ORIGINAL RESEARCH INVESTIGACIÓN ORIGINAL
Social determinants and inequalities in tuberculosis incidence in Latin America and the Caribbean
Determinantes sociales y desigualdades en la incidencia de tuberculosis en América Latina y el Caribe
César V. MunaycoI; Oscar J. MújicaII; Francisco X. LeónIII; Mirtha del GranadoIII; Marcos A. EspinalIV
IDepartment of Preventive Medicine and Biometrics, Uniformed Services University of the Health Sciences, Bethesda, Maryland, United States of America. Send correspondence to César V. Munayco, email: firstname.lastname@example.org
IISocial Epidemiology, Special Program on Sustainable Development and Health Equity, Pan American Health Organization (PAHO), Regional Office of the World Health Organization (WHO), Washington DC, United States of America
IIICommunicable Diseases and Health Analysis Department; HIV, Hepatitis, Tuberculosis, and Sexually-transmitted Infections Unit, PAHO/WHO, Washington DC, United States of America
IVCommunicable Diseases and Health Analysis Department, PAHO/WHO, Washington, DC, United States of America
OBJECTIVE: To identify key social determinants of tuberculosis (TB) incidence among countries in Latin America and the Caribbean (LAC), a geographic area regarded as one of the most socioeconomically unequal in the world
METHODS: An ecological study was conducted at the country level. Data were obtained from several institutional-based sources. Random-effects regression modeling was used to explore the relationship between several social determinants indicators and TB incidence rates in 20 LAC countries in 1995-2012. Standard gap and gradient metrics of social inequality in TB incidence among countries in 2000, 2005, and 2010 were then calculated.
RESULTS: TB incidence rate trends were significantly associated with health expenditure per capita and access to improved sanitation facilities, as well as with life expectancy at birth and TB detection rate, after adjusting for other socioeconomic, demographic, and health services variables. Absolute and relative inequality in TB incidence remained mostly unchanged: countries at the bottom 20% of both health expenditure and sanitation coverage distributions concentrated up to 40% of all TB incident cases, despite a considerable decline in the overall TB incidence mean rate during the period assessed.
CONCLUSIONS: Along with the intensity of TB control (reflected by TB detection rate), both access to sanitation (as a proxy of quality of living conditions) and health expenditure per capita (either as an indicator of the level of resources and/or commitment to health care) appear to be key determinants of TB incidence trends in LAC countries. Inequalities in both health expenditure per capita and access to sanitation seem to define profound and persistent inverse gradients in TB incidence among LAC countries.
Key words: Tuberculosis; social determinants of health; social inequality; equity in health; Latin America; Caribbean Region.
OBJETIVO: Establecer los determinantes sociales clave de la incidencia de tuberculosis (TB) en los países de América Latina y el Caribe (ALC), una zona geográfica consideraba como una de las más afectadas por las desigualdades socioeconómicas en el mundo.
MÉTODOS: Se llevó a cabo un estudio ecológico a nivel de país. Los datos se obtuvieron de diversas fuentes institucionales. Mediante un modelo de regresión de efectos aleatorios se exploró la relación entre varios indicadores de determinantes sociales y las tasas de incidencia de TB en 20 países de ALC durante el periodo de 1995 al 2012. A continuación, se calcularon los valores ordinarios de la brecha y el gradiente de desigualdad social en la incidencia de TB entre países en el 2000, el 2005 y el 2010.
RESULTADOS: Las tendencias en la tasa de incidencia de TB se asociaban significativamente con el gasto per cápita en salud y el acceso a mejores instalaciones de saneamiento, así como con la esperanza de vida al nacer y la tasa de detección de la TB, tras ajustar para otras variables socioeconómicas, demográficas y de servicios de salud. La desigualdad absoluta y relativa en la incidencia de TB se mantuvo prácticamente inalterada: los países que se distribuían en el 20% inferior del gasto en salud y la cobertura de saneamiento aglutinaban hasta un 40% de todos los casos nuevos de TB, a pesar de una considerable disminución de la tasa general media de incidencia de TB durante el período evaluado.
