INTRODUCTION
Peru is highly endemic for tuberculosis. There is an overall estimated incidence of 228 cases of tuberculosis for every 100,000 inhabitants,1 despite relatively low rates of infection with human immunodeficiency virus (HIV) (0.35%, with the vast majority of cases occurring among homosexual men and in cities).2 The geography of Peru is highly varied, ranging from high altitude sierra, to Amazonian basin jungle, to coastal desert, with gradations of intervening climatic zones. To date, little is known about whether these environmental differences affect the epidemiology of tuberculosis transmission.
There are two general mechanisms that might affect tuberculosis transmission at high altitude. These are 1) decreased viability of the organism due to, for example, lower humidity or more intense ultraviolet (UV) light exposure, and 2) decreasing susceptibility of humans to develop the disease, for example, because of reactive hypoxia, or genetic factors fixed within the particular population located at high altitude that allow for enhanced survival or its morbidity. It is doubtful whether the genetics of high altitude populations has a significant effect on the clinical manifestations of tuberculosis since when such individuals migrate to places such as Lima, they develop similar rates of tuberculosis as the local population. Anecdotal accounts including Thomas Mann’s The Magic Mountain and medical texts have described the preventive effect of acquiring active disease, citing the relative health of American trappers and explorers who found sanctuary in the Rocky Mountains or Pike’s Peak or Peruvian Andes villagers.3–5
Recent, more recent epidemiologic studies of tuberculosis disease rates in Africa have suggested that high altitude is associated with protection against symptomatic disease. In South Africa, the prevalence of sputum smear-positive tuberculosis was found to be approximately two times higher in low altitude villages than in high altitude villages.6 In Kenya (1988–1990) at altitudes of 1,000 meters (3,300 feet) or more, notification rates of tuberculosis were less than 30% of the rates in districts below 500 meters.7 Experimental studies of guinea pigs and rabbits showed that when these highly susceptible animals were infected with Mycobacterium tuberculosis, they survive longer when housed in lower oxygen tension than did controls.8
Population-based studies examining the effect of high altitude on M. tuberculosis infection are limited. A study in India published in 1951 found a 2–5-fold increase in tuberculin skin test (TST) positivity associated with living at a lower altitude (< 1,000 meters versus > 1,000 meters).9 The conclusions that can be drawn from this older study are limited because there was no control for potentially confounding variables such as contact with patients with active tuberculosis, age, and number of bacille Calmette-Guérin (BCG) vaccinations. In this study, we examined more closely the hypothesis that high altitude in Peru is associated with a lower prevalence of infection with M. tuberculosis, as assessed by the TST. Infection rates and clustering of tuberculosis infection were examined in two Peruvian highland villages using TST and compared with those of three communities at sea level: a shanty town located at the periphery of Lima, and two Amazonian jungle basin villages near Iquitos.
MATERIALS AND METHODS
This study was reviewed and approved by the ethical review boards of Asociación Benéfica PRISMA (Lima, Peru) and the Johns Hopkins Bloomberg School of Public Health (Baltimore, MD). The study was prompted by data obtained from a study surveying TST-positive prevalence in Vichaycocha (Figure 1), which showed an unexpectedly low (5%) TST-positive prevalence at an altitude of 3,500 meters. To confirm these findings, we conducted a follow-up study at another high altitude village, Quilcas, testing for both TST-positive prevalence and household clustering. As a case-control study, we compared positive prevalence and household clustering in three sites situated at sea level: Pampas, Buen Pastor, and San Carlos. The populations tested were mestizos, who are descendants of immigrants, mainly Spanish, and Native Americans.
Site descriptions.
The sites in the Amazon basin (Buen Pastor and San Carlos) are characterized by perennial, warm temperatures and higher humidities than found at sea level in Lima, or in the highlands (Table 1). Annual precipitation in Lima is negligible, while the Amazon jungle and the highlands receive higher levels (76–290 cm). All areas under study except Pampas are defined as rural according to the last national census (1993). The rate of HIV infection in the rural area is extremely low and in the Pampas is less than 1% (Gilman RH, unpublished data). The fieldwork in all five sites was carried out between July 1997 and August 2000.
High altitude sites.
Vichaycocha.
Vichaycocha (population = 517, altitude = 3,500 meters) is a rural village in the central Peruvian Andes situated 150 km northeast of the capital city of Lima. The main sources of employment are agriculture and animal husbandry. Village inhabitants live in adobe houses that are closely grouped in a ravine, with dirt floors and piped river water. Because of cold weather, houses have small windows and are poorly ventilated.
Quilcas.
