Under registration of occupational accidents in Brazil, 1997



Maria Cecília Pereira Binder; Ricardo Cordeiro

Faculdade de Medicina de Botucatu da Universidade Estadual Paulista. Botucatu, SP, Brazil





OBJECTIVES: To estimate the number of occupational accidents that occurred in a certain municipality during a specific period of time as well as the extent of sub-registration.
METHODS: The study sample was comprised of 4,782 households within the municipality of Botucatu, São Paulo occupied by a total 17,219 inhabitants on the 1st of July, 1997. In each household, an adult inhabitant was interviewed in order to identify the occurrence of occupational accidents in the three months preceding the interview. When such occurrences were identified, the injured workers were interviewed. The Cochran formula was utilized to calculate the confidence interval.
RESULTS: Seventy-six individuals confirmed that they had suffered occupational accidents during these three months. In 1997, there were approximately 1,810 occupational accidents in Botucatu, according to our estimates, and the incidence of work related injuries in the population was approximately 4.1% (CI 95% 3.0%-5.3%). Thirty-nine (51.3%, CI 95% 41.1%‑61.6%) of the above 76 workers were not covered by the Social Security System. Consequently, their injuries were not reported for there was no legally binding obligation to fill out and emit the official registration form Comunicação de Acidente do Trabalho (CAT) [the work injury report]. Included among the latter are civil servants and informal sector workers, such as self-employed, casual workers and others. Although the remaining 37 workers (48.7%) were covered by the Social Security System and emission of the work injury report was obligatory, 20 of the cases (54.1% CI 95% 39.4%-68.7%) had not been registered. A greater proportion of cases of sub-registration were found among those workers employed in micro, small and medium sized businesses than among those working in large firms. Only 22.4% (CI 13.8%-30.9%) of the occupational accidents reported in this study were registered by the Social Security System.
CONCLUSIONS: Research findings confirm that analyses based on the number of officially registered workplace accidents are limited. It highlights the importance of utilizing other sources of data, besides the registration form for occupational accidents used currently by the Social Security System, Comunicação de Acidente do Trabalho or CAT, in order to elaborate official statistics on occupational accidents.

Keywords: Accidents, occupational. Epidemiology. Underregistration. Occupational accidents registry.




Occupational accidents are socially determined phenomenon, indicative of the intense exploitation to which workers are, in large part, submitted. They constitute an important public health problem in Brazil, which mainly effects young adults and causes a large number of cases of permanent disability and deaths.4,6,14 According to Brazilian social security legislation,11 an occupational accident is an accident which occurs in the course of or for the purpose of work for a firm and results in personal injuries or functional disturbances that cause death, or loss, or a temporary or permanent reduction of the victim’s ability to perform the normal duties of work in the job or post occupied when the accident occurred. Accidents occurring on the habitual route, in either direction, between the place of work and the worker’s residence, are also considered occupational accidents.

Official Brazilian statistics of occupational accidents, considered precarious by several authors,1,4,10,12 are based on information registered in a document entitled Comunicação de Acidente de Trabalho (CAT) [Occupational Accident Report], which was devised by the Social Security System, for insurance purposes. These statistics exclude information concerning civil servants, military statutories, informal sector workers, autonomous workers who have social security, domestic servants, and proprietors. According to the Plano de Benefícios da Previdencia Sócial13[Social Security Benefits Plan], these workers are not covered by occupational accident insurance, and the employers are not obliged to emit this document.

For employees whose accidents are covered by occupational accident insurance, the flux of the CATs until these are registered by the Instituto Nacional de Seguridade Social (INSS) [National Institute of Social Security] depends upon the following factors:

  • In large part, the voluntary emission of this form by the employer, although, theoretically, it would be possible, in case the employer didn’t do so, for any other interested party to emit it, including: the victim of the accident, family members, employee’s union, physician who attended the victim, or any other public authority;11,12
  • A medical declaration, mentioned in item II of the CAT form, filled out by the physician attending the victim.

The form being forwarded to the INSS agency within the area where the accident occurred.

