ABSTRACT:

Objective:

To assess the completeness and consistency of reports describing viral hepatitis caused by work-related accidents in Brazil between 2007 and 2014.

Methods:

This is an analytical, epidemiological study evaluating the quality of data from the Information System for Notifiable Diseases (Sistema de Informação de Agravos de Notificação). Data were analyzed using absolute and relative frequencies, proportional percentage variation, and a linear χ² test.

Results:

The majority of mandatory and essential variables were classified with good completeness, despite growth during the study period. The occupation and clinical form variables were classified as normal when they had less than 25.1% incomplete data. Inconsistency was considered high among different variables above 15.0%, including, for example, serologic markers with the types of viral hepatitis and age with occupation and date of birth.

Conclusions:

We need to evaluate data quality periodically, in addition to train health professionals on the adequate way to completely fill out reports, because this contributes to the establishment of an efficient surveillance of communicable diseases and improves the population’s quality of life.

Keywords:
Communicable diseases; Data accuracy; Disease notification; Hepatitis; Work-related accidents

INTRODUCTION

The different types of viral hepatitis must be reported, as instituted by the Brazilian Ministry of Health, and these records are required be fed into the Information System for Notifiable Diseases (Sistema de Informação de Agravos de Notificação - SINAN). All suspected and confirmed cases, in addition to cases of outbreak, must be reported. Because it is a transmissible disease, viral hepatitis can be spread due to work-related accidents, which is another event that is mandatory to report¹1. Brasil. Portaria nº 204, de 17 de fevereiro de 2016. Define a lista nacional de notificação compulsória de doenças, agravos e eventos de saúde pública nos serviços de saúde públicos e privados em todo território nacional, nos termos do anexo, e dá outras providências. Diário Oficial da União. 2016; Seção 1 (32): 23-4..

As such, reports of viral hepatitis due to occupational accidents must be performed after establishing an epidemiological technical link between the two public health events. This link can occur in different work environments, which compromises the health conditions and quality of life of workers, and can generate economic and social damages to labor organizations.

Considering this context, the surveillance of communicable diseases is an efficient strategy for disease prevention and control, as it requires the compulsory reporting of suspected and confirmed cases, and a commitment on the part of health professionals to follow through with this public health strategy²2. Garcell HG, Hernandez TMF, Abdo EAB, Arias AV. Evaluation of the timeliness and completeness of communicable disease reporting: surveillance in the Cuban Hospital, Qatar. Qatar Med J. 2014; 2014(1): 50-6. DOI: 10.5339/qmj.2014.9
https://doi.org/10.5339/qmj.2014.9... . Health professionals’ commitment refers to early diagnosing and compulsory reporting, as well as to health education that aims to prevent diseases and promote health.

In order for surveillance to work efficiently, disease records need to be of high quality with regard to the information contained in the report form. Data quality can be verified from inconsistency between two variables that complement each other, and from the completeness of each variable, considering fields that are ignored or left blank³3. Brasil. Ministério da Saúde. Secretaria de Vigilância em Saúde. Roteiro para uso do Sinan Net, análise da qualidade da base de dados e cálculo de indicadores epidemiológicos e operacionais: Hepatites Virais. Brasília: Ministério da Saúde; 2008.^,44. Brasil. Ministério da Saúde. SINAN relatórios: manual de operações. Versão 4.5. Brasília: Ministério da Saúde ; 2014..

A review of the quality of the databases that are part of the health information system reveal the scarcity of studies on this subject in Brazil, mainly in the northeast region, where very few articles were found on the completeness of SINAN data, and where none were found on data consistency⁵5. Lima CRA, Schramm JMA, Coeli CM, Silva MEM. Revisão das dimensões de qualidade dos dados e métodos aplicados na avaliação dos sistemas de informação em saúde. Cad Saúde Pública. 2009; 25(10): 2095-109. DOI: 10.1590/S0102-311X2009001000002
https://doi.org/10.1590/S0102-311X200900... . The studies that are performed are limited to municipalities or states, and those with larger coverage in Brazil refer to other communicable diseases and only evaluate completeness⁶6. Correia LOS, Padilha BM, Vasconcelos SML. Métodos para avaliar a completitude dos dados dos sistemas de informação em saúde do Brasil: uma revisão sistemática. Ciênc Saúde Coletiva. 2014; 19(11): 4467-78. http://dx.doi.org/10.1590/1413-812320141911.02822013
http://dx.doi.org/10.1590/1413-812320141... .

