Microbial Populations in a Hospital Under Construction

Jack Sampling

Dr. Jack Gilbert collecting a sample from the floor of a future operating room.

On June 8th, 2012, members of the Hospital Microbiome Consortium were granted access to the construction site for the University of Chicago’s New Hospital Pavilion. Throughout the tour, we used swabs moistened with saline solution to collect microbial samples from the following 32 locations:

  • Person A, B, C, and D’s shoes before the tour.
  • Person A, B, C, and D’s shoes after the tour.
  • The road just outside the construction site.
  • Carpet, wood, and terrazzo flooring in the reception lobby.
  • Countertops of the reception desk and a nurse’s station.
  • An operating room floor.
  • The floor near a patient room’s bathroom sink.
  • Tap water from a drinking fountain and bathroom cold water faucet.
  • An ice machine’s spigot and ice reservoir.
  • Drainage holes from a patient room, bathroom, nurse station, and drinking fountain.
  • Light switch.
  • Oven panel.
  • Two toilet bowls.
  • Shower hose o-ring and showerhead.
  • A scrub sink’s faucet aerator.
  • Water from a cooling tower on the hospital’s roof.

DNA from these samples was then extracted, amplified, and sequenced at Argonne National Laboratory’s Next Generation Sequencing facility. The resultant DNA sequence reads were grouped together into operational taxonomic units (OTUs) based on sequence similarity, and searched against the greengenes database to determine which bacterial lineages were present in the original sample.

What Did We Find?

The most commonly found bacteria in this study belong to the following taxonomic orders:

  • Mycoplasmatales – In scientific terminology, the “myco-” prefix usually denotes an affiliation to fungal organisms, and in this case is applied to these bacteria not because they are closely related to fungi, but because these bacterial cells form filaments that appear similar to filamentous fungi. This is due to the fact that the Mycoplasmatales lineage is characterized by pleomorphism, that is, the cells are contained within a flexible three-layered membrane rather than a rigid cell wall, and so are able to take on any number of shapes. The absence of a cell wall also affords these bacteria immunity to antibiotics that inhibit cell wall biosynthesis. These cells are very small (0.1 – 0.25 um) and have one of the smallest genomes. In fact, This bacterium’s genome is so small that it has been completely articifially synthesized in the laboratory – producing the species Mycoplasma laboratorium.
  • Burkholderiales – In contrast to the minimal set of genes possessed by the Mycoplasmatales, bacteria in the Burkholderiales order encode a vast repertoire of enzymes for degrading a wide range of nutrients. As a result, they are able to grow almost anywhere, including in some cleaning products. Also in contrast to the Mycoplasmatales, which are non-motile, Burkholderiales cells have a single polar flagellum.
  • Actinomycetales – These bacteria grow end-to-end, forming filaments that appear to be fungal rather than prokaryotic in origin. This growth strategy is well-suited to soil environments, where this order of bacteria are commonly found, as it allows cells to better seek out sparse nutrients. Due to the fungal appearance, a genus within the Actinomycetales order has been named Mycobacterium, similarly to how Mycoplasmatales (above) was named.
  • Pseudomonadales – Like the Burkholderiales, Pseudomonadales are also capable of using a wide variety of compounds as their nutrient source, and are able to even grow in ammonia cleaning solutions. The natural environment for Pseudomonadales species, however, is the soil. In this environment, they are able to ‘breathe’ nitrate instead of oxygen, with the unfortunate net result of nitrogen fertilizer being released from soil as N2. However, Pseudomonadales are also environmentally beneficial due to their ability to degrade synthetic pesticides before those chemicals enter the food chain. When grown cultured in the laboratory, the pigments excreted by Pseudomonadales can turn their growth media blue/green, and some species in this order produce fluorescent chemicals that glow under UV light.
  • Lactobacillales – Since the discovery of Lactobacillus by Louis Pasteur in 1856, bacteria in the Lactobacillales order have been used in the commercial production of lactic acid. This useful property of the Lactobacillales is due to the fact that these cells grow using oxygen-independent reactions even when oxygen is available. Although this causes them to grow slower than oxygen-utilizing bacteria, the lactic acid produced by Lactobacillales inhibits of growth of other bacteria, thereby enabling Lactobacillales to thrive.
  • Bacteroidales – In the human intestinal tract, Bacteroidales of the Bacteroides genus are present in numbers approaching 100 billion cells per gram of feces. These bacteria are essential for the health of their human hosts, as they serve to break down complex organic compounds into simpler ones. Because these bacteria are abundant in fecal matter, and can not survive for long outside of a host, they are useful in environmental studies to determine the degree of fecal contamination in water. By sequencing the 16S gene from Bacteroides in the environment, it is even possible to determine which host they came from – human, livestock, or wildlife.

