How much heterogeneity is there in the spatial and temporal distribution of microbes in the built environment? Airborne fungi in different rooms vary across seasons in residences [
22]. The microbial exposure where infants crawl and young children toddle is distinct from that at the level where adults are breathing [
23,
24]. In a case study reported in this issue, Tang et al. find that individual samplers on different individuals varied from zero RSV RNA copies to a high of 2778/m
3. It is currently unknown how much heterogeneity is inside the cubic meter that contains thousands of genomes. At the extremes, the entire viral load could have been confined to one compact particle or, alternatively, it could have been evenly dispersed through the entire volume.
Rapid and focused public health actions motivated by precision data could lead to less overall disruption and more efficient prevention [
27]. For example, if infected animals can be reliably identified, there may be less need to cull larger groups. Individual cattle, pigs and chicken cages are already identified in current agricultural settings to allow the tracing of Bovine Spongiform Encephalopathy (BSE) [
28]. Globally, approximately 20% of livestock are lost to preventable diseases, costing around USD $2 billion per annum in Africa alone, with much of this loss in low- and lower-middle income settings [
29]. There is substantial potential for veterinary, human public health and economic benefit.
Heterogeneity and sampling
The optimum mode of sampling depends on the distribution of targets. If the target is rare and evenly distributed, then gathering as much of the environment as possible into each sample is best. However, if the target is unevenly distributed, then smaller sample sizes may be better so that rare targets are not diluted below the detection threshold. If the environment or reagents also contain inhibitors or decoy targets, then large scale concentration may decrease sensitivity. Taq polymerase and many other reagents contain bacterial DNA that is residual from their production. Assays that use minimal amounts of Taq decrease this background [
26] and small volumes assays should further help.
For the detection of Biowarfare agents that are normally absent and for which quick, specific, and sensitive detection are paramount, many thousands of liters of air rapidly collected are appropriate [
35]. On the other hand, to best understand the undisturbed environment the sampling method should be gentle, not itself create additional airflow, and it should be granular. Individual vacuum samplers are an intermediate methodology. They are worn or left near a patient and draw air onto a filter. Depending on the airflow rate they sample a variable-sized environment. The time resolution is typically one to several hours. As currently used the whole filter, which is typically 25 mm in diameter is extracted as a single sample [
36,
37].
Insight into heterogeneity of the air might be gained through a modification of the way that the sampling filter is currently processed. Instead of extracting the filter in one piece, suppose it were cut into a number of equal segments that are processed and assayed independently. If most segments are negative but a few contain a strong signal that would be clear indication of heterogeneity in the air volume sampled. This reasoning and experimental design take their inspiration from classic studies that determined the independence of bacterial mutation from its subsequent selection [
38,
39].
Individual passive sampling
If spatial and temporal variation are high, then smaller more localized and less mixed samples become relatively advantageous. The mouth and nose are the expected source and the reservoir for most airborne infectious agents. Dilution into the air volume predicts that the concentration of an airborne agent will decrease as a cubic function of distance from its source. Deactivation of viruses in aerosol may lead to a more dramatic drop-off of viable virus. Samplers should be unobtrusive and not alter the sampled environment. The assay of tiny volumes and the ability isolate rare positive samples recommend a new generation of small volume and/or passive samplers.
Clothing is a promising personal environmental sampling site [
40,
41] but difficult to standardize. One might build on the clothing idea to design defined samplers optimized for capture and preservation of the microbial and chemical environment. Sampling material might be housed in specially-designed holders, akin to radiation badges. For example, in hospitals, all patients, staff and visitors might wear standardized samplers.
The top of the shoulders appears to offer a minimally obtrusive yet maximally sensitive sampling site. Top-of-the-shoulder samplers would be closer to the mouth and nose than samplers worn as chest badges but relative sensitivity remains to be tested. Epaulettes such as those worn as military insignia might provide design inspiration. These approaches would potentially dovetail with simultaneously tracking movements within the environment of interest, and thus early consideration of the ethical issues and public or user concerns would be critical.
