A 20-year retrospective cohort study of TB infection among the Hill-tribe HIV/AIDS populations, Thailand

There were 3,130 Hill-tribe HIV/AIDS patients from 16 hospitals in Chaing Rai Province between 1990 and 2010 who were recuited for the study based on the criteria. Of these cases, 960 (30.7 %) had at least one episode of TB. The first case of HIV/AIDS in the Hill-tribe people was a 25-year-old Lisu, reported at Mae Suai Hospital, Chiang Rai Province, in 1990.

The majority of the tribes were Akha (46.0 %), followed by Lahu (19.7 %) and Yao (9.5 %). More than half were male (54.6 %), and 44.6 % were 26–35 years old, followed by 36–45 years old (25.2 %), and 16–25 years old (14.3 %). The peak period of HIV/AIDS infection among the Hill-tribes occurred from 2001–2005 (43.9 %), followed by 2006–2010 (33.7 %).

Regarding the habitat, 25.8 % of the people resided in the Mae Fah Laung district, followed by 18.8 % in the Mae Suai district, 16.2 % in the Muang Chiang Rai district, and 11.1 % in Mae Chan district. Regarding the occupations, 44.8 % were agricultural workers and 32.2 % were daily and temporary employees. The major risk factors of HIV infection were sexual intercourse (93.1 %), 6.4 % mother-to-child, and 0.5 % IDU. Most of the subjects did not recive ARV (66.7 %), only 297 had a measured CD4 level, 23.9 % were still alive at the date of data collection, and 30.7 % had TB disease documented at some time after the HIV/AIDS infection was diagnosed (Table 1).

Table 2 shows the proportion of TB-positive and-negative cases in various factors; eight variables had a statistically significant difference -- tribal groups, risk factors, survival status, sex, age, occupation, diagnosis, and receiving ARV at a significance level of 0.50. Regarding the tribes, 33.7 % of Akha-HIV and 32.6 % of Lahu-HIV had at least one episode of TB, which was greater than the other tribes. Female-HIV group had a greater proportion of TB disease (34.8 %) than the male-HIV group (27.3 %). HIV/AIDS patients who worked as trader and temporary employees groups had a 43.9 % and 41.0 % incidence of TB disease.

Table 3 shows the eight variables that were statistically significantly related to TB infection among the Hill-tribe HIV/AIDS patients in the simple logistic regression model at a significance level of 0.10. The Lisu and Akha groups had a greater chance of TB infection -- 1.40 and 1.46 times greater than that of the Yao, respectively. Being female was a factor associated with TB infection (1.42 times compared to being male), and the Hill-tribe HIV/AIDS patients age 26–35 and 46–55 had 1.85 and 1.92 times greater chance, respectively, of TB infection than did those aged less than 15 years.

Mother-to-child transmission of HIV/AIDS patients was less likely to result in TB infection among the HIV/AIDS Hill-tribe patients compared to sexual intercourse, with OR = 0.57 (90 % CI = 0.42-0.76). Moreover, those who did not receive ARV presented as a greater chance of TB infection among the HIV/AIDS Hill-tribe, with OR = 2.76 (90 % CI = 2.38-3.31).

In the multiple logistic regression model, five variables had a statistically significant association with TB infection among the Hill tribe HIV/AIDS individuals. The Akha Hill-tribe HIV/AIDS individuals had a greater chance for TB infection than did the Yao, with OR_adj = 1.50 (95 % CI = 1.01-1.92). Females had a greater chance of TB infection than did males, with OR_adj = 1.33 (95%CI = 1.11-1.59). The HIV and AIDS groups had a greater chance of TB infection among the HIV/AIDS Hill-tribe people, with OR_adj = 11.59 (95 % CI = 7.19-18.71) and OR_adj = 1.71 (95%CI = 1.03-2.87) compared to those in the asymptomatic group. For the Hill-tribe HIV/AIDS individuals, not receiving ARV was a factor associated with TB infection, with OR_adj = 2.59 (95 % CI = 2.09-3.22). Patients who had been diagnosed with HIV infection from 1990–1995 and 1996–2000 were less likely to have TB compared to those who were diagnosed from 1990–1995, with OR_adj = 0.04 (95 % CI = 0.01-0.14) and OR_adj = 0.11 (95 % CI = 0.07-0.17), respectively (Table 4).