What is Growing Where?

Below, Figure 1 shows the major taxonomic orders recovered from each of the surfaces we sampled. Additionally, comparing the taxonomic orders from one sample with those from another allowed us to create a tree to show which samples were most closely similar to one another. As might be expected, similar types of surfaces had similar types of bacteria: the eight shoe samples formed one cluster, floors and countertops another, and so on.

Community Structure Dendrogram


Figure 1: Sample Dendrogram

The taxonomic composition of each of the samples (shown on the right) were compared to one another to determine which samples were most similar. The resultant dendrogram, or tree, (shown on the left) illustrates that samples from similar types of surfaces have similar taxonomic compositions.
A detailed summary of the bacteria in each sample is also available.

Tracking the Microbes on Our Shoes

In Figure 1 above, notice how the samples from the bottom of our shoes before the tour formed one subgroup, and the shoe samples from after the tour formed another! Comparing the before- and after- tour bacterial communities from our shoes to the communities found on the floors, we can say with confidence (P = 0.002) that our shoes went home with some hospital bacteria. Perhaps equally interesting is that the grouping of shoes A+C and B+D remained the same both before and after the tour. This implies that even when the communities on one’s shoes are affected by where you are walking, you can still find the underlying signal that can identify who’s shoes they were from one hour to the next.

Samples Plotted Using PCoA


Figure 2: Principal Coordinate Analysis (PCoA)

A PCoA plot depicts the degree of taxonomic similarity between samples in multiple dimensions. As in Figure 1, Figure 2 above shows that samples from similar types of surfaces group together.
Additional 2-D and 3-D PCoA plots are also available online.

How Many Bacteria Did We See?

Although the type of analysis we did can’t say which surface had the most bacteria, what we can say is that some surfaces had thousands of different types of bacteria while others only had only a couple hundred. This type of calculation, alpha diversity, is shown in Figure 3 below. The bar chart summarizes the number of observed species on each of the different surfaces, while the rarefaction curves provide an idea of how close we came to observing all the species of those surfaces. Generally speaking, shoes were found to contain the greatest diversity of bacteria, followed by floors and countertops, then by water-associated environments. The shower head we sampled was found to contain the fewest different types of bacteria, but whether this was because the showerhead was covered in a biofilm composed of just a few species, or because (as we’d like to think) it was so clean that only a handful of bacterial colonies were present, the methods used here are unable to say.

Alpha Diversity


Figure 3: Alpha Diversity Plots

This pair of figures offer an estimate of how many different types of bacteria were present in each sample using the sequencing effort we applied. The rarefaction curves on the left indicate that some samples, like the one represented by the red line near the bottom of the graph (a sink), contained relatively few species and could be adequately sampled using 2000 sequencing reads. Other samples, such as the one represented by the brown line near the top of the graph (a shoe), contained more species than could be thoroughly sampled even with 16,000 reads. At the even sampling depth selected for this project (4,882 reads), the figure on the right shows the calculated number of observed species in each sample. The bottom of shoes before the tour had approximately 10 times as many different types of bacteria than the shower head.

Was Anything Dangerous Growing in the Hospital?

The identification of pathogenic bacteria wasn’t the goal of this study, and so our methods were not designed to answer this type of question. For people familiar with phylogenetic trees, our mention of finding Burkholderiales, Pseudomonadales, and Lactobacillales likely raised an eyebrow or two, as some species within these lineages are well-known pathogenic bacteria. However, other species within these lineages are harmless and common in the environment. Our 151 base-pair length DNA fragments were rarely sufficient to tell the difference between pathogenic and non-pathogenic species, and when it was, this DNA sequence did not provide information on whether the bacteria it came from encoded other genetic factors necessary for pathogenicity. Furthermore, pathogenicity is oftentimes dependent on the hosts immune system – several healthy individuals carry the dreaded MRSA bacterium without any ill effect. For these reasons, we are unable to say if any of the bacteria we observed were cause for concern. At any rate, these samples were collected while the hospital was under construction – before any of the several very thorough cleanings began.

Final Thoughts

This study was undertaken with the objective of quickly field-testing our sample collection and processing protocol while producing results that would be of general interest to indoor microbiologists. Nonetheless, our observations show that bacterial communities are largely driven by the surfaces they are found on, even in a building still being constructed. Our findings that microbial communities on shoes quickly mesh with their environment is a topic that may be an area ripe for future investigation, in order to determine the feasibility of this trait in forensics and the contribution of human foot traffic as a vector of bacteria. With this much information gleaned from 32 samples, we are now more than ever looking forward to beginning our 12,392 sampling project of this hospital in full operation.

Tagged with:
Posted in Findings