Variability in the recovery of a virus concentration procedure in water: Implications for QMRA

Highlights

• A large dataset for quantifying virus concentration procedures in water is presented.

• Recovery of mengovirus, adenovirus, and norovirus GI and GII were investigated.

• Recovery efficiencies were lower and more variable than expected from lab trials.

• Correlation between estimated recovery of spiked mengovirus and human viruses was low.

• Sample-specific controls mimicking human virus behaviour are needed.

Abstract

Methods for analysing water for viruses are known to have variable and relatively poor recovery efficiencies. Quantitative method recovery data are needed to correct virus enumeration results so that estimates of virus concentrations in surface waters for QMRA are not too low. Obtaining quantitative data representing method recoveries for different pathogenic viruses is a significant challenge. In this study, we investigated the use of mengovirus process control data for quantifying recovery efficiency of human adenovirus (AdV) and noroviruses GI (NoVGI) and GII (NoVGII) from surface waters. Samples were collected from the inlet to a drinking water treatment plant on the Glomma River, Norway. Performance of the sample concentration procedure was quantified by comparing the virus concentrations found in concentrated and unconcentrated samples. The mean recovery of viruses (1.2%, 0.31%, 0.15% and 0.053% for mengovirus (n = 86), AdV (n = 20), NoVGI (n = 33) and NoVGII (n = 21) respectively) estimated in this study were lower than expected, and the between sample variability in estimated recovery was very high, spanning around 6 orders of magnitude for mengovirus. Within-sample correlation between the estimated recovery of mengovirus and human viruses was poor, and therefore sample specific mengovirus data could not be used to correct all human virus concentrations. Instead beta distributions were fitted to human virus-specific recovery estimates. The magnitude and variability of virus concentration when corrected for the variable recovery efficiency was orders of magnitude higher than the uncorrected concentration. Better estimates of virus concentration could be achieved if a sample-specific spiking control could be developed that mimicked closely the behaviour of human viruses in environmental samples.

Introduction

Norovirus (NoV) is one of the most important agents of diarrhoea in humans worldwide (Patel et al., 2009) and has been implicated in several waterborne outbreaks in Scandinavia (Kukkula et al., 1999, Larsson et al., 2014, Maunula et al., 2005, Nygård et al., 2003a, Nygård et al., 2003b) and New Zealand (Hewitt et al., 2007), among others. The ongoing cost to the community is not insignificant: a recent study in Finland estimated the cost of lost workdays due to illness in a gastroenteritis outbreak to be, in addition to direct costs, 1.8–2.1 million euros (Halonen et al., 2012). Another norovirus outbreak in Lilla Edet, Sweden, in 2008 resulted in a total cost of 0.87 million euros (Larsson et al., 2014).

The current Guidelines for Drinking Water Quality (WHO, 2011) promote the provision of safe drinking water through the implementation of risk management strategies from catchment to consumer via the development of a Water Safety Plan (WSP). Within the WSP, Quantitative Microbial Risk Assessment (QMRA) (Haas et al., 1999, Medema et al., 2006) is being increasingly applied to assess whether the system is able to provide safe drinking water, and also to help identify key drivers of risk (including important hazards, their sources, and the level of treatment required to achieve safe water).

Quantifying pathogen concentrations in surface waters from environmental monitoring data is an important component of applying QMRA for drinking water systems. Enumeration of pathogens from surface waters typically requires concentrating the pathogens in a large volume (e.g. 1–1000 L) down to a smaller volume (e.g. 10 ml) in preparation for analysis. While the objective of any concentration procedure is to capture all the target microorganisms, methods are imperfect and inevitably a fraction of the microorganisms that were originally present is lost. Quantitative interpretation of virus enumeration data for QMRA must correct for the fraction that is lost during sample concentration, or the virus concentration, and hence the risk, will be underestimated.

In order to quantify recovery, it is necessary to have scientific data. For protozoa, spiking of a known number of (oo)cysts in the sample has been undertaken extensively in order to gain a quantitative estimate of overall recovery. Using these spiking data, incorporating method recovery in statistical models for quantifying pathogen concentrations for QMRA has become routine (Petterson et al., 2015, Petterson et al., 2007, Schmidt and Emelko, 2011, Schmidt et al., 2013, Teunis et al., 1999, Teunis et al., 1997). For viruses, however, there are few examples of where recovery performance has been quantitatively included in estimates of virus concentrations (van Heerden et al., 2005, Kundu et al., 2013), with most studies assuming 100% recovery efficiency (Åström et al., 2007, Masago et al., 2006, Oesterholt et al., 2007).

Obtaining quantitative data representing method recovery for viruses is a significant challenge. While surrogate (such as mengovirus) spikes are routinely used as process controls, these data are rarely used to correct concentration estimates. In this study, we investigated the use of mengovirus process control data for quantifying recovery efficiencies of human adenovirus, and noroviruses GI and GII from surface waters.

The samples used were collected as part of the Norwegian branch of a Scandinavian research project (VISK) concerned with viruses in water in Norway and Sweden. The samples were collected from the Glomma, which is the longest and largest river in Norway, with a length of 600 km, average water discharge of 700 m³/s, and a basin of 42,000 km² (Johansen, 2004). It is the water source for several drinking water treatment plants (DWTP), supplying approximately 230,000 people along its length. The Glomma also has a considerable human population living in its catchment area, and is a recipient of the effluent from several wastewater treatment plants (WWTP), as well as discharge from private septic tanks. It therefore provides a suitable model for investigation of a drinking water source contaminated by human viruses of faecal origin.