CONCLUSIONES: Junto con la intensidad de las actividades de control de la TB (reflejada por la tasa de detección de la TB), tanto el acceso al saneamiento (reflejo de la calidad de las condiciones de vida) como el gasto per cápita en salud (ya sea como indicador del nivel de recursos o del compromiso con la atención de salud) parecen ser determinantes clave de las tendencias en la incidencia de TB en los países de ALC. Las desigualdades tanto en el gasto per cápita en salud como en el acceso al saneamiento parecen definir los gradientes inversos profundos y persistentes en la incidencia de TB entre los países de ALC.
Palabras clave: Tuberculosis; determinantes sociales de salud; inequidad social; equidad en salud; América Latina; Región del Caribe
Tuberculosis (TB) is still a public health problem in the Region of the Americas, representing the second cause of death by a unique infectious agent (1). The Region accounted for 3.3% of TB incident cases worldwide in 2012 (2). The World Health Organization (WHO) estimated an incidence rate of 28 cases per 100 000 population in the Americas, ranging from 3.8 per 100 000 in the United States to 222 per 100 000 in Haiti. In the same year, the WHO best-estimate of the TB incidence rate for Latin America and the Caribbean (LAC) was 43 cases per 100 000 population, 11 times higher than the incidence rate for North America (3).
Since 1990, the TB incidence rate in the Americas has declined 2.6% per year on average, and the Region is on track to meet the corresponding Millennium Development Goals (MDG) target (2). This rate of decline, however, has decelerated since 2008 and is considered slow, despite successful implementation of both Directly Observed Treatment Short-course (DOTS) (1996-2005) and Stop TB (2006-present) strategies by national TB programs. In an area dubiously distinguished as one of the most inequitable in the world in goods distribution (4, 5), such deceleration and slow pace of progress toward TB elimination have been attributed to the persistence of prevailing factors linked to poverty, social inequity and exclusion, and rising urbanization. These factors, in turn, generate living conditions and circumstances favorable to TB transmission, regardless of disease control measures put in place (6, 7).
Historically regarded as the social disease par excellence (8, 9), TB plummeted drastically during the 19th century in Europe-a result of social reforms and improved living conditions prompted by the Industrial Revolution and despite the absence of curative chemotherapy (10, 11). Yet, modern TB control strategies have been dominated by the germ theory paradigm. More recently however, in part stirred by the work of the Commission on the Social Determinants of Health (12), and in part by a renewed focus on TB clustering around disadvantaged individuals and communities (13-15), the push has been toward addressing the disease's more distal determinants. These are the so-called 'causes-of-the-cause' that help explain the unjust inequalities (i.e., inequities) in TB burden across the social gradient. Addressing social determinants may be more effective for achieving long-term TB control.
In order to identify potential entry points for intervention, several researchers have recently proposed a number of convergent theoretical frameworks that attempt to integrate a comprehensive model of causal pathways, mediators, and effect modifiers between distal, upstream determinants and proximate, downstream TB risk factors (11, 16-20). In line with these conceptual models-in particular Lönnroth's (18), which highlights the causal role of globalization; migration and urbanization; demographic transition; weak and inequitable economic, social, and environmental policies; low socioeconomic status and poverty; low education; and weak health systems in TB incidence and long-term TB control-the present study conducted an exploratory analysis of potential determinants of TB incidence trends in LAC countries in 1995-2012 and quantified the magnitude of associated inequality. A number of variables from the socioeconomic, demographic, health services, and TB program management domains were studied. The hypothesis was that, unless national TB programs were universally successful in TB control, more distal social and economic determinants would be prominent in driving TB incidence trends and inequality gaps and gradients at the ecological level.
MATERIALS AND METHODS
This was an ecological, country-level, observational panel data study, from which random-effects regression models were run to determine the relationship between trends in TB incidence and indicators of social determinants across LAC countries; standard gap and gradient metrics of social inequality in TB incidence over time between countries were also measured.
