## The impact of living on the streets on latency children's friendships

### Impacto da vivência de rua nas amizades de crianças em idade escolar

Luís A. Rohde, Maria Helena M. Ferreira, Andréa Zomer, Letícia Forster and Heloisa Zimmermann
Departamento de Psiquiatria do Hospital de Clínicas de Porto Alegre da Universidade Federal do Rio Grande do Sul. Porto Alegre, RS - Brasil

INTRODUCTION

The importance of friendships to the normal social and emotional development of children has been widely demonstrated 2, 4, 15, 16. A conceptual landmark, widely accepted for understanding the development of peer relations, defines friendship as the progressive capacity to establish peer relations in an autonomous way, based on acceptance, understanding, support and trust. Its main aspects are appraisal of differences, reciprocity and intersubjectivity14.

Friendship is strongly influenced by the child's age13, sex11, cognitive development13, social class3, interaction with siblings8, mental disorder14 and, mostly, by quality of parent-child interaction5, 10, 17. Thus, for example, Park and Waters17 emphasize the fact that patterns of mother-child attachment are closely related to child-best friend dyad performance. Moreover, Russell & Radojevic20 show that the father is fundamental for the development of egalitarian relations at play, a basic condition for peer interaction.

Considering the importance of the family for an adequate development of friendship, reflection on the consequences of a catastrophic increase in the number of abandoned children seems relevant. According to the United Nations Children's Fund 22, there were 30 million street children around the world in 1991. In Brazil, epidemiologic data on street children are either unknown or inaccurate (Forster et al.9, 1992). In Porto Alegre, capital of Brazil's southernmost state, estimated street children prevalence (mostly male) is of about one per 3,000 inhabitants. It has been suggested that socioeconomic problems, such as poverty, hyperinflation and deterioration of social services, have contributed to the increasing numbers of children living on the streets in Brazil6. Rosa et al.19 showed that these children most commonly mention economic necessity as the major reason for becoming street children.

When referring to street children, it is fundamental to establish the difference between children on and of the street. Both carry out their activities on the streets, either selling articles, watching over cars, polishing shoes or begging. The boys on the street, however, go back home at night to sleep with their families, while boys of the street (homeless) have partially or totally broken family ties; they live on the streets, sleep either there or in shelters, and are thus exposed to greater risks 19, 21.

Research involving street children is made difficult by the migratory character of this population and by its deep distrust of unfamiliar adults. In Porto Alegre, a team of researchers is working exclusively with street children, thus some data on this population are known. Forster et al9, in a transversal study with 105 street children between 6 and 18 years old, showed that the main causes for their hospitalization were respiratory illnesses and accidents (being hit by cars). Concerning drug use, the study verified that 58% of the boys of the street smoked cigarettes, 26% consumed alcohol regularly, 42% used inhalants, 26% marijuana and 3% diverse medicines. Theft was an important behavior problem found within this group (22% of boys stole). However, theft within this population could be viewed as a means of subsistence or of maintaining status in the group. Concerning psychiatric diagnosis, DeSouza et al.6 could find no significant differences in the presence of depressive symptoms among a group of street children, one of children living in an orphanage, and a sample of middle-class children living with their families. As for exposure to trauma, many boys and almost all girls living on the streets of Brazil have been raped at least once (Hutz & Foster12).

Studies evaluating exclusively friendship among these children are not found in the literature. The aims of the present study were: a) To evaluate friendship in latency boys of the street (aged 7-11); b) To determine whether latency boys of the streets' friendships differ from friendships of latency boys living with their low income families and having no experience of street life. A priori hypotheses were: a) There are significant differences in friendships between these two groups and latency boys of the street present more problems involving friendships than boys in the other group; b) Latency boys of the street present significantly lower levels of self-esteem, developmental appropriateness and social skills than boys in the other group.

METHOD

#### Sample

Subjects were latency boys, users of "Projeto Girassol", living on the streets of Porto Alegre, and latency boys from Italia elementary public school living with their low income families. Porto Alegre has a population of approximately 1,350,000 inhabitants. "Projeto Girassol" is a government program offering street children recreation, food, medical and odontological care, plus group therapy. Children may use the program from 9 a.m. to 5 p.m., and are free to come and go whenever they wish.

