Impact of indoor ventilation on TB transmission risk: implications of climate change

Summary

Background:

Due to rising temperatures, individuals are predicted to spend more time in under-ventilated indoor spaces, increasing tuberculosis (TB) transmission risk. We studied the impact of indoor ventilation on TB transmission risk in homes of persons with TB (PWTB) and in healthcare facilities in Puducherry, India.

Methods:

We measured ventilation in air changes per hour (ACH) under different ventilation conditions using a carbon dioxide decay method. We estimated transmission risk using the Wells-Riley equation.

Results:

45 measurements were taken in 13 homes and 7 healthcare spaces. In the closed condition (doors and windows closed, fans off), ACH was low (mean 2.23, SD 2.27) and TB transmission risk was high at 62% (SD 31%). When air conditioning (AC) was on, ACH reduced to 0.75 (SD 0.51), and TB transmission was highest at 76% (SD 13%). Natural ventilation significantly improved ACH (mean 9.46, SD 3.90; p<0.001) and TB transmission risk to mean 20% (SD 14%; p<0.001) compared to the closed condition.

Conclusions:

TB transmission risk in homes and healthcare spaces is high, especially with AC on. Adapting to rising temperatures using novel methods of ventilation, cooling, and air purification is critical to TB infection control in the era of climate change.

INTRODUCTION

Despite efforts to mitigate climate change, temperatures are likely to rise 1.5°C above pre-industrial levels by 2035, the threshold established by the Paris Agreement to avoid the worst impacts of climate change.¹ The effects of rising global temperatures will disproportionately affect tropical and subtropical regions. For instance, severe heat waves are becoming increasingly common in South Asia, such as the heat wave that lasted over a month in 2022, reaching temperatures as high as 49.5°C (121.1°F).²

In tropical regions, most individuals rely on open doors, windows, and ceiling fans to mitigate heat. However, as temperatures rise, air conditioner (AC) use is rising dramatically in the tropics.^3,4 Indeed, increased electricity demands due to AC use led to a collapse of the electrical grid in parts of India during the 2022 heatwave.² While AC is an effective cooling method, the closing of windows and doors that accompanies AC use increases recirculation of indoor air and reduces air changes per hour (ACH).^5,6 This has particular implications for healthcare spaces which must balance cooling needs with infection control. Reduced ventilation in healthcare spaces and homes of persons with TB (PWTB) may increase transmission risk of respiratory pathogens.

We conducted a cross-sectional analysis to estimate the impact of ventilation on ACH and the probability of TB transmission in homes of PWTB and a hospital in South India where TB care is provided.

METHODS

Study design, participants, and setting:

We conducted this cross-sectional pilot study in Puducherry, South India from June - July 2024. We measured ventilation in the homes of persons with pulmonary TB and in healthcare spaces.

For home measurements, we recruited a convenience sample of participants enrolled in the TB-LEOPARD (Learning Effect of Parasites and Reinforcing Diets [NCT05048485]) study. Participants left their homes during measurements. Homes were classified as either “finished” or “semi-finished”. Finished homes were constructed from higher-quality materials such as cement. Semi-finished homes were constructed from rudimentary materials such as corrugated aluminum sheets, plywood, or mud or had gaps between the roof and walls.

For healthcare spaces, we used offices and patient care areas at the Jawaharlal Institute of Postgraduate Medical Education and Research (JIPMER), a tertiary hospital in Puducherry. Healthcare spaces were all finished settings. Staff and patients were not present during measurements.

Data Collection:

Home and Healthcare Space Measurements

In each space, we collected the following data: room volume; material and dimensions of windows and doors; presence of cross-ventilation; presence and number of fans and AC units. In keeping with previous studies, cross-ventilation was defined as pairs of opposing windows or windows across from doors.⁷

Environmental Measurements

We recorded indoor and outdoor temperature and humidity.

Carbon dioxide Measurements

We used a carbon-dioxide (CO₂) decay technique to measure ACH. With a Vernier GDX-CO₂ gas sensor, we first recorded the baseline CO₂ level in parts per million (ppm) as the average CO₂ level over 60 seconds.

We then took measurements in 3 configurations: (1) closed, (2) closed with AC, and (3) open. In (1), all doors and windows were closed, and fans and AC were turned off. In (2), all doors and windows remained closed, and AC was turned on. If the space did not have AC, this configuration was skipped. In (3), all doors and windows were opened, AC was turned off, and all fans were turned on.