The agricultural village of Quilcas (population = 1,278, altitude = 3,330 meters) is located one hour outside the urban center of Huancayo (population = 327,000, altitude = 3,249 meters) in the central Peruvian Andes. Housing is similar to that of Vichaycocha.
Sea level sites.
Pampas.
This Lima, sea level, peri-urban shantytown has a population of approximately 40,000 individuals, of whom 25% are stably employed. It is located close to the Pacific Ocean and has a desert climate. Households are built next to each other and made of woven thatch walls with roofs of plastic coverings, usually containing one or two bedrooms. Similar to those in the highlands, many homes lack windows and are therefore poorly ventilated.10 In the past five years, piped water has become available for most houses in this area.
Buen Pastor.
Buen Pastor is an Amazon jungle village located on a side road that comes off the main airport road 22 km from Iquitos, the urban center of the Amazon (population = 367,000), and Nauta. It has a population of 270 individuals living in 50 households. It is located on a dirt road impassable to vehicles during the rainy season. The inhabitants work primarily in agriculture, fishing, and animal husbandry. Houses are spaced apart from each other and are well ventilated. They are generally made of wood, and some are without walls, with palm leaf roofs usually open at the top.
San Carlos.
San Carlos is a small village located on the shores of the Itaya River, about five kilometers from Buen Pastor, with a population of 176 people living in 31 households. In most respects, San Carlos is similar to Buen Pastor, except access is only by boat or foot.
Quilcas, a high altitude village and jungle sites (Buen Pastor and San Carlos) are similar in being located 20–30 km from major cities. Traffic into both sites is limited. In both sites, much of the population needs to walk a mile or so to get to a road passable by a vehicle.
Sample selection and coverage.
Vichaycocha.
The TST was performed on 370 people received from attendees at the local health post from 1997 to 1998. Although this sample was obtained from a health post, overall coverage was 72% of the population.
Quilcas.
One hundred families were selected at random from a sampling universe of 204 households. Among the 100 selected families, there was a population of 677, of whom 67% were tested.
Pampas.
Within 397 families randomly selected from a census of 1,649 families living in the most recent and poorest part of the shantytown, there were 1250 (68%) individuals who agreed to participate in the study.
Buen Pastor.
A total of 188 (70%) of 270 village inhabitants participated in the study.
San Carlos.
A total of 149 (85%) of 176 village inhabitants participated in the study.
Field operations.
Written informed consent was obtained from all adult participants and from the parents or guardians of children. Subjects at each site were surveyed with structured questionnaires regarding recent (within the past five years) exposure to active tuberculosis cases and general conditions of health. As mentioned earlier, the data at Vichaycocha were collected at the local health post. At all other sites, the studies were designed as community-based evaluations of TST-positivity. Because of the different population sizes at the sites, the studies at Pampas and Quilcas were designed to estimate the prevalence of TST positivity, whereas in Buen Pastor and San Carlos we sought to sample the entire population. Questionnaires used in Vichaycocha data collection queried individuals, whereas surveys conducted in the other sites evaluated entire households.
Trained field workers performed a skin test using a 0.1-mL dose of five tuberculin units (Tubersol; Connaught Laboratories, Inc., North York, Ontario, Canada) on the volar surface of the left arm. Tuberculin vials (1 mL) were kept refrigerated and carried in a cooler box with ice packs before use. Induration produced by injections was measured 48–72 hours following administration using the pen method in the long axis and the transverse plane of the forearm.11 Induration greater than or equal to a mean of 10 mm was considered positive, according to American Thoracic Society recommendations for persons born in countries with high prevalence of tuberculosis, groups with poor access to health care,12 and the Peruvian Ministry of Health.13 All fieldworkers and medical personnel underwent standardized training for applying and reading the TST. Patients testing positive were evaluated for evidence of active tuberculosis and if suspected a chest radiograph and a sputum smear were obtained as clinically indicated. Children less than 15 years old with recent contact with an active tuberculosis case were referred to the national tuberculosis control program for prophylaxis with isoniazid. Neither M. avium complex nor M. kansasi has been isolated from HIV or non-HIV patients in our laboratory in Peru (Gilman RH, unpublished data).
Statistical analysis.
Differences in continuous variables such as rooms per household, people per household, and body mass index among study sites were evaluated using one-way analysis of variance applying Bonferroni’s adjustment for multiple comparisons. Chi-square tests or Fisher’s exact corrections were used to identify differences in categorical variables (sex, age groups, educational level, BCG vaccination status, cough, contact with tuberculosis patients, and presence of another TST-positive subject in the household).