Passive samplers have different characteristics that in some scenarios may make them superior to vacuum-enhanced air sampling: i) if the target distribution is highly heterogenous in space, then source proximity may give the highest signal to noise ratio. ii) the related ability to identify the source with less distortion due to the sampler’s altering airflow. Each sampling patch could contain internal standards. Research to design samplers might take inspiration from materials for replica plating of bacteria [
39,
42]. There may even be specific overlap of optimal materials such as velveteen (a velvet-like cloth) for capture.
Prospective identification of super-spreaders
Super-spreaders or super-shedders are individuals who disproportionately infect others and may dominate the epidemiology of infectious disease [
43‐
46]. In some contexts super-spreaders of infectious agents might be identified prospectively by molecular criteria rather than by
post facto epidemiology [
47,
48]. Asymptomatic super-spreaders of airborne infections might be identified as those whose samplers harbor the most signal for the agents in question [
48]. Super-spreaders or super-shedders have been documented in people and also among cattle and in mouse model systems [
49,
50].
There are important unknowns concerning super-spreaders. A study in cattle argues that super-shedding may be time rather than individual dependent, i.e. the same individuals may not be super-spreaders a mere few hours later [
51]. It is currently unclear- in fact the question appears to be unasked- whether super-spreaders are agent-specific or generalized. By further hypothesis generalized super-spreaders might excrete more of their own DNA as well as that of infectious agents. In this case the detection of spreading potential might be seen in two ways:
i) by the presence of more self-sequences on the individual’s samples and ii) by the presence of individual sequence-specific DNA from the sampling of others. The anticipation would be that those who shed either the “wrong”, i.e. potential pathogen, or perhaps simply too much, DNA may be considered candidates for a variety of control measures, up to and including confinement.
Rational discussions about policy and human rights should be informed by facts [
52]. Not all infections spread between individuals are harmful or neutral. Health-positive microbial spreading, e.g. the live attenuated oral polio and rotavirus vaccines, also occurs. Analogous to the way that some individuals are more influential than others in spreading ideas or memes [
53], some may be more influential in spreading health-promoting microbiomes. If social and environmental transmission of benign and of positively helpful microbiomes turns out to be the case, will it follow the same rules as the spread of detrimental infections? Depending on their resident microbiomes, the same individuals may turn out to be super-spreaders of benign, health-promoting, or harmful microbes.
Toward mapping the 3D topography of relevant variables
The survival and distribution of airborne infectious agents depends on temperature, humidity [
54‐
60] and airflow [
61,
62]. Improved mapping of these parameters would better inform building design, and real-time monitoring of these parameters in occupied buildings might open the way for their dynamic optimization. For example, there is generally a temperature difference between indoor and outdoor environments, which results in a temperature gradient across rooms whose windows or walls form a building’s exterior. Regions of every intermediate temperature and relative humidity will exist when the outside is below freezing and the inside is warm. Temperature gradients necessarily traverse the dew point where water vapor condenses into liquid. The stability of regions where water vapor liquifies, and the consequences for microbes are complex and situation-specific. Moreover, most things microbial in the built environment also depend on the behavior of occupants. It is therefore challenging for modelling to predict, for example, that a specific region of hospital room, such as the southeast corner near the ceiling, happens to be in a “sweet spot” for bacteria or virus.
An increasing number of ‘smart’ buildings, including hospitals, are fitted with sensors [
63‐
65] which monitor temperature and/or humidity at a given position and are networked in an Internet of Things (IoT) [
66,
67]. These IoT systems can provide 3D maps of the monitored parameters in real-time, and can be used e.g. to optimize heating, ventilation, and air conditioning (HVAC) systems [
68,
69] or to monitor the integrity of the building structure [
70]. Such IoT systems could be readily adapted for applications targeting the microbiome, with additional sensors placed in at-risk areas. Sensors can be installed during building construction, retrofitted permanently, or temporarily installed. The IoT systems could be augmented with other sensors to monitor e.g. levels of particular gas types or volatile organic compounds (VOC) relevant to microbial growth [
71‐
73]. Environmental sensing will interface with wearable technology and individual-identified information [
74].
