Profi le of tuberculosis patients progressing to death , city of São Paulo , Brazil , 2002

MÉTODOS: Estudo descritivo, abrangendo todos os óbitos por tuberculose (N=416) ocorridos em 2002, entre maiores de 15 anos. Os dados analisados foram obtidos do Sistema Municipal de Informações de Mortalidade, prontuários hospitalares, Serviço de Verifi cação de Óbitos e Sistema de Vigilância de Tuberculose. Os cálculos dos riscos relativos e intervalos de confi ança de 95% (IC 95%) tiveram como referência o sexo feminino, grupo de 15 a 29 anos, e os naturais do Estado de São Paulo. A análise comparativa usou o teste do qui-quadrado de Pearson e o exato de Fisher para variáveis categóricas e o teste Kruskal-Wallis para variáveis contínuas.


INTRODUCTION
Brazil ranks among the countries with the highest rates of tuberculosis (TB) morbidity and mortality with around 85,000 new TB cases and 6,000 deaths every year. 6From 1980 to 1995, the proportion of deaths associated to TB compared to total deaths due to infectious diseases (International Classifi cation of Diseases -10 th Revision [ICD-10]; Chapter I) increased from 10.1% to 15.5%. 13a For the city of São Paulo, TB morbidity and mortality rates are close to mean rates reported nationwide; however, they are higher than that seen in the interior of the State of São Paulo a (Galesi 1998).In 2000, the city of São Paulo had an incidence rate of 65 per 100,000 inhabitants, 60% of them new bacillary TB cases and a mortality rate of 5.7 per 100,000.Incidence rates varied widely by districts ranging from 18.7 to 261.1 per 100,000, while mortality rates were nil in some districts and as high as 20.9 per 100,000 in the poorest districts.b Data available on TB control activities in the city of São Paulo are consistent with the seriousness of this endemic, i.e., in 2000, there were about 70% of cure, 20% of treatment default and 13% of case-fatality rate.b The objective of the present study was to profi le adult patients dying of TB in the city of São Paulo with respect to biological, environmental and institutional factors.

METHODS
The city of São Paulo has approximately 10 million inhabitants and is characterized by a wide-ranging Human Development Index (HDI), from 0.245 in poorer districts to 0.811 in well-off ones.c This is a descriptive study including all patients, both males and females, aged 15 years or more, living in the city of São Paulo who progressed to death due to any clinical form of TB as a main cause of death (codes A15 to A19, ICD 10) between January and December 2002.