As such, the need for investment in the analysis of SINAN data is evident in order to improve the quality of its information and, consequently, efficient practices of communicable disease surveillance, which improves the quality of life of the population. Therefore, this study aimed to evaluate the completeness and consistency of reports of viral hepatitis that were due to work accidents in Brazil between 2007 and 2014.

METHODS

This is an analytical, epidemiological study, which evaluated the data quality of reports of viral hepatitis due to work-related accidents in SINAN. All cases of viral hepatitis were verified by laboratorial, clinical-epidemiological, and serological scar confirmation - closed cases - reported in Brazil from 2007 to 2014, and they had a source of infection from workplace accidents.

The variables analyzed were those contained in the report form for viral hepatitis, of which some are mandatory to include in the SINAN database and some are just essential for investigating the case. There were a total of 36 variables, of which 28 were mandatory and 8 were essential. In order to analyze the SINAN data quality, the completeness of the variables (incomplete, ignored, or blank fields) and inconsistency (the relationship between two selected variables) were evaluated. For completeness, the variables were dichotomized into complete and incomplete fields. For inconsistency, the same categories as used in the report form were utilized in the analysis.

The completeness of the data was evaluated in the reports according to the calendar year of the report under study, considering the criteria recommended and adapted from SINAN’s manual of operations⁴4. Brasil. Ministério da Saúde. SINAN relatórios: manual de operações. Versão 4.5. Brasília: Ministério da Saúde ; 2014., which categorizes completeness as: good (presents ≤ 25.0% incomplete fields); normal (between 25.1 and 50.0%); poor (between 50.1 and 75.0%); and very poor (those with ≥ 75.1% incomplete fields).

The consistency between two related variables was evaluated according to the script for using SinanNET, an analysis of the quality of the database, and a calculation of the epidemiological and operational indicators - viral hepatitis³3. Brasil. Ministério da Saúde. Secretaria de Vigilância em Saúde. Roteiro para uso do Sinan Net, análise da qualidade da base de dados e cálculo de indicadores epidemiológicos e operacionais: Hepatites Virais. Brasília: Ministério da Saúde; 2008.. The variables selected for this analysis were: final classification and etiological classification, final classification and clinical form, clinical form and etiological classification, serological results and etiological classification, age and etiological classification, age (<18 years) and occupation, age (<18 years) and exposure to an “accident with biological material”, age (<18 years) and date of birth; and age (<18 years) without date of birth and date of the accident. The age <18 years variable was selected upon considering Brazilian labor laws. And, when evaluating the relationship between other variables and this age group, the analyses were made case by case.

In order to perform the data analysis, the absolute and relative frequencies of completeness and consistency between the variables were calculated. For completeness, proportional percentage variation (PPV) - PPV = [(final year - initial year] / initial year]*100 was calculated, and the linear χ² test of the incomplete data was used in order to verify the trend of the time series, considering a statistical significance of p <0.05. The temporal trend was classified as decreasing when PPV was negative, increasing when PPV was positive, and stationary when there was no linear statistical significance. The variables that had ≥ 25.1% incomplete fields (normal, poor and very poor) were evaluated by region and by year.

This study complied with all ethical principles according to Resolution number 466/2012, when the database was authorized by the Department of Health Surveillance (Secretaria de Vigilância em Saúde) and approved by the Research Ethics Committee of the Bahia School of Medicine.

RESULTS

A total of 1,493 cases of viral hepatitis were reported due to work-related accidents in Brazil between 2007 and 2014. The Southeast (40.6%), South (28.1%) and Northeast (11.9%) reported the majority of cases in this period.

An analysis of the completeness of these data revealed that the variables that are considered mandatory when completing the reports were classified as having good completeness, that is, <25.0% incomplete fields. However, the majority presented a growth in incomplete data between 2007 and 2014, except for the institutionalized variables and HIV/AIDS-related diseases, which presented a reduction of 28.0 and 5.9%, respectively. The variables of exposure to: injectable drugs (PPV = 266.7%), hemodialysis (PPV = 233.4%), surgical treatment (PPV = 200.0%), injectable drugs (PPV = 175.0%), inhalable drugs (PPV = 175.0%) and transfusion (PPV = 140.0%) presented more than 100% growth in incomplete data. Despite the decreasing and increasing variation in the incomplete data, no obligatory variable presented statistical significance, remaining in a steady trend (Table 1).

