Impact of the GeneXpert® MTB/RIF rapid molecular test on tuberculosis detection: temporal trends and vulnerable territories*

Objective: to assess the impact of the GeneXpert® MTB/RIF rapid molecular test on tuberculosis detection, to analyze the temporal trend of the event and to identify vulnerable territories in a Brazilian municipality. Method: an ecological study carried out in Ribeirão Preto, São Paulo, Brazil, a municipality considered a priority in tuberculosis control due to the high number of cases. To classify the temporal trend, the Prais-Winsten method and the Interrupted Time Series were used to identify changes in the disease incidence. Kernel intensity analysis was applied to identify vulnerable areas. Results: the temporal trend of tuberculosis decreased by 18.1%/year and by 6.9%/year for children under 15 years old. The North District decreased by 6.67%/year and the East District increased by 17.5%/year in the incidence of tuberculosis. Resistant tuberculosis, after the implementation of the Rapid Molecular Test, increased by 0.6% per year. The South and West Districts showed a higher density of cases, with a range from 45 to 79 tuberculosis cases per square kilometer (km2). Conclusion: although resistant tuberculosis is not a problem in the scenario, the study showed an increase in its incidence, which puts it on alert. The use of spatial analysis enabled the identification of priority areas, putting them in evidence for health surveillance actions.


Introduction
Tuberculosis (TB), a communicable disease, is one of the top 10 causes of death worldwide and the predominant reason for death by a single infectious agent.
It is estimated that, approximately, one third of the world population is infected with Mycobacterium tuberculosis and at risk of developing the disease (1) .
According to the World Health Organization, in 2018 there were approximately 10 million new cases in the world, a number that has been maintained in recent years, with 57% of these cases being men, 32% women, and 11% people under 15 years of age. In that same year, an estimated 1.2 million deaths due to tuberculosis and 251,000 deaths due to tuberculosis and Human Immunodeficiency Virus (TB-HIV) coinfection (1) .
Brazil occupies the 20th position among the 30 countries that concentrate 90% of the global burden of tuberculosis. In 2018, more than 73,000 new cases were recorded in the country, indicating a detection rate of 35.0 cases/100,000 inhabitants, 13,610 cases of tuberculosis retreatments and just over 500 new cases of drug resistance (2) .
In general, between 2009, when the incidence was 38.3 cases/100,000 inhabitants, and the year 2018, it was possible to observe a mean annual drop of nearly 1% in the tuberculosis incidence rate in the country; however, it is worth noting that the incidence of the disease increased between the years 2017 and 2018, when compared to the years 2014 to 2016 (3) . The mortality rate, in the same period, was 2.2 cases/100,000 inhabitants (3) .
In view of the limitations of the conventional diagnostic tests, the Rapid Molecular Test for Tuberculosis (RMT-TB) was carried out using the GeneXpert ® MTB/ RIF system, used for the detection of Mycobacterium tuberculosis and resistance to rifampicin (4) . This system was recommended by the World Health Organization in 2010 (4) , being incorporated by various health systems and most Latin American countries for the diagnosis of tuberculosis. In Brazil, the test was approved by the National Commission for the Incorporation of Technologies year, with the acquisition of 160 pieces of equipment distributed throughout the territory (4) .
According to the World Health Organization, in samples where the sensitivity of smear microscopy was 65%, the RMT-TB showed a sensitivity of 88%, increasing by 23% the detection of tuberculosis. The World Health Organization recommends conducting operational research studies aimed at evaluating the contributions of the RMT-TB to the health systems such as expenses, impact for the patient and society (5)(6)(7) ; however, no studies were found in the literature that sought to assess the impact of the RMT-TB in the detection of tuberculosis cases in normal circumstances, of health service activities.
The study aims to assess the impact of the GeneXpert ® MTB/RIF rapid molecular test on tuberculosis detection, to analyze the temporal trend of the event and to identify vulnerable territories in a Brazilian municipality.