Risk factors for and clustering in TST positivity were evaluated using a separated random effects logistic regression model14 for each study site, defining the household as the group variable. Only data from Quilcas, Pampas, Buen Pastor, and San Carlos was analyzed using these models. Data from Vichaycocha was not included in the family cluster analysis because data on the covariates analyzed were lacking and the survey did not include all family members. The proportion of variance attributed to clustering effects in TST positivity was used to describe the amount of aggregation existing in tuberculosis infection within household units. A likelihood ratio test was used to assess whether this proportion was statistically different from zero. Additionally, the increase in risk attributed to clustering effects was assessed by adding a variable to the model representing the presence of an additional TST-positive person in the household and estimating its odds ratio (OR). Similar random effect models were fitted comparing the clustering OR between study sites.
Wald tests15 were used to evaluate the association of different covariates with TST positivity. One additional model combining all four sites (Quilcas, Pampas, Buen Pastor, and San Carlos) was fitted to compare the risk for TST positivity between Quilcas and each of the three sea level sites adjusting for covariates found to be statistically significant in the univariate analysis at the site level.
Differences in TST reactivity were further analyzed for the TST reaction size as a continuous variable using maximum-likelihood random effects linear regression models.16 A model combining data from all sites was fitted to assess the difference in reaction size between Quilcas and each of the sea level sites. All analyses were performed using Stata 7.0 (Stata Corporation, College Station, TX) and all confidence levels were set to 95%.
RESULTS
Distribution of demographic characteristics and risk factors.
The population studied in Vichaycocha was significantly older than that of Quilcas, Pampas and San Carlos (P < 0.001, P < 0.001, and P = 0.025 respectively; Table 1). Individuals from Buen Pastor were on average older than those in Pampas (P = 0.001) and Quilcas (P < 0.017). The highest levels of education above elementary school were found in Pampas (P < 0.01), although children less than five years of age were included in the “no education” category. Crowding levels varied among study sites. The average number of people per room in Quilcas households was significantly lower than in Buen Pastor (P = 0.004) and San Carlos (P < 0.001), but higher than in Pampas (P = 0.010). Pampas recorded higher rates of BCG immunization (Table 2), compared with all other sites (75% versus 52–60%; P < 0.001 in all cases). Pampas and San Carlos recorded higher numbers of contacts with cases of active tuberculosis (10% in Pampas, 11% in San Carlos versus 1–5% in other sites; P < 0.001 to P = 0.040). Vichaycocha had the lowest rates of contact with active cases among all sites (1%; P = 0.041 to P < 0.001). Finally, the crude rates of TST positivity were significantly lower (P < 0.001) in both high altitude sites (6.8% and 5.7%) compared with the jungle sites (33.0% and 30.9%) and Pampas (24.8%). The TST-positive rates in Pampas were statistically lower than in Buen Pastor (P = 0.017).
Risk factors and associations of TST positivity.
The adjusted ORs for risk factors associated with TST positivity from multiple regression models for each site are shown in Table 3. Results from variables not significant in the multiple regression models such as sex, nutritional status (body mass index), household crowding, and household characteristics are not presented. The prevalence of TST positivity in the sea level population (Pampas, San Carlos, and Buen Pastor) increased progressively with age until approximately 40–49 years of age, after which it showed a slight and non-significant decrease in subjects more than 50 years old (Figure 2). In high altitude villages, the prevalence of TST positivity peaked at an earlier age (30–39 years) and later again at more than 50 years of age. The TST positivity increased significantly with higher education in Quilcas and Pampas, most likely due to age-related residual confounding. When children not yet attending school were excluded from the analysis, the association between education and TST risk became non-significant. The BCG vaccination and contact with active tuberculosis cases were significantly associated with increased TST risk in Pampas. Prolonged coughing was associated with higher TST risk in Quilcas. In all sites studied, the same association between the above risk factors and TST positivity was observed, although sometimes failing to reach statistical significance (Table 3).
High altitude villages, risk for TST positivity and TST reaction size.
After adjusting for age, education, BCG immunization, prolonged coughing, contact with active TB cases, and household clustering, all three sea level sites showed a significantly higher risk for positive TST result. In Pampas, the relative risk of infection was 4.5 times (95% confidence interval [CI] = 2.9–7.0) that of Quilcas. The relative risk was 5.7 times (95% CI = 3.3–10.0) that in Buen Pastor and 6.0 times (95% CI = 3.3—10.9) that in San Carlos compared with Quilcas (Table 4). The mean TST reaction size was larger in all three sea level sites compared with Quilcas by approximately 4.1–4.8 mm (P < 0.001), as estimated by a multiple random effects linear regression model adjusting for the same risk factors (Table 4).
