Quality, opportunity, and treatment of data of tuberculosis and human immunodeficiency virus co-infection

Qualidade, oportunidade e tratamento de dados de tuberculose e coinfecção com vírus da imunodeficiência humana

Calidad, oportunidad y manejo de los datos de coinfección tuberculosis y virus de la inmunodeficiencia humana

Esta obra está bajo una Licencia Creative Commons Atribución 4.0 Internacional.

Tuberculosis (TB) is a respiratory infection caused by Mycobacterium tuberculosis (Mtb) which, with timely diagnosis and precise treatment, presents a high chance of cure and, consequently, a break in the chain of transmission ¹. However, it still represents a public health problem, and it is among the main causes of death in the world². Up until the new coronavirus pandemic (COVID-19, abbreviation for coronavirus disease 2019), TB was the main cause of death by a single infectious agent, causing more deaths than the infection by the human immunodeficiency virus (HIV) ¹.

In people living with HIV (PLWH), infection with Mtb presents a higher risk of developing active disease and unfavorable outcomes, such as developing multidrug-resistant TB, treatment dropout and death.^2-5

According to the World Health Organization (WHO), in 2021, approximately 10 million cases of TB occurred in the world, and a total of 1.2 million people died due to the disease, including 208,000 deaths among those with TB-HIV co-infection. Brazil is among the 30 countries with high burden of TB and co-infection TB-HIV¹. In 2019, 78,428 new cases of TB were notified, from which 8,070 presented TB-HIV co-infection, in other words, 10.3% of the cases.^6,7

Among the strategies for the elimination of TB, WHO has drawn recommendations to promote the universal health coverage policy and regulatory structures, such as case notification, vital registration, quality and correct use of medicines for the infection control¹.

TB and HIV case notification is mandatory in Brazil, and it must be done using the Notifiable Diseases Information System (SINAN - Sistema de Informação de Agravos de Notificação). However, studies show that the surveillance system presents flaws and case underreporting that can be due to underdiagnosis, typing errors by the responsible professional, and lack of communication among the health services.^8,9 Those inconsistencies might affect the quality of the data and information generated from it, which is deleterious for the planning and conduction of sanitary strategies⁹.

Thus, the identification, evaluation, and use of methodological resources for the analysis of data quality is a useful initiative to base the planning of actions for public policies for the control of TB and the co-infection TB-HIV. Therefore, the present study aimed to analyze the quality and timeliness of notification of TB and TB-HIV co-infection in SINAN (Notifiable Diseases Information System) in Espirito Santo State, from 2016 to 2018.

This is a cross-sectional study of data quality analysis of SINAN-TB according to the quality attributes defined in the methodology proposed by the Centers for Disease Control and Prevention (CDC), United States of America^10,11

The data quality analysis is performed by checking the attributes of completeness and consistency of data. ¹⁰ Data consistency is defined as the percentage of cases with the date of notification higher than the date of diagnosis, using as a threshold greater than or equal to 70%. Yet the analysis of completeness aims at verifying the percentage of filling of variables, in order to identify the frequency of variables “ignored” or “blank”. Completeness is considered satisfactory when the result of the analysis gets a value greater than or equal to 70% for essential variables and 100% of the fields for the mandatory variables (Figure 1).¹⁰ Both attributes were categorized as great (≥ 90%), regular (between 70 and 90%) and poor (<70%).¹² Furthermore, the attribute timeliness of notification, from CDC guidelines, was used. It is the proportion of cases from which the time between the date of notification and the date of diagnosis is less than or equal to seven days. ¹⁰ All cases of TB and TB-HIV co-infection, notified in SINAN-TB, from 2016 to 2018, of people residing in Espirito Santo State were included in the study. Both databases were made available by the State Health Secretariat (SESA). The time frame used in the study was chosen because of the inclusion of new variables in the version number five of the notification record for TB, such as special populations (persons deprived of freedom, homeless population, health care professionals and migrants), government benefit claimants, use of antiretroviral therapy (ART) during TB treatment, rapid molecular test for TB (RMT-TB), susceptibility testing, and, if it is a case of “transfer”, inclusion of the destination, and the alteration in the variables “if extrapulmonary” (removal of the second option), “sputum-smear bacilloscopy” (withdraw of the second bacilloscopy), “associated diseases and illnesses” (inclusion of illicit drug use and smoking) and “follow-up sputum-smear bacilloscopy” (inclusion after the 6^th month).¹³

Figure 1
Flow chart about the methodology of the process for data quality analysis in SINAN, from 2016 to 2018.

Y – year; D - day; F – female; M – male; M – month; N -number.

Data analysis was carried out in five steps (figure 1): the first was the data quality analysis; the second was the records standardization; the third step was the duplicity analysis; the fourth step the linkage; and the fifth step was the anonymization of the data.