RESULTS
A total of 416 deaths due to TB as primary cause were identifi ed in 2002.The most common clinical forms were pulmonary (77.9%) and disseminated (17.5%).Of them, 44.1% were untreated cases, 51.1% received regimen 1 (isoniazid, rifampicin and pyrazinamide) and 4.8% regimen 1R (isoniazid, rifampicin, pyrazinamide, and ethambutol); 20.2% and 38.5% were treated for at least one week or up to one month, respectively, before dying.
Among those treated patients, 82/232 (35.3%) had past history of TB, of which 50% had defaulted prior treatment and 34/82 (41.5%) had TB in the last two years.For 30.4% TB diagnosis was made only after death.Of all deaths identified, 86.0%, 11.1% and 2.9% took place in hospitals, at home and in a public road, respectively.Of those who died in a hospital, 20.1% and 43.1% died within the fi rst 24 and 72 hours after admission, respectively.Criteria for diagnostic confirmation were: bacteriological examination in 31.8%,anatomopathology (macroscopic) examination in 38.9%, clinical-radiology evaluation in 27.2% and histopathology examination in 2.1%.Of 416 deaths, 19.5% received home visits to search for TB cases among household contacts.Of those cases not reported (n=206; 49.5%), 187 were untreated.Address information was missing for 15 patients (3.6%), and personal identifi cation was also missing for six of them and none of these cases had been reported, suggesting they were homeless.
Median age was 51 years (16 to 98 years), 53 years in women (17 to 98 years) and 50 in men (16 to 93 years) (p> 0.05); and 75.5% of deaths were in males.
Of 202 deaths with schooling information, 51.9% had less than four years of schooling, 32.7% had four to seven, and 15.4% had more than seven.Schooling in those originally from the State of São Paulo was 37.2%, 43.0% and 19.8%, respectively; and in those born in other Brazilian states was 63.6%, 26.1% and 10.2% (p<0.005),respectively.
Overall TB mortality rate was 5.1 per 100,000 inhabitants per year, 8.3 in men and 2.3 in women, and increasing with age (Figure 1).Mortality rates in those originally from the State of São Paulo, in the south and central-west regions and in a combination of the north, northeast, and southeast regions, excluding the State of São Paulo, were 2.5, 3.7 and 6.1 per 100,000 inhabitants per year, respectively.Taking the former as reference, relative risks were 1.48 (95% CI: 0.79;2.66)and 2.48 (95% CI: 1.98;3.01),respectively.There were no deaths reported in 8/96 (8.3%) districts of the city, whereas in 11/96 (11.5%) mortality rates were equal to or higher than twice the mean rates in São Paulo, reaching as high as 34.3 per 100,000 inhabitants per year.Figure 2 illustrates the distribution of mortality rates and HDI per district.The percent distribution of deaths shows that 2.6%, 52.2%, 34.6% and 4.1% lived in districts with HDI below 0.40, 0.40 to 0.50, 0.51 to 0.69 and above 0.69, respectively.TB patients also had diabetes (16%), chronic obstructive pulmonary disease (19%), HIV infection (11%), smoking (71%) and alcohol abuse (64%).Of all cases, 84.6% had respiratory symptoms and 89.3% had weight loss.
When those who were untreated or treated for up to one week were classifi ed as not effectively treated, no signifi cant differences were found between treated and untreated patients by gender, age, marital status, schooling and ethnicity (p>0.05)(Table 1).
After a similar comparison was carried out for diabetes, cancer, chronic obstructive pulmonary disease, HIV co-infection, pulmonary symptoms at the time of diagnosis, weight loss, smoking and alcohol abuse, it was evidenced that untreated patients were more likely to have HIV infection (p<0.005) and history of alcohol abuse (p<0.01)(Table 2).