In all configurations, we slowly released CO₂ into the room with fans on to ensure even mixing. As gas was infused, we monitored CO₂ levels with measurements taken at 0.5 second intervals. Once the CO₂ level crossed 2000 ppm, we stopped infusing the gas and exited the room. If the test configuration required it, we turned off the fans. CO₂ measurements were collected either for 15 minutes or until CO₂ levels returned to the recorded baseline, whichever occurred first. We calculated the ACH for each configuration as the slope of the line of best-fit for the natural log of the CO₂ measurements plotted against time in hours.⁷

TB Transmission Probability

We used the Wells-Riley (WR) equation to calculate the probability of transmission:

$$P=1-e^{-Iqpt∕Q}$$

where $I$ is the number of untreated infectors, $q$ is the infectious quanta produced per hour per infectious individual, $p$ is the pulmonary ventilation rate of susceptible individuals, $t$ is time, and $Q$ is absolute room ventilation, i.e. ACH multiplied by the room volume.⁸ P is the probability of a new TB case. We use the calculated probabilities to elucidate relative comparisons across ventilation scenarios rather than precise predictions of infection risk.

Where possible, previously established values are used to facilitate comparisons to other studies. We assumed all studied spaces had 1 infectious individual. We assumed $t$ to be 0.6 m³/h.⁹ We assumed $p$ to be 10 hours – approximately the time susceptible individuals would spend with an infectious individual overnight or during a workday.⁷

Several papers have used a $q$ of 13, a value established from a TB outbreak in an office building in 1991.^7,9,10 In keeping with these studies, we used $q$ = 13 for our baseline estimate. However, studies have estimated a wide range of $q$ using two methodologies: (1) data from real TB outbreaks and (2) data from guinea pigs breathing in TB hospital ward air. We identified 3 outbreak studies and 4 guinea pig studies (Supplementary Table 1). ^9,11-16

To address heterogeneity in q, we used a modified WR equation described by Edwards et al. that incorporates quanta emission rate as an exponential probability distribution function:

$$P=1-\frac{\lambda }{\frac{pI}{Q}t+\lambda }$$

Here, $\lambda$ is the rate parameter (reciprocal of the mean q).¹⁷ Unlike the original WR equation, which is derived with a Poisson distribution, this approach is based on a susceptible-exposed (SE) model. We estimated mean values for each study type and calculated rate parameters for the real outbreak ($\lambda$ = 0.024), guinea pig ($\lambda$ = 0.126). We used the modified WR equation to estimate transmission risk based on these two study types.

Statistical Analysis Methods

We assessed normality using the Shapiro-Wilk test, and Levene's test evaluated variance equality. Where data did not meet normality or equal variance assumptions, we used non-parametric tests. To compare ACH and TB transmission risk between the close and open conditions only, we used Wilcoxon rank-sum test. To compare all three configurations (open, closed, and AC), we performed a Friedman test followed by post-hoc Wilcoxon tests with Bonferroni corrections. For group comparisons, we used the Mann-Whitney U test to compare finished and semi-finished homes.

Ethics:

We secured IRB approvals at Boston Medical Center (H-44289) and JIPMER. Written consent was obtained and participants were provided compensation.

RESULTS

Study Population

We measured ACH 45 times and estimated the TB transmission risk in 13 homes and 7 healthcare spaces in Puducherry, India (Table 1). Of 13 homes, 9 were categorized as 'finished' (constructed from improved materials) and 4 as 'semi-finished' (constructed from rudimentary materials). Of 7 healthcare spaces, 3 are in parts of the pulmonary department where PWTB wait and are seen by clinicians. Five of the spaces (one home and four healthcare) were equipped with AC.

Table 1:

Structural characteristics of the spaces assessed

Number of Windows, (%)	Homes (n=13)	Healthcare Spaces (n=7)
0	1 (8%)	0 (0%)
1	5 (38%)	1 (14%)
2	5 (38%)	1 (14%)
3	1 (8%)	2 (29%)
4	1 (8%)	2 (29%)
5	0 (0%)	0 (0%)
6	0 (0%)	1 (14%)
Window Area, mean (SD)
Single Window [m2]	0.5 (0.4)	1.1 (0.4)
Total Window Area [m2]	0.72 (0.79)	1.75 (1.68)
Number of Doors
1	4 (31%)	3 (43%)
2	9 (69%)	2 (29%)
3	0 (0%)	2 (29%)
Door Area [m2], mean (SD)	1.3 (0.28)	2.1 (0.43)
Cross Ventilation Present	6 (46%)	6 (86%)
Roof Material
Cement	8 (62%)	7 (100%)
Asbestos Sheets	4 (31%)	0 (100%)
Aluminum Sheets	1 (8%)	0 (100%)
Room Volume [m2], mean (SD)	40 (18)	139 (100)
Indoor Temperature [C°], mean (SD)	31.6 (2.5)	29.6 (1.7)
Outdoor Temperature [C°], mean (SD)	32.3 (1.5)	32.2 (4.9)
Indoor Relative Humidity [%], mean (SD)	63.0 (2.4)	39 (1.2)