According to official statistics, there has been a progressive decrease in the incidence of occupational accidents since 1976. Initially, this tendency was influenced by modifications in the social security legislation: the period of exemption from insurance payments increased from 2 to 15 days, which entailed an increase in the sub registration of minor accidents; and inclusion of domestic servants and autonomous workers in Social Security’s General Regiment, although this measure did not grant theses last two categories occupational accident insurance.6,13 Carmo et al4 indicate that the extinction of differential payment to private hospitals for attending workers suffering from occupational injuries, and the expansion of assistance within the public services, which occurred from the mid- eighties onward, are factors that have contributed towards the sub registration of these events. More recently, explanations for the observed trend of increasing under registration have included processes of productive restructuring, the introduction of new technologies, and retraction of the secondary sector with a concomitant expansion of the tertiary sector of the economy.15,16

It is believed that, in Brazil, there is a significant degree of sub registration of occupational accidents, particularly of those accidents which are not very grave, and in less developed regions of the country, of serious accidents as well.1, 4,6 Recently, a study conducted in several municipalities in the interior of the State of Sao Paulo with more than eighty thousand inhabitants, revealed a 42% rate of sub registration of occupational accidents among workers employed in the formal sector of the economy, and a 71% rate of sub registration for the economically active population at large.2

Confronted with the precarious quality of the data concerning occupational accidents, the Associação Brasileira de Pós-Graduação em Saúde Coletiva* [Brazilian Association of Graduate Studies in Collective Health] recommended, in 1996, that efforts should be made in order to measure the scope of occupational accidents in Brazil and to characterize them with greater precision.

In consonance with this recommendation, the objectives of this study were: to estimate the number of occupational accidents which occurred in a specific municipality, independently of the individuals’ insertion in the labor market; and to estimate the extent of sub registration of these events among workers whose accidents, according to the Plano de Benefícios da Previdencia Social [Social Security Benefits Plan], should have been notified and registered through the emission of a CAT, the official occupational accident form.



This study was conducted in 1997, in a sample population, within the municipality of Botucatu, São Paulo, inhabited, at the time, by approximately 102,494 people.

The sampling process, described by Cordeiro,8 was conducted in June and July of 1997. The first of July, 1997 was adopted as the date of reference and the 1996 Population Count was utilized as a source of information. According to this source, among the 33,900 residential households existing in the municipality in 1996, 31,604 were within the urban area and 2,296 were in the rural area.

In the urban area, 4,454 households were selected by means of random systematic sampling of conglomerates. In the rural area, 328 households were selected by means of convenient conglomerate sampling. A total of 17,219 inhabitants were included in both samples.

All the households within the samples were visited by trained interviewers and 10% of these were visited once again by the field coordinators in order to detect possible inaccuracies, systematic errors or attempted frauds committed by interviewers.

In the urban area, permanently vacant households were replaced by those which followed them directly and households which were closed on the first visit were visited two more times, on different days of the week and at periods of the day differing from that of the initial visit. If no one was found at home during these three visits, the household was excluded from the sample and was not replaced.

The interviewers asked an adult inhabitant (aged 18 years or more) how many individuals were living at that address at that time and if, in the last 90 days, any of them had had any kind of accident in the course of work. If there was a positive reply, the accident was confirmed by means of a questionnaire. The latter contemplated the following items: identification, occupational status (proprietor, autonomous worker, permanent or temporary employee, whether or not the worker is registered as an employee, civil servant specifying the kind of contract and working registration, domestic servant and others), name of the firm, concise description of the accident, incapacity expressed in calendar days of absence from work or leave of absence, type and location of injury, medical assistance received, whether or not the CAT was emitted and whether the victim received social security benefits.

The total number of occupational accidents in the municipality of Botucatu in 1997 was estimated by multiplying the proportion of victims of accidents in the workplace that occurred within 90 days preceding the interviews amongst the residents of the domestic units within our study sample by the total population of Botucatu in July 1997. The confidence interval was calculated by estimating the sample variance of the estimator of the proportion of occupational accident victims in the conglomerate sample according to the formula developed by Cochran:5

f = sample fraction
n = number of households within the sample
m = median of residents per household
ai = number of occupational accidents which occurred in the previous 90 days within the i household
p = proportion of workers within the sample who had occupational accidents within the last 90 days
mi = number of residents in household i

In order to estimate the number of occupational accidents that occurred in 1997, it is assumed that there is a homogeneous annual distribution of occupational accidents in the municipality of Botucatu and the estimate made for the 90 day period was then multiplied by 365/90.