Yearly data for 1995-2012 was collected from different institutional sources, including the WHO Global TB Database, the PAHO Core Health Data Initiative, the WHO/UNICEF Joint Monitoring Program for Water Supply and Sanitation, the United Nations Population Divi- sion World Population and Urbanization Prospects, and the World Bank World DataBank (21-26). A total of 12 independent variables related to economic policy, social development, infrastructure, labor, urbanization, health services, and TB program management, from 20 LAC countries, were included in the study (see Table 1 for detailed variable definitions and data sources). These countries were: Argentina, Bolivia, Brazil, Chile, Colombia, Costa Rica, the Dominican Republic, Ecuador, El Salvador, Guatemala, Haiti, Honduras, Mexico, Nicaragua, Panama, Paraguay, Peru, Trinidad and Tobago, Uruguay, and Venezuela. Collectively, they account for 99.8% of the estimated and notified TB caseload in LAC.
All variables were logarithmically transformed to achieve linearity and stabilize residual variation in the random- effects regression analysis. To determine which independent variables explained TB incidence, a panel data analysis was performed based on a random-effects regression model indexing by country and time. The decision for a random- effect, rather than a fixed-effects regression model, was informed by the Hausman test, which showed that the unique errors were uncorrelated with the regressors. To test the assumptions of this model, diagnostics tests for heteroskedasticity, serial correlation (Lagram-Multiplier [LM] test), and cross-sectional dependence (Breusch-Pagan LM test of independence) were performed (27).
Selection of variables for inclusion in the model was guided by Lönnroth's theoretical framework, described elsewhere (18). Improvement in the adjusted r-squared (r2) was used to compare the fit of two models. The statistical significance of the final model was defined by the minimal model containing only variables that were significant at P < 0.05 in 2-sided t-tests, those with the highest r2, and compliant with the theoretical framework (28). R language was used to perform all the analyses and data management, and its "plm package" was used specifically to do the random-effects regression model (29, 30).
After identifying the final variables in the model, the analyses explored inequalities in TB incidence between countries by using those variables as equity stratifiers and computing standard metrics of gap and gradient inequality. These were guided by WHO methodology, described in detail elsewhere (31, 32). The inequality analyses were done independently for 2000, 2005, and 2010, to assess changes over time. Unbiased (i.e., population-weighted) TB incidence rates by country-level quartiles of each equity stratifier were computed, followed by computations of the absolute (difference) and relative (ratio) gap between TB incidence rates of the top (the most advantaged) and bottom (the most disadvantaged) quartiles as metrics of gap inequality (i.e., Kuznets-like indexes). Estimates were also made of the slope index of inequality (SII) as the metric of absolute gradient inequality, by regressing country-level TB incidence rate on a relative scale of social position, as defined by the cumulative class interval mid-point of the population ranked by the equity stratifier. This used a weighted least-squares regression model to account for heteroscedasticity of aggregated data, by applying Maddala's procedure, described elsewhere (33). Finally, the health concentration index (HCI) was computed as the metric of relative gradient inequality, by fitting by non-linear optimization a Lorenz concentration curve equation (i.e., plotting the observed cumulative relative distributions of population ranked by the equity stratifier and TB incident cases, respectively) across the countries studied, and numerically integrating the area under the curve (34).
Because this study did not research human subjects whatsoever, no ethics review was sought or necessary.
Social determinants of TB incidence
In the bivariate random-effects regression analysis, the following variables were significantly associated with reduced TB incidence rates over time: higher gross domestic product (GDP); gross national income (GNI); health expenditure per capita; life expectancy at birth; TB detection rate; access to improved water source; and access to improved sanitation facilities. Variables significantly associated with increased TB incidence rates over time were: incarceration rate and urban population growth. In the multivariate random-effects regression analysis, higher health expenditure per capita (β= -0.24, P < 0.001); access to improved sanitation facilities (β= -0.24, P = 0.0156); life expectancy at birth (β= -3.02, P < 0.001); and TB detection rate (β= -0.25, P < 0.001) were statistically significantly associated with reduced TB incidence rates over time-after adjusting for the other variables (Table 2).