SD denotes standard deviation

ACH Across Ventilation Conditions

ACH varied significantly across ventilation conditions (Wilcoxon test, p < 0.001) (Figure 1). The closed condition in homes yielded a mean ACH of 2.31 (standard deviation [SD] 2.39), which significantly increased when doors and windows were opened (mean ACH = 9.34, SD 4.00). Similarly, in healthcare spaces, ACH significantly increased from a mean 2.07 (SD 2.19) in the closed condition to 9.69 (SD 4.00) in the open condition (Wilcoxon test, p < 0.001). However, the use of AC in healthcare spaces led to a marked decrease in ventilation, with a mean ACH of 0.62 (SD 0.49), significantly lower than even the closed condition (Friedman test, p = 0.039).

Figure 1: Air changes per hour and TB transmission risk under different ventilation conditions.

Box-plots of air changes per hour (ACH) (1A, 1B) and TB transmission risk using base estimate of q (q=13) (1C, 1D) in homes and healthcare spaces in the three ventilation conditions tested: (1) windows and doors closed with fans and air conditioning (AC) turned off, (2) windows and doors closed with AC on, (3) windows and doors open with fans turned on and AC turned off. Solid line at top indicates Wilcoxon test comparing closed and open conditions, showing a significant difference in both ACH and transmission risk. SD denotes standard deviation.

TB Transmission Risk

With our base estimate of infectious quanta per hour (q=13), transmission risk was highest when AC was used, with an average risk over 10 hours of 77.04% (SD 14.45) in healthcare spaces and an estimate of 74.27% in the one home with AC in our study. Comparatively, the average transmission risk was lower in the closed condition at 69.34% (SD 28.49) in homes and 49.73% (SD 33.94) in healthcare spaces. The open configuration significantly reduced transmission risk to 24.00% (SD 11.42%; p < 0.001) in homes and 12.26% (SD 14.64; p = 0.011) in healthcare spaces.

Sensitivity Analysis of Quanta Generation

Similar results were seen using the q-distributions based on outbreak and guinea pig studies (Figure 2). Transmission risk in homes and healthcare spaces was highest when AC was used, with average TB transmission risk of 82% (SD 6) and 48% (SD 10) for q-distributions from outbreak and guinea pig studies, respectively. In the closed condition, transmission risks dropped to 71.3% (SD 22) and 41.3% (SD 24) when using outbreak and guinea pig study q-distributions, respectively, although these were not significantly lower than the transmission risks with AC. As before, the open condition significantly reduced transmission risk to 37% (p<0.001) and 12% (p<0.001) for outbreak and guinea pig study types, respectively.

Figure 2: TB Transmission Risk for Different q-distributions.

Boxplot of TB Transmission Risk for q-distributions based on different study types. Solid lines underneath the boxplots indicate Wilcoxon tests between closed and open ventilation conditions.

Finished versus Semi-Finished Homes

In the closed configuration, finished homes had a significantly higher transmission risk (mean 82.2%, SD 21.7%) compared to semi-finished homes (mean 40.4%, SD 19.5%) (Mann-Whitney U test, p = 0.019) (Figure 3). This difference was not observed in the open condition (Supplementary table 2), where transmission risks between finished and semi-finished homes were comparable (26.5% versus 18.4%, p = 0.257).

Figure 3: TB transmission risk for finished vs. semi-finished homes.

Boxplots of TB Transmission Risk for finished and semi-finished homes in the closed and open ventilation configurations. Solid lines at the top indicate Wilcoxon test between finished and semi-finished home types for each ventilation condition. SD denotes standard deviation.

DISCUSSION:

To our knowledge, this is the first study to directly estimate ventilation rates and TB transmission risk in residential and healthcare spaces in India. The most critical finding is the dramatic reduction in ACH and associated increase in TB transmission risk when doors and windows are closed. In healthcare spaces, using AC with closed windows and doors provided the greatest risk of TB transmission among the configurations tested. Semi-finished homes had a lower TB transmission risk even in closed configurations, likely due to passive ventilation from gaps between walls and ceiling. Opening doors and windows to facilitate natural ventilation reduced transmission risk threefold. Our findings are particularly concerning in the context of climate change, as rising temperatures are expected to drive more individuals to spend extended periods in poorly ventilated, air-conditioned environments.

