Congratulations! You’ve survived an influenza pandemic! Okay, that might be old news to you, but you might not appreciate how rare of an event they are. Typically, there are about 30 years between influenza pandemics, however, having had one a few years back doesn’t preclude the possibility that another isn’t just around the corner. Unfortunately, some strains of influenza with ‘pandemic potential’ have been identified in recent years, but so far (fortunately) they have rather poor person-to-person infectious properties. Today’s immediate concerns are arguably more focused on Ebola & MERS, both with some degree of potential for a pandemic if not managed well. Having said that, it’s worth noting that only moments before the H1N1 pandemic in 2009, the world was panicked by a strain of H5N1, so it would be prudent to remain vigilant about new and old foes.
Although the 2009-10 pandemic was relatively mild (by comparison to the ‘Spanish Influenza H1N1’), it was nevertheless, a real test of national and global preparedness plans. Since the mid-2000’s, antivirals–specifically Tamiflu–has been central to most national preparedness plans. This new class of antiviral (neuraminidase inhibitor) was intended to slow the person-to-person spread of influenza, with secondary benefits of decreasing the severity of the symptoms and ultimately lowering the risk of contracting bacterial pneumonia. It is widely acknowledged that the majority of deaths in the 1918–1919 influenza pandemic were a direct result of secondary bacterial pneumonia, hence, the pandemic preparedness plans necessitated clinical management plans for antibiotic use to treat these bacterial infections.
If you’re new to environmental chemistry, it might come as surprise to discover that many of the drugs proposed for use during the pandemic exit the body in the urine and feces in the bioactive form. Another way to say this, is that life contained within our sewage works (i.e., mostly microbes) and our rivers (i.e., microbes, protists, invertebrate, vertebrates) are being exposed to most of the drugs we take. It is a tribute to bacteria’s ability to adapt to toxins that our sewage works continue to work–but the question remains, ‘Is there a Tipping Point’ or will microbes continue to evolve to the seemingly endless chemical challenges we flush down our toilets?
It is widely understood that societies around the world consume enormous quantities of antibiotics, and that an influenza pandemic would simply add to this pre-existing quantity of drug. It is not possible to know how the increased use of antibiotics during a pandemic will impact society’s existing problems with antibiotic resistance and its development, but what is clear, it’s not going to improve the situation.
Unlike antibiotics, antivirals are not typically prescribed for influenza–in large part because people typically recover from influenza on their own–for most it is a ‘self-limiting disease’. Hence, if you don’t have to spend money on a drug to get better, typically, governments don’t. However, Japan is the only country where Tamiflu is prescribed to a very large fraction of those impacted by seasonal influenza.
The potential environmental implications of a medical response to an influenza pandemic have been discussed in the literature prior to the pandemic, but this discussion was mostly focused on the uncertainty associated with the potential effects of Tamiflu on microorganisms found in sewage works and rivers. A second concern was the potential for Tamiflu that is released to rivers to select for antiviral resistance in wildfowl. It turns out that ducks and the like, happen to prefer to hang out around the nice nutrient rich, warm sewage outflows. Unfortunately, wildfowl are the natural host for influenza. Should a duck living in a solution of Tamiflu be infected with influenza, it has a chance of developing a Tamiflu-resistant strain of influenza, which might ultimately increase the prevalence of Tamiflu resistance in ‘the wild type’ influenza making Tamiflu an ineffective drug, much like its predecessors: amantadine and rimantadine. Recent studies have demonstrated it is an entirely possible scenario and likely to have happened already (again something for another day).
As previously mentioned, a pandemic would potentially increase associated environmental and human health risks stemming from increased antibiotic use (i.e., the generation, maintenance and spread of existing and possibly novel antimicrobial resistance genes as well as the inhibition of critical microbial-driven functions of a sewage works). Although we can estimate antibiotic use, it is much more difficult (as will be discussed later) to predict environmental concentrations of these drugs. So, what was lacking from this dialogue was actual measurements of antibiotics in major rivers during an influenza pandemic.