The case group (latency boys of the street) was obtained from "Projeto Girassol" through systematic sampling, according to the following inclusion criteria: boys aged 7-11, living on the streets of Porto Alegre, having either infrequent or no contact with their families. Exclusion criteria were: a) being intoxicated at the time of the interview; b) clinic suspicion of moderate to severe mental retardation, psychosis or organic brain syndrome; c) being physically handicapped.

The control group (latency boys from low income families) comprised all Italia elementary school students fulfilling the following inclusion criteria: boys from low income families, aged 7-11, living in Porto Alegre with their biological parents and at least one sibling. The crite-rion used to define low income was a monthly family inco-me of less than U$80 per family member aged 10 and ol-der and U$ 40 per member under age 10. The criteria for exclusion were the same as those used for the case group.

An a priori definition of the necessary sampling size for this study was impossible, since: a) the street children population of Porto Alegre is not accurately known; b) the relation between study factor (street boys) and the main outcome variable (problems in friendships) has not been determined, as studies on friendship among street children are not to be found in the literature. However, considering that the street child population of Porto Alegre is estimated at from 400 to 500, that a large majority of it are male and that 60% of them range from 11 to 14 years old, the collected sample should be highly representative of Porto Alegre's latency boys of the street population.

#### Instrument

The instrument chosen to evaluate friendship in the two groups was the Cornell Interview of Peers and Friends (CIPF)14. An instrument designed to evaluate friendships in children aged 7-11, it consists of 82 objective questions with dichotomous answers (yes/no), 5 open-ended questions and 3 subjective subscales (Level of developmental appropriateness, Level of self-esteem and Level of social skills). A higher global score (GS) on objective questions means more difficulties in terms of friendships; a higher score on subjective subscales indicates more developmental appropriateness, self-esteem and social skills.

One of the authors of this study (LAR) was first trained in the U.S., by the CIPF devisers, to administer the instrument. Then the instrument was translated into Portuguese and the backtranslation was made independently. The original version and that resulting from backtranslation were compared. A small pilot study was made to test acceptance of the Portuguese version by Brazilian children and its reliability among Brazilian researchers. A fine adjustment of the Portuguese version was performed through consecutive administration of the original and the Portuguese versions to bilingual children at the Panamerican School of Porto Alegre. Next, the instrument was adjusted, in terms of both structure and language, to be used in street children. A last small pilot study was made, involving the boys from case and control groups, to test reliability between the researcher trained in the U.S. and the researcher who would administer the CIPF (AZ). The Kappa coefficient for objective questions was 0.99. The weighted Kappas (Kw) for subscales were: a) Level of Developmental Appropriateness, Kw=0.72: b) Level of self-esteem, Kw=0.60; c) Level of social skills, Kw=0.79.

#### Data Collection

Data collection took place from October 1994 until January 1995. The project was discussed and approved by all the institutions involved. The verbal agreement of all children was requested. Personal identity was preserved.

The boys of the street sample was collected at the "Projeto Girassol"; the boys from the low income families sample were collected at Italia Elementary Public School.

The interviewer administered to the two groups: a) CIPF; b) a questionnaire on personal data and potential confounding factors: age, race, education, length of time on the streets and presence of contact with family (the last two items were considered only for the boys of the street group).

#### Data Analysis

Global Score (GS) comparison between groups was performed through the Student T test. Scores for subjective subscales in the two groups did not show a normal distribution. Thus, non-parametric statistics were utilized in this data comparative analysis (Mann-Whitney U test).

The confounding effects of length of time living on the streets and of contact with family on the boys of the street GS were controlled by the Spearman Rank Coefficient and Mann-Whitney U test, respectively.

The control of other possible confounding factors (age, race, education) for intergroup GS comparison was carried out through Analyses of Variance (ANOVA), using group status, race and education as main effects and age as a covariable.

The comparison of all objective CIPF items between boys of the street and boys from low income families was performed through chi-square test. Since multiple comparisons were performed, a significance level of 1% was fixed to deal with the possibility of Type I error.

RESULTS

The sample consisted of 83 boys, with a loss of 2 (2,4%). The boys of the street group comprised 32 boys, with a loss of 2 (6.2%). The boys group of with a family consisted of 51 boys, with no sample loss. In the group of boys of the street, the reason for sample loss was voluntary withdrawal during CIPF administration.

Demographic characteristics of boys in each group are shown in Table 1. The most outstanding aspect is the clear disparity between the groups.