Social inequalities in TB incidence
TB incidence was about 4-5 times greater in the quartile of countries with lowest health expenditure per capita than in the better-off quartile; this excess incidence amounted to 110 cases per 100 000 population in 2000 and around 85 cases per 100 000 in 2010. Similar pattern and trends were observed between extreme quartiles of access to improved sanitation facilities. Figure 1-A and Figure 1-B depict these inequality gaps, showing that the most disadvantaged countries in terms of health expenditure per capita (Haiti, Bolivia, Nicaragua, Guatemala, and Honduras) and access to improved sanitation facilities (Haiti, Bolivia, Guatemala, Trinidad and Tobago, and El Salvador) had the highest average TB incidence rates in the three time points studied. The inequality regression curves (2000, 2005, and 2010) showed an inverse non-linear gradient between TB incidence rate and the social position defined by both health expenditure and access to improved sanitation facilities (Figures 2-A and 2-B, respectively). This means that countries with lower health expenditure per capita and less access to improved sanitation facilities not only had higher TB incidence rates than better-off countries, but that this health inequality was disproportionately concentrated within the most socially disadvantaged countries. A moderate reduction of this absolute inequality was observed between 2000 and 2010; the SII went down from 88.1 excess TB incident cases per 100 000 population to 58.0 in the country gradient defined by health expenditure per capita, and from 85.8 to 63.2 in the country gradient defined by access to improved sanitation facilities. Table 3 also shows smaller inequalities in TB incidence among countries (and smaller changes over time in these inequalities as well) according to gradients of life expectancy. Interestingly, no statistically significant inequalities in TB incidence, nor changes over time among them, were observed by gradients of TB detection rates.
Finally, the health inequality concentration curves showed that the bottom 20% of the most disadvantaged countries, in terms of health expenditure per capita, concentrated up to 40% of burden of TB incidence in LAC; this situation remained unchanged in the period studied. The same social gradient in TB incidence was seen with access to improved sanitation facilities (Figure 3). In fact, the corresponding health concentration indices were around -0.30, showing very little variation between 2000 and 2010.
This study showed that in the first decade of the 21st century, LAC countries with lower health expenditure, lower access to improved sanitation facilities, lower life expectancy at birth, and lower TB detection rate have higher TB incidence rates.
Despite a limited ability to draw meaningful comparisons given the scarcity of regional studies on this topic at the country level, the findings of this study are consistent with those of Dye and colleagues who examined 134 countries globally in 1997-2006 (35). They found a reduction in TB incidence associated with higher GDP per capita, improved water source, improved sanitation, lower mortality rate among those < 5 years of age, and higher new smear+ TB cases detected. More broadly, our findings concur with those of other studies that have established an association between TB morbidity and mortality and the human development index (36, 37), a composite measure of wealth, education, and longevity.
When social determinants of health are not adequately addressed, or just plainly neglected, they cause or reinforce social stratification in the population, leading to unequal distribution, inequity, and social exclusion (4, 12). Inequalities in TB are thus driven by the unequal distribution of key social determinants, such as wealth, education, healthy nutrition, adequate housing, environmental conditions, access to employment, cultural barriers to health care, etc. (16, 38). In turn, these unequal social conditions facilitate TB transmission in the community (15, 19).
Our study also showed the presence and persistence of cross-country, non- trivial inequalities in TB incidence systematically associated to the unequal societal distribution of some key social determinants. The more disadvantaged a country was in terms of health expenditure per capita, access to improved sanitation facilities, and life expectancy at birth, and to a lesser extent, TB detection rate, the higher its TB incidence rate would be. The presence of this pervasive inequality gradient adds to the disproportionate concentration of TB incident cases at the bottom of the social hierarchy among countries: 40% of all new TB cases in LAC were focalized in the lowest quintile of most disadvantaged countries. Moreover, this study provides ecological evidence of the endurance of these social inequities in TB incidence over time.