Method
An ecological study (8) carried out in Ribeirão Preto, located 314 kilometers (km) from the capital of the State of São Paulo, Brazil, which has an approximate area of 650 km² and a demographic density of 995.3 inhabitants/ km² (9) . It is noteworthy that this city was selected as the setting for the study because it is considered one of the priority municipalities, due to the high number of tuberculosis cases. As for the care for tuberculosis patients in the municipality, the BHU are responsible for conducting an active search for respiratory symptoms, performing sputum smear microscopy and/or x-ray request (11) . The treatment and follow-up of tuberculosis cases are carried out in specialized infectology outpatient clinics and do not occur in a decentralized manner in the municipality (11) .
The study population consisted of tuberculosis cases reported in the Tuberculosis Patient Control System (TBWeb), which consists of an online system in which municipal health managers report tuberculosis cases.
Data collection was carried out with the approval of the People whose diagnosis were made in another municipality were also excluded.
Tuberculosis cases were separated, in order to verify the behavior of the time series in different groups, but it is worth mentioning that the same case may be inserted in more than one group, according to www.eerp.usp.br/rlae 3 Berra TZ, Bruce ATI, Alves YM, Ramos ACV, Giacomet CL, Arcêncio RA. Regarding the analysis plan, it is worth noting that time series are characterized as a set of observations obtained, sequentially, over time (12) . Thus, the annual incidence rate of the tuberculosis groups and the notified cases grouped by administrative regions of the municipality were first calculated, considering the number of cases in the numerator, the population and the multiplication factor per 100,000 inhabitants in the denominator. Subsequently, the rates were converted to logarithmic notation (log10) in order to reduce data amplitude, using Microsoft Office The purpose of this technique is to assess whether, when an intervention occurs, there is an immediate impact (change in level) and/or a progressive impact (change in trend) in the values of the series (12-13) . The software used for this analysis was also STATA version 14. The incidence rates were calculated month by month, the level was called "intervention" and the impact of the implementation of the RMT-TB was called "progressive post-intervention".
The diagnosis of tuberculosis in Ribeirão Preto was performed through sputum smear microscopy and culture.
In November 2014, following the recommendation of the Ministry of Health, the diagnosis of the disease was initiated in the priority municipalities through the RMT-TB carried out by the GeneXpert ® MTB/RIF (14) system, which is an automated, simple, quick and easy-to-perform test in laboratories.
Therefore, 2014 was the cutoff point considered in the study, in order to verify whether, after the implementation of the diagnosis of tuberculosis through this new test, there was a change in the incidence of this disease in the municipality studied.
In the spatial analysis stage of the study, primarily with the objective of verifying the spatial dependence of the analyzed events, the Global Moran Index (GMI) was performed using the ArcGis software, version 10.5, based on inferential statistics, whose null hypothesis states that the event is randomly distributed in space, that is, there is no spatial dependence. If the result is statistically significant (p<0.05), the alternative hypothesis that indicates spatial dependence of the analyzed events is accepted. In numerical terms, the GMI can vary between -1 and +1, with negative values indicating the existence of inverse autocorrelation and positive values indicating direct correlation (15) .
Subsequently, the geographical coordinates (latitude and longitude) of the tuberculosis cases were obtained using the Google Earth Pro tool and the georeferencing was performed using the ArcGis software, version 10.5, the census sector of the municipality being used as a unit of analysis. The same procedure was carried out for the 49 health units in the city.

Subsequently, a point density analysis defined as a
Kernel intensity estimator was performed, which consists of an exploratory interpolation method based on the definition of circular areas of influence around points of occurrence of a phenomenon, generating a density surface for the identification of vulnerable areas (16)(17)(18) .
The Kernel estimator has as its basic parameters the radius of influence, which defines the neighborhood of the point to be interpolated and controls the smoothness of the generated surface and an estimation function, with smoothing properties of the phenomenon.
Thus, the Kernel estimator is very useful to provide an overview of the distribution of the sample points, as well as indicating the occurrence of clusters (17)(18) . In this way, thematic maps of the density distribution of the  were classified as stationary, as shown in Table 1.
The temporal trend of tuberculosis in the municipality was classified as decreasing, with a reduction of 18.1% per year (95% CI=-1.14 to -32.23) and, also, decreasing for Table 1