Table 1 - Demographic characteristics of the two groups.

 Characteristics Boys of the street Boys with family N (%) N (%) Race* White 13 (44.8) 45 (88.2) x2=15.36; p<0.0001 Non-white 16 (55.2) 6 (11.8) Education* Up to 1st grade 21 (72.4) 10 (19.6) x2=19.55; p<0.00001 More than 1st grade 8 (27.6) 41 (80.4) Age Median (range) 10 (8-11) 9 (7-11) U=319.5; p<0.0001

*Lost data for one boy from the case group.

The group of boys of the street had a significantly higher GS on the CIPF compared with the group of boys with a family. The mean for the boys of the street group was 102.00 (SD=5.6; N=30), and the mean for the boys with a family group was 90.3 (SD=6.0; N=51) (t=8.7; p < 0.0001).

ANOVA was carried out taking into consideration group status, race and education as main effects, and age as a covariable. There was no main effect for race, F(1.70)=0,009; p > 0.05, but there was a significant difference between boys of the street and boys with a family, F(1.70)=22.4; p < 0.0001. Also, there was a main effect for education, F(1.70)=7.9; p=0.006. There was no effect for age (covariable), F(1.70)=0.3; p > 0.05. Not of the two-way and the three way interactions was significant. The mean score for boys who attended more than the first year of grade school was significantly lower (more healthy) than for boys who either did not study or attended only the first year of grade school (90.8; SD=6.4 vs 100.6; SD=6.9) (t=6.5, p < 0.0001).

The Mann-Whitney U Test showed that all three subscale scores were significantly different between boys of the street and boys with a family, the first group being consistently more healthy on this bases than the latter (Table 2).

Table 2 - Group Differences for Subscale Scores.

 N Median Range U p Developmental appropriateness Boys of the street 30 3.50 2 - 5.5 252.0 < 0.0001 Boys with family 51 4.50 3.5 - 6 Self-Esteem Boys of the street 30 4.00 3 - 5 263.5 < 0.0001 Boys with family 51 4.50 3.5 - 6 Social skills Boys of the street 30 4.00 3.5 - 5 374.0 = 0.0001 Boys with family 51 4.50 3.5-5.5

The Spearman Rank Coefficient did not show any significant correlation between CIPF global score and length of time living on the streets for the boys of the street group (N=30; rs=-0.08, p=0.69). For this group, the median length of time spent living on the streets was 24 months (range: 5-96 months). Also, the Mann-Whitney U Test showed no significant association between CIPF global score and maintenance of contact with the families for the boys of the street group (N=29: U=72.5, p=0.16). For this group, median length of time since the last contact with the family was less than one month (range: 0-36 months).

Chi-square was utilized to compare the answers to all objective questions between to boys of the street and boys with a family. Table 3 lists only the questions where the comparison between the two groups showed a Yates corrected chi-square equal to or higher than 6,61 (p £ 0,01).

Table 3 - The discriminability of questions for the boys of the street vs. boys with family groups expressed as percentages*

 Question Answer Boys of the Street Boys with family How old are your Same age 25.9 62.5 x2=7.84, p<0.01 friends? Older 74.1 37.5 How long have you had Three or more years 43.3 75.5 x2=6.96, p<0.01 the same friend? Less than three years 46.7 24.5 Do you have a best Up till three 53.3 84.3 x2=7.66, p<0.01 friend? More than three 46.7 15.7 How old is this best Same age 23.1 73.9 x2=15.39, p<0.00001 friend? Older 76.9 26.1 Do you have a girl No 53.3 90.0 x2=12.32, p<0.001 friend? Yes 46.7 10.0 Do you have enemies? No 33.3 74.5 x2=11.61, p<0.001 Yes 66.7 25.5 Have you lost friends? No 51.7 91.8 x2=14.33, p<0.001 Yes, through fights 48.3 8.3 Do you make friends by No 40.0 86.3 x2=16.85, p<0.0001 giving them things? Yes 60.0 13.7 Do you sell things to No 56.7 98.0 x2=19.81, p<0.00001 your friends? Yes 43.3 2.0 Do you fight with Yes 96.7 41.2 x2=22.33, p<0.00001 other kids? No 3.3 58.8 Did you make up yet? Yes 48.3 95.2 x2=10.28, p<0.01 No 51.7 4.8 When you are with your friends do you do No 26.7 80.4 x2=20.62, p<0.00001 things that get you into trouble? Yes 73.3 19.6 Do kids hit you or throw things at you on No 36.7 68.6 x2=6.61, p=0.01 purpose when you havent done anything to start it? Yes 63.3 31.4 Do you lie to your No 36.7 92.2 x2=25.94,p<0.000001 close friends? Yes 63.3 7.8 How is it for you to Easy 36.7 70.6 x2=7.59, p<0.01 make friends? Hard 63.3 29.4 Do you tell your Yes 80.0 45.1 x2=8.07, p<0.01 friends how you feel about them? No 20.0 54.9 Do you share worries Yes 80.0 29.4 x2=17.39, p<0.0001 with your friends? No 20.0 60.6 Do they share worries Yes 90.0 43.1 x2=15.45, p<0.0001 with you? No 10.0 56.9 No matter which friends Yes 70.0 92.2 x2=5.34, p<0.01** you are with, are you pretty much the same person? No 30.0 7.8