When considering the TB database in the first step, the analysis of the mandatory filling variables completeness was used, those variables whose field is considered mandatory to fill 100% of the records in any notification, such as: “Number of notification”, “Illness/disease”, “Date of current notification”, “Municipality of current notification”, “Current health unity”, “Date of diagnosis”, “Age”, “Sex” (male, female), “Pregnant” (mandatory filling if patient is a female), “Current State of residence”, and “Current municipality of residence”, “Country”, “Type of entry”, “Form” (pulmonary, extrapulmonary or extrapulmonary plus pulmonary), “Sputum-smear bacilloscopy”, “Sputum culture”, “Susceptibility testing” (when the culture is positive), “HIV”, “Total number of contacts identified”.¹³

In addition to the mandatory variables analysis, the completeness of essential variables was evaluated. It was considered those variables with filling in up to 70% of the records, such as: “Special populations” (persons deprived of freedom, homeless population, health care professionals and migrants), “Government benefit claimants”, “Associated diseases and illnesses” (aids, alcoholism, diabetes, mental disorder, illicit drug use, smoking, and others), “ART during TB treatment”, “RMT-TB”, “Date of start of current treatment”, “Bacilloscopy from the 1^st to the 6^th month”, “Bacilloscopy after the 6^th month”, “Directly observed treatment (DOT) performed”, “Total number of contacts assessed”, “Closure situation”, “If transferred”, “Federation Unity” (State), “Municipality of residence”, and “Date of closure”. ¹³

We searched to verify the completeness of the following additional variables: “Race/skin color” (yellow/Asian, white, indigenous, mixed, black, and ignored) and “Schooling” (illiterate, 1^st to incomplete 4^th grade of elementary school, complete 4^th grade of elementary school, 5^th to incomplete 8^th grade of elementary school, complete elementary school, incomplete high school, complete high school, incomplete higher education, complete higher education, does not apply, and ignored).

In the second step, it was carried out the standardization of the variables “Patient name” and “Mother’s name”, with the removal of the accents and replacing the capital letters by the correspondent small letters. Following, the standardization of “Date of notification”, “Date of diagnosis”, “Date of birth”, “Date of treatment start”, “Date of closure” in the format day/month/year; and the codification of the variable “Sex” assigning the value 1 for “F” – female, and 2 for “M” – male (figure 1).

During the third step (figure 1), the duplicity analysis was performed in four stages. In the first stage, records were selected by searching for the name, date of birth and sex; the second stage included searching for the name, age, mother’s name, and phonetic search, marking the duplicities; in the third stage, initially we checked the duplicity in the notification dates before carrying out the data linkage, if duplicity was identified, we generated a new record with the notification information and investigation of the first record, adding the information from the second record, complementing the missing data, and last we performed the deduplication, that is, the exclusion of duplicated records.¹³ After the duplicity analysis, data quality analysis was carried out again, with regards to consistency and completeness as well as notification timeliness.

In the fourth stage, we performed the linkage with SINAN-HIV database for cases notified between 2007and 2018, with the purpose of identifying cases notified in SINAN-TB database without the diagnosis of the co-infection with HIV, but that were registered in SINAN-HIV database. After that, the exclusion of TB cases data with a negative diagnosis for HIV was done and solely the records of TB-HIV co-infection cases were kept. After that, analysis of completeness and data consistency of TB-HIV co-infection cases only was performed.

In the last step, we carried out the exclusion of identification data, which is called the data anonymization process, using technical means that are reasonable and available for preventing the possibility of direct or indirect association with the individual (figure 1).

During the methodological process we carried out a descriptive analysis of the data and the results were entered into frequency tables in the software STATA, version 16.

The study was approved by the Research Ethics Committee from the Centro de Ciências da Saúde da Universidade Federal do Espírito Santo (CEP/CCS/UFES) under the number CAAE: 05675618.4.0000.5060 and report number 4022892 in 12/05/2020, according to the resolution 466/12 from the Brazilian Nacional Health Council (CNS – Conselho Nacional de Saúde).

The analysis of 4,428 records from SINAN-TB database from Espírito Santo State was performed and 375 duplicated records were identified ad removed, resulting in a total of 4,053 cases. At the end of the five-step process of analysis of data treatment, a total of 332 cases presented TB-HIV coinfection, from which three cases were found in the SINAN-HIV database with negative serology for HIV in the SINAN-TB database. We identified that the timeliness of notification was regular, showing that around 25% of the cases were not notified in up to seven days after the confirmation of the disease.

