Risk factors associated with multidrug-resistant tuberculosis in Espírito Santo, Brazil

ABSTRACT OBJECTIVE To analyze the prevalence and factors associated with multidrug-resistant tuberculosis in Espírito Santo, Brazil. METHODS This is a cross-sectional study of cases of tuberculosis tested for first-line drugs (isoniazid, rifampicin, pyrazinamide, ethambutol, and streptomycin) in Espírito Santo between 2002 and 2012. We have used laboratory data and registration of cases of tuberculosis – from the Sistema Nacional de Agravos de Notificação and Sistema para Tratamentos Especiais de Tuberculose. Individuals have been classified as resistant and non-resistant and compared in relation to the sociodemographic, clinical, and epidemiological variables. Some variables have been included in a logistic regression model to establish the factors associated with resistance. RESULTS In the study period, 1,669 individuals underwent anti-tuberculosis drug susceptibility testing. Of these individuals, 10.6% showed resistance to any anti-tuberculosis drug. The rate of multidrug resistance observed, that is, to rifampicin and isoniazid, has been 5%. After multiple analysis, we have identified as independent factors associated with resistant tuberculosis: history of previous treatment of tuberculosis [recurrence (OR = 7.72; 95%CI 4.24–14.05) and re-entry after abandonment (OR = 3.91; 95%CI 1.81–8.43)], smoking (OR = 3.93; 95%CI 1.98–7.79), and positive culture for Mycobacterium tuberculosis at the time of notification of the case (OR = 3.22; 95%CI 1.15–8.99). CONCLUSIONS The partnership between tuberculosis control programs and health teams working in the network of Primary Health Care needs to be strengthened. This would allow the identification and monitoring of individuals with a history of previous treatment of tuberculosis and smoking. Moreover, the expansion of the offer of the culture of tuberculosis and anti-tuberculosis drug susceptibility testing would provide greater diagnostic capacity for the resistant types in Espírito Santo.


INTRODUCTION
The emergence of drug-resistant strains of Mycobacterium tuberculosis (Mtb) is a great challenge for the elimination of tuberculosis (TB) worldwide. The World Health Organization (WHO) estimates that 3.5% of new cases and 20.5% of previously treated cases of TB are multidrug-resistant tuberculosis (MDR-TB). This is defined as in vitro resistance to at least two of the most potent drugs used in the conventional treatment of the disease, rifampicin (R) and isoniazid (I) 24 .
Mortality and incidence of cases of TB have decreased in Brazil over the last decade. However, Brazil remains among the 22 countries with high burden of the disease in the world and faces major difficulties related to public health systems in promoting an effective control of new cases of TB 16 . Approximately 4,064 cases of MDR-TB were reported in Brazil between 2000 and 2008. In 2014, 260 new cases of monoresistance (resistance to one anti-tuberculosis drug), 81 cases of polyresistance (resistance to two or more anti-tuberculosis drugs, except R and I), 374 cases of MDR-TB, and 56 cases of XDR-TB (cases which, in addition to resistance to R and I, present additional resistance to a fluoroquinolone and an injectable second-line drug 16 ) were notified in the Sistema de Informação de Tratamentos Especiais de Tuberculose (SITETB -Information System of Special Treatment of Tuberculosis). Although Brazil has an undervalued number compared to other countries 24 , the problem of resistance in the country is more focused on its potential to be spread than on its magnitude 6 .
The State of Espírito Santo has presented a 13% reduction in the incidence rates of cases of TB in the last 10 years. The incidence rate was 40.6/100,000 inhabitants in 2003 and 35.2/100,000 inhabitants in 2012. The rates of cure and abandonment of treatment in the same period were, on average, 78% and 8%, respectively 16 . Most cases (60%) are concentrated in the Metropolitan Area of Greater Vitória (RMGV), formed by the capital, Vitória, and six other municipalities of great political and economic importance. Since 2002, laboratory data for anti-tuberculosis Drug Sensitivity Testing (DST) are interconnected and available to public health services. In addition, four municipalities of Greater Vitória offer universal culture for pulmonary samples, which has allowed the increase of 23.0% of the diagnosis of TB in the region 19 . However, data on resistance to Mtb are little explored 22 .
Thus, this study has aimed to describe cases of TB resistant to first-line drugs and identify factors associated with the presence of resistance in Espírito Santo.  We have classified as "resistant" individuals who had isolated strains of Mtb with in vitro resistance to at least one of the first-line drugs and as "non-resistant" individuals who did not show resistance to any of the drugs tested. We have excluded cases diagnosed as non-tuberculous mycobacteria and individuals not living in Espírito Santo.