Most of the essential variables for follow-up of reported cases were classified as good completeness, except for occupation and clinical form, which were classified as normal (> 25.1% of incomplete data). Of these, the etiological classification (p = 0.002) and clinical form (p = 0.006) variables revealed a statistically significant growth in incomplete data during the study period. The occupation (PPV = -16.7%, p = 0.006), patient referral (PPV = -28.3%, p = 0.028) and genotype for hepatitis C virus (PPV = -41.8%; = 0.001) variables had a reduction in incomplete data with statistical significance, that is, a decreasing trend during the period (Table 1).

For the occupancy and clinical form variables, completeness was also analyzed by region in Brazil. Occupation was classified as normal with regard to completeness in all of the regions and had a growth in incomplete data between 2007 and 2014 in the northeast region (PPV = 4.1%), but with a steady trend. Only the Midwest (PPV = -43.7%, p = 0.042) and Southeast (PPV = -19.5%; p = 0.033) showed a decreasing trend of incomplete data. A reduction of incomplete data was verified in the North (PPV = -62.5%) and Northeast (PPV = -28.7%), although it was not statistically significant. A growing trend was observed in the Southeast (p = 0.006), South (p = 0.020) and Midwest (p = 0.020) (Table 2).

With regard to the consistency analysis, in the relationship between final classification and etiological classification, an inconsistency of 6.0% was verified, whereas in the final classification and clinical form, an inconsistency of 26.5% was observed. Between clinical form and etiological classification, there was an inconsistency of 26.0%, which was greater than the hepatitis B virus (19.2%). The serological results and the serological classification presented divergent data for all types of hepatitis virus, with an inconsistency of more than 15.0%. Age and etiological classification were inconsistent mostly among children under 14 years old (50.0%) (Table 3).

Consistency between age <18 years and other variables was verified on a case-by-case basis, individually (n = 21). Among cases of patients <14 years old, there was an inconsistency of 100.0% in relation to occupation and 42.9% in relation to exposure to an accident with biological material. For those patients between 14 and 17 years old, the inconsistencies were 50.0 and 57.1%, respectively. The age <18 years old variable presented an inconsistency of 31.6% in relation to date of birth and 100.0% in relation to age (<18 years old), without date of birth and with the date of the accident (Table 3).

DISCUSSION

The results point to the relevance of reporting work-related viral hepatitis for the field of occupational health, as it provides consistent data that allows for the construction of epidemiological indicators that are able to portray Brazilian workers’ health situations and show how these two associated diseases are preventable in the arena of public health. The analyses presented proportional differences between reporting by Brazilian region. The largest proportion of mandatory and essential fields showed a growth in incomplete data and an inconsistency of greater than 15%.

In the case of mandatory variables in the report form, other studies in Brazil⁷7. Barbosa DA, Barbosa AMF. Avaliação da completitude e consistência do banco de dados das hepatites virais no estado de Pernambuco, Brasil, no período de 2007 a 2010. Epidemiol Serv Saúde. 2013; 22(1): 49-58. DOI: 10.5123/S1679-49742013000100005
https://doi.org/10.5123/S1679-4974201300... and in the United States⁸8. Woodruff RSY, Pratt RH, Armstrong LR. The US national tuberculosis surveillance system: a descriptive assessment of the completeness and consistency of data report from 2008 to 2012. JMIR Public Health Surveill. 2015; 1(2): e15. DOI: 10.2196/publichealth.4991
https://doi.org/10.2196/publichealth.499... also indicated good and excellent completeness. While with the essential variables, where filling out of the fields is necessary but not mandatory, such as in work-related reporting, incompleteness has varied between 0 and 98%⁹9. Alvares JK, Pinheiro TMM, Santos AF, Oliveira GL. Avaliação da completitude das notificações compulsórias relacionadas ao trabalho registradas por município pólo industrial no Brasil, 2007-2011. Rev Bras Epidemiol. 2015; 18(1): 123-36. http://dx.doi.org/10.1590/1980-5497201500010010
http://dx.doi.org/10.1590/1980-549720150... .

Exposure factors presented a growth in incompleteness during the study period, but authors¹⁰10. Heisey-Grove DM, Church DR, Haney GA, Demaria A Jr. Enhancing surveillance for Hepatitis C through public health informatics. Public Health Rep. 2011; 126(1): 13-8. https://dx.doi.org/10.1177%2F003335491112600105
https://dx.doi.org/10.1177%2F00333549111... have considered that at least one risk factor that is filled out completely is adequate, thus obtaining 81%. As such, this variable could be re-developed according to type of viral hepatitis, or even designed in such a way that allows for one or more risk factor to be chosen, without the obligation of filling out all of the variables related to exposure.