Mandatory filling variables, that is, those with 100% completeness, in the first data quality analysis (1^st step) in SINAN-TB, showed that from a total of 19 variables, 17 presented a completeness percentage of 100%, differently from the variable “Extrapulmonary type” with 17.2% completeness, and the variable “Total number of contacts identified” that presented 98.9% of filling. In the third data quality analysis, with the TB-HIV co-infection cases, the variable “Extrapulmonary type” presented an increase in the frequency of completeness to 33.8% when we considered solely the TB-HIV coinfection cases (table 1).

Table 1
Distribution of the frequency of completeness in mandatory variables performed before and after deduplication and after linkage of SINANTB database with SINANHIV database in Espirito Santo State between 2016 and 2018

Abbreviations: HIV - human immunodeficiency virus; N – number; TB – tuberculosis; SINAN – sistema de informação de agravos de notificação (Notifiable Diseases Information System)

As for the completeness analysis of essential variables (with 70% of filling), in the first analysis, from 23 variables analyzed, a total of 21 variables presented a completeness percentage above 70%, except for the variables “Bacilloscopy 6th month”, which presented 67.5% of completeness and the variable “Antiretroviral therapy during TB treatment”, with 66.5% of filling.

In the second data quality analysis, after the exclusion of the duplicated cases, we observed an increase in the frequency of completeness for the variable “Bacilloscopy 6th month” to 71.1%. On the other hand, the variable “Antiretroviral therapy during TB treatment” remained presenting a low frequency of completeness after the deduplication (67.7%).

In the third data quality analysis, with the TB-HIV co-infection cases, we observed that the percentage of completeness of the variable “Antiretroviral therapy during TB treatment” reached 91.3% and a decrease in the completeness frequency in the variables of bacilloscopy per month, once it was possible to verify that from the second month the frequency of completeness was below 70% for those variables, as well as for the variable “Total number of contacts assessed”, which also presented a reduction to 40.9% (table 2).

Table 2
Distribution of the frequency of completeness in essential variables performed before and after deduplication and after linkage of SINANTB database with SINANHIV database in Espirito Santo State between 2016 and 2018

The additional variables “Race/skin color” and “Schooling”, included in the study, presented a completeness of more than 70% in all three steps of evaluation (table 3).

Table 3
Distribution of the frequency of completeness in additional variables performed before and after deduplication and after linkage of SINANTB database with SINANHIV database in Espirito Santo State between 2016 and 2018

Abbreviations: HIV - human immunodeficiency virus; N – number; TB – tuberculosis.

In all the consistency analysis of the notification data, that is, the percentage of cases with the date of notification higher than the date of diagnosis was considered excellent, and most of the dates of notification were higher than the date of diagnosis in all the steps.

As for the analysis of timeliness of notification in SINAN-TB, which correlates the date of notifications with the date of diagnosis, we identified that among the 4,428 records assessed in the first analysis, 2,483 (56.0%) were notified within a minimum interval of one day, 815 (18.4%) up to seven days, totalling 3,298 (74.4%) records with the notification occurring in up to seven days, and records with more than seven days were 1,130 (25.6%).

In a second analysis of timeliness, performed during the step 3, from a total of 4,053 records analyzed, 2,317 (57.0%) were notified within an interval of one day, 755 (18.6%) in up to seven days, totalling 3,072 (75%), and 981 (24.4%) were notified more than seven days later. And in the analysis of opportunity of case notification in SINAN TB-HIV, 176 (54.2%) were identified and notified within an interval of one day, 74 (22.7%) were identified and notified in up to seven days, totalling 250 (76.9%) records with notification in up to seven days from the date of diagnosis, and 75 (23.1%) were notified with more than seven days.

Timeliness of notification in SINAN-TB was considered regular in Espirito Santo State, from 2016 to 2018. When considering the methodological process of quality analysis of data from SINAN-TB, it was possible to verify an increase in the percentage of completeness after duplicity analysis and at the end of the steps from the study. When considering solely TB-HIV co-infection data, completeness of the variables related to TB treatment follow up was lower than the recommended. As for the data consistency, we observed that the majority of the notification dates were later than the diagnosis date in all steps.

The frequency of data completeness for most of the variables was higher than 70%, for both mandatory filling and the essential variables in the three steps evaluated. This can be considered a progress in the quality analysis process of health information systems since previous studies presented low completeness in the filling of mandatory and essential variables. ^14,15,16 Low completeness of data can be deleterious to the surveillance system because it directly impacts the interventions and elaboration of public health policies for disease control⁹.

