Tuberculosis and HIV co-infection in European Union and European Economic Area countries

Study characteristics

All 55 documents providing information on the burden of TB–HIV co-infection reported prevalence of HIV in TB patients, with one also measuring incidence [11]. The majority of papers provided national estimates (37 (67%) out of 55), the remaining regional or local. The greatest number of studies was from Spain, followed by France. For 19 countries the only sources of data were from four international surveillance reports [8, 12–14]. Information on sociodemographic characteristics of patients was provided in 26 studies.

The seven risk factor studies were from Spain (five out of seven), France and the Netherlands. Most were longitudinal studies (four out of seven), one was cross-sectional and two were obtained from surveillance studies. The majority of papers (five out of seven) presented risk factors for co-infection, with PLWHA as the baseline population.

Overall burden of TB–HIV co-infection

Data on HIV co-infection among TB patients was identified for 23 out of the 30 EU/EEA countries. The most recent national estimates are shown in table 1 and figure 2. Prevalence varied between 0% in Bulgaria, Slovenia and Slovakia to 15% in Portugal and Iceland. In general, a very low level of co-infection was observed in central EU/EEA countries. Information on the burden of co-infection in capital cities was available for four countries. The prevalence was 7.3% in London (in 2005), 7.5% in Paris (in 2001), 8.6% in Brussels (in 2008) and 14.3% in Madrid (in 2003). Except for London, where prevalence was the same as the national average, the co-infected proportion in these cities was higher than the national average.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2–

Map of the most recent data on proportions of tuberculosis (TB) patients co-infected with HIV in European Union (EU)/European Economic Area (EEA) countries. All data were based on national surveillance (2001–2008). The burden in some countries with low numbers of TB cases, e.g. Iceland, must be interpreted with caution.

Trends in TB–HIV co-infection

Between 2000 and 2008, Estonia, Latvia, Lithuania, Belgium, Denmark and England and Wales, as well as the Czech Republic and Romania, reported increasing proportions of HIV co-infection among TB patients, although in some this reached a plateau in the latest years of observation (fig. 3). Data for France are also suggestive of an increase in co-infection, but additional time-points are required to confirm this. Spain and Portugal show a decline in co-infection levels, and Finland may be exhibiting a downward trend.

• Download figure
• Open in new tab
• Download powerpoint

Figure 3–

Trends in the prevalence of HIV co-infection in tuberculosis (TB) patients in European Union/European Economic Area countries by year of reporting.

Characteristics of co-infected cases

Demographic characteristics were reported for few countries, and the most recent data are shown in figure 4. The majority of co-infected cases were male and young adults (fig. 4a and b), and their proportion was larger than in non-co-infected patients. In northwestern EU countries, cases were mostly foreign born while in Spain and Italy around 30% were foreign born.

• Download figure
• Open in new tab
• Download powerpoint

Figure 4–

Sociodemographic characteristics of cases co-infected with HIV (▪) compared with mono-infected tuberculosis (TB) cases (▓) (most recent data). Year(s) from which data were acquired and reference numbers are given for each country. All data are from surveillance apart from reference [28], which is from a cross-sectional study in more than one hospital in Madrid (Spain). nd: no data; IDU: intravenous drug users. ^#: unknown category not included; ^¶: only those originating from high TB prevalence countries were included as foreign born.

In Spain, a large proportion of co-infected cases were intravenous drug users (IDU) compared with all TB cases (59% versus 8%, respectively). Although no comparable mono-infected data were available, Latvia showed a similarly high proportion of IDU co-infected cases.

Risk factors for TB–HIV co-infection

In the seven papers that statistically examined risk factors for TB/HIV co-infection, sex, age, CD4 count at HIV diagnosis, HIV transmission route and TB risk groups were most often studied (tables 2–4). Co-infection was associated with male sex in three out of five studies (table 2). Younger adults (aged 25–30 to 45 yrs) tended to have a higher risk of TB co-infection compared with children and older adults, but this was statistically significant in only two out of the four relevant studies (table 2).

The risk of TB disease was higher in PLWHA who presented with lower CD4 cell counts (<200 cells·mm⁻³) at HIV diagnosis in three studies (table 3). A fourth study among AIDS patients found that TB was more likely to be reported as the AIDS-defining illness at higher CD4 counts (200–350 cells·mm⁻³) than other AIDS-defining illnesses.

Acquiring HIV through injecting drugs was significantly associated with higher risk of TB co-infection than acquisition through sexual contact in three out of the four studies investigating probable route of HIV infection [29, 33, 35].

Combinations of risk behaviours and lifestyles, including homelessness, imprisonment, alcohol abuse, sharing syringes and close contact with a TB case, were associated with an increased risk of being TB–HIV co-infected (table 4).

Two studies investigated the risk of co-infection by place of birth or citizenship, showing contradictory evidence of whether being foreign born conferred increased risk of co-infection. One study from France [29] found an increased risk of TB in foreign-born PLWHA. Conversely, a Dutch study demonstrated that TB patients of Asian and eastern EU/EEA citizenship had a decreased risk of HIV diagnosis among TB patients compared with Dutch nationals [21]. When comparing such data it should be considered that the definition of citizenship might vary between countries and the ethnic distribution may vary as well. Both studies also looked at area of residence, showing a higher risk of co-infection in urban areas (p<0.001) [21, 29].