This
The data on DST have been collected using software which stores laboratory data, called TB Notes. This software is present in two reference laboratories of the State to carry out culture and DST, being they the mycobacteriology laboratory of the Núcleo de Doenças Infecciosas of the Universidade Federal do Espírito Santo (NDI-UFES) and the LACEN-ES. Both are certified to perform these procedures. The LACEN-ES is under the supervision of the Centro de Referência Professor Hélio Fraga of Rio de Janeiro (CRPHF-RJ), a national reference for the performance of DST for first-and second-line anti-tuberculosis drugs.
For the description of the socio-demographic and clinical/epidemiological characteristics and for the analysis of risk factors for resistance, the data were researched on the main information systems (IS) that register cases of TB: the Sistema de Informação de Agravos de Notificação (SINAN -Information System of Reportable Diseases) and SITETB.
The sociodemographic variables used were: gender (male, female), race (white, non-white), years of study (illiterate, one to four years, five to eight years, nine to twelve years, twelve years or more), age (in years: < 20, 20 to 39, 40 to 59, 60 or more), and institutionalization (yes, no). Regarding health history, we analyzed the presence/absence of HIV infection, alcoholism, diabetes mellitus, mental illness, smoking, and use of illicit drugs.
We selected for the analysis of characteristics of the disease and its treatment the variables: type of case (new case, recurrence, re-entry after abandonment, transfer), number of prior treatments of TB (none, one, two, three or more), radiological examination of the thorax (suspected, normal), tuberculin test (no reaction, weak reaction, strong reaction), clinical type (pulmonary, extrapulmonary, pulmonary + extrapulmonary), sputum smear microscopy in the diagnosis (positive, negative), sputum culture in the diagnosis (positive, negative), other material culture in the diagnosis (positive, negative), and ending (cure, abandonment, death by TB, death by other causes, transfer, MDR-TB). We verified the existence of more than one record of treatment for the same individual in the study period to ensure the reliability of the information about history of prior treatment of TB and number of treatments carried out.
The rates of resistance (to any drug and MDR-TB) were expressed as the proportion of resistant individuals among those tested. We used Pearson' s Chi-square test for the categorical variables and t-test for the numeric variables to observe statistical differences between the groups of cases of resistant and non-resistant TB. The multiple logistic regression analysis was used to calculate the adjusted odds ratios (OR). We considered 5% significance for the input of variables in the model and the method ENTER to choose the variables (which considers all variables included in the model). The significance level adopted in all analyses was 5%.
We created a database as a spreadsheet in Microsoft Office Excel  , and we summarized the data in the statistical program SPSS, version 18 (Chicago, IL, USA).
This study has been approved by the research ethics committee of the Centro de Ciências da Saúde of the Universidade Federal do Espírito Santo (Process 201.111/2013).

RESULTS
Approximately 15,851 cases of tuberculosis were reported in Espírito Santo between 2002 and 2012. We have found 1,669 individuals in the TB Notes with DST performed in the same period. Of these individuals, 89% had no in vitro resistance to first-line drugs, and 10.6% presented some type of resistance (Figure).
The monoresistance to isoniazid followed by streptomycin were the most common (2.1% and 0.9% of the cases tested, respectively). The rate of multidrug resistance was 5%, and most of them showed resistance to isoniazid + rifampicin (2.3%) ( Table 1). Of the individuals who underwent DST, 93 were not located in the IS for the treatment of tuberculosis, and of these three were classified as resistant. Therefore, we have collected the sociodemographic, clinical, and epidemiological information of 1,576 individuals ( Figure).
Most individuals (84%) were from the nine priority municipalities for TB control, and 67% were from the four municipalities that have universal culture for pulmonary samples (Cariacica, Serra, Vila Velha, and Vitória). In the comparison between the resistant and non-resistant groups, no statistically significant differences were found for the sociodemographic variables. There was a predominance of males, non-whites, and low education level -most (39%) had complete elementary school ( Table 2). Average age (in years) in the non-resistant group was 36.8 (SD = 14.5) and in the resistant group it was 37.6 (SD = 13.5; p = 0.482).
There was no difference in the distribution of cases of HIV infection between the groups. Statistical differences were observed for the variables alcohol consumption (p = 0.021) and smoking (p < 0.001), both more common among the resistant than the non-resistant individuals (45% versus 34% and 34% versus 9%, respectively) ( Table 2).
Statistical differences were observed for the clinical and epidemiological variables. For the variable type of case (p < 0.001), the non-resistant group showed the highest percentage of new cases of TB compared to the resistant group (77% versus 45%). On the other hand, the number of cases of retreatment was higher in the resistant group (53% versus 21%).
The registration of one previous episode of TB was more common among resistant individuals than among the non-resistant individuals (30% versus 11%). The positive culture at the time of diagnosis (p = 0.023) was more frequent in the resistant group than in the non-resistant group, 95% versus 85%. The proportion of cured cases among the non-resistant group was greater than among the resistant group (83% versus 77%) ( Type of case, smoking, and sputum culture at the time of diagnosis were important associated factors for resistance to anti-tuberculosis drugs after the analysis adjusted for variables included in the logistic regression model (