In terms of ecological evidence on the social determination of TB incidence, our study exposed some interesting and even debatable issues. Despite the well-known correlation between health and GDP/GNI per capita (11-13, 16-19), these two highly collinear proxies of income and wealth were not retained in our final model. Instead, health expenditure per capita showed a better association with TB incidence rate over time. In spite of a likewise high collinearity between health expenditure and GDP/GNI, this may suggest that, over the period observed, the absolute level of economic resources in LAC countries may not have been as relevant for TB incidence as the specific commitment to health care, as conveyed by the level of health expenditure per capita. Indeed, a distinctive feature of our analysis was the exploitation of a panel data regression approach, which provides superior estimates in describing change over time (27, 28). In this regard, a desirable improvement to our model would have been the inclusion of precise economic expenses incurred by TB control programs in each country, had such data been available. On a more upstream, distal level, access to improved sanitation facilities and improved water sources are essential development indicators (16): it is just in those environments lacking access to these services where TB clusters and propagates. Our final model suggests that only the former remains a significant environmental determinant of TB incidence in LAC countries over the period assessed. This is consistent with the 2012 WHO/UNICEF Joint Monitoring Program assessment on progress towards the MDG 7 on access to water and sanitation, declaring that LAC had already attained target 7-C for access to safe drinking water (24).
The reported non-association between HIV/AIDS and TB incidence at the aggregated level may be explained by the fact that in LAC, most countries have concentrated HIV epidemics that have a milder effect on TB incidence. The high coverage of highly active antiretroviral therapy (HAART; roughly 70%) in LAC has also contributed to this observation (2, 39). With regard to unemployment, incarceration, and urban growth-shown elsewhere to be associated with TB incidence (11, 13, 15, 19, 20)-the fact that they were excluded from our final model despite their significance in the unadjusted one, may reflect the inability of the current level of aggregation to detect relevant contextual effects, or else differing definitions across countries.
This study had some limitations, mostly inherent to its ecological design: since country-level aggregated data were used in the analyses, these results cannot be inferred at the individual level, and we cannot claim causal relationships between TB incidence and the socioeconomic independent variables explored. Another study limitation was shaped by data unavailability and data paucity, i.e., other variables could not be included that have been shown to be related with TB: e.g., human development, gender, alcohol consumption, drug abuse, diabetes prevalence, and nutritional status. Moreover, the panel design carries the caveat of any observational study, and therefore, endogeneity and omitted variable bias cannot be ruled out. Despite these limitations, these study findings are useful for informing health decision- makers on the need to improve public policies that include reducing the social inequity gaps and gradients in TB incidence at the collective level.
Currently, countries affected by this public health problem have the challenge of promptly stopping TB transmission in the community by applying interventions aimed at tackling comorbidities and social determinants of health, building upon the current strategy of diagnosis and curative treatment (11). In LAC, more resources need to be designated to improving and strengthening access to health services, and at the same time, to improving the conditions in which people are born, grow, live, work, and age-such as better housing, jobs and wages, nutrition, water and sanitation, transportation, and the like. To the same extent to which TB has been regarded a social disease (8, 9), we contend that TB prevention and control should also be viewed as a paradigmatic example of why health matters for economic growth: a TB-free workforce adds to the human capital of a nation, and a healthier society, in turn, results in growth (40, 41).
These results support the call to go beyond what we are doing now to control TB and aim at the elimination of unfair social inequities in TB transmission. Without undermining ongoing efforts on the more proximal determinants-especially those affecting quality access to care-TB control and prevention efforts should address TB root causes, i.e., its social determinants. Despite scant evidence on cost-effective interventions to address the social determinants of TB, we believe that sound social policies sensitive to health equity-such as preferentially targeting the most vulnerable and disadvantaged (42), coupled with more interdisciplinary action and scaled-up innovation, can positively impact TB control in Latin America and the Caribbean, and elsewhere.
Acknowledgements. Financial support was provided by PAHO/WHO and the United States Agency for International Development (USAID) through Grant AID-LAC IO-11-00001. The views expressed are solely those of the authors and do not necessarily reflect the official views of any of the funding or affiliated organizations.
The authors thank Jorge Victoria, Rafael Lopez, Anna Volz, and Vanessa Gutierrez who provided valuable contributions to earlier versions of this manuscript. We are also indebted to Chris Dye for his helpful suggestions and advice.
Conflicts of interest. None.
Disclaimer. Authors hold sole responsibility for the views expressed in the manuscript, which may not necessarily reflect the opinion or policy of the RPSP/PAJPH and/or PAHO.