High altitude villages and household clustering in TST risk.
Significant levels of household clustering in TST positivity, as measured by the higher OR of TST positivity due to presence of another TST-positive household member, were observed in Quilcas (OR = 3.7, P = 0.005, Table 4), Pampas (OR = 2.1, P < 0.001), and Buen Pastor (OR = 2.8, P = 0.040). Non-significant clustering levels were found in San Carlos. The household clustering OR was only significantly different between Quilcas and San Carlos (P = 0.022). The proportion of the variance of TST positivity explained by household clustering was higher in Quilcas (61%) than in Pampas (41%). Family clustering did not explain a significant proportion of the variance for TST positivity in the jungle sites.
DISCUSSION
In this study, the TST-positive prevalence rates in two poor high altitude villages in Peru were found to be significantly lower than those found in three different but similarly poor sea level communities. These differences remained statistically significant even after adjusting for risk factors known to be associated with tuberculosis, such as contact with persons with active tuberculosis, household crowding, and education status. Ventilation, which is protective against tuberculosis transmission, was best in the jungle where TST-positive rates were highest. The low prevalence of tuberculosis infection in these highland village populations is not likely to be explained by racial or genetic differences between the study sites since the majority of the population of Peruvian shantytown in Lima is made up of recent immigrants from the Peruvian Andes.17 Furthermore, in high altitude villages, TST-positive prevalence remained relatively low until late in adolescence, before increasing sharply to almost double the level in children. In contrast, TST positivity in low altitude regions increases progressively starting at five years of age, and continues to increase throughout adulthood. Finally, household clustering of TST positivity was found to be higher in the high altitude villages than in the sea level communities studied.
While historical evidence has suggested an association between high altitude and decreased tuberculosis risk, no studies have addressed this hypothesis in a systematic and comprehensive way. Our findings support previous observations of a lower prevalence of tuberculosis infection in high altitude villages.9 In our study, sea level areas located in two very different geographic environments (coast and Amazon basin jungle) were compared with two high altitude villages also from separate areas. Despite their differences in setting and transmission dynamics, the TST-positive prevalence rates determined at the two high altitude villages were almost identical and also significantly lower than those of the three sea level sites, which conversely had highly comparable TST positivity rates. In spite of the historical consideration that tuberculosis is a disease of crowded urban areas, the crude TST-positive prevalence in jungle villages is actually higher than in our Lima shantytown. The reason for the high rates in the jungle population is not clear. The proximity to urban centers was similar in both the low and high altitude rural villages, yet Quilcas had a low TST-positive prevalence rate in contrast to the high rate in the jungle population.
The protection effect afforded by high altitude appears to be relative. Crowding, exposure to increased tuberculosis carriers, and increased travel appear to be able to overcome this protective effect. Preliminary data from a high altitude shantytown of Cusco, located at an altitude of 3,240 meters, has demonstrated TST-positive prevalence rates similar to those found in sea level populations and three times higher than the rural high altitude villages tested in this survey. (Saito M, Gilman RH, unpublished data). The direct affect of increased contact with known patients with active tuberculosis is accounted for statistically in this study, but this might not account for an increased risk of acquiring tuberculosis from travelers, visitors, or other transient and unknown contacts. The data showing that residence clustering of TST positivity is greater in high altitude village is also consistent with the idea that outsider contact might be driving tuberculosis infection in lowland communities.18
There are several potential mechanisms by which a single environmental factor, high altitude, may affect M. tuberculosis transmission. First, M. tuberculosis grows in vitro at a slower rate at lower oxygen tensions, which may also be true in vivo.19 Second, UV light has been shown to increase the death of microorganisms, including M. tuberculosis.20,21 Ultraviolet light is increased at high altitude. This effect may result in a stronger germicidal effect than that which occurs at sea level.22 However, this is less likely to be a factor since UV rays do not reach inside houses, where most transmission is likely to occur. Third, studies have suggested a link between humidity and the susceptibility of Mycobacterium to UV light. Ko and others recently found that UV resistance of Serratia marcescens and Mycobacterium bovis BCG increased dramatically in conditions of high relative humidity (≥ 85%). These investigators have postulated that high relative humidity may exercise its protective effect on Mycobacterium via attenuation of the UV beam, maintenance of the microorganism’s water coat, or by increasing particle size through water absorption.22 In environments of low relative humidity, bacterial susceptibility to UV radiation may be elevated, potentially leading to decreased viability of microorganisms, and therefore decreased rates of infection.23 The above considerations suggest that high altitude environmental conditions may reduce tuberculosis transmission. However, further detailed studies are needed to determine which mechanism(s) are primarily responsible for the decreased transmission of tuberculosis in the highlands. However, our data strongly suggest that high altitude is associated with decreased tuberculosis transmission.