Table 5 shows the nine variables in the simple logistic regression analysis that had a statistically significant association with the patients’ life status at the date of data collection (alive or dead from the medical record) of the Hill-tribe HIV/AIDS individuals. The Karen and Kmong had less chance of being still alive at the date of data collection compared to the Yao tribe, with OR = 0.58 (90 % CI = 0.43-0.77) and OR = 0.55 (90 % CI = 0.41-0.74), respectively. Patients aged 16 years and above had less chance of being still alive compared to those aged less than 15 years, with OR = 0.51 (90 % CI = 0.37-0.69), OR = 0.60 (90 % CI = 0.46-0.79), OR = 0.60 (90 % CI = 0.45-0.80), and OR = 0.33 (90 % CI = 0.20-0.53) respectively. Those who had been diagnosed with HIV and AIDS had less chance of being still alive compare to those asymptomatic group, with OR = 0.06 (95 % CI = 0.04-0.12), and OR = 0.25 (90 % CI = 0.10-0.47) respectively. Those who had been infected through mother-to-child transmission and IDU had a better chance of being still alive at the date of data collection, with OR = 2.05 (90 % CI = 1.56-2.68) and OR = 8.45 (90 % CI = 1.55-46.13), respectively. Patients who did not receive ARV had less chance of being still alive at the date of data collection than those who received ARV, with OR = 0.15 ( 90 % CI = 0.12-0.19), and those who had an occupation in the categories of “temporary employee,” “housewife,” “students”, and “others” had a better chance of being still alive at the date of data collection compared to “agriculture,” with OR = 1.88 (90 % CI = 1.66-2.19), OR = 1.61 (90 % CI = 1.10-2.36), OR = 5.46 (90 % CI = 3.21-9.30), and OR = 1.74 (90 % CI = 1.36-2.23), respectively. Patients who had been diagnosed from 1990–1995, 1996–2000, and 2001–2005 had less chance of being still alive at the date of data collection compared with those who had been diagnosed between 2006 and 2010, with OR = 0.03 (90 % CI = 0.01-0.08), OR = 0.06 (90 % CI = 0.04-0.09), and OR = 0.82 (90%CI = 0.71-0.95) respectively.

From the multiple logistic regression model at a significance level of 0.05, five variables were statistically significantly associated with patients’ life status at the date of data collection. The Karen and Kmong had less chance of being still alive compared to the Yao, with OR_adj = 0.58 (95 % CI = 0.43-0.77), and OR_adj = 0.55 (95 % CI = 0.41-0.74), respectively. Females had a better chance of being still alive than males, with OR_adj = 1.41 (95 % CI = 1.19-1.66). Patients who were diagnosed with “HIV” and “AIDS” had less chance of being still alive than those who were diagnosed with “Asymptomatic,” with OR_adj = 0.05 (95 % CI = 0.03-0.09), and OR_adj = 0.34 (95 % CI = 0.17-0.64), respectively. Those who had a defined route of transmission in the category of “mother-to-child” and “IDU” had a better chance of being still alive compared to those who contracted HIV from “sexual intercourse,” with OR_adj = 2.05 (95 % CI = 1.56-2.68) and OR_adj = 8.45 (95 % CI = 1.55-46.13), respectively. Finally, patients who did not receive ARV had less chance of being still alive compared to those who received ARV, with OR_adj = 0.11 (95 % CI = 0.08-0.14) (Table 6).

Our study illustrates that among 3,130 members of the Hill-tribe HIV/AIDS population, major factors associated with TB infection were being from the Akha tribe, being female, HIV and AIDS stages, and non-receiving ARV compared to the others tribes, being male, asymptomatic stage, and receiving ARV, respectively. Regarding factors related to patients’ survival status, we found that being of the Karen and Kmong tribes, non-receiving ARV, and HIV and AIDS stages had less of an opportunity to survive compare to those of being from other tribes, receiving ARV, and asymptomatic stage, respectively.