1. World Health Organization. Tuberculosis fact sheet no. 104. Available from: www.who.int/mediacentre/factsheets/fs104/en/ Accessed on 4 January 2015.
2. World Health Organization. Global Tuberculosis Report 2013. Geneva: WHO; 2013. Available from: http://apps.who.int/iris/bitstream/10665/91355/1/9789241564656_eng.pdf?ua=1 Accessed on 4 January 2015.
3. Pan American Health Organization. Tuberculosis in the Americas. Regional Report 2012: epidemiology, control, and financing. Washington DC: PAHO; 2013. Available from: www.paho.org/hq/index.php?option=com_docman&task=doc_download&Itemid=270&gid=22953&lang=en Accessed on 4 January 2015.
4. Pan American Health Organization. Health in the Americas, 2012 edition. Washington DC: PAHO; 2012. Available from: www.paho.org/healthintheamericas Accessed on 4 January 2015.
5. Belizan JM, Cafferata ML, Belizan M, Althabe F. Health inequality in Latin America. Lancet. 2007;370(9599):1599-600.
6. Inter-American Development Bank. Sustainability Report 2013. Washington DC: IDB; 2014. Available from: www.iadb.org/en/topics/sustainability/sustainability-report,1510.html Accessed on 4 January 2015.
7. Biggs B, King L, Basu S, Stuckler D. Is wealthier always healthier? The impact of national income level, inequality, and poverty on public health in Latin America. Soc Sci Med. 2010;71:266-73.
8. Raviglione M, Krech R. Tuberculosis: still a social disease. Int J Tuberc Lung Dis. 2011;15(6):56-8.
9. Ali M. Treating tuberculosis as a social disease. Lancet. 2014;383:2195.
10. Blower SM, McLean AR, Porco TC, Small PM, Hopewell PC, Sanchez MA, et al. The intrinsic transmission dynamics of tuberculosis epidemics. Nat Med. 1995;1:815-21.
11. Rasanathan K, Sivasankara Kurup A, Jaramillo E, Lönnroth K. The social determinants of health: key to global tuberculosis control. Int J Tuberc Lung Dis. 2011;15(suppl 2):S30-6.
12. World Health Organization, Commission on Social Determinants of Health. Closing the gap in a generation: health equity through action on the social determinants of health. Final report of the Commission on Social Determinants of Health. Geneva: WHO; 2008.
13. Ximenes RAA, de Albuquerque MFPM, Souza WV, Montarroyos UR, Diniz GTN, Luna CF, et al. Is it better to be rich in a poor area or poor in a rich area? A multilevel analysis of a case-control study of social determinants of tuberculosis. Int J Epidemiol. 2009;38(5):1285-96.
14. Hino P, Villa TCS, da Cunha TN, dos Santos CB. Spatial patterns of tuberculosis and its association with living conditions in the city of Ribeirão Preto in the state of São Paulo. Ciênc Saúde Coletiva. 2011;16:4795-802.
15. Kamper-Jørgensen Z, Andersen AB, Kok-Jensen A, Kamper-Jørgensen M, Bygbjerg C, Andersen PH, et al. Migrant tuberculosis: the extent of transmission in a low burden country. BMC Infect Dis. 2012; 12:60:1-8.
16. Lienhardt C. From exposure to disease: the role of environmental factors in susceptibility to and development of tuberculosis. Epidemiol Rev. 2001;23(2):288-301.
17. Harling G, Ehrlich R, Myer L. The social epidemiology of tuberculosis in South Africa: a multilevel analysis. Soc Sci Med. 2008;66(2):492-505.
18. Lönnroth K, Jaramillo E, Williams BG, Dye C, Raviglione M. Drivers of tuberculosis epidemics: the role of risk factors and social determinants. Soc Sci Med. 2009;68(12):2240-6.
19. Hargreaves JR, Boccia D, Evans CA, Adato M, Petticrew M, Porter JDH. The social determinants of tuberculosis: from evidence to action. Am J Public Health. 2011; 101(4):654-62.
20. van Hest NA, Aldridge RW, de Vries G, Sandgren A, Hauer B, Hayward A, et al. Tuberculosis control in big cities and urban risk groups in the European Union: a consensus statement. EuroSurveill. 2014;19(9):42-54.