In addition to the temperature and humidity information provided by an IoT network of sensors, there is a need for devices that provide portability, remote sensing, and high spatial resolution. These could be used to rapidly assess large areas without having to install an IoT system, monitor targets otherwise difficult to access, or provide the cm- or mm-scale spatial resolution required to localize problem sources.
High resolution, remote imaging of temperature is routinely performed using infrared (IR) cameras for radiometry, exploiting the dependence of the blackbody radiation intensity at a given wavelength on temperature. This IR thermal imaging can be used e.g. to identify thermal leakage between indoor and outdoor environments [
75]. Blackbody radiation is also emitted at microwave frequencies. Although the emission intensity from room temperature bodies is much lower at microwave frequencies than at IR frequencies, microwaves can penetrate through dielectric materials such as walls to assess spaces hidden to visible or IR frequencies, and can still provide spatial resolution on the mm- to cm-scale. Whilst promising, microwave thermometry for applications in the built-environment is at a much earlier stage of development than IR cameras and requires further development, with single-channel sensors of moderate sensitivity and spatial resolution reported [
76,
77].
Water in all its forms (humidity, moisture, condensation, etc.) and its distributions are often key variables for the health of a building’s occupants [
78‐
80]. Microwave radiometers are used in the large scale environment for remote measurement of atmospheric humidity. Water vapor can be identified from its distinct absorption peak at a frequency of 22.235 GHz in its microwave spectrum [
81], and multi-frequency measurements can be used to detect liquid water (e.g. suspended microdrops). However, atmospheric measurements integrate over 100 s to 1000s of meters of air, and the research has not been done to determine if indoor microwave radiometry, integrating over much smaller volumes, would provide a sufficient signal for practical humidity measurements.
It may be more immediately practical to look for water in materials, such as walls with water leakage, which can both drive local humidity and host infectious agents themselves. Wet material can be indirectly detected with IR imaging, through the resulting change in surface temperature [
75]. Direct remote sensing techniques for indoor water are less advanced, and would require research and development effort to become practical.
Microwave-based sensing is again promising, due to its through-wall imaging capability and ability to provide sufficient spatial resolution [
82] (unlike radiation of longer wavelengths). Microwave sensors for water can be based on the scattering of microwaves generated by an external source, as discussed in this issue [
83], with holographic techniques able to provide 3D reconstructions of scattering objects [
84,
85]. In the short to medium term, single-channel point-sensors could be developed, with imaging performed by scanning the sensor. As a long-term aspiration, one could envisage a camera providing images of 3D moisture levels in real time, similar to temperature imaging with IR cameras. Schlieren and shadowgraph optics allow 3D imaging of air movements at distance [
86]. It remains to be seen if these and/or other methods can be developed to noninvasively monitor the full range of airflows relevant to building occupants. A strength of microwaves are their ability to penetrate many building materials even though they are strongly absorbed by water and reflected by metals. Infrared has the advantage of better specific spectra for many materials of interest.
Real time carbon dioxide (CO
2) imaging in the built environment seems a reasonable approach to become both practical and informative. The atmospheric concentration of CO
2 is about 400 ppm (ppm), 0.04% and increasing at an annual rate of approximately 2-4 ppm per year [
87]. Directly exhaled human breath is between 4 and 5% CO
2, approximately 100 fold more than the atmosphere. Increased ambient CO
2 correlates well with subjective reports of stuffy air [
88] and local areas of increased concentration correlate well with increased airborne concentrations of
Mycobacterium tuberculosis [
89]. CO
2 quantification is carried out with compact devices that measure concentration via specific IR spectral absorption. Wavelength-specific IR is used for earth [
90] and extraterrestrial [
91] quantification of atmospheric CO
2. To our knowledge IR imaging of CO
2 has not yet been used in the built environment. If there is clear motivation to do it, then it can be done. Real time IR imaging of CO
2 might, for example, be used to direct a small fan into regions of high CO
2 thereby dispelling patches of stagnant air in an energy efficient manner.