Rationale of New Study in Plos One
A collaboration between me, my colleague at CEH (Mike Bowes), Swedish environmental chemists (Hanna Söderström, Jerker Fick, Richard H. Lindberg), and avian influenza scientists (Josef D. Järhult and Björn Olsen), and Czech environmental chemists (Roman Grabic, Ghazanfar A. Khan, Ganna Fedorova) has just published the first report of eleven antibiotics (trimethoprim, oxytetracycline, ciprofloxacin, azithromycin, cefotaxime, doxycycline, sulfamethoxazole, erythromycin, clarithromycin, ofloxacin, norfloxacin), three decongestants (naphazoline, oxymetazoline, xylometazoline) and the antiviral drug oseltamivir’s (Tamiflu®) active metabolite, oseltamivir carboxylate (OC), at 21 locations within the River Thames catchment in England (see Figure; CEH Thames Initiative) during the autumnal peak of the influenza A[H1N1]pdm09 pandemic.
The aim was to quantify the pharmaceutical response to the pandemic and compare this to drug use during the late pandemic (March 2010) and the inter-pandemic periods (May 2011). A large and small sewage works were sampled in November 2009 to understand the differential fate of the drugs in the two sewage works prior to their entry in the river and to estimate drug users using a wastewater epidemiology approach. In short, wastewater epidemiology aims to determine attributes about a population based on chemicals measured in sewage. This approach was recently used in our paper entitled “Compliance to Oseltamivir among Two Populations in Oxfordshire, United Kingdom Affected by Influenza A(H1N1)pdm09, November 2009 – A Waste Water Epidemiology Study,” where we demonstrated evidence for poor adherence to the prescribed Tamiflu regime (i.e., people didn’t start or finish the course as prescribed), such that approximately 50% of the prescribed doses of Tamiflu in the study region of England were not consumed (i.e., we recovered approximately 50% less Tamiflu in the sewage than one would predict based on the number of Tamiflu prescriptions).
Tamiflu (i.e., in this case I mean oseltamivir carboxylate (OC) the active antiviral) Mean hourly Tamiflu concentrations in the small and large sewage works influent (i.e., the raw sewage as it enters the works) were 208 and 350 ng/L (nanograms per litre), with a maximum concentration of: 2070 and 550 ng/L, respectively. The Tamiflu concentration of 2070 ng/L is the highest reported in the literature (during a pandemic). Consistent with previous research, Tamiflu did not biodegrade in the larger sewage works. There is an indication that Tamiflu did degrade in the smaller sewage works, but as the number of Tamiflu users was quite low, this might simply be an artifact of inadequate sampling. A higher sampling frequency would be needed for such small populations–something to take into consideration in future efforts! An excellent reference for why this is the case can be found here. It’s worth a read if you’re not familiar with it (but it’s not open access). Tamiflu was found in 73% of river samples during the weekly November sampling period, achieving a maximum of: 193 ng/L. Notably, only 5% and 0% of the samples taken in the late- and inter-pandemic period contained Tamiflu above the limit of detection (2 ng/L). This result clearly indicates that Tamiflu is a drug that is exclusively used during a pandemic in the UK.
Given the mild nature of the pandemic, measured environmental concentrations of Tamiflu (33-62 ng/L) in the River Thames were 2-3 orders of magnitude lower than predicted environmental concentrations during a severe pandemic, i.e., R0 >2.0, but was consistent with the lower end of concentration predicted for a mild pandemic within the Thames catchment, England (27-11,000 ng/L). These concentrations provide further support to the hypothesis that widespread Tamiflu use during a pandemic could increase the risk of generating OC-resistance in avian influenza-infected wildfowl, as the concentrations are within the same order of magnitude needed to achieve this phenomenon.
Antibiotics Erythromycin was the most concentrated antibiotic measured in sewage influent (max=6,870 and 2,930 ng/L, for the small and large sewage works, respectively). The mean river concentration of each antibiotic during the pandemic largely fell between 17-74 ng/L, with clarithromycin (max=292 ng/L) and erythromycin (max=448 ng/L) yielding the highest single measure. Fewer different antibiotics were recovered in the river during the late-pandemic (March 15, 2010) and the inter-pandemic (May 11, 2011) period than in the peak pandemic period (November 2009).