21. World Health Organization. Global Tuberculosis Database. Available from: www.who.int/tb/country/data/download/en/ Accessed 6 November 2013.
22. World Bank. World DataBank. Available from: http://databank.worldbank.org/data/home.aspx Accessed on 28 January 2014.
23. Pan American Health Organization. Regional core health data initiative. Available from: www.paho.org/hq/index.php?option=com_tabs&view=article&id=2151&Itemid=3632&lang=en Accessed on 28 January 2014.
24. World Health Organization, United Nations Children's Fund. Joint Monitoring Program for Water Supply and Sanitation. Available from: www.wssinfo.org/ Accessed on 6 November 2013.
25. United Nations Population Division. World urbanization prospects: the 2014 revision. Available from: http://esa.un.org/unpd/wup/ Accessed on 12 October 2014.
26. Walmsley R. World Prison Population List, Tenth Edition. The International Centre for Prison Studies. Available from: www.prisonstudies.org/research-publications?shs_term_node_tid_depth=27 Accessed on 20 September 2013.
27. Hsiao C. Analysis of panel data. Cambridge: Cambridge University Press; 2003.
28. Frees EW. Longitudinal and panel data: analysis and applications in the social sciences. Cambridge: Cambridge University Press; 2004.
29. R Core Team. R: A language and envi- ronment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. 2013. Available from: www.R-project.org/ Accessed on 29 June 2015.
30. Croissant Y, Millo G. Panel data econometrics in R: the plm Package. J Stat Software. 2008;27(2):1-43.
31. World Health Organization. Hosseinpoor AR, ed. Handbook on health inequality monitoring with a special focus on low- and middle-income countries. Luxembourg: WHO; 2013. Available from: http://apps.who.int/iris/bitstream/10665/85345/1/9789241548632_eng.pdf Accessed on 4 January 2015.
32. Wagstaff A, Paci P, van Doorslaer E. On the measurement of inequalities in health. Soc Sci Med. 1991;33(5):545-57.
33. Maddala GS, Lahiri K. Introduction to econometrics. Chichester: Wiley; 2009.
34. Murray C, Lopez A. The global burden of disease. Boston: Harvard University Press; 1996.
35. Dye C, Lönnroth K, Jaramillo E, Williams BG, Raviglione M. Trends in tuberculosis incidence and their determinants in 134 countries. Bull World Health Organ. 2009; 87(9):683-91.
36. Castañeda-Hernández DM, Tobón-García D, Rodríguez-Morales AJ. Asociación entre incidencia de tuberculosis e índice de desarrollo humano en 165 países del mundo. Rev Peru Med Exp Salud Publica. 2013;30(4):560-8.
37. Rodríguez-Morales AJ, Castañeda-Hernández DM. Relationships between morbidity and mortality from tuberculosis and the human development index (HDI) in Venezuela, 1998-2008. Int J Infect Dis. 2012;16:e704-5.
38. Odone A, Crampin AC, Mwinuka V, Malema S, Mwaungulu JN, Munthali L, et al. Association between socioeconomic position and tuberculosis in a large population-based study in rural Malawi. PLoS One. 2013;8:e77740.
39. Millet J-P, Moreno A, Fina L, del Baño L, Orcau A, de Olalla PG, et al. Factors that influence current tuberculosis epidemiology. Eur Spine J. 2013;22(suppl 4):539-48.
40. Weil DN. Accounting for the effect of health on economic growth. Q J Econ 2007;122(3):1265-1306.
41. Aghion P, Howitt P, Murtin F. The relationship between health and growth: when Lucas meets Nelson-Phelps. National Bureau of Economic Research Working Paper 15813. March 2010. Available from: www.nber.org/papers/w15813 Accessed on 4 March 2015.
42. Andrews JR, Basu S, Dowdy DW, Murray MB. The epidemiological advantage of preferential targeting of tuberculosis control at the poor. Int J Tuberc Lung Dis. 2015;19(4):375-80.
Manuscript received on 5 April 2015.
Revised version accepted for publication on 9 June 2015.