This study has a major limitation in the quality of the available secondary data. The identification and confirmation of a patient’s Hill-tribe status was a significant challenge. Fortunately, in identifying which patient on the hospital lists was a Hill-tribe person, we could rely on the characteristics of their first and last names, which provide a reliable clue to whether they met the inclusion criteria. The naming tradition is that the child takes the father’s last name (which is recorded on the hospital registration) even if he/she has grown up and chosen later to belong to a different tribe than that of his/her parents. The local ARV clinic staff had been extremely helpful in this respect, as most of them have worked in the area for many years and know the patients, their families and their villages intimately from providing counseling and health checks and giving out medication on a monthly basis. A few subjects had changed their first name, but it was not a problem for classification because most still lived in their original village. Missing of the specific date of last visited among the subjects, as a result could not use a better statistical model such as survival analysis including hazard model.

Prior to 2000, the majority of the Hill-tribe people did not have Thai identification cards, which is one reason for the poor attendance to clinics of this section of the study population during this period. Another important reason is that, for several years after HIV/AIDS diagnosis became recognized, the condition did not entitle the person to free medical care under the government’s universal medical care scheme. It is certain that many sufferers had never been diagnosed or that if they had been initially, they became lost to follow up due to financial constraints for treatment. The overall result of this study is that the recorded number of HIV patients who we were able to trace from records must have been lower than the true number in the beginning of the survey periods.

Other reasons that affected the number of recorded cases included social stigmatization in the early days, which deterred patients from seeking hospital care. Apidechkul in 2009 found that Akha youths, members of one of the Hill-tribes, preferred to travel to large cities to seek employment to support their families, and some contracted HIV/AIDS during their time away from home [11].

Previously available information from the Chiang Rai Public Health Office reported that between 1987 and 2009, the highest number of TB cases of HIV/AIDS in the province was recorded [12], 44.0 % of which were smear-positive pulmonary TB; among the TB infection patients, 24.0 % tested positive for HIV/AIDS. The Thai Public Health aimed at a TB treatment success rate of 85 %, but the actual success rate was 73.2 %, while it was only 64.7 % in those with HIV. Those who lived in remote areas had a greater default rate than those who lived near health centers [12], and default during TB treatment was a significant problem in HIV-infected patients [13]. Apidechkul in 2012 [14] reported that most of the TB infections in the Hill-tribe population, at 84.6 %, were pulmonary TB, and only 22.7 % were cured, while the HIV/AIDS prevalence among the Hill-tribe TB patients was 17.2 %. This report supports the results of the study that the Chain Rai province has a high prevalence of TB, particularly in HIV/AIDS patients.

Due to heavy stigmatization in the early period of HIV epidemic in Thailand, HIV with TB among the Hill-tribes would have been under-reported during the time. Amarita [15] reported that the confluence of the two epidemics rendered TB symbolic and symptomatic of HIV and enhanced the visibility of AIDS. Dual illness thus introduced a paradox to patients’ identity constructions and produced a unique, overlapping double stigma.

A study of Jimenez et al. (2006) found that men were more likely to get TB infection than women in southern Mexico [16]. However, the reports of Thorson et al. (2001), Young et al. (2009) and Holmes et al. (1998) found that women and girls were more likely to get TB and more serious outcomes because of delay in diagnosis [17‐19].

Vermund et al. [20] reported that Southeast Asia faced the problem of TB-HIV/AIDS co-infection, particularly among MSM and IDU. However, in this study, the author found that the main route of transmission was sexual intercourse. Julien Zwang et al. [21] reported that the factors supporting the TB epidemic in Southeast Asia were poverty, living in crowded areas, and the quality and quantity of health care services and educational systems.

The number of TB reported cases increased steadily from 1990 to 2010, possibly due to inaccessibility of health care service and lack of awareness and technologies for diagnosis, and the policy of the Thai government of identifying new TB patients based on WHO guidelines [22]. The study also showed that youth (≤15 years old) had a smaller proportion of TB disease compared to those who ≥ 16. This might reflect the limited diagnostic options and that the numbers of HIV/AIDS patients in this group was less than in the others.

Kingkaew et al. [23] found that those with HIV/AIDS, especially males, were at greater risk for extrapulmonary TB; also in HIV patients, TB meningitis and low CD4 levels were risk factors for death. This result is inconsistent with our study, which found that the group without a diagnosis of TB had a greater chance of death compared to those with TB diagnosis. This paradox is most likely explained by the limited access to treatment of a marginalized population due to financial, social and/or legal problems, especially during the early period of HIV/AIDS. Mo et al. [24] also reported that patients with multiple infections with HIV, HBV, and HCV were at risk of developing abnormal LFTs and death during TB treatment. Unfortunately, the documentation in our study was not adequate for analyzing such associations.