At low altitudes, in addition to having increased TST-positive rates, the exposure of populations to tuberculosis seems to begin at an earlier age than in the high altitude villages. Age-stratified analyses showed that for any given age range, the levels of TST positivity are significantly lower in the high altitude villages, compared with the low altitude locations. In low altitude areas TST positivity rates begin to increase at a younger age (10–19 years of age) than at higher altitudes, where TST positivity rates remain relatively low and stable until adulthood (20–29 years of age). In the low altitude communities, the early increase in TST positivity is probably attributable to the increased level of social interaction that begins when children enter school. In high altitude villages, no large increase was observed in individuals less than 30 years old, suggesting low levels of tuberculosis transmission within these communities. Higher rates of TST positivity after 30 years of age are likely to be the result of increased travel to areas where the prevalence of tuberculosis transmission is high. Economic activities may expose inhabitants of the highlands to higher risk of tuberculosis transmission due to travel to or work in lower altitude communities.
There was a much higher number of individuals having contact with a known case of tuberculosis in a sea level shantytown than at any other location. The small number of individuals with contact with a known case of tuberculosis in both jungle and high altitude villages may explain the lack of a significant association for this variable. More difficult to explain is the effect high education had on increased TST rates. It is possible that families with higher education travel more and thus have increased exposure to non-family tuberculosis carriers.
Clustering analysis in Quilcas, Pampas, and Buen Pastor revealed differences in household aggregation of infection, implying that familial risk of infection may be higher in these low TST-positive prevalence areas than in high prevalence areas. Indeed, significant levels of intra-household clustering of infection were found in all sites, but were particularly strong in Quilcas. Household clustering of tuberculosis can arise from a common source of infection, similarities in behavior within the family, or intra-familial transmission. Evidence of higher clustering in high altitude villages suggests that familial risk of infection is increased in even low TST prevalence areas. Given the higher household clustering OR observed in high altitude villages, family-based screening and prophylaxis may offer a more effective prevention strategy in these areas as opposed to individual-targeted tuberculosis control measures.18 This finding is particularly relevant because peri-urban in contrast to high altitude, rural villages, is influenced more by elements in the community than by household-levels factors.24
This study has demonstrated that the protective effect in village populations of high altitude for tuberculosis transmission appears to be related to both their relative isolation and the protective effect of high altitude. The identification of high altitude villages as zones of increased household clustering of tuberculosis infection may help to develop more effective interventions in these areas.
Site description and demographic characteristics of the study population*
| High altitude | Sea level | ||||
|---|---|---|---|---|---|
| Feature | Quilcas | Vichaycocha | Pampas | Buen Pastor | San Carlos |
| * min = minimum; max = maximum; NA = not available. | |||||
| †Weather information from Iquitos was used as an approximation for that of San Carlos and Buen Pastor. | |||||
| ‡Participation was requested at the health post. | |||||
| §Children not attending school yet were included in the “No education” category. | |||||
| ¶P < 0.01 compared to all other sites. | |||||
| Region | Highlands | Highlands | Coast | Jungle | Jungle |
| Altitude (meters) | 3,330 | 3,500 | 140 | 110 | 110 |
| Area | Rural | Rural | Urban | Rural | Rural |
| Climate | |||||
| Temperature (°C) [min–max] | [10–15] | NA | [17–22] | [23–32]† | |
| Relative humidity (%) [min–max] | 53 [49–82] | NA | 84 [83–86] | 89 [87–91]† | |