Discussion
The study aimed to assess the impact of GeneXpert ® MTB/RIF in the detection of tuberculosis, to analyze the temporal trend of the event, and to identify vulnerable territories in Ribeirão Preto, State of São Paulo, Brazil, a city considered a priority in the control of the disease.
The temporal trend of tuberculosis in this endemic municipality in the inland of São Paulo, was classified as decreasing. According to the World Health Organization, globally, the incidence and mortality rates due to tuberculosis are falling; however, the disease remains an important public health issue (1) . This drop in its rates can be the reflection of the strategies launched to eliminate the disease, such as the End TB Strategy, which is based on three pillars: integrated care and prevention, centered on the patient; bold policies and support systems, in addition to intensifying research and innovation (19) .  (21)(22)(23) .
The big challenge related to childhood tuberculosis is its diagnosis (21)(22)(23) , hindered by the absence of an exam  Figure 2. that can be considered as gold standard, once again, raising the question about whether the disease indices are really decreasing.
According to the results observed in another study (24) , the sensitivity of RMT-TB in children was 80%, with induced sputum when compared to culture (gold standard); even if gastric lavage is considered, sensitivity reaches 90%, with culture as the standard test. In contrast, the sensitivity of smear microscopy in children is close to zero when compared to culture; if gastric lavage is used, the sensitivity of sputum smear increases slightly, around 20%.
Another hypothesis that is raised in view of the results obtained is that, even not showing a significant of studies (4)(5)(6) show that, in samples with negative smear microscopy, the sensitivity of the RMT-TB for a sputum sample is 72.5% and that, for three samples, it reaches almost 91%. Specificity reaches 99%. The test also detects resistance to rifampicin with 99.1% sensitivity and excludes resistance with 100% specificity.
Another investigation (27) showed the high positive predictive value for resistance to rifampicin (90.2%) in countries where the prevalence of tuberculosis is low. In addition, it also demonstrated, that 82% of the cases of resistance to rifampin diagnosed using RMT-TB were subsequently confirmed as cases of drug-resistant tuberculosis.
Based on the assumption that there can be underdetection of cases, geotechnologies can be useful tools to assist in the identification of priority areas and, thus, improve the effectiveness of control measures and, also, reduce operating costs (28) .
The GMI confirmed the spatial dependence of  to rifampicin, the main drug used to treat the disease (3)(4)(5) .
In addition, it can be operated in the same physical space where smear microscopy is performed and does not require special conditions for biosafety (3)(4)(5) .
Two bacilloscopies are recommended by the National Tuberculosis Control Program, to achieve a sensitivity of 70% and, therefore, represent 80% of the value of a test performed by the Xpert ® MTB/RIF system, which has 88% sensitivity. Therefore, the Xpert ® MTB/RIF system is considered a technology with an accurate result, costeffective in diagnosis and faster than the conventional tests (36) .
This study advances knowledge as it presents the impact of RMT-TB on the routine of the health service, It is worth mentioning that, with the use of spatial analysis tools, it is possible to define priority areas for control actions in the territory and allow for the early diagnosis of the disease, since tuberculosis still persists as the main cause of death due to infectious diseases in the world, and social policies directed at the most disadvantaged groups, combined with multi-sectoral and interdisciplinary actions, are indispensable for the control of the disease.

Conclusion
It was not possible to identify changes in the incidence of sensitive tuberculosis after the implementation of RMT-TB in the municipality. Although 3 years is a relatively short time to indicate the effect of this new technology, the evidence points to the impact of GeneXpert ® MTB/RIF system in the detection of resistant tuberculosis, which at first appeared not to be problematic in the scenario.
However, the present study indicated an increase in the rates of this condition after the beginning of the diagnosis by means of RMT-TB, which automatically detects the resistance of the bacillus to rifampicin.
With the use of spatial analysis tools, it was possible to identify areas that must be considered a priority in the municipality, so that the managers have to prioritize these regions in actions to combat tuberculosis, with an active search for respiratory symptoms aiming at breaking the chain of bacillus transmission and at disease control.