The study also found that the Karen and Kmong tribes had a greater chance of death than the Yao. This may be due to their socio-economic status, which limits access to health care and causes them to be vulnerable to disease infections [25‐27]. Karen and Kmong also widely use traditional medicine [28].

Lawn et al. [29] reported that the death rate of TB and HIV/AIDS co-infection dramatically decreased after scaling up HIV/AIDS testing among the TB-positive and improved the survival of those with HIV-associated TB, reducing mortality rates by 64 % to 95 %. This result agrees with our study, which found that patients receiving ARV had a greater chance of surviving. Arbor et al. [30] also reported that receiving ARV and co-trimoxazol and having CD4 greater than 50 were supportive factors for survival among HIV/AIDS patients. However, Boettigeretal et al. [31] reported that not everyone having ARV could control the viral load, as those patients who had a history of opportunistic infection tended to develop virological failure. A meta-analysis conducted by Jiang et al. [32] found that ARV could be associated with adverse events and drug interaction with antiTBs. However, Rangaka et al. [33] suggested that isoniazid preventive therapy should be recommended to all patients receiving antiretroviral therapy in moderate- or high-incidence areas. Mfinanga et al. [34] suggested that ART could be delayed until after the completion of six months of tuberculosis treatment for HIV-positive patients with tuberculosis who have CD4 cell counts greater than 220 cells per μL. Harries et al. [35] supported the early diagnosis and treatment of HIV infection in affected communities and proposed that the urgent assessment of frequent testing for HIV and early start of antiretroviral treatment (ART) should result in short-term and long-term declines in tuberculosis incidence through individual immune reconstitution and reduced HIV transmission. Yang et al. [36] reported that having started HAART during TB treatment was significantly associated with better survival. These regimens should be considered by the Thai public health services for implementation to improve the survival of HIV-TB co-infection in the Hill-tribe people.

Grenfell et al. [37] reported that the latent TB infection prevalence was high and that the active disease was more common among HIV-positive people with IDU. However, our study found only a small proportion of people with IDU and HIV/AIDS had TB infection among the Hill-tribes. The IDU is not a common route of choice among Hill-tribe drug abusers who prefer oral methamphetamine tablets [14].

We also found that being HIV and AIDS stages had a greater chance for death than those an asymptomatic stage. This is the natural history of HIV/AIDS disease in human; those at the HIV and AIDS stages have a greater chance of death than those at the early infection stage [38‐40].

However, early in the HIV/AIDS epidemic in Thailand, people who knew their HIV status very often decided to commit suicide. Therefore, HIV positive and AIDS stages had a greater proportion of death than the asymptomatic stage. This is consistent with a report by Lotrakul [41] that the major cause of death among the HIV/AIDS patients in Northern Thailand was suicide. Porapakkham et al. [42], Guadamuz et al. [43], Schlebusch et al. [44], Bundhamcharoen et al. [45], and Chutinantakul et al. [46] also reported that suicide was most prevalent in the Upper Northern Thailand region, where HIV infection was related to the high prevalence of suicide and was the major cause of premature death.

The Hill-tribes of Thailand are minority groups with specific needs due to their language, cultural and social differences. Public health interventions to reduce the number of new TB and HIV/AIDS cases must take such needs into consideration in the planning and delivery of services, especially in health education. Thomas et al. [47] reported that using specific community-based social and behavioral counseling, as well as voluntary testing, could reduce the HIV incidence in Thailand and Africa. Thato et el. [48] also reported that a brief Peer-Led HIV Prevention Program significantly increased knowledge of preventive behaviors and motivated participants to have a better attitude toward preventive behaviors, better subjective norms, and greater intention to practice preventive behavior among school children in Bangkok, Thailand. Rojanawiwat et al. [49] reported that the National Access to Antiretroviral Program reduced the number of opportunistic infections, especially PCP and TB, in Thai people living with HIV/AIDS.

In conclusion, the present study indicates that being Akha and female and lacking access to ARV were the major causes of TB infection among the HIV/AIDS Hill-tribe population. Our study suggests an urgent need to consistently implement TB and HIV control programs for the Thai Hill-tribe.