| Annual rainfall (cm) | 76 | 78 | 1 | 290† | |
| Population | 1,278 | 517 | 40,000 | 270 | 176 |
| No. of households selected | 100 | NA | 397 | 50 | 31 |
| No. of participants tested/targeted (%) | 453/677 (67) | 370/517 (72)‡ | 1,250/1,847 (68) | 188/270 (70) | 149/176 (85) |
| Study period | Jul–Sep 1999 | Jul 1997 | Mar–Aug 2000 | May–Jun 2000 | Apr 2000 |
| Female/male ratio | 1.1 | 1.1 | 1.2 | 1.0 | 1.0 |
| Mean (SD) age in years | 20 (17) | 27 (22) | 19 (15) | 24 (20) | 21 (17) |
| Education | |||||
| No education (%)§ | 122 (27) | 21 (7) | 207 (25) | 36 (19) | 28 (19) |
| Primary (%) | 248 (55) | 219 (74) | 529 (42) | 109 (58) | 90 (62) |
| Secondary or more (%) | 81 (18) | 57 (19) | 409 (33) | 43 (23) | 27 (18) |
| Mean (SD) number of people per room | 1.9 (1.0) | NA | 1.5 (0.9) | 2.7 (1.8) | 3.3 (2.0) |
Distribution of risk factors associated with tuberculin skin test (TST) positivity by study site*
| High altitude | Sea level | ||||
|---|---|---|---|---|---|
| Variables | Quilcas | Vichaycocha | Pampas | Buen Pastor | San Carlos |
| *NA = not available. | |||||
| †P < 0.001. | |||||
| ‡P < 0.05. | |||||
| Bacillus Calmette-Guérin vaccinated (%) | 264/453 (58) | 191/370 (52) | 937/1,248 (75)† | 97/182 (53) | 87/146 (60) |
| History of living with a know active tuberculosis patient (%) | 19/453 (4) | 4/370 (1)‡ | 118/1,241 (10) | 9/187 (5) | 16/149 (11) |
| Coughing for two weeks or more (%) | 126/453 (28)‡ | 26/370 (7) | 145/1,247 (12) | 35/188 (19) | 12/149 (8) |
| Presence of another TST-positive person in the household (%) | 107/448 (24)† | NA | 728/1,226 (59) | 141/183 (77) | 124/148 (84) |
| Participants tested for TST positivity | 453 | 370 | 1,250 | 188 | 149 |
Adjusted odds ratios (ORs) for risk factors associated with tuberculin skin test (TST) positivity within study site*
| High altitude | Sea level | |||
|---|---|---|---|---|
| Quilcas | Pampas | Buen Pastor | San Carlos | |
| Variables | OR (95% CI) | OR (95% CI) | OR (95% CI) | OR (95% CI) |
| * CI = confidence interval; BCG = Bacillus Calmette-Guérin; TB = tuberculosis. The categories in parenthesis describe the baseline group for odds ratio calculation. All odds ratios were calculated within each site. | ||||
| †P < 0.05. | ||||
| ‡P < 0.001. | ||||
| §P < 0.01. | ||||
| ¶ Children not attending school yet were included in the “No education” category. The association of education with higher TST positivity became nonsignificant when those children were excluded from the analysis. | ||||
| Age 20–29 years (age 0–19 years) | 4.0 (1.0–16.3) | 4.3 (2.8–6.7)† | 11.7 (3.4–40.4)† | 3.7 (0.8–18.2) |
| Age 30–39 years (age 0–19 years) | 18.6 (5.1–68.4)‡ | 7.2 (4.7–11.0)† | 15.9 (3.8–66.0)† | 3.4 (0.9–12.8) |
| Age 40–49 years (age 0–19 years) | 14.1 (1.7–120.0)† | 16.2 (8.0–32.7)† | 60.9 (10.8–344.4)† | 14.4 (1.8–115.0)‡ |
| Age ≥50 years (age 0–19 years) | 53.0 (9.0–314.0)‡ | 14.5 (6.4–32.9)† | 30.6 (6.6–142.3)† | 50.5 (4.6–554.8)§ |
| Some primary education (no education)¶ | 7.9 (0.9–65.3) | 2.0 (1.2–3.4)§ | 1.4 (0.4–5.1) | 5.1 (0.9–29.3) |
| Completed high school or more (no education)¶ | 11.4 (1.2–110.7)† | 4.1 (2.4–7.0)§ | 1.5 (0.4–6.4) | 5.3 (0.8–33.7) |
| Having BCG scar (no BCG scar) | 2.2 (0.6–7.5) | 1.6 (1.0–2.5)‡ | 2.1 (0.6–7.2) | 2.3 (0.7–8.1) |
| History of living with a known active TB patient (no history) | 1.4 (0.3–5.9) | 2.7 (1.7–4.2)† | 4.5 (0.6–32.6) | 2.7 (0.4–17.6) |
| Coughing for two weeks or more (not coughing or coughing less than 2 weeks) | 3.0 (1.2–7.4)‡ | 1.4 (0.9–2.2) | 2.5 (0.9–7.3) | 1.3 (0.3–7.1) |
Differences in tuberculin skin test (TST) risk and TST reaction size, household clustering for TST positivity, and reaction size between low altitude communities and high altitude village in Peru*
| High altitude | Sea level | |||
|---|---|---|---|---|
| Outcome | Quilcas | Pampas | Buen Pastor | San Carlos |
| * OR = odds ratio adjusted for age, education, Bacillus Calmette-Guérin vaccination status, contact with active tuberculosis cases, and prolonged productive coughing; CI = confidence interval. | ||||