Information concerning the emission of CATs obtained in the interviews was confronted with information available in the data bank generated by means of these documents, as described by Binder et al. 3

In order to estimate sub registration of occupational accidents, the domestic unit was considered the sample unit of interest and the ratio between the observed number of occupational accidents, verified by means of the domestic visits, that should have been registered obligatorily in the Social Security System (INSS) and the number of accidents that were in fact notified..

Cochran’s formula5 was utilized in order to establish the confidence interval:

f = sample fraction
n = number of households within the sample
x = median of occupational accidents of mandatory notification per household within the sample
yi = number of occupational accidents of mandatory notification not notified within the i household
R = sample ratio of occupational accidents of mandatory notification not notified and of occupational accidents of mandatory notification.
xi = number of occupational accidents of mandatory notification found within household i

The data was processed utilizing the 6.0 version of the Epi-Info program.



Visits to the domestic households within the sample revealed that there were 776 permanently vacant households, all of which were substituted; 111 households were closed during the three stipulated visits, implying in a loss of 2.3% of the sample; there were eight refusals, which implied in the exclusion of the household without replacements, which signified a loss of 0.2% of the sample.

The estimated number of permanently vacant households in the municipality was 5,432. This number was obtained by multiplying the total of 776 unoccupied households in the sample by seven because one in seven houses was included in the sample. Thus, the number of inhabited households was estimated in 28,468 (33,900-5,432). 4,782 of these households were visited, which corresponds to the sample fraction of 0.680.

During the initial interviews, the information obtained from the adult residents which answered the questionnaires revealed that 141 occupational accidents had presumably occurred in the last 90 days. In 14 cases, the subsequent interview could not be undertaken for the following reasons: the person was not at home in at least three scheduled visits (eight cases); the interviewee was not located because he/she moved (two cases); the person refused to be interviewed (three cases); and the person died after the first visit and the interviewer’s scheduled return visit to the household (one case).

Thus, 127 individuals were interviewed, of which: 19 denied they had suffered any kind of accident; 3 revealed they had had an accident which was in no way related to work; 6 stated they were victims of occupational diseases; 23 confirmed they had had occupational accidents in the past, in a period prior to the one defined by the study that varied from 7 months to eight years beforehand; 76 confirmed they had had occupational accidents during the study period, being that 64 were typical accidents and 12 were commuting accidents.

The distribution of the accident victims according to some of their characteristics may be observed on Table 1. The majority, as can be noted, were males (82.9%), aged 20 to 39 years old (67.9%), and married (63.2%). As to their formal education, 47.4% had not completed junior high school and 26.3% had not completed high school. As to their income, five of the workers stated they earned one minimal wage and 51, the majority (67.11%), earned from two to five minimal wages. Among those workers who agreed to inform us of their income, the mean income was US$560.00 per month**

The total number of occupational accidents that occurred in Botucatu during the year 1997 was estimated as approximately 1,810 (CI 95% 1,367-2,252), according to the procedures described in the section on Methodology. The proportion of incidences of occupational accidents in the same year (1997) was estimated as 4.1% (CI95% 3.0%-5.3%) of the Economically Active Population, which is defined as the population ten years of age and older,9 and estimated in 43,970 workers (CI 95% 42,842-45,097).8

Among the 76 workers who had occupational accidents during the study period, it was found that, in 39 cases (51.3%, CI95% 42,842-45,097),8 the emission of the CAT was not obligatory for they were statutory civil servants, autonomous wage laborers working in the informal labor market, proprietors and others. The remaining 37 victims of occupational accidents (48.7%, CI 95% 38.4%-58.9%) were composed of formally employed wage laborers, for which the emission of the CAT was mandatory (see Table 2).

For the 37 occupational accidents, identified by interviewers, in which emission of the CAT was mandatory, it was verified that 20, that is, 54.1% (39.4%-68.7%) had not been registered in the local branch of the INSS, [National Social Security Institute]. These results indicate that only 22.4% (13.8%-30.9%) of the occupational accidents verified in the residential interviews conducted in this study were registered by the Social Security System, as shown in the Figure below.