* All analyses done with actual numbers (Yates Corrected chi-squares)
** Calculated by Fisher exact test.

DISCUSSION

Child abandonment is a typical phenomenon of underdeveloped countries and, in Brazil, it is a growing trend of the utmost importance in contemporary society. Groups of children living on the streets today constitute a significant portion of a population presenting peculiar and still little known characteristics21. This study sets out to research one aspect of these children's lives, i.e., their friendships. Its clinical importance resides in the fact that the relation between friendship and emotional, social and cognitive child development has been positively established8, 15.

There was a significant difference in the CIPF global score between the two groups. As expected, boys of the street showed more problems in terms of friendships than boys in the other group. Also, boys of the street showed significantly lower scores on the CIPF's subscales of self-esteem, social skills and developmental appropriateness than boys with a family. Thus, living on the streets seems to interfere significantly with the construction of self-esteem and social skills, as well as with appropriate development.

Age, race and education did not seem influence on the difference in CIPF global score between groups. Since there have been reports of the influence of age on children's friendships13, only children at the same developmental stage, i.e., in latency (aged 7-11) have been assessed. Since some influence of social class3 and sex11,14 on friendships has also been reported, the control group of boys with a family was composed exclusively of boys from low income families, and only boys were used in this study. It is important also to stress that the large majority of street children are male.

Education was significantly associated with the GS on CIPF. Lower education seems to be related to greater difficulties in friendship. This finding agrees with that of Keller and Wood's13, which shows a relation between friendship and cognitive development. It is possible, however, that dimensions of friendship more closely related to verbal capacity are privileged by CIPF, which would imply a bias in favour of boys with higher education.

There is no correlation between the CIPF global score and the length of time spent living on the streets for the boys of the street group. As the median length of time spent living on the streets was 24 months, it is improbable that these findings on friendships represent only a transitory state, related to the extreme stresses of leaving home and becoming a street child. It is important to stress, however, how difficult it is to evaluate any measure of time in this group, since street children's concept of time is not always clear. Also, there is no association between the CIPF global score and maintenance of contact with families in the boys of the street group. It was evaluated only if there was contact with the family; neither frequency of contact nor family interactions were assessed. As the median length of time since the last contact with the family was less than one month, boys of the street, although living on the streets, keep some contact with their own families. Apparently, some contact with the family, implying the permanence of bonding, could make boys of the streets better qualified to establish peer relations. However, their original families are generally unstructured. Most boys report that they left home when violence and lack of support became unbearable, and many report having left home for fear of being killed21. In this respect, Forster et al.9, in a study involving 31 Porto Alegre boys of the street, show that 39% of them come from broken homes, 19% have no father and 3% have neither parent. Even more important, 42% report fights and violent discussions occurring at home. These disturbed family interactions probably work as a model for the highly-aggressive relationships which these children establish with their peers on the streets.

Latency boys of the street have more older friends and best friends than boys with a family. Probably, at the age level assessed they look for friends who will give them some kind of protection to make them able to survive on the streets. They have more girlfriends. Early sexuality is a characteristic of this population, probably as an attempt to compensate for the lack of early affective relations with parental figures. They report that it is like trying to establish an affective bond without the flaws observed in their parents' relation21. These children, frequently organized in gangs, have more enemies, maybe due to the intense rivalry existing among these gangs. Friendships are more recent (less than three years), probably resulting from the constant loss of friends, either killed, arrested or institutionalized.