DISCUSSION
We have found strong association between the number of previous treatments of TB, smoking, and positive culture at the time of diagnosis with cases of resistant TB. There is a consensus in the literature that previous treatment of TB is a strong risk factor for bacterial resistance to anti-tuberculosis drugs 8,9,11,17 . Our results have shown that cases of recurrence have twice the odds ratio for the occurrence of resistance in relation to retreatment by re-entry after abandonment. This can be explained by the increased contact with anti-tuberculosis drugs.
A study conducted in eleven countries has shown that the longer the time of exposure to anti-tuberculosis drugs, the greater the chance of occurrence of resistance 8 .
Another risk factor presented in this study was the habit of smoking. Despite increasing the risk of active TB 12,20 , we have found no sufficient evidence in the literature that smoking is an important risk factor for resistance 7 . However, studies show smoking as a factor associated with the failure of treatment of TB. The time to sputum culture conversion from positive to negative among smokers is greater than among nonsmokers after the second month of the start of treatment with both first-line 13,23 and second-line drugs 1,14 .
There is a strong association between smoking and recurrence of TB. A consistent study has shown that smokers are 2.5 times more likely to have a recurrence than nonsmokers 2 . Thus, the relationship between smoking, previous treatment of TB (recurrence), and resistance to anti-tuberculosis drugs deserve to be better investigated.
The occurrence of one previous treatment was more frequent among previously treated individuals. The presence of universal culture covering areas of high population concentration in Espírito Santo may have contributed to the earlier suspicion and confirmation of resistant cases 22 . These data are probably related to the greater number of positive cultures at diagnosis among resistant cases. The recent seizure of a new technology based on a rapid molecular testing for TB (RMT-TB), as an important tool in detecting the resistance to R for new cases of TB, does not exclude the culture and DST by conventional methods in Brazil 16 .
Universal culture is a recommendation of the WHO, although it is not yet a reality in many places in Brazil and in the world 24 .
Among the resistant cases, 45% had never been treated for TB. This may suggest the evidence of active transmission in the population, as we also have not found association with other variables commonly reported in the literature that may point to groups at higher risk for resistant TB, such as: age, gender, HIV infection, and history of institutionalization 8,18 . Similar result has been described in an European study, which has assessed cases of MDR-TB in sixteen countries, showing that 52.4% of them had never received anti-tuberculosis treatment (59.2%, 74.4%, and 38.7% in countries with low, intermediate, and high incidence of TB, respectively) 11 . This can be explained by migratory movements of individuals in search of better living conditions. Espírito Santo is the main route linking the Northeast and Southeast regions of the country and is located between two large States that have high incidence rates of TB, Rio de Janeiro and Bahia 16 .
For this hypothesis, we must consider the possibility of underreporting of cases. This study is based on secondary data and no interview or recent contact was performed with the patients reported in the period of study to investigate the history of treatment of TB.
The rate of resistance to any drug was 10.6% and for MDR-TB it was 5%. Among the Brazilian studies with secondary data, there is great variability of results, given the distinct characteristics of the populations studied. In these studies, the rate of resistance to any drug has varied from 9.4% to 19.2% and for MDR-TB it has varied from 3.4% to 15% 4,5,10,15,21 .
Data from the first national resistance survey conducted between 1995 and 1997, carried out prospectively, presented the same rate of resistance to any drug of 10.6%. However, the rate of multidrug resistance was 2.2%, half of that observed in this study 3 . A retrospective study in Israel has found results very close to those observed herein. The rate of resistance to any drug was 12.5% and for MDR-TB it was 5.8% 18 .
There was a predominance of isolated resistance to isoniazid followed by streptomycin. For the cases of MDR-TB, the highest frequency was observed for isoniazid + rifampicin. Other Brazilian studies have found a similar situation [3][4][5]15,21 . We believe that this result is a reflection of a consolidated policy of notification and treatment of cases of TB made available free of charge and offered exclusively by public health services. On the other hand, perhaps there is a different profile of resistance in regions where there has not been an effective policy of control and treatment of cases 1,6 .
Of the 1,669 cases of TB with DST, 5.6% were not located in the IS. We have not considered it a statistically significant loss for the final analyses, as we have obtained 94.4% of the data. However, this fact raises some possible explanations. The first of them is that these cases may have been diagnosed in Espírito Santo, but they were sent to other States for treatment. Another possibility would be of these individuals being outside the study period and not identified in the IS, or yet, they could have been lost in the public health system without access to appropriate treatment, characterizing primary abandonment of treatment.
Espírito Santo, following the rules of the PNCT, does not perform DST in all new cases of TB a . Therefore, the rates of resistance presented here may be underestimated. The missing information in the databases has been classified as missing in the bivariate analyses. For the logistic regression, we have used 30% of the available data. However, we consider the number of 428 individuals as enough for the final analysis.
The results presented in this study show that the partnership between Tb control programs (PCT) and health teams working in Primary Health Care needs to be strengthened. Individuals with a history of previous treatment of TB and smoking should be identified and monitored. This can contribute to reduce the adverse outcomes in the treatment of TB. Another important point is the orientation to stop smoking among individuals under treatment of TB. In addition to other health benefits, it can prevent the emergence of cases of resistant TB.
It is essential to expand the provision of culture and DST, providing greater diagnostic capacity for the types of resistant TB in Espírito Santo. The early diagnosis and treatment of these cases prevent the circulation of strains of Mtb, reducing the number of primary cases, i.e., individuals who have never been treated for TB and who get sick with resistant strains.