Strengths and limitations

This review used a sensitive approach, identifying a large number of publications to ensure all information was captured. Although we were not able to check for evidence of publication bias, the comprehensive search undertaken minimised the potential for missing studies. A limitation is the inherent observational nature of surveillance studies associated with biases and confounding. Comments in our survey of national HIV and TB contacts to the ECDC suggest that surveillance systems may under-represent marginalised populations [9, 10]. The majority of studies presenting data on risk factors and demographic characteristics were conducted in Spain, and data were only available for a small number of countries. The heterogeneous nature of the populations and infrastructures within the EU/EEA hinders the generalisation of this information to the whole of Europe. Furthermore, differences in the design of and populations covered by these studies only allowed for a descriptive synthesis of the evidence available. We included both demographic data from surveillance reports and risk factor data from observational studies, which increases the heterogeneity of this data. However, because of the limited number of observational studies, the demographic information was obtained to strengthen our results. Despite these limitations, findings were in agreement with knowledge on risk groups for TB and/or HIV separately, and/or studies from outside the EU/EEA, and countries with similar populations and infrastructure may be found to have similar risk groups.

This is the first comprehensive overview of this subject in the EU/EEA and it provides useful data for planning and evaluation of control measures.

Trends in TB–HIV co-infection

The distribution of co-infection across the EU/EEA is in agreement with what is reported by ECDC and in our survey of national contacts [8, 10]. The burden of co-infection follows the national prevalence of HIV in these countries [8] and the country-specific prevalence of drug users. Some of the observed time-trends may be due to changes in testing activity and reporting practices, and may not represent an increase in the true number of HIV-infected TB patients, but rather in the diagnosed burden. However, it is interesting to note that countries such as Spain and Italy, with higher levels of co-infection in IDU and homeless populations, show decreasing rates of TB–HIV co-infection, while countries with higher levels of HIV infection in foreign-born people had increasing rates. This suggests that the increases in some countries may be related to migration and decreases to strengthened control in IDU and homeless populations. Further efforts are required to control TB in other parts of the world to reduce the risk in migrant populations in these countries.

HIV prevalence in TB patients was generally higher in capital cities, and urban areas were found to be associated with an increased risk of co-infection. This is in agreement with previous data from non-EU countries [36] and may be related to the greater number of high-risk individuals for TB and/or HIV co-infection in these areas [37, 38]. Urban conditions such as poverty and overcrowding have also long been associated with TB [39, 40]. Studies of hospital in-patients also tended to show a higher prevalence of HIV co-infection than the national average. This is to be expected as TB–HIV co-infected patients are more likely to have complex needs, drug interactions and toxicities, and are therefore more likely to be in contact with hospitals.

Risk factors/groups for TB–HIV co-infection

The higher risk of co-infection in males found in four studies is consistent with the observation that both TB [8] and HIV [41] are more common among males in Europe [13]. This reflects the more prominent transmission of HIV through males having sex with males and via injection of drugs, since a higher proportion of drug users is male [31, 32]. Co-infection was more prevalent in the 15–45-yr age group in five studies, reflecting the higher proportion of HIV in this group, or more frequent HIV testing in young adult TB cases [21, 29, 30]. Five out of seven countries where the data were available showed a majority of the TB–HIV co-infected patients to be born abroad. These countries have high levels of migration from parts of the world with generalised TB and HIV epidemics. Countries where the majority of migrants originate from areas with concentrated epidemics, for example Spain, which receives more migrants from Latin-American countries, show a different pattern of co-infection. Based on three Spanish studies, PLWHA with lower CD4 counts, mainly as a result of late diagnosis, were more likely to have TB disease compared with those with a higher CD4 count [32]. Among persons diagnosed with an AIDS-defining illness in France, TB was diagnosed at higher CD4 cell counts than other AIDS-defining illnesses. This is consistent with findings that TB occurs earlier in HIV infection than other opportunistic diseases [42, 43], highlighting the need to monitor PLWHA for TB throughout the course of their infection.

Several countries have highlighted the diversity in the level of co-infection across population groups. These findings provide evidence for more robust monitoring of co-infection in EU/EEA countries in order to develop targeted strategies to reduce HIV and TB infection. Although demographic and clinical data pertaining to persons living with HIV and/or TB are available in many countries, they are not always reported. Reporting of this information and analyses to determine statistical associations between co-infection and demographics would provide much needed information to tailor effective control strategies.

Clinical implications

The results of this review highlight how some population groups affected by TB and HIV co-infection are particularly vulnerable, suggesting that efforts for identifying, treating and raising awareness for both diseases should be considered by countries with high levels of co-infection in these populations. The survey of TB and HIV contact points shows limited overlap in risk groups targeted for TB and HIV screening; IDU and prisoners are screened for both infections in eight countries, and migrants in nine [10]. In order to capture and effectively treat all co-infected patients, some countries with higher co-infection burdens might benefit from screening risk groups for both infections. In addition, various studies have shown that many TB patients are not tested for HIV [10, 44–46]. Further improvements are needed in offering HIV testing to all TB patients, as recommended by the World Health Organization [47].

Conclusion

Co-infection levels in EU/EEA countries ranged from 0% to 15% and were in the same range as those reported from comparable parts of the world (e.g. the USA had 12% HIV-positive TB cases in 2007, Canada had 5.7%, Australia had 3.2% and Japan had 0.5% [48–51]). The increasing levels observed in a number of EU/EEA countries are concerning. Efforts to scale up control of both infections as well as further monitoring of the trends of co-infection are therefore required.

Males, young adults, foreign-born people and those living in urban areas, as well as injecting drug users, the homeless and prisoners, were identified as high-risk groups for TB–HIV co-infection. The information obtained in this review indicates that a diverse and varied approach is required to deal with the burden and population groups at risk of TB–HIV co-infection in EU/EEA countries. Greater use of available demographic and clinical data, as well as further, standardised, Europe-wide research would help increase our understanding of co-infection risk.