Table 3 indicates that, among the 37 victims of occupational accidents, 48.6% (34.0%-63.3%) were employees of large firms, and the remaining 51.4% (36.7%-66.0%), worked in medium sized, small and micro-firms. Among the cases that occurred in the large companies, 61.1% (44.7%-77.5%) of the accidents were registered in the Social Security System, and 38.9% (22.5%-55.3%) were not. On the other hand, among the accidents that occurred in small firms, only 31.6% (3.5%-59.6%) had been registered, while 68.4% (40.4%-96.5%) had not been registered. Companies which employed a total of five hundred or more workers were considered large firms, even if the unit of the company located in the municipality of Botucatu had a smaller number of employees. It was not possible to differentiate micro-firms from small and medium-sized ones because, within the existing records at our disposal, there was no up-to-date, reliable data concerning these firms available which would allow us to make this distinction.

As to medical care, it was verified that 92.1% (86.6%-97.6%) of the total number of victims were attended by health professionals: 100% of those victims for whom emission of the CAT was mandatory and 84.6% (74.3%-94.9%) of those for whom this was not mandatory received medical assistance.

Absence form work as a consequence of occupational accidents, as indicated in Table 4, occurred in 69.7% (60.3%-79.2%) of the cases reported in this study, being that 78.4% (66.3%-90.5%) occurred among workers for whom the emission of CATs were mandatory and 61.5% (47.6%-75.5%) occurred among workers for whom the emission of CATs were not mandatory.

Among the 76 occupational accident victims, the leave of absence from work, on an average, lasted 18.4 days. Among those for whom emission of CATs were mandatory, this average leave of absence lasted 19 days and, among those for whom the emission of CATs were not mandatory, leave of absence lasted, on the average, 22.1 days. In 14 of the 76 cases, the leave of absence lasted for a month or longer. Eight of the victims who had registered working contracts, that is, who were employed in the formal labor market, continued working without taking a leave of absence. Five of these declared they had suffered light injuries (superficial wounds, small burns, and foreign objects in the eye). However, three workers who had received recommendations from physicians to take a leave of absence stated they preferred to continue working for fear of loosing their jobs.

Table 5 presents the distribution of victims according to the sector of the economy in which they worked. Among the 76 cases, 9.2% (3.3%-15.1%) worked in the primary sector, 42.1% (32.0%-52.2%) in the secondary sector and 48.7% (38.4%-58.0%) in the tertiary sector of the economy. Considering those engaged in the secondary sector of the economy, 43.7% (28.1%-59.4%) of the cases identified occurred among workers of the building industry, being that, within this branch of activities, 78.6% (69.9%-94.3%) of the victims’ occupational status was such (they did not have formal contracts or were autonomous workers) that the emission of CATs was not mandatory.

Considering those occupational accidents for which the emission of CATs was mandatory, it was verified that: only one of the five cases in which workers were engaged in activities within the primary sector was registered in the INSS [Social Security System]; 11 of the 17 cases (64.7%, 44.0%-85.4%) involving workers engaged in the secondary sector were registered; and only 5 of the 15 cases (33.3%, 11.6%-55.1%) in the tertiary sector were registered.

Furthermore, 80.3% (72.1%-88.4%) of the 76 individuals who suffered occupational accidents did not know what the Comunicação de Acidente de Trabalho [work injury report] was and what its function was, being that this proportion was similar amongst those whose current occupational status made the emission of CATs mandatory and those for whom it was not mandatory.

Based upon the mean income of the victims of occupational accidents, the equivalent of US$ 560.00*** per month, the annual cost, in salaries, of leaves of absence due to accidents for the municipality of Botucatu was estimated as the equivalent of US$ 621,550.00*. This estimate was calculated by multiplying the median period of absence (in months) by the estimated number of accidents occurring that year, which was then multiplied by the median monthly income. Considering that, besides the expenditures with wages mentioned above, the expenditures with medical assistance, with eventual material losses and decrease in productivity after the accident, it may be presumed that, besides the human costs, the financial costs of accidents are very high.