DISCUSSION
A decline in TB mortality has been seen in the city of São Paulo since 1996, which could be in part attributed to the introduction of new antiretroviral therapies.These treatments have reverted the growing trends started from mid-80s due to the impact of TB-HIV co-infection. 1,2espite this mitigation, TB has remained a major cause of death in the city, affecting mostly those living in lower HDI districts, evidencing a strong impact of socioeconomic factors, as seen in other Brazilian capitals. 14t, the data found in the present study reveal only part of the problem.If TB deaths as an associated cause were also included, the observed magnitude would be dramatically greater. 17In addition, the results are likely to be underestimated since the number of deaths among those people presumably homeless was relatively low based on social indicators of the city of São Paulo.a On the other hand, no reporting of deaths among patients originally from other Latin American countries contrasts with the current growing migration fl ows from areas with high TB prevalence. 16close examination of the characteristics of patients progressing to death revealed that mortality rates by gender and age groups found in the present study were corroborated in the literature. 14,17,20Higher TB risk seen among the elderly was probably due their lower immunity, more diffi cult diagnosis of TB as well as to the fact that older cohorts were more exposed to infection in the past. 22gher TB death rates found among migrants from other Brazilian regions may be because they are usually older, 2,5 belong to the poorest segments of society and come from areas with high TB prevalence. 3gh TB prevalence among patients with diabetes, chronic obstructive pulmonary disease, smokers and those with past history of alcohol abuse suggest that biological, socioeconomic, and behavioral factors played a role in increasing their vulnerability and favored TB progressing to more severe forms and death. 3,5,12,22e prevalence of TB-HIV co-infection found in the study is lower than that reported in studies conducted in 1990s, 2,20 but similar to Oliveira et al 15 recent fi ndings of reduced mortality in TB-HIV co-infected people probably due to the introduction of new highly active antiretroviral therapy.
The high proportion of cases diagnosed after death or untreated, treated for less than a month and progressing to death right after hospital admission make clear the failure of health services to identify and timely treat a substantial number of TB patients in the city of São Paulo.These characteristics make these cases potentially preventable and they should be a priority in targeted public health interventions.
Allied to that, the predominance of bacillary forms and the small proportion of cases identifi ed in primary care services and of those receiving home visit to improve treatment compliance scale up the risk of disease transmission among those exposed, reducing or neutralizing the impact of TB control activities. 8,11Similarly to that seen in other Brazilian regions, 20   Low schooling of patients progressing to death may have contributed to their inability to perceive the disease.But a recent study in a Brazilian capital city did not show any association between diffi cult access to health services and delay in TB diagnosis and treatment. 18This fi nding is consistent with free and universal access to TB diagnosis and treatment in Brazil and broad coverage provided in primary care services in the city of São Paulo.Since a large proportion of patients here studied showed TB typical signs and symptoms, it suggests that health providers failed to suspect TB in high-risk groups or in those with clinical presentations indicative of infection.
The high proportion of patients with past history of TB in the two years prior to their death and treatment default, and the fact that some of them have been treated with 1R regimen (isoniazid, rifampicin, pyrazinamide, and ethambu- tol) allow to assuming that, at least, part of these deaths may be associated to multidrug-resistant Mycobacterium tuberculosis. 5,21In addition, it suggests inadequate follow-up of patients and their close contacts during treatment and for at least two years after treatment.
The high prevalence of patients infected with Mycobacterium tuberculosis hinders a signifi cant decline in TB rates in the short run in Brazil.However, the study fi ndings indicate that widely implementing the Directly Observed Therapy Short-Course (DOTS) strategy and prioritizing migrant population, districts with the lowest HDI as well as targeting those at higher risk of progressing to severe disease will favor a rapidly reduction of TB mortality as achieved by other countries where TB was also a serious public health concern. 6,7,19ee and universal access to TB diagnosis and treatment, broad coverage of primary care services and low prevalence of multidrug-resistant Mycobacterium tuberculosis 4,21 will create favorable grounds for rapidly reducing TB mortality in Brazil.
Hence, it would be advisable to change the DOTS strategy in São Paulo since it has aimed at providing universal coverage to TB patients but lacks a special focus on high-risk groups for TB death.On the other hand, there is a need for further studies to better understanding TB death predictors in the Brazilian scenario.From an operational perspective, given Brazil's large population size, huge poverty-stricken areas and high incidence rates of TB, it is crucial to provide adequate infrastructure for diagnosis which could facilitate systematic search of TB cases among patients with respiratory symptoms seeking care at primary health units.
about half of TB deaths were underreported cases and thus gone unidentifi ed by TB Control Program before their deaths.a Department of Development, Labor and Solidarity.Human Development Index -HDI.São Paulo; 2002.

Figure 1 .Figure 2 .
Figure 1.Tuberculosis mortality rate due to all clinical forms in individuals aged 15 or more living in the city of São Paulo, according to age group and gender.São Paulo, Brazil, 2002.

Table 1 .
Distribution of deaths due to all tuberculosis clinical forms, in untreated and treated patients, among individuals aged 15 or more, according to sociodemographic characteristics.São Paulo, Brazil, 2002.N=416 * TB patients progressing to death ** Untreated patients or those who did receive treatment for less than seven days

Table 2 .
Distribution of deaths due to all tuberculosis clinical forms, in untreated and treated patients, among individuals aged 15 or more, according to comorbidities, clinical conditions and presentations.São Paulo, Brazil, 2002.N=416