Occupation was one of the essential variables that presented regular completeness. The low quality of variables related to occupation was also found in reports of communicable diseases in Cuba.²2. Garcell HG, Hernandez TMF, Abdo EAB, Arias AV. Evaluation of the timeliness and completeness of communicable disease reporting: surveillance in the Cuban Hospital, Qatar. Qatar Med J. 2014; 2014(1): 50-6. DOI: 10.5339/qmj.2014.9
https://doi.org/10.5339/qmj.2014.9... In various locations in Brazil, poor quality was observed in malaria records¹¹11. Braz RM, Tauil PL, Santelli ACFS, Fontes CJF. Avaliação da completude e da oportunidade das notificações de malária na Amazônica Brasileira, 2003-2012. Epidemiol Serv Saúde. 2016; 25(1): 21-32. DOI: 10.5123/S1679-49742016000100003
https://doi.org/10.5123/S1679-4974201600... and mortality due to occupational accidents¹²12. Alves MMM, Nomellini PF, Pranchevicius MCS. Mortalidade por acidente de trabalho no Estado do Tocantins, Brasil: estudo descritivo, 2000-2010. Epidemiol Serv Saúde. 2013; 22(2): 243-54. http://dx.doi.org/10.5123/S1679-49742013000200006
http://dx.doi.org/10.5123/S1679-49742013... .

Even in reports of work-related injuries, in which filling out an occupation is mandatory, there was not 100% completeness⁹9. Alvares JK, Pinheiro TMM, Santos AF, Oliveira GL. Avaliação da completitude das notificações compulsórias relacionadas ao trabalho registradas por município pólo industrial no Brasil, 2007-2011. Rev Bras Epidemiol. 2015; 18(1): 123-36. http://dx.doi.org/10.1590/1980-5497201500010010
http://dx.doi.org/10.1590/1980-549720150... . The occupation variable presents a number of methodological problems in reporting, such as consistency, completeness, definition and coding, considering that the instructions for filling out the field is vague in relation to the time of the activity, and does not specify that the Brazilian Occupation Classification (Classificação Brasileira de Ocupação) should be used in the updated version¹³13. Romero DE, Cunha CB. Avaliação da qualidade das variáveis epidemiológicas e demográficas do Sistema de Informação sobre Nascidos Vivos, 2002. Cad Saúde Pública. 2007; 23 (3): 701-14. http://dx.doi.org/10.1590/S0102-311X2007000300028
http://dx.doi.org/10.1590/S0102-311X2007... .

The clinical form showed normal completeness, and the proportion of incomplete data in the southeastern, mid-western and southern regions of the country increased. The state of Pernambuco⁷7. Barbosa DA, Barbosa AMF. Avaliação da completitude e consistência do banco de dados das hepatites virais no estado de Pernambuco, Brasil, no período de 2007 a 2010. Epidemiol Serv Saúde. 2013; 22(1): 49-58. DOI: 10.5123/S1679-49742013000100005
https://doi.org/10.5123/S1679-4974201300... , which is located in the northeast region, presented an incompleteness of 6.9% of the reports of viral hepatitis, confirming the results of the present study. It was noticed that because the variables are classified as essential, professionals have stopped filling them out, thus compromising the system of disease surveillance.

In addition to incompleteness, there were several inconsistencies between reporting variables, such as clinical aspects that did not match the characteristics of a particular case of viral hepatitis. Reactive and non-reactive serological markers were also found to be incompatible with hepatitis types, mainly for type B hepatitis. Similar data was also reported in another study⁷7. Barbosa DA, Barbosa AMF. Avaliação da completitude e consistência do banco de dados das hepatites virais no estado de Pernambuco, Brasil, no período de 2007 a 2010. Epidemiol Serv Saúde. 2013; 22(1): 49-58. DOI: 10.5123/S1679-49742013000100005
https://doi.org/10.5123/S1679-4974201300... with an inconsistency of 32.6%.

A problem of inconsistency was found in the variable clinical form with regard to filling out the final classification, in order to confirm the nature of the case. The form states that, when filling out the final classification field under the category “serological scar”, the system will automatically fill out the clinical form under the category “inconclusive”. However, this did not occur in most of the reports, and the field on the clinical form was left blank or ignored.

Another relevant fact refers to the source of infection, a variable used to select the cases caused by an occupational accident in the present study. The category “other” appeared in innumerous reports. However, in the space provided to specify the source, some of the categories present were the same as in the variables, only written using different words.

