form, it is important to provide continuing education for the responsible professionals, sensitizing them to the importance of generating quality information. 30The lack of notification channels in places where services are provided and where it is consolidated negatively influences data quality.In addition, many data are obtained from medical records, which sometimes may not contain all the necessary information. 30limitation of this study is the use of unidentified SINAN data in the assessment of completeness and consistency, which makes it impossible to observe duplicate notifications.Another limitation to highlight is the absence of stratification of indigenous peoples by ethnicity in the violence notification/investigation form, which makes it impossible to analyze this variable in identifying the ethnic group most vulnerable to violence, which would contribute to targeting specific actions according to each group.This study was crucial in identifying the need to improve the quality of notifications of violence against indigenous women in Dourados, Mato Grosso do Sul.However, the ethnic and cultural diversity of the Brazilian indigenous population, as well as the regionalization of health service management, prevent the extrapolation of the findings to other Brazilian regions.Nevertheless, the study serves as a guidance for further investigations into the quality of reported data in other indigenous territories.
It can be concluded that the data evaluated in this study are adequate for making inferences and epidemiological analysis of cases of violence in the region, although improvements in the completeness of the variables "schooling", "occupation" and "time of occurrence" are needed.In order to obtain consistent and highquality data, it is necessary for the Ministry of Health and the SESAI, along with municipal and state health departments, to adopt measures aimed at improving the completion of violence notification/investigation form, such as continuing education for professionals, training on how to deal with intercultural situations, raising awareness about the importance of correctly filling out the forms, epidemiological surveillance actions through case investigation, and periodic analysis of databases to identify inconsistencies and incompleteness.Generating quality data can contribute to assessing the magnitude and determining violence against indigenous women in the health macro-region of Dourados, Mato Grosso do Sul, as well as assisting in the development of public policies addressing this issue.

Figure 1 -
Figure 1 -Distribution of the number of notifications of violence against indigenous women in the Notifiable Health Conditions Information System (SINAN) and institutions that reported violence, health macro-region OF Dourados, Mato Grosso do Sul state, Brazil 2009-2020

Table 2 -
Number, percentage and temporal trend of completeness of fields in the notification/investigation form for violence against indigenous women aged 10 years and older, health macro-region Dourados, Mato Grosso do Sul state, Brazil,

Table 2 -
Number, percentage and temporal trend of completeness of fields in the notification/investigation form for violence against indigenous women aged 10 years and older, health macro-region Dourados, Mato Grosso do Sul state, Brazil, 2009-2020 APC: Annual percentage change; b) 95%CI: 95% confidence interval; c) P-value: P-value, estimated by means of the Prais-Winsten regression; d) STD: Sexually transmitted disease; e) HIV: Human immunodeficiency virus; f) CRAS (Centro de Referência de Assistência Social): Social Assistance Reference Center; g) IML (Instituto Médico Legal).Forensic Medical Institute.
Research Ethics Committee of the Escola Nacional de Saúde Pública Sergio Arouca/ Fundação Oswaldo Cruz (CEP/ENSP/Fiocruz), Opinion No. 5,274,177; and by the National Research Ethics Committee/National Health Council (Comissão Nacional de Ética em Pesquisa/Conselho Nacional de Saúde -CONEP/CNS), Opinion No. 5,469,695.Mato Grosso do Sul State Health Department (SES/ MS) signed the consent letter for this research to be conducted.

Table 2 -Number, percentage and temporal trend of completeness of fields in the notification/investigation form for violence against indigenous women aged 10 years and older, health macro-region Dourados, Mato Grosso do Sul state, Brazil, 2009-2020
Epidemiologia e Serviços de Saúde, Brasília, 33:e20231075, 20248ORIGINAL ARTICLEAnalysis of the SINAN database quality

Table 3 -Number and percentage of consistency of variables selected in the notification/investigation form for violence against indigenous women aged 10 years and older, according to biennium, health macro-region of Dourados, Mato Grosso do Sul state, Brazil, 2009-2020 period
, there was an increasing trend in the completeness for 31 variables.