In order to obtain a good estimation of the occurrence of occupational accidents in Botucatu, it was indispensable to contemplate the rural area of the municipality within the sampling distribution. If the sample population was restricted to the urban areas of the city, the occupations associated with the primary sector of the economy would be underestimated. The greatest difficulty encountered in planning the sample for this study was that it was not possible to effectuate a randomized sampling technique for conglomerations within the rural zone. The decision to include residences from the rural area within our sample (representing 6.3% of the sample population), according to accessibility rather than proceeding with a sampling technique involving probabilistic algorithms, was due to the fact that this was the only feasible way to go about sampling in this zone with the resources available, for there were no rural district census maps of Botucatu. Designing such a map, due to the dispersion of homes as well as to the difficulties in getting access to residences within the rural district, would consume practically all the resources available for conducting this research project. However, since the proportion of rural households included in the sample is rather small, it is presumed that imprecisions in the estimation of Confidence Intervals presented which may originate from such an inclusion are negligible.

This study made it possible to estimate the number of occupational accidents which occurred in 1997 in Botucatu in 1,810 and the average leave of absence in 18.4 days. This data allows us to estimate that 33,304 days or 91.2 years of work were lost in that municipality during the year 1997. The results also indicate that, in Botucatu, in the year 1997, only 22.4% of the occupational accidents mentioned in the residential interviews were registered by the social security system.

An adult inhabitant, present during the first visit to the residence in question, was responsible for identifying the worker presumed to have had an occupational accident. The latter was interviewed subsequently. In the majority of cases this first person interviewed was a housewife or a retired person due to the fact that these interviews occurred, preferentially, during commercial hours, when the economically active population cannot be found at home. Thus, in the majority of cases, the initial information obtained depended on the knowledge and memory of the first person interviewed within the household. For this reason, the estimates made by this study, with respect to under registration of accidents, may be underestimated.

As to those cases in which the emission of CATs was mandatory, no investigation of the employees’ records was undertaken to verify if this document had or had not been emitted by the firm. Considering that the physician’s report must be filled out before this document is dispatched to the local social security office to be registered, it is possible that, in the midst of this flow, some CATs emitted by the companies had not arrived at their destination. This possibility is particularly strong in those cases that do not result in social security benefits, that is, those in which the worker was on leave for a period of 15 days or less.

The flow of data to the social security system concerning the occurrence of occupational accidents initiates with the emission of the CAT by the employer. In Botucatu, the emission of CATs was mandatory in a little less than half of the cases reported in the interviews. Thus, even if there was an improvement in the registration of cases in which this procedure is mandatory, many occupational accidents would still not be identified by the current system. Furthermore, the increasing precariousness of labor relations that has characterized the past few decades, particularly in recent years, leads us to formulate the hypothesis that the proportion of workers covered by benefits from social security, in cases of occupational accidents, will decrease progressively. This, in turn, will generate repercussions as to official statistics’ capacity to identify cases of occupational accidents, if these continue to be elaborated on the basis of the CATs.

The results obtained by this study are similar to those observed by Barata et al,2 for the year 1994, in municipalities with more than eighty thousand inhabitants, even though the method utilized by the latter was not similar to the one utilized in this study.

On the other hand, 92% of the total number of victims of occupational accidents identified in this study and 100% of those victims for whom the emission of CATs was mandatory received medical care. This result indicates that, in Botucatu, the health services responsible for caring for the victims of occupational accidents may be considered a good source of information with respect to the occurrence of these accidents.

Data presented in this study indicate that only a little more than one fifth of the occupational accidents which occurred in the municipality of Botucatu during the study period were registered. Observed under registration, amounting to almost 80% of the cases, is due, in part, to the Brazilian social security legislation, which excludes almost half the labor force from the system. Furthermore, the vulnerability of the information system itself, which is very heavily dependent upon the employers’ voluntary act of emitting CATs, must be taken into consideration. There is nothing to suggest that the under registration observed in Botucatu constitutes a particularity of that municipality. On the contrary, it probably reflects a situation which is common throughout the country.

The results of this study indicate that it is necessary to construct local data systems on occupational accidents which are capable of identifying more adequately the occurrence of these phenomenon, so as to supply municipal health administrators with the information they need to plan preventive actions and, to enable them to evaluate the impact of such actions.