Age was a variable that presented the greatest inconsistency with different variables, including occupation and accidents with biological material, both of which required a high level of schooling to perform the activity, which did not correspond to people under 14 years of age. In addition, work performed by children under 14 years of age is prohibited under Brazilian law. Legislation allows for people over 14 years old to participate in a young apprentice program. Teenagers between the ages of 16 and 18 are allowed to work, except at night and in hazardous or unhealthy places¹⁴14. Brasil. Congresso Nacional. Constituição da República Federativa do Brasil. Diário Oficial da União. 1988 [citado 14 maio 2016]; 191 A: 1. Disponível em: Disponível em: https://www.planalto.gov.br/ccivil_03/constituicao/constituicao.htm
https://www.planalto.gov.br/ccivil_03/co... .

This result was also demonstrated in reports of work-related mortality, with records of 0.9% among children under 14 years of age and 7.7% among children aged 15 to 19 years old¹²12. Alves MMM, Nomellini PF, Pranchevicius MCS. Mortalidade por acidente de trabalho no Estado do Tocantins, Brasil: estudo descritivo, 2000-2010. Epidemiol Serv Saúde. 2013; 22(2): 243-54. http://dx.doi.org/10.5123/S1679-49742013000200006
http://dx.doi.org/10.5123/S1679-49742013... . In addition, date of birth also appeared inconsistent with age in the same report. In another study, we found inconsistency in 465 records of people older than 120 years old, which was considered improbable.

Consistency between diagnostic suspicion and etiological classification was not evaluated because these are fields for which a strict consistency is not expected, because there is no way to change the field referring to the diagnostic suspicion after the report. This is due to the fact that it is possible to change the fields “laboratory data” and “case completion after an epidemiological investigation” in the SINAN database, as all reports of suspected or confirmed cases should be completed within 180 days of the report date.

Data quality involves many factors, such as the number of questions and the number of pages to be filled out, which can contribute to whether the report is filled out or not. In Massachusetts, USA, modifications were made to the surveillance system for hepatitis C virus infection, reducing the form to one page on a computer system, which contributed to the increase of reports and a reduction in incomplete fields¹⁰10. Heisey-Grove DM, Church DR, Haney GA, Demaria A Jr. Enhancing surveillance for Hepatitis C through public health informatics. Public Health Rep. 2011; 126(1): 13-8. https://dx.doi.org/10.1177%2F003335491112600105
https://dx.doi.org/10.1177%2F00333549111... .

Most of the report systems perform data cleaning procedures and verify the duplication of records, which is good for data quality. However, data consistency is not verified by most systems. Therefore, greater care is required in the records and in the investigation of each reported case⁸8. Woodruff RSY, Pratt RH, Armstrong LR. The US national tuberculosis surveillance system: a descriptive assessment of the completeness and consistency of data report from 2008 to 2012. JMIR Public Health Surveill. 2015; 1(2): e15. DOI: 10.2196/publichealth.4991
https://doi.org/10.2196/publichealth.499... . One strategy for improving this information and data quality is an annual evaluation of the performance indicators of the information system²2. Garcell HG, Hernandez TMF, Abdo EAB, Arias AV. Evaluation of the timeliness and completeness of communicable disease reporting: surveillance in the Cuban Hospital, Qatar. Qatar Med J. 2014; 2014(1): 50-6. DOI: 10.5339/qmj.2014.9
https://doi.org/10.5339/qmj.2014.9... , which should be carried out by each administrator or by partnerships with universities.

It is worth noting the possible limitations of this study. For one, an analysis of duplicate data, which was prevented when the patient and mother’s names were not present. Second, we were unable to establish a causal link between viral hepatitis and work-related accidents of people with different professions. Third, it was not possible to analyze data consistency by region and calendar year due to the number of values below five, even considering the two-year periods. Finally, because there is a lack of studies testing similar variables, it was impossible to compare results. However, this study presents crucial points for health professional interventions with regard to the appropriate and careful filling out of disease reports, and the improvement of the information system, the reporting form, and the form instructions.

CONCLUSION

The data quality of SINAN’s reports should be evaluated frequently in order to ensure the best possible analysis of the distribution of cases of viral hepatitis due to occupational accidents and, consequently, the effective monitoring of communicable diseases. Essential variables in some cases should be mandatory. Furthermore, it is necessary to review the compatibility of the system with instructions for completing the reports. A more careful evaluation is necessary in cases with children under 18 years of age, because if there is no inconsistency, it means that child labor is being exploited illegally. Finally, in order for SINAN to operate with quality and contribute to improve the population’s quality of life, health professionals should be trained periodically to fill out the report form (with correct and consistent knowledge of the concepts in each variable and category) and to understand the ethical commitment they must establish with regard to public health.

References

Publication Dates

History