| † From two multiple random effects regression models pooling all sites data using Quilcas as the comparison group. The logistic regression for TST positivity was additionally adjusted for the presence of another TST-positive person in the household. | ||||
| ‡P < 0.001. | ||||
| § From separated multiple random effects logistic regression models for each study site. The clustering odds ratio is calculated by adding to the model a variable representing the presence of another TST-positive individual living in the household. | ||||
| ¶P < 0.01. | ||||
| #P < 0.05. | ||||
| Differences in TST risk and TST reaction size† | ||||
| OR for TST positivity between sea level communities vs. high altitude village (95% CI) | Reference | 4.5 (2.9–7.0)‡ | 5.7 (3.3–10.0)‡ | 6.0 (3.4–10.9)‡ |
| Difference in reaction size (mm) between sea level communities vs. high altitude village (95% CI) | Reference | 4.1 (3.2–4.9)‡ | 4.8 (3.4–6.1)‡ | 4.8 (3.3–6.2)‡ |
| Differences in household clustering for TST positivity§ | ||||
| OR for the presence of another TST-positive person in household (95% CI) | 3.7 (1.5–9.0)¶ | 2.1 (1.5–2.9)‡ | 2.8 (1.0–7.7)# | 0.2 (0.0–3.3) |
| Percentage of the variance in TST positivity attributed to household clustering (95% CI) | 61 (28–87)# | 41 (25–60)‡ | 29 (2–88) | 0 |
Study sites and natural geographic regions of Peru. Small dots represent study sites. Large dots with names in uppercase letters represent important cities near the study sites.
Citation: The American Journal of Tropical Medicine and Hygiene Am J Trop Med Hyg 68, 6; 10.4269/ajtmh.2003.68.721
Study sites and natural geographic regions of Peru. Small dots represent study sites. Large dots with names in uppercase letters represent important cities near the study sites.
Citation: The American Journal of Tropical Medicine and Hygiene Am J Trop Med Hyg 68, 6; 10.4269/ajtmh.2003.68.721
Study sites and natural geographic regions of Peru. Small dots represent study sites. Large dots with names in uppercase letters represent important cities near the study sites.
Citation: The American Journal of Tropical Medicine and Hygiene Am J Trop Med Hyg 68, 6; 10.4269/ajtmh.2003.68.721
Prevalence of tuberculin skin test (TST) positivity by age group and study site in Peru.
Citation: The American Journal of Tropical Medicine and Hygiene Am J Trop Med Hyg 68, 6; 10.4269/ajtmh.2003.68.721
Prevalence of tuberculin skin test (TST) positivity by age group and study site in Peru.
Citation: The American Journal of Tropical Medicine and Hygiene Am J Trop Med Hyg 68, 6; 10.4269/ajtmh.2003.68.721
Prevalence of tuberculin skin test (TST) positivity by age group and study site in Peru.
Citation: The American Journal of Tropical Medicine and Hygiene Am J Trop Med Hyg 68, 6; 10.4269/ajtmh.2003.68.721
Authors’ addresses: Susan Olender, Mayuko Saito, Jane Apgar, Kari Guillenwater, Christian T. Bautista, Andres G. Lescano, Pedro Moro, Luz Caviedes, and Evelyn J. Hsieh, A.B. PRISMA Carlos Gonzales 251, Urb. Maranga, San Miguel, Lima 32, Peru, Telephone: 51-1-464-0221, Fax: 51-1-464-0781. Robert H. Gilman, Department of International Health, The Johns Hopkins School of Public Health, 615 North Wolfe Street, Room W3503, Baltimore, MD 21205, Telephone: 410-614-3959, Fax: 410-614-6060, E-mail: rgilman@jhsph.edu.
Acknowledgments: We thank Drs. Carlton Evans, Lawrence Moulton, Richard Oberhelman, Yutaka Aoki, and Dimitris Placantonakis for their advice on this paper and comments on the manuscript; Jenny Centeno and Pamela Limo for preparation of the map; Marco Varela for data management; and J. B. Phu, D. Sara, E. Talula for technical support. We also thank the communities of Quilcas, Vichaycocha, Las Pampas de San Juan de Miraflores, San Carlos, and Buen Pastor for their cooperation.
Financial support: This study was supported by the United States Agency for International Development Tuberculosis Award HRN-5986-A-00-6006-00, National Institutes of Health (NIH) International Training and Research in Emerging Infectious Diseases grant 5D43-TW00910, Fogarty-NIH AIDS training program 3T22-TW00016-05S3, National Institute of Allergy and Infectious Diseases (NIAID) tutorial training grant 5T35-AI-07646-02, NIAID predoctoral training grant 5T32AI-007526, and the anonymous RG-ER fund for the advancement of tropical medicine research.