Our results are consistent with previous research from South African hospitals that demonstrated the importance of natural ventilation for reducing TB transmission risks in clinical settings.¹⁸ Similarly, studies in Indian hospitals have identified AC use as a potential risk factor for healthcare-associated infections due to inadequate filtration, rebreathing of contaminated indoor air, and reduced ventilation due to closed windows with AC use.^19,20 The cooling effect of AC likely has minimal effect on air circulation.²¹ Given that most clinical centers that care for PWTB in India are finished spaces resembling those in our study, there is a clear need to prioritize infection prevention and control measures in these settings, such as environmental controls (e.g. ultraviolet germicidal irradiation, air purifiers, high-efficiency particulate air filters) and administrative controls (e.g. masks and crowd control).^22-24

Architectural interventions, such as implementing natural ventilation or germicidal ultraviolet lamps to sanitize air, may provide more cost-effective solutions for reducing disease transmission.^10,25,26 Indeed, Lygizos et al. (2013) showed that cross-ventilation can drastically increase ACH, effectively halving the risk of TB transmission in rural Zulu homes.⁷ Incorporating elements of traditional architecture may improve passive cooling in modern buildings, particularly in regions with dry heat in India. The wooden lattice screens over windows at the Amber Fort, for example, effectively blocked sunlight, increased cross-ventilation, and humidified dry air.^27,28 Where traditional buildings relied on central bodies of water for evaporative cooling, modern homes may use desert coolers instead of AC. However, the risk of arboviral illnesses like dengue must be weighed against the risk of TB transmission.²⁹ Vegetation can also ameliorate extreme temperatures in urban areas.³⁰

An average decrease in ACH was observed in the four healthcare spaces with AC. It is conceivable that similar findings would be observed in finished homes and non-healthcare workspaces. Indeed, two healthcare spaces in our study are typical of Indian workspaces. One is occupied by approximately 30 staff members daily who operate AC in the summer despite numerous windows. This space had one of the worst ventilation rates noted in our study, highlighting the role of occupational health in reducing transmission risk. Recently, there has been an interest by businesses and corporate entities in joining the fight against TB.³¹ Improving ventilation and cooling in workspaces can be a concrete step towards doing so.

AC use in residential spaces is expected to increase 4-fold by 2040 in India, and this rise may not be restricted to wealthy households.^3,4 Impoverished individuals may share air-conditioned spaces with others, spending more time in cramped, under-ventilated rooms. AC use is likely to rise in workspaces where employees spend the bulk of their waking hours. Indeed, 80% of TB transmission globally is expected to occur outside the home.³² While the impact of climate change on mixing behaviors is not well understood, the greater adoption of AC and reduction in natural ventilation in homes and congregate settings alike has important implications for transmission of respiratory pathogens.^6,10

One strength of this study is the use of direct ventilation measurements across various conditions, providing a more nuanced understanding of how environmental risk factors impact transmission, particularly AC, a key knowledge gap. Additionally, the integration of WR modeling with sensitivity analyses using quanta emission rate distributions further enhances our transmission risk estimates by accounting for heterogeneity in production of infectious material.

We recognize several limitations. First, the small sample size limits the generalizability of our findings. While our study provides valuable pilot data, larger studies are needed to confirm these results across diverse settings, particularly with AC in homes and during different seasons. Indeed, limited evidence suggests that low ambient humidity in winter may increase release of respiratory aerosols and facilitate TB transmission.³³ Second, our use of a CO₂ decay method requires assumption of steady-state conditions prior to decay, which is challenging to obtain in leaky spaces, preventing determination of ventilation in absolute terms. Third, the WR equation, though widely used, is a theoretical model that assumes constant infectiousness and does not account for individual variability in quanta generation, individual risk factors for disease progression, or repeat exposures.⁸ Moreover, our exposure time estimates, while consistent with previous studies, may underestimate cumulative risk in homes where prolonged contact is common. Fourth, for the sensitivity analysis, it should be cautioned that the Edwards modification of the WR equation utilizes a different interpretation of quanta, limiting one’s ability to draw comparisons between probabilities calculated from each of the models. Lastly, our transmission risk estimate does not account for individual behaviors, such as adherence to mask-wearing in clinical settings, which could modify risk.

This study identifies a key mechanism through which climate change may exacerbate the TB pandemic in countries facing both high TB and climate change burdens. Prioritizing architectural interventions that enhance natural ventilation, such as optimizing building designs to increase airflow and incorporating passive ventilation systems, could significantly reduce the risk of TB transmission.