Imprimir Republish

public policy

Risky algorithm

Project in public policy measures situations of violence of the São Paulo population

The vast majority of acts of violence that result in injuries caused by firearms in the city of São Paulo occur at the weekends, mainly on Saturdays. Sexual aggressions towards women are more frequent during the weekdays, with their peak being on Thursdays, when they are a long way from the family, going to work, to school or shopping. Self-aggressions or suicides concentrate themselves around Tuesdays or Wednesdays, suggesting a strong relationship with the situation of work, and as well on Saturdays, a possible motive being anxiety and loneliness. Few take their lives on Fridays.

The average age of the victims is thirty years. This data, which is from 7,073 cases that were attended to between January and December of 2002 at the Jabaquara Hospital – a trauma reference center and with a specialized service for the victims of aggressions -, makes it clear that it is possible to establish risk calculations in cases of violence that demand a hospitalized system – and which therefore imply a risk of death – and to implement prevention policies on a Micro-scale in the city of São Paulo.

“With a little investment it is possible to develop and implement a system of monitoring capable of capturing the information regarding violent incidents and the most common accidents”, says Paulo Saldiva, a professor at the Medical School of the University of São Paulo (USP) and the coordinator of the project Epidemiology of the Violent Criminal in the City of São Paulo: A Study on the Macro and Micro Scales, developed in partnership with the Municipal Secretary of Health, the Secretary of Prison Administration and the Mathematics and Statistics Institute of USP, under the umbrella of the Public Policies Program funded by FAPESP.

Saldiva, a specialist in the study of the effects of atmospheric pollution on human health, decided to make use of the criteria of epidemiological research to study violence, using statistics for risk assessment. “Violence, different from pollution, is not homogeneously distributed. It is necessary to take into account variables of time and space and to create new theoretical techniques to assess risk”, he recognizes. The studies, the same as had happened with the researchers from the Center for Studies of Violence (USP), got bogged down in the difficulties of collecting reliable data in the official organs of public security. The option chosen was to collect information about the theme within the health system.

In the case of the data from the Saboya Hospital, “in spite of the efforts and dedication of the Center for Attention to Violence Victims”, he emphasizes, the analysis of the data was made difficult by the lack of standardization of some important variables. For example, the researchers had to generate some variables such as the classification of the occurrence, taking as a base the information registered in the reason for registration of the complaint and the description of the occurrence. “The quality of the information is an essential item in whatever research study”, he observes. In order to define the classification of the occurrence, the researchers fell back on the International Classification of Diseases, frequency of determined occurrences and specific interests of others, in order to reach the following main categories: accidents, aggressions, legal interventions, medical and surgical complications and events of ignored intention.

The aggressions were classified into self-aggression, sexual aggression, other types of aggression using a firearm, with the use of other bruising instruments, and other aggressions with the use of physical force or non-specified. Standardized and classified, the registers revealed precious information. For example, they indicate that men and women show similar distributions in relation to the age of the victim at the moment of registering the occurrence, a tendency for a drop of violence in the months of January to July, followed by a continuous increase from August to December, and the prevalence of victims of the female sex during this period. In the distribution of the occurrences by age groups and location of the fact, the researchers observed a greater frequency between sixteen and sixty five -ears (9.2%) in violence registered in the work location and a greater incidence of victims in the one to five years age bracket when the aggression occurred in the home itself. In the other occurrence locations – commerce and services areas, public streets and roads, in someone else’s home, not informed and others -, the distribution of the cases was similar in all of the age groups.

Promiscuity and violence
Saldiva’s team also collected data about violence from the Coordination of Prison Health System (Cosaspe in the Portuguese acronym), which brings together all of the records related to hospital attention given to inmates in the city of São Paulo. “Cosaspe did not have any computerized information. Thus the first stage of the work was to develop, in partnership with the Secretary of Prison Administration, the structure of a data bank to make a beginning in digital records and then make an analysis of them”, says Liliam Pereira de Lima, the statistician who participated in the project.

Here also the data was revealing. Violence in the prisons shows itself in the form of illnesses: HIV and tuberculosis, with occurrence of 67.5% and 35.1% respectively in the patients detained in the Hospital Center of the Carandiru Prison, where a large part of the cases of illness and violence occur in the São Paulo penitentiary system. Gun wounds did not surpass 5.2%, and those with a side arm, 1.3%. In both cases, the victims were sent to the hospital by the Secretary of Public Security, responsible for attending to the wounded in the police precincts. These aggressions, nevertheless, could have occurred as the moment of arrest. Violence, in the case of the penitentiary system fails to control contagious infectious diseases. “The prisons are unhealthy and the prisoners live in promiscuity”, emphasizes Saldiva. “The solution is in the reduction of the possibility of contagion and in the distribution dispensing medicines to control these diseases”, he adds.

Risk Algorithms
Having concluded the diagnosis, the research project is now entering into the second phase. “We are going to propose to the penitentiary system a codification for the data and the management of the information based on logical criteria”, Saldiva says. In the area of health, the proposal will be to implement a unified system for the completion of critical information that will feed a data bank on violence in the city.

“Qualified information, for example, will allow us to discover the cost of violence related to medical procedures, the number of lost working hours, among other information relevant to implementation of public policies”, Saldiva says. This data will allow the elaboration of a geo-referenced map in the study of morbidity. However, Saldiva wants to go further with his project and reach an algorithm on risk calculation that could be translated into an alert to society: a woman, walking alone, in the region of such and such a street during the period of X hours until Y hours, has a high percentage of risk to becoming a victim of sexual violence.