The variable “antiretroviral therapy” (ART) presented a completeness percentage lower than 70% among the cases of TB, but the percentage increased when we analyzed only TB-HIV co-infection cases. The early identification of the use of ART, during the diagnosis of TB, reduces the risk of complications related to the treatment, and the adherence difficulties due to personal matters or issues related to the health service. A study carried out in Brazil, from 2006 to 2016, found that early start of ART reduces in up to 94% the risk of dying of aids during the first six months.^10,17

A filling percentage lower than 70% was also observed for variables related to the treatment follow-up among the TB-HIV co-infection cases. The low frequency of variables related to the treatment follow-up in people with TB and the co-infection TB-HIV is suggestive of a poor treatment follow-up and can compromise the surveillance of the cases following the guidelines of the Ministry of Health (MoH) as an effective strategy for the control of the disease’s chain of transmission, with the performance of a monthly bacilloscopy as from the second month of treatment. Furthermore, contact tracing allows the early detection of a case, and consequently, minimizes the chance of evolving to a severe form of the disease and drug resistance, in addition to create an opportunity to screen for latent infection with Mtb (LTBI) and the prescription of the TB preventive treatment.^{16, 19}

A retrospective cross-sectional study, including 354 individuals, carried out in Ethiopia during 2016 and 2017, highlighted the importance of case follow-up for the treatment success and the consequent reduction in the mortality rate among TB-HIV coinfection cases¹⁹. Besides the reduction of the unfavorable outcome “death”, with a proper follow-up it is possible to minimize the treatment dropout and the recurrence of the disease. And that is because the risk of unfavorable outcomes is three times more likely to occur in people with TB-HIV co-infection when compared to patients without the co-infection.^15,21

The notification flow occurs by case recording in the notification record by the healthcare professional using the laboratory and clinical-epidemiologic criteria, filling the record with precise identification information from the user, from the service, and clinical and epidemiological data. This data can be collected by direct interview with the user, from TB record books and case follow-up or via medical record, or else, using other systems such as the Mortality Information System (SIM - Sistema de Informações sobre Mortalidade), the Hospital Information System (SIH - Sistema de Informação Hospitalar) or the Laboratory Environment Manager (GAL - Gerenciador de Ambiente Laboratorial). ^13,19

After the notification process, the municipality epidemiologic surveillance division generates a monthly report called the “Follow-up Report” (Boletim de acompanhamento), in which the cases notified with closure date blank and under treatment for 30 days after the date of diagnosis are listed by health unity. The report is forwarded to the health services for the updating of the data and return for the feedback of the information in the system. Then, the municipality epidemiologic surveillance division consolidates the data and forwards to the State epidemiologic surveillance division, that after its evaluation, will forward it to the federal level, which will then be made available in SINAN.^13,19

The health surveillance is responsible for the investigation of cases suspected of having TB, the notification of confirmed cases and assuring the generation of treatment and follow-up data of confirmed cases up until their outcome, so as to plan and implement the actions needed for the control of TB and HIV/aids. The data quality assurance is also dependent on the correct typing, analysis, case follow-up and data evaluation by the surveillance professionals in the municipality, with the aim of reaching the targets established in the State and federal levels.^9,8,15

For achieving that, it is needed training, qualification and to sensitize the healthcare professionals involved in the process of notification, in addition to the correct filling of the forms, the improvement of the surveillance and the data evaluation, once the information generated is essential for the planning and monitoring of the actions and services related to the control of TB.^{16, 21,22, 23}

The timeliness of TB data notification was considered satisfactory according to the CDC¹⁰ guidelines, once more than 70% of the notifications occurred in up until seven days after the diagnosis. It differs from the findings of other studies carried out in the country, in which the timeliness of notification was classified as regular and poor for cases of TB and HIV co-infection.^24,12,15

The study presents a few limitations, such as not considering separately for analysis the cases that had multiple notifications. As well as using guidelines for the evaluation of general health information systems, not a specific one for TB, and that are outdated (from 2001). In addition to the indiscriminate use of different analysis attributes of the database, without taking into consideration the importance of the data preparation or treatment process for the evaluation. In many circumstances, the easiness of use of the attribute limits the interpretation and comparability of the results.

Despite those limitations, the data quality analysis from TB-HIV coi-nfection cases registered in SINAN may guide the strategies for the improvement in the notification process and TB-HIV case follow-up and favor the interruption of the chain of transmission. The present study showed that the methodological process used was effective to improve the quality and timeliness of the data.

It is noteworthy that most of the information systems already perform some methodological procedures, but some inconsistencies are not identified in those processes.¹⁶. For that, an improved elaboration of methodological paths for the data evaluation is necessary, databases integration, in addition to publicize the results of this yearly analysis of the performance indicators from the information systems, as well as expanding the partnership with the academia, aiming to generate improvements and facilitate the analysis process.¹⁶

We recommend broadening the dialogue among managers, the health services²⁵and the academia for the revision of technical documents, classification of mandatory variables in SINAN, professional qualification, and the implementation of strategies in the methodological processes of qualified information production, searching to minimize unfavorable outcomes with the aim of contributing to the control and elimination of TB.