REFERENCES
- 1↑
World Health Organization, 2001. WHO Report 2001: Global TB Control. Geneva: World Health Organization.
- 2↑
World Health Organization, 2000. Peru: Epidemiological Fact Sheet on HIV/AIDS and Sexually Transmitted Infections: 2000 Update. Geneva: World Health Organization.
- 3↑
Rogers FB, 1969. The rise and decline of the altitude therapy of tuberculosis. Bull Hist Med 43 :1–16.
- 4
Smith A, 1840. Practical observations on the diseases of Peru, described as they occur on the coast and in the sierra. Edinburgh Med Surg J 54 :1–19.
- 6↑
Fourie PB, 1983. The prevalence and annual rate of tuberculous infection in South Africa. Tubercle 64 :181–192.
- 7↑
Mansoer JR, Kibuga DK, Borgdorff MW, 1999. Altitude: a determinant for tuberculosis in Kenya? Int J Tuberc Lung Dis 3 :156–161.
- 8↑
Rich AR, Follis RH, 1942. The effect of low oxygen tension upon the development of experimental tuberculosis. Bull Johns Hopkins Hosp 71 :193–202, 345–363.
- 9↑
Bates LR, Busk T, Palmer CE, 1951. Tuberculin Sensitivity at Different Altitudes of Residence. Copenhagen, Denmark: World Health Organization. Tuberculosis Research Office. Research Contributions of BCG Vaccination Programs.
- 10↑
Getchell WS, Davis CE, Gilman J, Urueta G, Ruiz-Huidobro E, Gilman RH, 1992. Basic epidemiology of tuberculosis in Peru: a prevalence study of tuberculin sensitivity in a Pueblo Joven. Am J Trop Med Hyg 47 :721–729.
- 11↑
Jordan TL, Sunderam G, Thomas L, Reichman LB, 1987. Tuberculin reaction size measurement by the pen method compared to traditional palpation. Chest 92 :234–236.
- 12↑
American Thoracic Society and the Center for Disease Control and Prevention, 2000. Diagnostic Standards and Classification of Tuberculosis in Adults and Children. Am J Respir Crit Care Med 161 :1376–1395.
- 13↑
Ministry of Health, 2000. Dirección del programa nacional de control de enfermedades transmisibles control de la tuberculosis. Actualizacion en la Tecnica de Aplicación, Lectura e Interpretación de la Prueba de Tuberculina. Directiva No. 003 –2000 -PCT. Lima, Peru: Ministry of Health.
- 15↑
Hauck WW, Donner A, 1977. Wald’s test as applied to hypotheses in logit analysis. J Am Stat Assoc 72 :851–853.
- 16↑
Gould W, Sribney W, 1999. Maximum Likelihood Estimation with STATA. College Station, TX: Stata Press.
- 17↑
Moro PL, Checkley W, Gilman RH, Cabrera L, Lescano AG, Bonilla JJ, Silva B, 2000. Gallstone disease in Peruvian costal natives and highland migrants. Gut 46 :569–573.
- 18↑
Madico G, Gilman RH, Checkley W, Cabrera L, Kohlstadt I, Kacena K, Diaz JF, Black R, 1995. Community infection ratio as an indicator for tuberculosis control. Lancet 345 :416–419.
- 19↑
Sever JL, Youmans GP, 1957. The relation of oxygen tension to virulence of tuberculosis bacilli and to acquired resistance in tuberculosis. J Infect Dis 101 :1427–1442.
- 20↑
Riley RL, Knight M, Middlebrook G, 1976. Ultraviolet susceptibility of BCG and virulent tubercle bacilli. Am Rev Respir Dis 113 :413–418.
- 21↑
Menzies D, Fanning A, Yuan L, Fitzgerald M, 1995. Tuberculosis among health care workers. N Engl J Med 332 :92–98.
- 22↑
Ko G, First MW, Burge HA, 2000. Influence of relative humidity on particle size and UV sensitivity of Serratia marcescens and Mycobacterium bovis BCG aerosols. Tuber Lung Dis 80 :217–228.
- 23↑
Riley RL, Kaufman JE, 1972. Effect of relative humidity on the inactivation of airborne Serratia marcescens by ultraviolet radiation. Appl Microbiol 23 :1113–1120.
- 24↑
Comstock GW, Ferebee SH, Hammes LM, 1967. A controlled trial of community-wide isoniazid prophylaxis in Alaska. Am Rev Respir Dis 95 :935–943.