Decongestants Napthazoline and oxymetazoline were the most frequently detected and concentrated decongestant in the small sewage works influent (1650 and 67 ng/L) and effluent (696 and 307 ng/L), respectively, but were below detection in the large sewage works. None of these drugs were recorded in the River Thames, indicative of either their ability to biodegrade relatively quickly, propensity to adsorb to suspended organic particles (and potentially settle out from the water column, where we were sampling), and/or they were diluted beyond our detection limits. It’s almost certain all three played a role.
Wastewater epidemiology Tamiflu was uniquely well-suited for the wastewater epidemiology approach owing to its nature as a prodrug, recalcitrance and temporally- and spatially-resolved prescription statistics. Although it would be very helpful to gain some understanding of the compliance to antibiotics and decongestants, this is not easily achieved. Antibiotics can be biodegraded in wastewater as well as adsorb to particulates making it difficult to take full account of all the antibiotics in the waste stream. The same antibiotic going through the same sewage works will degrade to different extents on different days and seasons (depending on residence time in the sewage works, rainfall (dilution), and temperature).
Because our use of water as a society is non-unform within the week and day, it should be expected that the same drug will have a different fate depending on the day and time it was ‘flushed down the toilet’. Given that each kind of sewage work will introduce yet another level of complexity to this calculation, it becomes near impossible to have confidence about estimating the load (i.e., mass/time) of antibiotics. What is exceedingly helpful when attempting wastewater epidemiology is to have highly resolved (spatially and temporally) prescription statistics. This can then provide you with a starting point from which you can make estimates of actual users, i.e., compliance. Unfortunately, statistics on antibiotic prescriptions is not publicly available at the local level–and is only available at a national level on an annual basis. Hence, significant ‘massaging’ of these annual drug use estimates is needed–a necessity that introduces potentially significant uncertainty into the wastewater epidemiology approach–especially in the smaller population catchments. With regard to decongestants, these are primarily acquired over-the-counter and are thus not subject to National Health Service accounting. Hence, estimating decongestant use is a guessing game that is not easily verified. The fact that these drugs are fairly easily degraded and might adsorb to particulates further ‘muddies’ the water. In conclusion, wastewater epidemiology is quite an interesting and potentially useful approach, but you need to pick your battles wisely and measure appropriately.
Final Thoughts: In hindsight, the 2009 influenza A(H1N1)pdm09a virus generated a relatively small number of fatalities as compared to severe pandemics like the 1918 ‘Spanish flu’, which meant that the medical response was proportionately lower than would have been expected in a moderate or severe influenza pandemic. Hence, the potential negative effects to sewage works function and the environment proposed to occur in a moderate and severe pandemic were not reported. This study provides the first evidence that antibiotic and antiviral use was elevated during the pandemic. Theoretically, the antiviral recorded in the River Thames was of sufficient concentration to select for antiviral resistance in wildfowl, however, it remains to be demonstrated whether this had occurred (it turns out we have this data, but it’s yet unpublished, which indicates there was resistance in the wildfowl in the Thames–watch this space).There remains a great deal of uncertainty with regard to pharmaceutical use patterns during a pandemic, as a result of poor adherence to prescribed drugs and the widespread use of over-the-counter medications. The focus on Tamiflu here and in the literature is unlikely to reflect future antiviral practices, as preparedness plans might, in fact, employ a combination therapy of two or more antivirals in an effort to combat resistance. However, as the efficacy of the antivirals has been put into question, it remains to be seen whether national stockpiles are replenished with the same degree of enthusiasm next time around.
Opportunities to ground truth model predictions for ‘black swan events’ such as influenza pandemics are, by definition, very rare (every 30 years), making this study conducted during the last influenza pandemic a unique window into public health practice, human behaviour, and drug adherence in the UK. It represents the first study to measure antibiotics and decongestants in influent and effluent and receiving rivers during a public health emergency, thereby establishing a baseline from which future modeling and risk assessments can be built in preparation for more severe public health emergencies.