Compared with boys with a family, boys of the street: a) fight more; b) lose more friends through fights; c) lie more to close friends; d) give things to make friends, sell things to get friends and get themselves into trouble with friends more frequently; e) are more frequently a target of friends' unreasonable aggression. Clearly aggression is exacerbated in these children, probably associated with the experience of living in hostile surroundings, where surviving is the main issue. Dinwiddie et al.7 emphasize the fact that the environment in which these children grow and develop is quite favorable to the development of conduct disorders. However, the conduct disorder diagnosis should be applied very cautiously to these children. The DSM - IV1 (1994) stresses that `the Conduct disorder diagnosis should be applied only when the behaviour in question is symptomatic of an underlying dysfunction within the individual and not simply a reaction to the immediate social context." It also points out that such as diagnosis is not necessarily applicable in situations where aggressive behaviour is indispensable to survival. It is also important to bear in mind that these children's first reaction to the possibility of an affective bond tends to be one of aggressiveness calculate to test the real possibilities of that bond (Silva et al21., 1991).

They also have more difficulty in making friends, and make up less frequently, possibly due to the infe-rence of hostile surroundings with peer relations. Surprisingly, however, they tell friends how they feel about them, share worries on friends and have friends share worries with them more frequently. This could be related to the fact that, according to their own reports, these children feel they can count only on each other.

Finally they report, more frequently than boys with a family, that they are not the same person with different friends. This is probably related to a precocious failure in self development and an urge to adapt to the unstable conditions of the streets.

Some limitations of the present study must be emphasized. First, the interviewer was not blind to group status, since such a condition would be impossible, except through the criation of a highly artificial research milieu. Second, the reliability of information obtained from street boys is doubtful, mainly when related to the concept of time. Furthermore, the comparison of all objective questions between groups implied multiple comparisons, increasing considerably the possibility of Type I error. Thus, a fixed for these comparisons was 1%. Finally, no diagnostic screening to detect mental disorders was undertaken. Therefore, the differences in the friendships between boys of the street and the other group may be due to the higher prevalence of mental disorders in the former group. Dinwiddie et al.7 & Rafferty & Shinn18 show an increased prevalence of mental disorders in street children, and Kernberg et al.14 demonstrate an association between mental disorders and friendship difficulties. In Kernberg et al.14 study, 35 items distinguish between children with a mental disorder and normal children. In the present study, 19 items distinguish between boys of the street and boys with a family who were quite probably normal. However, only 7 items were present in both lists. Thus other factors than psychopathology, such as the experience of living on the streets', probably determine difficulties in boys of the streets friendships.

In Brazil, strategies of mental health intervention for the street child population are presently performed mainly for boys on the street. However, the present study demonstrates that boys of the street show significant problems involving friendships, especially greater aggressiveness, as well as diminished self-esteem, social skills and developmental appropriateness.

Knowledge of peer relations in the boys of the street population is clinically fundamental, since individual interventions in this group are not very effective. Obviously, interventions in this group depend on a solid background knowledge as to how these children organize their relationships.

It is always difficult to speculate about the long term consequences of children's stress, problems or disorders. However, there is no doubt that the absence of minimal conditions implies impairments in several areas of streetchildrens' functioning, including friendships. It puts these children at risk for mental disorders, behavioral and drug problems, and most of all for later criminality and social exclusion. One of the main fears expressed by streetchildren, not suprisingly, is reaching adulthood, since at that time they will become legally responsible.

Finally, it is important to stress the need for further research, not only into friendships but also into other relevant issues involving street boys in order to under-stand and help this group in the best possible way.

ACKNOWLEDGEMENTS

To Dr. Kernberg's research team for allowing them to use CIPF in Brazil; To Lidia do Carmo
Nascimento and Jonathan Cohen, Ph.D. for advice on statistics, and Dr. Kernberg, M.D. for reviewing and making suggestions on the manuscript.

REFERENCES

1. American Psychiatric Association. Diagnostic and statistical manual of mental disorders - DSM-VI. Washington, DC: American Psychiatric Association, 1994.