1. Alves S, Luchesi G. Acidentes do trabalho e doenças profissionais no Brasil: a precariedade das informações. Inf Epidemiol SUS 1992;3:5-20.        

2. Barata RCB, Ribeiro MCSA, Moraes JC. Acidentes do trabalho referidos por trabalhadores moradores em área urbana no interior do estado de São Paulo em 1994. Inf Epidemiol SUS 2000;9:199-210.        

3. Binder MCP, Wludarski SL, Almeida IM. Estudo da evolução dos acidentes do trabalho registrados pela previdência social no período de 1995 a 1999, em Botucatu, SP. Cad Saúde Pública 2001;17:915-24.        

4. Carmo JC, Almeida IM, Binder MCP, Settimi MM. Acidentes do Trabalho. In: Mendes, R. Patologia do trabalho. Rio de Janeiro: Editora Ateneu; 1995. p. 431-55.        

5. Cochran WG. Sampling techniques. 3rd ed. New York: John Wiley & Sons; 1977.        

6. Cohn A, Karsh US, Hirano S, Sato AK. Acidentes do trabalho. Uma forma de violência. São Paulo: Ed. Brasiliense; 1985.        

7. Cordeiro R, Olivencia ERP, Cardoso CF, Cortez DB, Kakinami E, Souza JJG et al. Desigualdade de indicadores de mortalidade no sudeste do Brasil. Rev Saúde Pública 1999;33:593-601.        

8. Cordeiro R. Efeito do desenho em amostragem de conglomerados para estimar a distribuição de ocupações entre trabalhadores. Rev Saúde Pública 2001;35:10-5.        

9. Fundação Instituto Brasileiro de Geografia e Estatística. Contagem da população, 1996. Rio de Janeiro: Fundação IBGE; 1997. vol. 1.        

10. Hirata HS, Salerno MS. L'implantation d'outils statistiques sur l'organisation et les conditions de travail dans les pays dits 'semi-développés' – Le cas du Brésil. In: France, Ministère du Travail, de l'Emploi et de la Formation Professionelle. L'usage des méthodes statistiques dans l'étude du travail; 1995. p. 117-29. (Serie Cahier Travail et Emploi).        

11. Ministério da Previdência e Assistência Social. Plano de Benefícios da Previdência Social. Lei 8.213, de 24 de julho de 1991. Diário Oficial da União, Brasília, 14 de agosto de 1998, seção I .        

12. Possas C. Avaliação da situação atual do sistema de informação sobre doenças e acidentes do trabalho no âmbito da Previdência Social Brasileira e propostas para sua reformulação. Rev Bras Saúde Ocup 1987;15(60):43-67.        

13. Possas C. Saúde e trabalho. A crise na previdência social. São Paulo: Hucitec; 1981.        

14. Ribeiro HP, Lacaz FAC. Acidentes de trabalho. In: Departamento Intersindical de Estudos e Pesquisa de Saúde e dos Ambientes de Trabalho, editor. De que adoecem e morrem os trabalhadores. São Paulo: Diesat; 1984.        

15. Ribeiro HP. O número de acidentes do trabalho no Brasil continua caindo: sonegação ou realidade? SOS Saúde Ocup Segur 1994;20:14-21.        

16. Wünsch Filho V. Reestruturação produtiva e acidentes de trabalho no Brasil: estrutura e tendências. Cad Saúde Pública 1999;15:41-51.        



Correspondence to
Maria Cecília Pereira Binder
Departamento de Saúde Pública – Faculdade de Medicina de Botucatu (Unesp)
18618-000 Botucatu, SP, Brazil
E-mail: binder@laser.com.br
E-mail: cordeiro@fmb.unesp.br

Received on 11/29/2001.
Review on 19/11/2002.
Aproved on 2/10/2003.



This research received funding from the Fundação de Amparo a Pesquisa do Estado de São Paulo (Fapesp – Process n. 97/12.782-9).
* Associação Brasileira de Pós-Graduação em Saúde Coletiva. I Seminário Nacional de Pesquisa em Saúde do Trabalhador. Rio de Janeiro, 5-7 de agosto de 1996.
** US$1.00 = R$1.077 in 1997.
*** US$1.00 = R$1.077 in 1997.

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