2. Bierman, K.L.; Smoot, D.L.; Aumiller, K. Characteristics of aggressive-rejected, aggressive (nonrejected) and rejected (nonaggressive) boys. Child Dev., 64, 139-51, 1993.

3. Bjoernsson, S.; Edelstein, W.; Kreppner, K. Explorations in social inequality: stratification dynamics in social and individual development in Iceland. Berlin, Max Planck Institute for Human Development and Education, 1977 (Studien und Berichte, 38).

4. Bullock, J.R. Children without friends: Who are they and how can teachers help? Child Educ., (Winter): 92-7, 1992.

5. Cassidy, J.; Parke, R.D.; Butkovsky, L. Braungart, JM. Family-peer connections: within the family and children's understanding of emotions. Child Dev., 63: 603-18, 1992.

6. Desouza, E.; Koller, S.; Hutz, C.; Forster, L. Preventing depression among Brazilian streetchildren. Interam. J. Psychol., 29: 261-5, 1995.

7. Dinwiddie, S.H.; Reich, T.; Cloninger, C.R. Solvent use and psychiatric comorbidity. Br. J. Addic, 5: 1647-56, 1990.

8. Dunn, J. & Mcguire, S. Sibling and peer relationships in childhood. J. Child Psychol. Psychiat., 33: 67-105, 1992.

9. Forster, L.M.K.; Barros, H.M.T.; Tannhauser, S.L.; Tannhauser, M. Meninos de rua: relação entre abuso de drogas e atividade ilícitas. Rev. ABP-APAL 14: 115-120, 1992.

10. Hart, C.H.; Dewolf, D.M.; Wozniak, P.; Burts, D.C. Maternal and paternal disciplinary styles: relations with preschoolars' playground behavioral orientations and peer status. Child Dev., 63: 879-892, 1992.

11. Hartup, W.W.; French, D.C.; Laursen, B.; Johnston, M.K.; Ogawa, J.R. Conflict and friendship relations in middle childhood: behavior in a closed-field situation. Child Dev., 64: 445-54, 1993.

12. Hutz, C.S. & FORSTER, L. Comportamento e atitudes se-xuais de crianças de rua. Psicol. Refl. Crit., 9: 209-29, 1996.

13. Keller, M. & Wood, M. Development of friendship reasoning: a study of interindividual differences in intraindividual change. Dev. Psychol., 25: 820-6, 1989.

14. Kernberg, P.F.; Clarkin, A.J.; Greenblatt, E.; Cohen, T. The Cornell interview of peers and friends: development and validation. J. Am. Acad. Child Adolesc. Psychiatry, 31: 483-9, 1992.

15. Newcomb, A.F.; Bukowsky, W.M.; Pattee, L. Children's peer relations: a meta-analytic review of popular, rejected, neglected, controversial and average sociometric status. Psychol. Bull., 113: 99-128, 1993.

16. Ollendick, T.H.; Weist, M.D.; Borden, M.C.; Greene, W. Sociometric status and academic behavioral and psychological adjustment: a five-year longitudinal study. J. Consult. Clin. Psychol., 60: 80-7, 1992.

17. Park, K.A. & Waters, E. Security of attachment and preschool friendships. Child Dev., 60: 1076-81, 1989.

18. Rafferty, Y. & Shinn, M. The impact of homelessness on children. Am. Psychol., 46: 1170-9, 1991.

19. Rosa, C.S.; Borba, R.E.S.; Ebrahim. G.J. The street children of Recife: a study of their background. J. Trop. Pediatrics, 8: 34-40, 1992 .

20. Russell, G. & Radojevic, M. The changing role of Fathers. Current understandings and future directions for research and practice. Infant Ment. Health J., 13: 296-311, 1992.

21. Silva, A.S.; Gorayeb, R.; Lebensztojn, B.; Kosin, M.D.; Snitcovsky, M. Observação sobre o universo mental das crianças de rua. Rev. ABP-APAL, 13:85-96,1991.

22. Unicef. The state of the world's children. New York. Oxford University Press, 1991.

Correspondence to: Luís A. Rohde - Av. Ramiro Barcello, 2350 - 90035-007 Porto Alegre, RS - Brasil - E-mail: lrohde@zaz.com.br
Submitted on 28.7.1997. Approved on 10.3.1998.

Faculdade de Saúde Pública da Universidade de São Paulo São Paulo - SP - Brazil
E-mail: revsp@org.usp.br